Environmental Problem Solving With Geographic Information Systems 1994 and 1999 Conference Proceedings

               United States                 Office of Research and          EPA/625/R-00/010
               Environmental Protection         Development                 August 2000
               Agency	Cincinnati, OH 45268	

v>EPA     Environmental  Problem  Solving with
               Geographic Information  Systems

               1994 and 1999 Conference Proceedings
               About this CD:

               This CD, EPA Document Number EPA/625/R-00/010, contains conference proceedings
               documents from the 1994 and 1999 Environmental Problem Solving with Geographic
               Information Systems conferences.  The 1994 papers are also contained in a Seminar
               Publication, EPA Document Number EPA/625/R-95/004, available from EPA.
               Accessing CD contents:

               All of the files contained on this CD are in Adobe Acrobat 4.0 format. A copy of Adobe
               Acrobat Reader 4.05 for Windows has been included on this CD for your convenience if
               you do not have this software.  The software is also available free of charge on the
               Adobe homepage at: http://www.adobe.com/.  If an older version of the Acrobat Reader
               is used, error messages will appear when files are  opened.

               To install Adobe Acrobat from the file on this CD, double-click the file entitled
               rs405eng.exe and follow the installation instructions that will appear on the screen. You
               will be asked to read and accept the Electronic End-User License Agreement as part of
               this installation.

               Adobe, Acrobat, and the Acrobat logo are trademarks of Adobe Systems Incorporated.
               Contents of CD:

               •  1999 Agenda.pdf- This file contains the agenda of the 1999 conference with links
                  to the available presentation papers. It is organized by date and time of presentation
                  at the conference.

               •  1994 Contents.pdf- This file contains a list of presentations from the 1994
                  conference with links to the available presentation papers. It is organized by topic

               •  1999 Attendees.pdf- This file contains the list of attendees from the 1999
                  conference, with links to papers by attending authors.

   Papers - This folder contains PDF files of all of the available papers from the 1994
   and 1999 conferences. Every paper in this folder may be accessed through the links
   in the 1999 Agenda and 1994 Contents documents.

   Readme.pdf - This file contains information about the CD contents, access, and
   navigation. Start here when using this CD for the first time.

   rs405eng.exe - This file contains the Adobe Acrobat 4.05 installation files.
Navigating between files:

The 1994 Contents, 1999 Agenda, and 1999 Attendees List may all be accessed from
this file. All the papers may be accessed by using the links from the 1994 Contents,
1999 Agenda, and 1999 Attendees List. When a link is followed to a paper, the
originating file is closed.  To return to the contents or agenda file, click on the title of the
open paper.  Every PDF document on this CD may also be accessed using your
Windows Explorer or File Manager.
For more information:

Contact the United States Environmental Protection Agency at:  www.epa.gov

              Environmental  Problem  Solving
        with Geographic  Information Systems
                     September 21-23,  1994
                          Cincinnati, Ohio
GIS Concepts
GIS Uncertainty and Policy: Where Do We Draw the 25-Inch Line?
      James E. Mitchell

Data Quality Issues Affecting GIS Use for Environmental Problem-Solving
      Carol B. Griff in

You Can't Do That With These Data! Or: Uses and Abuses of Tap Water Monitoring
      Michael R. Schock and Jonathan A. Clement

Ground-Water Applications

Using GI5/GP5 in the Design and Operation of Minnesota's Ground Water
Monitoring and Assessment Program
      Tom Clark, Yuan-Ming Hsu, Jennifer Schlotthauer, Don Jakes, and
      Georgianna Myers

Use of GIS in Modeling Ground-Water Flow in the Memphis, Tennessee, Area
      James Outlaw and Michael Clay Brown

MODRISI: A PC Approach to GIS and Ground-Water Modeling
      Randall R. Ross and Milovan 5. Beljin

GIS in Statewide Ground-Water Vulnerability Evaluation to Pollution Potential
      Navulur Kumar and Bernard A. Engel

Verification of Contaminant Flow Estimation With GIS and Aerial Photography
      Thomas M. Williams

Geology of Will and Southern Cook Counties, Illinois
      Edward Caldwell Smith

Watershed Applications

The Watershed Assessment Project: Tools for Regional Problem Area
      Christine Adamus

Watershed Stressors and Environmental Monitoring and Assessment Program
Estuarine Indicators for South Shore Rhode Island
      John F. Paul and George E. Morrison

(515 Watershed Applications in the Analysis of Nonpoint Source Pollution
      Thomas H. Cahill, Wesley R. Horner, and Joel 5. Mc(5uire

Using (515 To Examine Linkages Between Landscapes and Stream Ecosystems
      Carl Richards, Lucinda Johnson, and George Host

Nonpoint Source Water Quality Impacts in an Urbanizing Watershed
      Peter Coffin, Andrea Dorlester, and Julius Fabos

A (515 for the Ohio River Basin
      Walter M. Grayman, Sudhir R. Kshirsagar, Richard M. Males, James A.
      Goodrich, and Jason P. Heath

Nonpoint Source Pesticide Pollution of the Pequa Creek Watershed, Lancaster
County, Pennsylvania: An Approach Linking  Probabilistic Transport Modeling and (515
      Robert T. Paulsen and Allan Moose

Integration of (515 With the Agricultural Nonpoint Source Pollution Model: The
Effect of Resolution and Soils Data Sources on Model Input and Output
      Suzanne R. Perlitsh

X(5RCWP, a Knowledge- and <5IS-Based System for Selection, Evaluation, and
Design of Water Quality Control Practices in Agricultural Watersheds
      Runxuan Zhao, Michael A. Foster, Paul t>. Robillard, and David W. Lehning

Integration of EPA Mainframe Graphics and (515 in a UNIX Workstation
Environment To Solve Environmental Problems
      William B. Samuels, Phillip Taylor, Paul Evenhouse, and Robert  King

Wetlands Applications

Wetlands Mapping and Assessment in Coastal North Carolina: A GIS-Based
      Lori  Sutter and James Wuenscher

Decision Support System for Multiobjective Riparian/Wetland Corridor Planning
      Margaret A. Fast and Tina K. Rajala

Design of GI5 Analysis To Compare Wetland Impacts on Runoff in Upstream Basins
of the Mississippi and Volga Rivers
      Tatiana B. Nawrocki

Water Quality  Applications

Vulnerability Assessment of Missouri Drinking Water to Chemical Contamination
      Christopher J. Barnett, Steven J. Vance, and Christopher L. Fulcher

Reach File 3 Hydrologic Network and the Development of GIS  Water Quality Tools
      Stephen  Bevington

EPA's Reach Indexing Project:  Using GI5 To Improve Water Quality Assessment
      Jack Clifford, William D. Wheaton, and Ross J. Curry

Environmental Management Applications

Ecological Land Units, GIS, and Remote Sensing: Gap Analysis in the Central
      Ree  Brannon, Charles B. Yuill, and Sue A. Perry

A GIS Strategy for Lake Management Issues
      Michael F. Troge

A Watershed-Oriented Database for Regional Cumulative Impact Assessment and
Land Use Planning
      Steven J. Stichter

A GIS Demonstration for Greenbelt Land Use Analysis
      Joanna J. Becker

GIS as a Tool for Predicting Urban Growth Patterns and Risks From Accidental
Release of Industrial Toxins
      Samuel V. Noe

Integration of GIS and Hydrologic Models for Nutrient Management Planning
      Clyde W. Fraisse, Kenneth L. Campbell, James W. Jones, William &. Boggess,
      and Babak Negahban

Other GIS Applications

Expedition of Water-Surface-Profile Computations Using GIS
      Ralph J. Haefner, K. Scott Jackson, and James M.  Sherwood

Small Is Beautiful: GIS and Small Native American Reservations—Approach,
Problems, Pitfalls, and Advantages
      Jeff Besougloff

A GIS-Based Approach to Characterizing Chemical Compounds in Soil and Modeling
of Remedial System Design
      Leslie L. Chau, Charles R. Comstock, and R. Frank Keyser

Polygon  Development Improvement Techniques for Hazardous Waste Environmental
Impact Analysis
      David A. Padgett

Comparing Experiences in the British and  U.S. Virgin Islands in Implementing GIS
for Environmental Problem-Solving
      Louis Potter and Bruce Potter

Application of GIS for Environmental Impact Analysis in a Traffic Relief Study
      Bruce Stauffer and Xinhao Wang

                     Environmental  Problem Solving

               With Geographic Information  Systems

                        Tuesday, September  21, 1999

               Pre-registration and Cash Bar Reception (5:00 PM - 8:00 PM)

Day 1 - Wednesday, September 22, 1999

                                Grand Ballroom A-B

                                PLENARY SESSION

7:30 - 9:00      Registration and Name  Badge Pickup

9:00 - 9:15      Welcome & Overview
                      Sue Schock,  USEPA,  ORD, NRMRL
                      Daniel J. Murray, USEPA, ORD, NRMRL

9:15 - 10:00     New Directions in Environmental Problem Solving
                      Michael F. Goodchild, Ph. D., Chair
                      National Center  for Geographic Information and Analysis, and
                      Department of Geography, University of California - Santa Barbara

10:00 - 10:20    BREAK

10:20 - 10:50    6IS Workgroup: An Overview 6IS -  QA
                      George M.Brills,  J.D., USEPA, ORD, NERL

10:50 - 11:30    Environmental Visioning  with Geographic Information Systems
                      Sudhir R. Kshirsagar, Ph. D., Global Quality Corporation, and
                      Paul Koch, Pacific Environmental Services Inc.

11:30 - 1:00     LUNCH

Day 1  -  Wednesday, September 22, 1999 (continued)
1:00 -
1:25 -
1:50 -
2:15 -
2:40 -
3:00 -
3:25 -
3:50 -
4:15 -
Grand Ballroom A
Lyn Kirschner
Conservation Technology Informal-ion Center
&IS Watershed Delineation
Nonpoint Pollutant Loading
Application for ArcView GIS
Application of DEM and Land
Cover Data in Estimating
Atmospheric Deposition to the
Northeast and Mid-Atlantic
Regions: Model Development
and Applications
Assessing the Impact of
Landuse/Landcover on Stream
Chemistry in Maryland
James Goodrich,
Laurens van der
Tak, P.E.,
James A. Lynch,
State University
Gabriel Senay,
.Grand Ballroom B
Jill Neal, USEPA, ORt), NRMRL
GIS and GPS in Environmental
Remediation Oversight at
Federal Facilities in Ohio
The Impact of Spatial
Aggregation on Environmental
Modeling: A GIS Approach
Characterizing the Hydrogeology
of Acid Mine Discharges from
the Kempten Mine Complex,
West Virginia and Maryland
The GIS Connection to
Residential Yard Soil
Kelly Kaletsky
and Bill Lohner,
Ohio EPA
Lin Liu, Ph.D.,
University of
Benjamin R.
Hayes, Bucknell
Jennifer Deis,
Black & Veatch
Assessing the Long-Term
Impact of Land Use Change On
Runoff and Non-Point Source
Pollution Using a GIS-NPS
A Web- Based &IS Model for
Assessing the Long-Term
Hydrologic Impacts of Land
Use Change (L-THIA GIS
WWW): Motivation and
Using a Geographic Information
System for Cost-Effective
Reductions in Nonpoint Source
Pollution: The Case of
Conservation Buffers
Putting Geospatial Information
Into the Hands of the "Real"
Natural Resource Managers:
Lessons from the NEMO
Project in Educating Local Land
Use Decision Makers
Bhaduri, Oak
Ridge National
Jon Harbor,
Purdue University
(Bernie Engel,
Mark S. Landry,
Virginia Tech
Joel Stacker,
University of
GIS in the Confirmation Process
Using a Geographic Information
Systems Application to
Implement Risk Based Decisions
in Corrective Action
Determining the Accuracy of
Geographic Coordinates for
NPDES Permittees in the State
of Ohio
No presentation
Raymond E.
Bailey, Ph.D.,
Lesley Hay
Wilson, P.E.,
The University
of Texas -

4:30 - 6:30   Reception (cash bar)

Day 2 - Thursday, September 23, 1999
8:00 - 9:00   Registration and Name Badge Pickup
8:30 -
8:55 -
9:20 -
9:45 -
10:10 -
10:30 -
10:55 -
11:20 -
11:45 -
Grand Ballroom A
boug Grosse. USEPA, ORb, NRMRL
Evaluating Soil Erosion Parameter
Estimates from Different Data
A Planning Strategy for Siting
Animal Confinement Facilities: The
Integrated Use of &IS and
Digital Image Simulation
Lake Superior Decision Support
Systems: GIS Databases and
Decision Support Systems for
Land Use Planning
Update of GIS Land Use
Attributes from Land Surface
Texture Information Using
SIR-C Images
Gabriel Senay,
Thora Cartlidge,
University of
George E. Host,
Ph.D., Natural
Francisco J.
Artigas, Ph.D.,
Grand Ballroom B
Jim Kreissl, USEPA, CERI, TTB
Merging Transportation and
Enviromental Planning Using
Use of GIS Tools for
Conducting Community On-
Site Septic Management
Management and Reuse of
Contaminated Soil -- The
SoilTrak Method
Using GIS to Rank
Environmentally Sensitive Land
in Orange County, Florida
Dept. of
David Healy,
Rogers, Jr.,
BEM Systems,
Michael J.
Gilbrook, HDR
Onsite Wastewater Management
Program in Hamilton County,
Ohio- -An Integrated Approach to
Improving Water Quality and
Preventing Disease
Modeling Combined Sewer
Overflow (CSO) Impact: The Use
of a Regional GIS in Facilities
Building a Shared and Integrated
GIS to Support Environmental
Regulatory Activities in South
Timothy I.
Ingram, Hamilton
County General
Health District,
Michael D.
Witwer, Metcalf
& Eddy, Inc.
Guang Zhao,
Ph.D., South
Carolina Dept. of
Health & Env.
Use of GIS for the
Investigation and
Classification of Land Being
Redeveloped Under the Ohio
Voluntary Action Program
Assessing and Managing the
Impacts of Urban Sprawl on
Environmentally Critical
Areas: A Case Study of
Portage County, Ohio
Building a Brownfield Sitebank
With Internet Map Server
Jay Lee,
Ph.D., Kent
Alan Rao,
Brustlin, Inc

Day 2 - Thursday, September 23, 1999 (Continued)
1:30 -
1:55 -
2:20 -
2:45 -
3:10 -
Grand Ballroom A
Scott Minamyer, USEPA, CERI, TTB
Targeting the Knowledge
Assembly Process of the Flora of
North America (FNA): Biological
Resource Problem Solving Using
GIS Standards for Environmental
Restoration and Compliance
Reporting on the Development of
an Environmental &IS Application
- Wetlands Restoration in the
Central Valley of California
Habitat Filters, GIS. and
Riverine Fish Assemblages:
Sifting Through the Relationships
Between Fishes and Their
Leila M.
University and
Utah State
Bobby G.
Carpenter, P.E.,
Tri -Service
Tech. Center
David Hansen,
US Bureau of
Douglas A.
Grand Ballroom B
Tom Brennan, USEPA, OPPT
Using GIS to Analyze the
Spatial Distribution of
Environmental, Human Health,
and Socio- Economic
Characteristics in Cincinnati
Public Participation GIS
Applications for Environmental
Justice Research and
Community Sustainability
Quantifying Risk in Watershed
Assessment Using GIS &
Stochastic Field-Scale
Methodological Issues in GIS-
Based Environmental Justice
Xinhao Wang,
Ph.D. and Chris
University of
David Padgett,
Ph.D., P.E.,
Virginia Tech
Jeremy Mennis,

Day 2 -  Thursday, September 23, 1999 (Continued)
3:30 -
3:55 -
4:20 -
Grand Ballroom A
Scott Minamyer, USEPA, CERI, TTB
Using a GIS Model to Predict the
Extent of Common Reed
Encroachment into Two Tidal
Wetland Areas in Northeastern
New Jersey
The Application of GIS in the
Development of Regional
Restoration Goals for Wetland
Resources in the Greater Los
Angeles Drainage Area
Fractal Dimension as an Indicator
of Human Disturbance in Galveston
Bay, Texas
Karla Hyde and
Robin Dingle,
Associates, Inc.
Charles Rairdan,
US Army Corps
of Engineers
Amy Liu,
Grand Ballroom B
Tom Brennan, USEPA, OPPT
Using GIS to Evaluate the
Effects of Flood Risk on
Residential Property Values
Environmental Justice in
Kentucky: Examining the
Relationships Between Low-
Income and Minority
Communities and the Location
of Landfills, and TSD
Application of GIS to Address
Environmental Justice: Needs
and Issues
Bartosova and
David E.
Clark, Ph.D.,
Larisa J.
Kentucky Area
Babafemi A.
4:30 - 6:30   Reception (cash bar)

Day 3 - Friday, September 24,  1999
8:00 - 9:00  Registration and Name Badge Pickup
8:30 -
8:55 -
9:20 -
9:45 -
10:10 -
10:30 -
10:55 -
Grand Ballroom A
Mike Troyer, Ph.D., USEPA, ORD, NRMRL
The National Hydrography
Dataset - Status and Applications
Sustainable Developments:
Definition, Location, and
Development of a National
Watershed Boundaries Dataset
A Watershed -Based Approach to
Source Water Assessment and
Protection Utilizing GIS- Based
Inventories: A Case Study in
South Carolina
Thomas &. Dewald,
USEPA, and Keven
s. Roth, uses
Michael E. Troyer,
Alan Rea, US6S
James M. Rine,
Ph.D., Earth
Sciences Research
Grand Ballroom B
Randall Ross, Ph.D., USEPA, ORD, NRMRL
Strategic Planning for
No More 3 -Ring Binders!
Pollution Exposure Index
Model Measures Airborne
Pollutants in National
A GIS- Based Approach to
Predicting Wetland
Drainage & Wildlife
Habitat Loss in the
Prairie Pothole Region of
South-Central Canada
Swear ingen,
Robins AFB
Margaret B.
Martin, P.E.,
US Army Corps
of Engineers
Michael V.
Institute for
Wetland and
Using an ARC/INFO 6IS to
Analyze Forest Patches for
Watershed -Based Conservation
and to Present Data on a Web
Application of a Water Balance
Model and &IS for Sustainable
Watershed Management
Lonnie Darr,
County, MD, Dept.
of Env. Protection
Thomas H. Cahill,
P.E., and Susan
Pagano, Cahill
Application of a
Geographic Information
System for Containment
System Leak Detection
A High -Resolution
Hydrometeorological Data
System for Environmental
Modeling and Monitoring
Randall R.
Ross, Ph.D.,
David R.
Legates, Ph.D.,
University of

 Environmental Problem Solving with Geographic Information Systems Conference
 September 22-24,1999
     Eric Adams
            OFFICE: 614-644-2752
              FAX: 614-644-2909
List of Attendees
                     Cincinnati, Ohio
       Ohio EPA
         P.O. Box 1049
         Columbus  OH
     Babafemi A. Adesanya
            OFFICE:  7578658950
       Environmental Equity Information Institute
         P.O.Box 189
         Hampton VA  23669
     Larry Alber
            OFFICE:  518-457-3143
              FAX:  518-457-0738
            lalber@gw. dec. state, ny. us
       NYS Dept. of Environmental Conservation
         50 Wolf Road
         Albany NY  12233
     Lora Alberto
            OFFICE:  513-861-7666
              FAX:   513-559-3155
       Mill Creek Restoration Project
         42 Calhoun Street
         Cincinnati OH  45219
     Melanie Allamby
            OFFICE: 216-881-6600x446
              FAX: 216-881-2738
       N.E. Ohio Regional Sewer District
         4415 Euclid Avenue
         Cleveland  OH  44103
     Dr. James K. Andreasen (Jim)
            OFFICE: 202-564-3293
              FAX: 202-565-0076
       USEPA ORD Natl. Center for Env. Assessment
         401 M Street, SW, Mail Code (8623D)
         Washington DC   20460
     Roger Anzzolin
            OFFICE: 202-260-7282
              FAX: 202-401-3041
            anzzolin. roger@epa.gov
       U.S. EPA OGWDW 4606
         401 M. St. SW
         Washington DC  20460
     Gary Arnold
            OFFICE:  541-686-7838x247
              FAX:  541-686-7551
            arnold.gary@deq. state, or. us
         1102 Lincoln
         Eugene OR  97401
     Francisco J. Artigas
            OFFICE:  9733531069
            artigas@cimic. rutgers. edu
       Center for Info. Mgmt., Integration & Connectivity
         Rutgers University
         180 University Avenue
         Newark NJ   07102   USA
     Chris Auffrey
            OFFICE:  513-556-0579
              FAX:  513-556-1274
            chris. auffrey@uc.edu
       University of Cincinnati
         P.O. Box210016
         Cincinnati OH  45221-0016
     Margaret Ay cock
            OFFICE: 409-880-8897
              FAX: 409-880-1837
            aycockma@hal. lamar.edu
       Gulf Coast Hazardous Substance Research Center
         P.O. Box 10671
         Beaumont TX  77710-0671
                                 As of: Thursday, September 30, 1999

 Environmental Problem Solving with Geographic Information Systems Conference
 September 22-24,1999
     Russ Baier
            OFFICE: 512-239-1746
               FAX:  512-239-5687
            rbaier@tnrcc. state, tx. us
List of Attendees
Cincinnati, Ohio
       Texas Natural Resource Conservation Comm.
         12100 Park 35 Circle
         Austin TX 78753
     Jeanne M. Bailey
            OFFICE:  202-628-8303
               FAX:  202-628-2846
            jbailey@awwa. org
       American Water Works Association
         1401 New York Ave., NW#640
         Washington  DC  20005
     Raymond E. Bailey
            OFFICE:  6364418086
            ray_bailey@wssrap-host. wssrap. com
         7295 Highway 94 South
         St. Charles MO  63304
     Thomas E. Bailey
            OFFICE:  513-287-2596
               FAX:  513-287-3499
            tbailey@cinergy. com
       Cinergy Corp.
         139 E. Fourth St., Room 552a
         Cincinnati OH  45201
     Kim Baker
            OFFICE:  614-265-6411
               FAX:  614-267-2981
            kim. baker@dnr. state.oh. us
       Ohio Dept. of Natural Resources
         1952 Belcher Drive, Bldg. C-2
         Columbus OH   43224-1386
     M.C. Baldwin
            OFFICE:  520-871-7690
               FAX:  520-871-7599
       Navajo Nation Environmental Protection Agency
         Water Quality Program - PO Box 339
         Window Rock AZ  86515
     Todd Baldwin
            OFFICE:  202-232-7933
               FAX:  202-234-1328
            tbaldwin@islandpress. org
       Island Press
         1718 Connecticut Avenue, NW, Suite 300
         Washington DC  20009
     Brian Balsley
            OFFICE:  513-326-1500
               FAX:  513-326-1550
       BHE Environmental, Inc.
         Cincinnati OH  45246
     Robert Bamford
             OFFICE:  775-687-4670
               FAX:  775-687-6396
            rbamford@ndep. carson-city. nv. us
       Nevada Division of Environmental Protection
         333 West Nye Lane
         Carson City NV  89706
     Quinn Barker
            OFFICE: 765-285-2327
               FAX:  765-285-2606
            jeflinl@gw. bsu. edu
       Department of Natural Resources and Environmental Management
         Ball State University
         Muncie IN  47306
     Chris Barnett
            OFFICE: 573-882-9291
               FAX:  573-884-2199
            barnett@cares.missouri. edu
       CARES - University of Missouri
         130Mumford Road
         Columbia MO  65211-6200
                                  As of: Thursday, September 30, 1999

 Environmental Problem Solving with Geographic Information Systems Conference
 September 22-24,1999
     Ralph P. Barr
            OFFICE:  301-258-8651
               FAX:  301-258-8679
            barrR @ttnus. com
List of Attendees
       Tetra Tech NUS, Inc.
         910 Clopper Road, Suite 400
         Gaithersburg MD 20878
Cincinnati, Ohio
     Alena Bartosova
            OFFICE:  4142885128
       Dept. of Civil & Env. Engineering
         Marquette University
         P.O. Box1881
         Milwaukee Wl  53201-1881   USA
     Bruce Battles
            OFFICE: 405-744-8974
               FAX:  405-744-7008
            bruce@seic. he. okstate.edu
       Oklahoma State University
         201 CITD
         Stillwater OK  74078
     Bruce Bauch
            OFFICE:  502-493-1945
       U.S. Geological Survey
         9818 Bluegrass Parkway
         Louisville KY  40299
     Stephanie Beak
            OFFICE:  614-644-4852
               FAX:  614-728-1245
            Stephanie. beak@epa. state.oh. us
       Ohio EPA
         122 South Front Street
         Columbus OH   43215
     Michael L. Bechdol
             OFFICE:  214-665-7133
               FAX:  214-665-2191
             bechdol.michael@epamail. epa.gov
       U.S. EPA - Region 6WQ-SG
         1445 Ross Avenue
         Dallas TX  75202
     Glynn Beck
            OFFICE:  270-827-3414
               FAX:  270-827-1117
            ebeck@kgs. mm. uky. edu
       Kentucky Geological Survey
         P.O. Box 653
         Henderson KY  42419
     Tara L. Beckman
            OFFICE:  4129218358
            beckmant@ttnus. com
       TetraTech NUS, Inc.
         Foster Plaza 7,  661 Andersen Dr
         Pittsburgh PA  15220
     Bob Bednar
            OFFICE:  405-702-8197
               FAX:  405-702-8101
            bobby. bednar@deqmail.state.ok.us
       Oklahoma Dept. of Environmental Quality
         707 N. Robinson Ave.
         Oklahoma City OK  73102
     Brian Begley
            OFFICE:  502-564-6716
               FAX:  502-564-2705
            brian. begley@mail.state.ky. us
       Kentucky Division of Waste Management
         14 Reilly Road
         Frankfort KY  40601
     Dick Behr
            OFFICE:  207-287-6828
               FAX:  207-287-7826
            richard. s. behr@state.me. us
       Maine Department of Environmental Protection
         17 State House Station
         Augusta ME  04333
                                  As of: Thursday, September 30, 1999

 Environmental Problem Solving with Geographic Information Systems Conference
 September 22-24,1999
     Milovan Beljin
            OFFICE:  513-729-1602
List of Attendees
       M.S. Beljin and Assoc.
         Cincinnati OH  45231
Cincinnati, Ohio
     Kevin Benck
            OFFICE:  605-688-4776
            kbenck@brookings. net
       Water Resources Institute-South Dakota St. Univ.
         Brookings SD  57007
     Malcolm Bender
            OFFICE:  214-665-8378
              FAX:  214-665-6660
       U.S. EPA / SEE Program
         1445 Ross Avenue
         Dallas TX  75202
     Clif Benoit
            OFFICE:  801-625-5594
               FAX:  801-625-5483
            cbenoit/r4@fs.fed. us
       US Forest Service
         Federal Building
         140 E. 3275 N.
         N.Ogden UT  84414
     Jerry Bernard
            OFFICE:  202-720-5356
               FAX:  202-720-0428
            jerry, bernard@usda.gov
       USDA-Natural Resource Conservation Service
         P.O. Box 2890, Room 6123
         Washington DC   20013
     Budhendra Bhaduri
            OFFICE:  423-241-9272
               FAX:  423-241-6261
       Oak Ridge National Laboratory
         P.O. Box 2008, MS 6237
         Oak Ridge TN  37831-6237
     Taher Bishr
            OFFICE:  0020-12-2166800
               FAX:  00203-4835337
         20 Salah Salem Str. D. Toun
         Alex - EGYPT
     Ben Blaney
            OFFICE:  513-569-7852
               FAX:  513-569-7680
            blaney. ben@epa.gov
         26 West Martin Luther King Dr.
         Cincinnati OH  45268
     Brian Bohl
            OFFICE:  513-271-4182
               FAX:  513-569-7160
            bohl. brian@epamail. epa.gov
       U.S. EPA
         26W. Martin Luther King Drive
         Cincinnati OH  45268
     Steve Bolssen
            OFFICE: 502-564-3410
               FAX: 502-564-4245
            Steven. bolssen@mail.state.ky. us
       KY Dept. of Env. Protection, Division of Water
         14 Reilly Road
         Frankfort KY  40601-1189
     Bill Boria
            OFFICE:  716-753-4481
              FAX:  716-753-4344
            billb@health. co. chatauqua. ny. us
       Chautauqua County Health Department
         7 North Erie Street
         Mayville NY  14757
                                  As of: Thursday, September 30, 1999

 Environmental Problem Solving with Geographic Information Systems Conference
 September 22-24,1999                List Of Attendees                     Cincinnati, Ohio
     Joshua Bowen
            OFFICE:  800-355-9429
              FAX:  770-399-0535
            jbowen@ftr. com
United Capitol Insurance Company
  400 Perimeter Center Terrace, Suite 345
  Atlanta GA  30346
     Virgil Brack
            OFFICE:  513-236-1163
              FAX:  513-236-1178
BHE Environmental, Inc.
  11733 Chesterdale Road
  Cincinnati OH
     John Bradley
            OFFICE:  517-335-3146
              FAX:  517-373-9657
            bradlejn@state.mi. us
Michigan DEQ
  300 South Washington Square
  Lansing Ml  48933
     Tom Brankamp
            OFFICE: 606-431-8579
              FAX: 606-431-8581
     Tom Brennan
            OFFICE:  (202)260-3920
            brennan. thomas@epa.gov
Woolpert LLP
  525 West Fifth Street, Suite 213
  Covington KY  41011

Don Brannen
OFFICE: 513-681-8247
FAX: 513-681-1594
Cincinnati Recreation Commission
1655 Chase Avenue
Cincinnati OH 45206
  Washington DC
     Timothy Bricker
            OFFICE:  765-214-0088
            tjbricker@bsuvc. edu
Ball State Univeristy
  3556 North Tillotson #205
  Muncie IN  47304
     James Bridges
            OFFICE:  513-489-6611
              FAX:  513-489-6619
           jbridges@cin.pes. com
Pacific Environmental Services, Inc.
  7209 E. Kemper Rd.
  Cincinnati OH  45247
     Jan W. Briede
            OFFICE:  513-247-8000
              FAX:  513-247-8010
            jbriede@scientech. com
Scientech, NES, Inc.
  11400 Grooms Road
  Cincinnati  OH  45242
     George M. Brilis
            OFFICE:  702-798-3128
  P.O. Box 93478
  Las Vegas NV  89193
     Linda Briscoe
            OFFICE:  513-641-3081
              FAX:  513-641-0508
            LBri938500@aol. com
Ohio/Cincinnati Women's Health Project
  4860 Winneste Ave.
  Cincinnati OH  45232
                           As of: Thursday, September 30, 1999

 Environmental Problem Solving with Geographic Information Systems Conference
 September 22-24,1999
     Thomas M. Brody
            OFFICE:  312-353-8340
               FAX:  312-353-4755
List of Attendees
       Region 5 Office of Information services
         77 W. Jackson Blvd.
         Chicago IL 60604
Cincinnati, Ohio
     Joyce A. Broka
            OFFICE:  517-686-8025x8371
              FAX:  517-684-9799
            brokaj@state.mi. us
       Michigan Department of Environmental Quality
         503 N. Euclid Ave.
         Bay City Ml  48706
     Hugh J. Brown
             OFFICE:  765-285-5788
               FAX:  765-285-2606
            hbrown@gw. bsu. edu
       Ball State University
         2000 University
         Muncie IN  47306
     Mike Bruening
            OFFICE:  513-3261500
       BHE Environmental
         Cincinnati OH
     Jason Buck
            OFFICE:  765-285-2327
               FAX:  765-285-2606
            jeflinl@gw. bsu. edu
       Department of Natural Resources and Environmental
         Ball State University
         Muncie IN  47306
     John Burckle
            OFFICE:  513-569-7496
               FAX:  513-569-7471
            burckle.john@epamail. epa.gov
       U.S. EPA / NRMRL
         26 West Martin Luther King Drive
         Cincinnati OH  45220
     David Burden
            OFFICE: 580-436-8606
               FAX:  580-436-8614
            burden, david@epa.gov
         P.O. Box 1198
         ADA OK  74820
     Tracy Burke
            OFFICE:  717-236-3006
               FAX:  717-233-0994
       GTS Technologies
         851 South 19th Street
         Harrisburg PA  17104
     David Butler
            OFFICE:  502-564-6716x339
               FAX:  502-564-2705
            david. butler@mail.state.ky. us
       Kentucky Division of Waste Management
         14 Reilly Road
         Frankfurt KY  40601
     Neill Cade
            OFFICE:  513-357-7211
               FAX:  513-357-7262
            neill. cade@chdburn. rcc. org
       Cincinnati Health Dept.
         3101 BurnetAve. Room 324
         Cincinnati OH  45229
     Thomas H. Cahill, P.E.
            OFFICE:  610-696-4150
               FAX:  610-696-8608
            tcahill@thcahill. com
       Cahill Associates
         104 South High Street
         West Chester PA  19382
                                  As of: Thursday, September 30, 1999

 Environmental Problem Solving with Geographic Information Systems Conference
 September 22-24,1999
     Tom Cain
            OFFICE: 304-636-1800x289
              FAX: 304-636-1875
            tcain/r9_monong@fs.fed. us
List of Attendees
       Monongahela National Forest
         200 Sycamore St.
         Elkins WV  26241
Cincinnati, Ohio
     Leslie M. Calderon
            OFFICE:  817-608-2341
               FAX:  817-695-9191
            lcalder@dfwinfo. com
       North Central Texas Council of Governments
         616 Six Flags Drive, Suite 200, Centerpoint Two
         Suite 200, Centerpoint Two
         Arlington TX   76005-5888
     Guy Cameron
            OFFICE:  513-556-9740
       University of Cincinnati
         Department of Biological Science
         Cincinnati OH  45221
     John Capillo
            OFFICE:  606-986-0868
              FAX:  606-986-2695
            kefcapil@acs. eku. edu
       KY Environmental Foundation
         PO Box 467
         Berea  KY  40403
     Bobby G. Carpenter
            OFFICE:  601-634-4572
              FAX:  601-634-4584
            carpenb@wes. army, mil
       Tri-Service CADD/GIS Technology Center
         USAGE Waterways Experiment Station, 3909 Halls Ferry Road
         Vicksburg MS  39180-6199
     Thora Cartlidge
            OFFICE:  612-624-9273
              FAX:  612-624-1704
            crd@tc. umn.edu
       University of MN-Centerfor Rural Design
         217-1518 Cleveland Avenue
         St. Paul MN  55108
     Erman Caudill
            OFFICE:  606-257-4093
            elcaudOO@pop. uky.edu
       University of Kentucky
         342 Waller Avenue #3C
         Lexington KY  40504
     Jean Caudill
            OFFICE:  614-644-7181
              FAX:  614-466-4556
            jcaudill@gw. ohd. state, oh. us
       Ohio Dept. of Health
         P.O.Box 118
         Columbus OH  43266-0118
     Stephane Chalifoux
            OFFICE:  514-287-8606
              FAX:  514-287-8643
            s. chalifoux@tecsult. com
         85 St. Catherine Street West
         Montreal, Quebec
         CANADA H2X3P4
     Yu-mei Chang
            OFFICE:  513-558-2744
       University of Cincinnati
         202 Ruth Lyons Way, Suite #261
         Cincinnati OH  45267-0840
     James Chapman
            OFFICE:  765-213-1269
           jdchapman31@hotmail. com
       Delaware County Indiana
         100 W. Main Room 206
         Muncie IN 47305
                                 As of: Thursday, September 30, 1999

 Environmental Problem Solving with Geographic Information Systems Conference
 September 22-24,1999
     Jonathan W. Chapman
List of Attendees
Cincinnati, Ohio
            OFFICE: 410-260-8514
               FAX:  410-260-8595
            jchapman@dnr. state, md. us
       MD Dept. of Natural Resources-Forest Service
         Tawes State Office Building E1 580 Taylor Avenue
         Annapolis MD  21401
     David Christenson
            OFFICE: 303-312-3345
               FAX:  303-312-6065
            christenson. dave@epa.gov
       USEPA Region VIM
         999 18th Street, Suite 500
         Denver CO  80202
     David Christian
            OFFICE: 613-787-3879
               FAX:  613-787-3884
       ATCO Frontec Corporation
         100-170 Laurier Avenue West
         Ottawa, Ontario K1P5V5
     Kendra Cipollini
            OFFICE: 312-886-1432
               FAX:  312-886-9697
            cipollini. kendra@epa. gov
       USEPA Region 5, Critical Ecosystems Team
         77 West Jackson Avenue T13J
         Chicago IL  60604
     John Ckmanec
            OFFICE: 513-569-7481
               FAX:  513-569-7585
       U.S. EPA (MS G-75)
         26 Martin Luther King Drive
         Cincinnati OH  45268
     Tara Clapp
            OFFICE: 502-852-8152
               FAX:  502-852-4558
            tlclapp@rcf.usc. edu
       University of Louisville
         426 W. Bloom Street, 202
         Louisville KY  40208
     David E. Clark
            OFFICE: 414-288-3339
               FAX:  414-288-5757
            clarkde@marquette. edu
       Marquette University, Dept. of Economics
         P.O. Box 1881
         Milwaukee Wl  53201-1881
     Tim Clarke
            OFFICE: 502-573-2886
               FAX:  502-573-2355
       Kentucky State Nature Preserves Commision
         801 Schenkel Lane
         Frankfort KY  40601-1403
     Richard Cochran
            OFFICE: 615-532-0997
               FAX:  615-532-0046
            RCochran2@mail. state, tn.us
       Tennessee Department of Environmental and Conservation
         401 Church Street, 7th Floor L & C Annex, Div of Water Pollution Control
         Nashville TN  37243
     James C. Coleman II
            OFFICE: 513-583-1249
               FAX:  513-583-1250
            jcoleman@environcorp. com
       ENVIRON International Corp.
         6443 Lewis Road
         Loveland OH  45140
     Jim Coon
            OFFICE: 937-285-6038
               FAX:  937-285-6404
            jim.coon@epa. state.oh. us
       Ohio EPA
         401 E. Fifth St.
         Dayton Oh  45402
                                  As of: Thursday, September 30, 1999

 Environmental Problem Solving with Geographic Information Systems Conference
 September 22-24,1999                List Of Attendees                     Cincinnati, Ohio
     Amy Covert
            OFFICE:  502-573-2886
               FAX:  502-573-2355
            amy.covert@mail. state, ky. us
Kentucky State Nature Preserves Commission
  801 Schenkel Lane
  Frankfort KY  40601-1403
     Christopher Cox
            OFFICE:  309-782-2887
               FAX:  309-782-5038
            wildmanf@ria. army.mil
Rock Island Arsenal
  510R1-AO, Bldg220
  Rock Island IL  61299
     Michael J. Cramer
             OFFICE:  513-556-9740
               FAX:  513-556-5299
            michaeljcramer@hotmail. com
University of Cincinnati
  P.O. Box210006
  Cincinnati OH  45221-0006
     Irene M. Crawford
            OFFICE:  513-569-7167
SoBran, Inc.
  26 West Martin Luther King Drive
  Cincinnati OH  45268
     Pat Curley
            OFFICE:  770-673-3640
               FAX:  770-396-9495
            pcurley@brwncald. com
Brown and Caldwell
  41 Perimeter Center East, Suite 400
  Atlanta GA  30346
     Darrin L. Curtis
            OFFICE:  501-973-0760
            dlc2@engr. uark.edu
  2335 East Yvonne Drive
  Fayetteville AR  72703
     Bernie Daniel
            OFFICE: 513-569-7401
               FAX:  513-569-7609
            daniel. bernie@epa.gov
  26 West Martin Luther King Drive
  Cincinnati OH  45242
     Lonnie Darr
             OFFICE:  2407777703
            darrl@co.mo.md. us
Watershed Management Div., Montgomery Cty Dept. of Env. Protection
  Suite 120, 155 Rockville Pike
  Rockville MD  20850
     Bruce De Young
            OFFICE:  616-336-3234
              FAX:  616-336-2436
Kent County Health Department
  700 Fuller, N.E.
  Grand Rapids Ml  49503
     Kevin L. DeFosset
            OFFICE:  606-624-4471
            studefok@acs. eku.edu
Ewers Water Consultants / Eastern KY University
  326-10 Lancaster Avenue
  Richmond KY  40475
     Jennifer Deis, P.G.
            OFFICE:  913-458-6585
               FAX:  913-458-2934
            deisj@bv. com
Black & Veatch Corporation
  11401 Lamar Avenue
  Overland Park  KS  66211
                           As of: Thursday, September 30, 1999

 Environmental Problem Solving with Geographic Information Systems Conference
 September 22-24,1999
     Ed Delisio
            OFFICE: 312-886-1303
              FAX: 312-353-5374
            delisio. edward@epa. gov
List of Attendees
       US EPA Region 5
         77 West Jackson Blvd., B-19J
         Chicago IL  60604
                     Cincinnati, Ohio
     Phil Dennis
            OFFICE:  765-285-2327
              FAX:  765-285-2606
            jeflinl@gw. bsu. edu
       Department of Natural Resources and Environmental
         Ball State University
         Muncie IN  47306
     Tommy G. Dewald
            OFFICE:  2022602488
       Office of Water
         401 "M" Street, SW (4503F)
         Washington DC  20460
     Martin Diaz-Zorita
            OFFICE:  606-257-3655
              FAX:  606-257-2185
            mdzori2@pop. uky.edu
       University of Kentucky - Agronomy Department
         N122-Agric.Sci.Center North
         Lexington KY  40546-0091
     Robin Dingle
            OFFICE:  2078799496
       Northern Ecological Associates, Inc.
         386 Fore St.
         Portland OR  04101
     Harold B. Dirschl
            OFFICE:  613-787-9611
              FAX:  613-787-3884
            hkdir@sympatico. ca
       ATCO Frontec Corporation
         100-170 Laurier Avenue West
         Ottawa, Ontario K1P5V5
     David Dixon
            OFFICE:  513-247-8000
              FAX:  513-247-8010
            ddixon@scientech. com
       Scientech, NES, Inc.
         11400 Grooms Road
         Cincinnati OH  45242
     Mohsen Dkhili
            OFFICE:  573-751-1300
              FAX:  573-526-5797
            nrdkhim@mail. dnr.state. mo. us
         P.O. Box 176
         Jefferson City MO  65102
     Doug Dobransky
            OFFICE:  614-644-2752
              FAX:  614-644-2909
       Ohio EPA
         P.O. Box1049
         Columbus OH
     Kevin Doniere
            OFFICE:  513-281-2211
              FAX:  513-281-2243
       Human Nature
         990 St. Paul Place
         Cincinnati OH  45206
     Damon Dougherty
            OFFICE:  361-883-6016
              FAX:  361-883-7417
            damon@moorhousecc. com
       Moorhouse Associates, Inc.
         5826 Bear Lane
         Corpus Christ! TX  78405
           As of: Thursday, September 30, 1999

 Environmental Problem Solving with Geographic Information Systems Conference
 September 22-24,1999
     Joan Douglas
            OFFICE:  352-378-6517
               FAX:  352-338-8247
List of Attendees
       Florida Rural Community Assistance Project
         6212 NW 43 Street, Suite A
         Gainsville FL  32653
          Cincinnati, Ohio
     Mark E. Duewell
            OFFICE:  573-526-5214
               FAX:  573-751-6417
       Missouri Department of Health
         930 Wildwood, P.O. Box 570
         Jefferson City MO  65102-0570
     Jeffrey Duke
            OFFICE:  216-881-6600x456
               FAX:  216-881-2738
            dukej@neodrsd. org
       N. E. Ohio Regional Sewer District
         4415 Euclid Avenue
         Cleveland OH  44103
     John Dunham R.S.
            OFFICE:  513-564-1788
               FAX:  513-564-1776
            John.Dunham@igwmail. rcc. org
       Cincinnati Health Dept.
         3845 W.P. Dooley By-Pass
         Cincinnati OH  42223
     Scott Dyer
            OFFICE:  513-627-1163
               FAX:  513-627-1208
            dyer. sd@pg. com
       Procter & Gamble Co.
         P.O. Box 538707
         Cincinnati OH  45253-8707
     Don Ebert
            OFFICE:  702-798-2158
               FAX:  702-798-2158
            ebert. donald@epa.gov
       U.S. EPA
         944 E. Harmon Avenue
         Las Vegas NV 89119
     T. J. Edwards
            OFFICE:  765-285-2327
               FAX:  765-285-2606
            jeflinl@gw. bsu. edu
       Department of Natural Resources and Environmental Management
         Ball State University
         Muncie IN  47306
     James Eflin
            OFFICE:  765-285-2327
               FAX:  765-285-2606
            jeflinl@gw. bsu. edu
       Department of Natural Resources and Environmental Management
         Ball State University
         Muncie IN  47306
     Keith Egan
            OFFICE:  513-576-0009
               FAX:  513-574-9756
            egank@srwenvironmental. com
       SRW Environmental Services
         55 West TechneCenter, Suite C
         Milford OH   45150
     Lisa E. Enderle
             OFFICE:  703-645-6950
               FAX:  703-698-6101
            enderlel@saic. com
         2222 Gallows Road, Suite 300
         Dunn Loring VA  22027
     Bernie Engel
            OFFICE:  765-494-1198
               FAX:  765-496-1115
       Purdue University
         1146 ABE
         W. Lafayette IN  47907-1146
As of: Thursday, September 30, 1999

 Environmental Problem Solving with Geographic Information Systems Conference
 September 22-24,1999                List Of Attendees                      Cincinnati, Ohio
     Sendar Ertep
            OFFICE:  404-562-9683
               FAX:  404-562-9598
            ertep.sendar@epamail. epa.gov
                           USEPA Region 4
                             61 Forsyth St., SW
                             Atlanta GA  30303
     Alan Evereson
            OFFICE:  513-569-7046
                           U.S. EPA/ORD/NRMRL
                             26 W. Martin Luther King Drive
                             Cincinnati OH  45268
     Dr. Ralph O. Ewers
            OFFICE:  606-623-8464
              FAX:  606-623-6464
            glyewers@acs. eku. edu
                           Dept. of Earth Science Eastern KY University
                             160 Redwood
                             Richmond KY  40475
     Susan Pagan
            OFFICE:  202-260-9477
               FAX:  202-260-2941
                           U.S. EPA, Office of Water
                             401 -M Street, SW - Mailcode 4501F
                             Washington DC  20460
     Todd Falter
            OFFICE:  402-471-6571
               FAX:  402-471-6436
            tfalter@hhs. state, ne.us
                           NE Health and Human Services
                             P.O. Box 95007
                             Lincoln NE  68509
     Ian Farrar
            OFFICE:  304-545-4388
                           Collumbia Gas
                             1700 MacConkle Ave.
                             Charleston WV
     Terry Felkerson
            OFFICE:  573-308-3725
               FAX:  573-308-3652
                           U.S. Geological Survey
                             1400 Independence Road
                             Rolla MO  65401
     Don Ficklen
            OFFICE:  210-671-4844
               FAX:  210-671-2241
            holmes.fwklen@lockland. af.mil
                           U.S. Air Force
                             Lockland Air Force Base
     William S. Fischer
            OFFICE:  513-564-1787
               FAX:  513-564-1776
            wfish@mailcity. com
                           Cincinnati Health Dept.
                             3845 William P. Doley By-Pass
                             Cincinnati OH  45223
     Jeff Flege
            OFFICE:  614-265-6686
            jeff.flege@dnr. state.oh. us
                           Ohio Department of Natural Resources
                             Fountain Square, Building C-2
                             Columbus OH  43224-1386
     Terry Flum
  26 W. Martin Luther King Drive
  Cincinnati OH  45268
                           As of: Thursday, September 30, 1999

 Environmental Problem Solving with Geographic Information Systems Conference
 September 22-24,1999
     Charles Frank
            OFFICE: 610-935-4703
               FAX: 610-935-5583
            cfrank@envstd. com
List of Attendees
       Environmental Standards Inc.
         1140 Valley Forge Road
         Valley Forge PA  19482-0810
          Cincinnati, Ohio
     William T. Frederick
            OFFICE:  716-942-2563
               FAX:  716-942-2247
            frederw@wv. doe.gov
       Dames & Moore
         10282 Rock Springs Road
         West Valley NY  14171-9799
     Dr. Robert C. Frey
            OFFICE:  614-466-1069
               FAX:  614-644-7440
            rfrey@gw.odh.state.oh. us
       Ohio Department of Health
         246 North High Street
         Columbus OH  43266-0588
     Lawrence Friedl
            OFFICE:  202-564-6933
               FAX:  202-565-2431
       US EPA
         401 MSt.,SW(8101R)
         Washington DC   20460
     Joseph Frizado
            OFFICE:  419-372-7202
               FAX:  419-372-7205
            frizado@bgnet. bgsu.edu
       Bowling Green State University
         Bowling Green State University
         Dept. of Geology
         Bowling Green OH  43403
     Kevin Frysinger
            OFFICE:  610-935-5577
               FAX:  610-935-5583
       Environmental Standards
         1140 Valley Forge Road
         Valley Forge PA  19482
     Florence Fulk
         26 W. Martin Luther King Drive
         Cincinnati OH  45268
     Richard Futrell
            OFFICE:  606-622-1581
            antfutre @acs. eku. edu.
       Eastern Kentucky University Sociology Dept.
         223 Keith
         Richmond KY 40475
     Robert Galbraith
             OFFICE:  513-576-0009
               FAX:  513-576-9756
            galbraib@srw environmental, com
       SRW Environmental Services
         55 West TechneCenter, Suite C
         Milford OH   45150
     Dr. Achal Garg
            OFFICE:  513-357-7209
              FAX:  513-357-7262
            achal. garg@chdburn. rcc. org
       Cincinnati Health Department
         3101 BurnetAve.
         Cincinnati OH  45229
     Donald Gasper
       WV Department of Natural Resources
         4 Ritchie Street
         Buchannon WV   26201
As of: Thursday, September 30, 1999

 Environmental Problem Solving with Geographic Information Systems Conference
 September 22-24,1999                List Of Attendees                     Cincinnati, Ohio
     Glee H. Gasper
            OFFICE: 304-942-3704
Upshur County Litter Control Bd.
  4 Ritchie St.
  Buckhannon WV  26201
     David R. German
            OFFICE: 605-688-5611
              FAX: 605-688-4917
            wrisdsu@mg.sdstate. edu
South Dakota Water Resources Institute
  Brookings SD  57007
     Patricia Germany
            OFFICE: 502-574-3645
              FAX: 502-574-1389
            pgermany@louky. org
City of Louisville / Health & Environment
  1514 Hale Avenue
  Louisville KY  40210
     Tim Gessner
BHE Envir.
  Cincinnati  OH  45204
     Nick Giardino
            OFFICE: 210-536-6128
              FAX: 210-536-1130
  2513 Kennedy Circle
  San Antonio TX  78235-5123
     Michael J. Gilbrook
            OFFICE: 407-872-7801
              FAX: 407-872-0603
            mgilbroo@hdrinc. com
HDR Engineering, Inc.
  201  S. Orange Avenue, Suite 925
  Orlando FL  32801-3413
     Rebecca Glos
            OFFICE: 703-318-4797
              FAX: 703-736-0826
  11251 Roger Bacon Drive
  Reston VA  20190
     Haynes Goddard
            OFFICE: 513-569-7685
              FAX: 513-569-7111
            goodard. haynes@epamail. epa.gov
  26 West Martin Luther King Drive
  Cincinnati OH  45268
     Maria A. Gomez-Balandra
            OFFICE: 52-73-19-4000x407 or x410
              FAX: 52-73-20-8638
            magomez@tlaloc. imta. mx
Water Technology Mexican Institute
  Paseo Cuauhnahuac 8532
  Morelos, Mexico 62550
     James A. Goodrich
            OFFICE: 513-569-7605
              FAX: 513-569-7185
            goodrich.jam es@epa.gov
  26 W. Martin Luther King Drive
  Cincinnati OH  45268
     Walter M. Grayman
            OFFICE: 513-281-6138
              FAX: 513-281-6139
W.M. Grayman Consulting Engineer
  730 Avon Fields Lane
  Cincinnati  OH  45229
As of: Thursday, September 30, 1999

 Environmental Problem Solving with Geographic Information Systems Conference
 September 22-24,1999
     Don Green
            OFFICE: 615-837-5492
               FAX:  615-837-5025
List of Attendees
          Cincinnati, Ohio
       TN Nonpoint Source Program
         Ellinton Ag. Center, Box 40627, Holeman Bldg.
         Nashville TN  37204
     Chris Griffith
            OFFICE:  513-732-8036
       Clermont Co. General Health District
         2275 Bauer Road Suite 300
         Batavia OH  45103
     Corey Gullion
            OFFICE:  765-288-4057
               FAX:  765-288-4057
       Ball State Univ. Natural Resource Env. Mgr.
         1616 W.Gilbert G-100
         Muncie IN  47303
     Beth Hailstock
            OFFICE:  513-357-7206
               FAX:  513-357-7262
            beth. hailstock@chdburn. rcc. org
       Cincinnati Health Dept.
         3101 BurnetAve
         Cincinnati OH  45229
     Tonia Hampton
            OFFICE:  765-285-2327
               FAX:  765-285-2606
            jeflinl@gw. bsu. edu
       Department of Natural Resources and Environmental Management
         Ball State University
         Muncie IN  47306
     David T. Hansen
            OFFICE:  916-978-5268
              FAX:  916-978-5290
            dhansen@mp. usbr.gov
       U.S. Bureau of Reclamation-MPGIS
         2800 Cottage Way
         Sacramento CA 95825-1898
     Todd Hanson
            OFFICE: 218-335-7415
               FAX:  218-335-7430
       Leech Lake Reservation Water Resources
         6530 Hwy 2 NW
         Cass Lake MN  56633
     Jon Harbor
            OFFICE:  7654949610
       Dept. of Earth & Atmospheric Sciences
         Purdue University
         West Lafayette IN  47906-1397
     Deborah D. Harris
            OFFICE:  513-791-8330
              FAX:  513-791-7335
            logictree@aol. com
       Natl. Technology Assoc.-Cincinnati
         P.O. Box 42356
         Cincinnati OH  45242
     Jasper L. Harris
             OFFICE:  919-530-6394
              FAX:  919-530-7966
       North Carolina Central University
         1801 Fayetteville Street
         Durham NC 27707
     Roderick L. Harris, R.S.
            OFFICE:  615-353-8363
       Meharry Medical College
         1408 Mountain Valley Bend
         Nashville TN  37209
As of: Thursday, September 30, 1999

 Environmental Problem Solving with Geographic Information Systems Conference
 September 22-24,1999                List Of Attendees                      Cincinnati, Ohio
     Joel Harrison
            OFFICE:  304-637-0245
               FAX:  304-637-0250
            jharrison@dnr.state.wv. us
WV Division of Natural Resources
  Ward Road
  Elkins WV 26241
     Robin Harrover
            OFFICE:  425-649-7232
               FAX:  425-649-7098
            rhar461 @ecy.wa. gov
Washington State Department of Ecology
  3190 160th Avenue, SE
  Bellevue WA  98008-5452
     Paul Marten
            OFFICE:  513-569-7045
               FAX:  513-569-7471
            harten.paul@epamail. epa.gov
  26 W. Martin Luther King Drive
  Cincinnati OH  45268
     Megan Hartgrove-Del Gaudio
             OFFICE:  410-974-7276
               FAX:  410-974-7200
Maryland Environmental Service
  2011 Commerce Park Drive
  Annapolis MD  21401
     Patrick L. Havens
            OFFICE:  317-337-3465
              FAX:  317-337-3235
            phavens@dowagro. com
Dow AgroSciences LLC
  9330 Zionsville Road 306/A2
  Indianapolis  IN  46468
     Fred Hayden
            OFFICE:  510-526-7140
            Haydens@Earthlink. net
  639 Madison Street
  Albany CA  94706
     Benjamin R. Hayes
            OFFICE:  570-372-4215
              FAX:  570-372-2726
Dept. of Geo. and Envir. Sc. Susquehanna U.
  514 University Ave.
  Selinsgrove PA   17870-1001
     Linda Haynie
            OFFICE:  512-239-6821
               FAX:  512-239-5687
TX Natural Resource Conservation Commission
  P.O. Box 13087
  Austin TX  78711
     David J. Healy
            OFFICE: 802-229-1879
               FAX: 802-229-5417
            dhealy@stone-env. com
Stone Environmental, Inc.
  58 East State Street
  Montpelier VT  05602
     Richard H. Heath
            OFFICE:  207-287-7637
               FAX:  207-287-7826
            richard. h. heath@state.me. us
Maine Department of Environmental Protection
  State House Station #17
  Augusta ME  04333
     Conrad Heatwole
            OFFICE:  5402314858
Biological Systems Engineering
Virginia Tech
  Blacksburg VA  24061-0303  USA
As of: Thursday, September 30, 1999

 Environmental Problem Solving with Geographic Information Systems Conference
 September 22-24,1999
     Pat Heider
            OFFICE: 614-644-2752
               FAX:  614-644-2909
List of Attendees
                      Cincinnati, Ohio
       Ohio EPA
         P.O. Box 1049
         Columbus OH
     Beverly Henderson
             OFFICE:  614-644-8065
               FAX:  614-644-7740
            bhenders@gw. odh. state, oh. us
       Ohio Dept. of Health Assessment Section
         246 N. High Street, 8th Floor
         Columbus OH   43215
     Nereida Hernandez
            OFFICE:  787-764-8824
               FAX:  787-766-0150
            jcaemer@prtc. net
       Puerto Rico Environmental Quality Board
         P.O. Box 11488
         San Juan PR  00910
     Bob Hilbert
            OFFICE:  716-942-2417
            hilbert@wv. doe.gov
       West Valley Nuclear Services
         10282 Rock Springs Road
         West Valley NY   19171
     Sara Hines
            OFFICE:  502-573-2886
               FAX:  502-573-2355
            sara. hines(a),mail.state.ky. us
       Kentucky State Nature Preserves Commission
         801 Schenkel Lane
         Frankfort KY  40601-1403
     Terri Hoagland
            OFFICE:  513-569-7783
               FAX:  513-569-7111
            hoagland. theresa@epamail. epa.gov
       U.S. EPA
         26 W. Martin Luther King Drive MS-466
         Cincinnati OH  45268
     Molly Hodgson
            OFFICE:  216-910-1941
               FAX:  216-910-2010
       Metcalf& Eddy, Inc.
         Suite 1215, 1300 E. Ninth Stree
         Cleveland  OH   44114
     Matthew Hopton
            hopton@toast. net
       Dept. of Biological Sc., University of Cincinnati
         Cincinnati OH  45221-0006
     George E. Host, Ph.D.
            OFFICE:  218-720-4264
               FAX:  218-720-4328
            ghost@sage. nrri. umn.edu
       Natural Resources Research Institute; U of MN
         5013 Miller Truck Highway
         Duluth MN  55811
     Kevin House
            OFFICE:  502-564-3080
               FAX:  502-564-9195
            kevin.house(fl),mail.state.ky. us
       Kentucky Division of Conservation
         663 Teton Trail
         Frankfort KY  40601
     David Howerter
            OFFICE:  204-467-3292
               FAX:  204-467-9426
            d howerter(fl),dveb.com
       Institute for Wetland and Waterfowl Research
         P.O. Box1160
         Oak Hammock Marsh
         Stonewall MB  ROC2ZO
           As of: Thursday, September 30, 1999

 Environmental Problem Solving with Geographic Information Systems Conference
 September 22-24,1999
     Hsuan-Tsung (Sean) Hsieh
            OFFICE: 405-744-8974
               FAX:  405-744-7008
            hsieh@seic. he. okstate. edu
List of Attendees
       Oklahoma State University-SEIC
         Oklahoma State University
         201 CITD
         Stillwater OK  74074
                     Cincinnati, Ohio
     Karla Hyde
            OFFICE: 2078799496
       Northern Ecological Associates, Inc.
         386 Fore St.  Ste. 40
         Portland  OR  04101
     Peter J. Idstein
             OFFICE: 606-624-8722
               FAX:  606-622-2876
            idstein@gateway. net
       Ewers Water Consultants
         971 Villa Drive, Apartment #27
         Richmond KY  40475
     Timothy I. Ingram
            OFFICE: 513-326-4503
               FAX:  513-772-6405
       Health Commissioner, Hamilton County General Health District
         Chester Towers, Suite 600, 11499 Chester Road          PAPER
         Cincinnati OH  45246
     Brian Jacobson
            OFFICE: 724-349-5733
       Penn State University
         905 F W. Aaron Drive
         State College PA  16803
     Karen Jarocki
            OFFICE: 505-768-7706
               FAX:  505-768-7601
            jarocki@mrcabq. com
       Mission Research Company
         5001 Indian School Road NE
         Albuquerque NM   87110
     Tony Jasek
            OFFICE: 210-536-5448
               FAX:  210-536-1130
            tony.jasek@brooks. af.mil
       US Air Force
         2903 Oak Falls
         San Antonio TX
     Bruce Jeffries
            OFFICE: 517-335-0183
               FAX:  517-335-6565
         401 S. Washington Square
         Lansing Ml  48933
     Becky Jenkins
            OFFICE: 614-265-6631
               FAX:  614-263-8144
       Ohio Division of Wildlife
         1840 Belcher Dr., Bldg. G-2
         Columbus OH   43224
     Ralph Johanson
            OFFICE: 502-584-4244
               FAX:  502-584-4246
            rjohanson@grwinc. com
       GRW Engineers
         433 South 5th Street
         Louisville KY  40202
     Brian C. Johnson
            OFFICE: 518-474-5488
               FAX:  518-473-2534
            brian.johnson@oag.state. ny. us
       NY State Attorney Gen. - Env. Protection Bureau
         The Capitol
         Albany NY   12224
          As of: Thursday, September 30, 1999

 Environmental Problem Solving with Geographic Information Systems Conference
 September 22-24,1999
     Robert L. Johnson
            OFFICE: 614-466-1060
              FAX:  614-564-2410
            bjohnson@gw. odh. state, oh. us
List of Attendees
       Ohio Department of Health
         246 North High Street
         Columbus OH  43215
           Cincinnati, Ohio
     J. Kent Johnson, PhD
            OFFICE:  319-626-6053
               FAX:  319-626-6053
            JKENT@inav. net
       Iowa Institute of Hydraulic Research
         University of Iowa
         Iowa City IO  52242
     Rick Jones
            OFFICE:  317-247-3105
               FAX:  317-247-3414
            jonesrw@in-arng. ngb. army.mil
       Military Dept. of Indiana
         2002 S. Holt Road
         Indianapolis IN  46241
     Terri Justice
            OFFICE:  719-567-4035
               FAX:  719-567-4036
       US Air Force 50 Ceslcecr
         300 O'Malley Ave., Suite 19
         SchrieverAFB CO   80912-5091
     Jeff Kaczka
            OFFICE:  765-285-2327
               FAX:  765-285-2606
            jeflinl@gw. bsu. edu
       Department of Natural Resources and Environmental Management;
         Ball State University
         Muncie IN  47306
     K.E. (Kim) Kalen
            OFFICE:  613-992-0757
               FAX:  613-996-3925
       Dept. of National Defense-North WarningSystem
         219 Laurier Avenue West
         Ottawa, Ontario K1AOK2
     Kelly Kaletsky
            OFFICE:  5132856075
            kelly.kaletsky@epa.state. oh. us
       Office of Federal Facilities Oversight - Ohio EPA
         401 East Fifth Street
         Dayton OH  45402-2911
     Sumana Keener
            OFFICE:  513-556-2542
               FAX:  513-556-2522
            skeener@uceng. uc. edu
       University of Cincinnati
         P.O. Box 210071
         Cincinnati OH  45227
     Bud Keesee
            OFFICE:  410-278-6755
               FAX:  410-278-6779
            bkeesee@dshe. apg. army.mil
       Div. of Safety, Health, and Environment
         STEAP-SH-ER - Bldg 5650
         Aberdeen Proving Group Garrison
         MD  21005
     Larisa J. Keith
            larisaj_k@yahoo. com
       Northern Kentucky Area Planning Commission in Kenton County
         146 Stafford Ridge Rd.                              PAPER
         Sanders KY  41083
     Jay H. Kim
            OFFICE:  304-285-6140
               FAX:  304-285-6111
       CDC I NIOSH
         1095 Willowdale Road
         Morgantown WV  26505
As of: Thursday, September 30, 1999

 Environmental Problem Solving with Geographic Information Systems Conference
 September 22-24,1999
     Carl Kinkade
            OFFICE: 402-441-6246
               FAX:  402-441-8323
            ckinkade@ci. lincoln. ne. us
List of Attendees
           Cincinnati, Ohio
       Lincoln-Lancaster County Health Dept.
         3140 N Street
         Lincoln  NE  68510
     Lyn Kirschner
            OFFICE: (765)494-9555
            kirschner@ctic.purdue. edu
       Conservation Technology Information Center
         1220 Potter Dr., Ste. 170
         West Lafayette IN  47906
     Karen KM ma
            OFFICE: 202-260-7087
              FAX:  202-260-7024
       USEPA/Office of Water
         401 M Street, SW Mail Code 4503F
         Washington DC  20460
     Randy Knight
            OFFICE: 319-398-5893
               FAX:  319-398-5894
            rknight@kirkwood. cc. ia. us
       Kirkwood Community College
         6301 Kirkwood Blvd. SW
         Cedar Rapids IA  52406
     Jill Korach
            OFFICE: 513-921-5124
               FAX:  513-921-5136
            imago@one. net
       IMAGO, Inc.
         553 Enright Avenue
         Cincinnati  OH  45205
     Angel J. Kosfiszer
            OFFICE: 214-665-2187
               FAX:  214-665-6490
            kosfiszer. angel@epamail. epa.gov
       US EPA Region 6
         1445 Ross Avenue
         Dallas TX  75202-2733
     Jeff Kreider
            OFFICE: 608-266-0856
               FAX:  608-267-2800
            kreidj@dnr. state, wi.us
       Wisconsin Dept. of Natural Resources
         101 S. Webster Street
         Madison Wl   53707
     Jim Kreissl
            OFFICE: (513)569-7611
       U.S. EPA, CERI, TTB
         26 W. Martin Luther King Dr.
         Cincinnati OH  45268
     Dennis Kreitzburg
            OFFICE: 216-739-0555
               FAX:  216-739-0560
            DJK&HaleyAldrich. com
       Haley & Aldrich, Inc.
         5755 Granger Road, Suite 100
         Independence OH  44131
     Eric J. Kroger
            OFFICE: 513-648-4473
               FAX:  513-648-4473
            eric. kroger@fernald. gov
       Fluor Daniel Fernald, Inc.
         P.O. Box 538704
         Cincinnati OH  45253
     Rosanne Kruzich
            OFFICE: 502-895-4559
               FAX:  502-895-4559
            rkruzich@bellsouth. net
       RKX Consulting, Inc.
         750 Zorn Avenue #56
         Louisville KY  40206
As of: Thursday, September 30, 1999

 Environmental Problem Solving with Geographic Information Systems Conference
 September 22-24,1999
     Dr. Sudhir R. Kshirsagar
            OFFICE:  513-474-9780
               FAX:  513-474-9781
            president@gqc. com
List of Attendees
       Global Quality Corporation
         7828 Beechmont Avenue, Suite 202
         Cincinnati OH  45255
           Cincinnati, Ohio
     Tony Lafferty
            OFFICE: 314-949-6620
               FAX:  314-949-6735
            tlafferty@esri. com
       ESRI St. Louis
         207 South Main St.
         St. Charles  MO  63301
     Mark S. Landry
            OFFICE:  540-951-4899
               FAX:  540-951-4896
            m -landry@vt. e du
       VA Polytechnic Institute and State University
         Dept. of Agric. & Applied Economics
         Blacksburg VA  24061-0401
     Robert Langstroth
             OFFICE:  608-264-8801
               FAX:  608-264-8795
       Wisconsin Dept. of Commerce
         201 W. Washington Ave.
         Madison Wl  53701
     Elizabeth L. Lanzer
            OFFICE:  360-705-7476
              FAX:  360-705-6833
       WA Department of Transportation
         P.O. Box 47331
         Olympia WA  98504-7331
     Jill Leale
            OFFICE:  616-696-1606
            lealej@river. it.gvsu. edu
       Grand Valley State University
         14990 Simmons Ave.
         Cedar Springs Ml  49319
     Lawrence A. Lee
            OFFICE:  919-560-6344
               FAX:  919-530-7973
            larrylee@wpo. nccu. edu
       North Carolina Central University
         1801 Fayetteville Street
         Durham NC  27707
     Sang-Suk Lee
            OFFICE:  614-292-0585
               FAX:  614-292-7688
       The Ohio State University
         223 Mendenhall Lab., 125 South Oval Mall
         Columbus OH   43210
     Jay Lee, Ph.D.
            OFFICE:  3306723222
       Dept. of Geography
         Kent State University
         Kent OH  44242-0001
     David R. Legates, Ph.D.
            OFFICE:  302-831-2294
              FAX:  302-831-6654
            legates@udel. edu
       University of Delaware, Center for Climatic Research
         216 Pearson Hall
         Newark DE  19716-2541
     Geoff Leking
            OFFICE:  419-373-3092
               FAX:  419-373-3125
            geoff.leking@epa.state. oh. us
       OHIO EPA
         347 N. Dunbridge Rd.
         Bowling Green OH  42402
As of: Thursday, September 30, 1999

 Environmental Problem Solving with Geographic Information Systems Conference
 September 22-24,1999                List Of Attendees                      Cincinnati, Ohio
     April Lewis
            OFFICE: 937-255-5654x3512
               FAX:  937-255-4645
            april. lewis@afit.af.mil
Air Force Institute of Technology
  AFITICEL 2950 P Street Bldg. 643
  Wright Patterson AFB OH  45433-7765
     Norma Lewis
             OFFICE:  513-569-7665
               FAX:  513-569-7471
            lewis, norma. @epamail. epa. gov
  26 W. Martin Luther King Drive
  Cincinnati OH  45268
     Xaan Li
               FAX:  614-292-7688
            li@geology. ohio-state, edu
Dept. of Geological Studies Ohio State U.
  Columbus OH  63210
     Jun Liang
            OFFICE: 513-569-7619
            liangjun@hotmail. com
  435 Riddle Rd. Apt.9
  Cincinnati OH  45220
     Susan Licher
Ball State Graduate Student
            sslicher@hotmail. com
     Jason Linn
            OFFICE: 765-285-2327
               FAX:  765-285-2606
            jeflinl@gw. bsu. edu
Department of Natural Resources and Environmental Management
  Ball State University
  Muncie IN  47306
     Beth Linton
            OFFICE: 770-677-0330
               FAX:  770-399-0535
United Capitol Insurance Company
  400 Perimeter Center Terrace, Suite 400
  Atlanta GA  30346
     Amy Liu
            OFFICE: 5135697171
            liu. amy@epamail. epa.gov
  MS 690, 26 W. Martin Luther King Dr
  Cincinnati OH   45268
     Lin Liu, Ph.D.
            OFFICE: 513-556-3429
               FAX:  513-556-3370
University of Cincinnati
  Department of Geography
  714 Swift Hall
  Cincinnati OH  45221-0131
     Bill Lohner
            OFFICE: 937-285-6051
               FAX:  937-285-6404
            bill. lohner@epa.state.oh. us
Ohio EPA
  401 E. 5th Street
  Dayton OH  45402
     Cornell Long
            OFFICE:  210-536-6121
               FAX:  210-536-1130
            Cornell, long@brooks.af.mil
USAF Institute for ESOH Risk Analysis
  2513 Kennedy Circle
  Brooks AFB TX  78235-5123
As of: Thursday, September 30, 1999

 Environmental Problem Solving with Geographic Information Systems Conference
 September 22-24,1999
     Colleen Lovett
            OFFICE:  937-656-3637
              FAX:  937-656-3663
            colleen. lovett@wpafb. af.mil
List of Attendees
       U.S. Air Force
         4225 Logistics Ave., Suite 23
         Wright - Patterson AFB OH  45433-5762
          Cincinnati, Ohio
     Damon Lowe
            OFFICE:  317-598-8742
            dlowe@worldnet. att. net
       Ball State University
         6836 Challenge Lane
         Indianapolis IN  46250
     James A. Lynch, Ph.D.
            OFFICE:  814-865-8830
              FAX:  814-863-7193
       Penn State University
         311 Forest Resource Lab
         University Park PA  16802
     Stan Lynn
            OFFICE:  513-564-8349
              FAX:  513-241-0354
            lynns@ttemi. com
       Tetra Tech EM Inc.
         625 Eden Park Drive, Suite 100
         Cincinnati OH  45202
     Debbie E. Maes
            OFFICE:  505-846-8568
              FAX:  505-853-1793
            debbie.maes@ao. dtra.mil
       Defense Threat Reduction Agency
         1680 Texas Street, SE
         Kirtland AFB NM  87117-5669
     Robert Magai
            OFFICE:  573-522-3779
              FAX:  573-525-5797
            nrmagar@mail. dnr.state.mo. us
       Missouri DNR
         205 Jefferson Street
         Jefferson City MO  65102
     Laura Mahoney
            OFFICE:  615-255-2370x406
              FAX:  615-256-8332
            lmahoney@brwncald. com
       Brown & Caldwell
         227 French Landing Drive
         Nashville TN  37228
     Sarada Majumder
            majumder. sarada@epa.gov
       SBI EnvAISEPA
         26 W. Martin Luther King Drive
         Cincinnati OH  45219
     Miguel A. Maldonado
            OFFICE:  787-764-8824
              FAX:  787-766-0150
           jcaemer@prtc. net
         P.O. Box 11488
         San Juan PR  00910
     Richard M. Males
            OFFICE:  513-871-8566
       RMM Technical Services, Inc.
         3319 Eastside Avenue
         Cincinnati OH  45208
     Christopher Mantia
            OFFICE:  937-776-7724
              FAX:  937-586-3059
       Corbus, LLC
         33 W. First Street, Suite 300
         Dayton OH
As of: Thursday, September 30, 1999

 Environmental Problem Solving with Geographic Information Systems Conference
 September 22-24,1999
     Renante Marante
            OFFICE: 312-742-0123
              FAX: 312-744-6451
            rmarante@ci.chi.il. us
List of Attendees
       Dept. of Environment Chicago Brownfields
         30 N. La Salle, Room 2500
         Chicago IL  6060
          Cincinnati, Ohio
     Joanne Marker!
            OFFICE: 3607057444
       Washington Dept. of Transportation
         310 Maple Park Ave., SE
         Olympia WA  98504-7331
     Lawrence Martin
            OFFICE: 202-564-6497
               FAX: 202-565-2926
         401 M Street. NW (8103R)
         Washington DC  20460
     Robert C. Martin, Jr.
            OFFICE: 404-894-8446
              FAX: 404-894-2184
       Georgia Tech Research Institute - Electro-Optics
         Environment and Materials Laboratory
         Atlanta GA 30332-0837
     Margaret B. Martin, P.E.
            OFFICE: 410-962-3500
               FAX:  410-962-2318
            margaret. b.martin@nab02. usace.army
       USAGE / HTRW Design Center
         P.O. Box 1715
         ATTN: CENAB-EN_
         Baltimore MD  21203
     Dana Martin-Hayden
            OFFICE: 419-373-3067
              FAX: 419-352-8468
            dana.martin-hayden@epa. state, oh. us
       OHIO - EPA
         347 North Dunbridge Road
         Bowling Green OH   43402
     Carmen Maso
            OFFICE: 312-886-1070
              FAX: 312-353-6519
            maso. carmen@epa.gov
       US EPA
         77 West Jackson
         Chicago IL  60604
     Jason Masoner
            OFFICE: 580-436-8508
         Ada  OK  74820
     Gregg Matthews
            OFFICE: 613-541-6000x6099
              FAX: 613-541-6596
            matthews-g@rmc. ca
       Environmental Sciences Group, RMC
         P.O. Box 17000 Stn. Forces, Bldg. 62
         Kingston - Canada ON  K7K7B4
     Robert Matzner
            OFFICE: 703-305-5975
              FAX: 703-305-6309
            matzner. robert@epa.gov
         401 M St., SW
         Mail Code 7507c
         Washigton DC  20460
     Dr. Eric F. Maurer
            OFFICE: 606-257-7652
               FAX: 606-257-1717
            efmaur@pop. uky.edu
       Univ. of Kentucky, Biological Sciences
         101 Morgan Building
         Lexington KY  40506
As of: Thursday, September 30, 1999

 Environmental Problem Solving with Geographic Information Systems Conference
 September 22-24,1999
     Chris Mazza
            OFFICE: 765-285-2327
              FAX: 765-285-2606
            jeflinl@gw. bsu. edu
List of Attendees
          Cincinnati, Ohio
       Department of Natural Resources and Environmental
         Ball State University
         Muncie IN 47306
     R.B. (Rusty) McAllister
            OFFICE:  501-682-0022
              FAX:  501-682-0010
            mcallister@adeq. state, ar. us
       Arkansas Dept. of Environmental Quality
         8001 National Drive
         Little Rock AR  72209
     Andy McCammack
            OFFICE:  765-285-2327
              FAX:  765-285-2606
            jeflinl@gw. bsu. edu
       Department of Natural Resources and Environmental
         Ball State University
         Muncie IN 47306
     Larry McCanoless
            OFFICE:  513-648-3772
              FAX:  513-648-4029
       Flour Daniel Fernald
         P.O. Box 538704
         Cincinnati OH  45253
     Jeff McCormack
            OFFICE:  810-239-1154
              FAX:  810-239-1180
            jmccormack@ssoe. com
       SSOE, Inc.
         111 E. Court St.
         Flint Ml   48502
     Seth McCoy
            OFFICE:  765-285-2327
              FAX:  765-285-2606
            jeflinl@gw. bsu. edu
       Department of Natural Resources and Environmental
         Ball State University
         Muncie IN 47306
     George McKee
            OFFICE:  513-684-3073
              FAX:  513-684-3039
            george. m. mckee@lrdor. usace. army.mil
       Army Corps of Engineers
         5717 Nickview Drive
         Cincinnati OH  45247
     Iver McLeod
            OFFICE:  207-287-8010
              FAX:  207-287-7826
            iver.j.mcleod@state. me. us
       Maine Department of Environmental Protection
         State House Station #17
         Augusta ME  04333
     Christine McMahon
            OFFICE:  6186921478
            mcmahon(a)icon-stl. net
         209 N. Fillmore Street
         Edwardville IL  62025
     Robert B. McMaster
            OFFICE:  6126259883
            mcmaster(fl),atlas.socsci. umn. edu
       Dept. of Geography
         414 Social Sciences Bldg.
         Minneapolis MN   55455
     Stephanie McSpirit
            OFFICE:  606-622-3070
            antmcspi@acs. eku
       Eastern Ky. University
         223 Keith
         Richmond KY  40475
As of: Thursday, September 30, 1999

 Environmental Problem Solving with Geographic Information Systems Conference
 September 22-24,1999
     Jeremy Mennis
List of Attendees
             Cincinnati, Ohio
       Dept. of Geography, Pennsylvania State Univ.
         302 Walker Building
         University Park PA   16802
     Jim Metzger
            OFFICE:  812-665-2207
              FAX:  812-665-5041
       Indiana DNR/Div. of Reclamation
         RR 2 Box 129
         Jascksonville IN  47438
     Jim Meyer
            OFFICE:  573-882-9291
              FAX:  573-884-2199
            jcmeyer@showme.missouri. edu
       CARES - University of Missouri
         Columbia MO  65211-6200
     Michael V. Miller
            OFFICE:  5417524271
       CH2M Hill
         2300 NW Walnut Blvd.
         Corvallis OR  97330
     Bruce Milligan
            OFFICE: 317-290-3200x347
              FAX: 317-290-3225
            bruce.milligan@in. usda.gov
       Natural Resources Conservation Service
         6013 Lakeside Blvd.
         Indianapolis IN  46278
     Scott Minamyer
            OFFICE:  (513)569-7175
              FAX:  (513)569-7585
       U.S. EPA, CERI, TTB
         26 W. Martin Luther King Dr.
         Cincinnati OH   45268
     Ken Mix
            OFFICE:  765-285-2327
              FAX:  765-285-2606
            jeflinl@gw. bsu. edu
       Department of Natural Resources and Environmental Management
         Ball State University
         Muncie IN  47306
     Sara Moola
            smoola@esri. com
         2205 A West Balboa Blvd.
         Newport Beach CA  92663
     Mike Moore
            OFFICE:  717-783-7258
              FAX:  717-783-7267
            mmoore@dcnr.state.pa. us
       PA Geological Survey
         1500 North 3rd St.
         Harrisburg PA  17102-1910
     Shawn Moore
            OFFICE:  937-255-5654
              FAX:  937-255-4645
         2950 P Street Bldg. 643
         Wright Patterson AFB OH
     Maggie Moorhouse
            OFFICE:  361-883-6016
              FAX:  361-883-7417
       Moorhouse Associates, Inc.
         5826 Bear Lane
         Corpus Christ! TX  78405
  As of: Thursday, September 30, 1999

 Environmental Problem Solving with Geographic Information Systems Conference
 September 22-24,1999                List Of Attendees                      Cincinnati, Ohio
     Bruce Motsch
            OFFICE:  614-265-6772
            bruce.motsch@,dnr.state. oh. us
Ohio Department of Natural Resources
  1952 Belcher Drive C-2
  Columbus OH   43224
     Erin T. Mutch
            OFFICE:  209-532-0361
              FAX:  209-532-0773
Condor Earth Technologies, Inc.
  21663 Brian Lane
  Sonora CA  95370
     Wolf Naegeli
            OFFICE:  423-584-4806
            wnn@utk. edu
Energy, Environment & Resource Center, U of TN
  4425 Balraj Ln
  Knoxville TN  37921-2938
     Jill Neal
            OFFICE:  (513)569-7277
            niel.jill@epamail. epa.gov
  26 W. Martin Luther King Dr.
  Cincinnati OH  45268
     Dean Nelson
            OFFICE:  423-483-3191
              FAX:  423-483-8617
            dnelson@usit. net
Peer Consultants, P.C.
  78 Mitchell Road
  Oak Ridge TN  37830
     Tamas Nemeth

            lacus@rissac. hu
GIS Lab, Research Inst. For Soil Science & Agri. Chemistry
  Hungarian Academy of Sciences
  H-1022 Bud Herman Otto ut 15
     Bruce Nielsen
            OFFICE:  317-290-3200
              FAX:  317-290-3225
USDA-Natural Resources Conservation Service
  6013 Lakeside Blvd.
  Indianapolis IN  46278
     Douglas A. Nieman
            Dnieman@normandeau. com
Normandeau Associates
  3450 Schuylkill Rd.
  Spring City PA  19475
     Eugene O'Brien
            OFFICE:  765-285-2327
              FAX:  765-285-2606
            jeflinl@gw. bsu. edu
Department of Natural Resources and Environmental Management
  Ball State University
  Muncie IN  47306
     Toby Obenauer
            OFFICE:  305-242-7892
              FAX:  305-242-7836
National Park Service - Everglades
  40001 SR 9336
  Homestead FL  33034-6733
     Ramon A. Olivero
            OFFICE:  919-572-2764
               FAX:  919-572-2765
            rolivero@lmepo. com
Lockheed Martin Environmental Services
  100 Capitola Drive, Suite 111
  Durham NC  27713
As of: Thursday, September 30, 1999

 Environmental Problem Solving with Geographic Information Systems Conference
 September 22-24,1999
     Deborah Olszowka
            OFFICE: 513-231-7719
              FAX:  513-231-7761
List of Attendees
           Cincinnati, Ohio
         5735 Kellogg Avenue
         Cincinnati OH  45228
     Lindell Ormsbee
             OFFICE:  606-257-1302
            lormsbee@engr. uky.edu
       University of Kentucky
         Dept. of Civil Engineering
         Lexington KY  40506-0281
     Steven D. Ortiz
            OFFICE:  512-239-2008
               FAX:  512-239-4007
            sortiz@tnrcc. state, tx. us
       TX Natural Resource Conservation Commission
         12100 Park 35 Circle, MC -108
         Austin TX 78753
     Kurt Overmyer
            OFFICE:  616-336-3086
               FAX:  616-336-2436
            overmyer@iserv. net
       Kent County Health Department
         700 Fuller, N.E.
         Grand Rapids Ml  49503
     David A. Padgett
            OFFICE:  615-963-5471
              FAX:  734-939-5813
            padgettdavid@netscape. net
       Tennessee State University
         3500 John A. Merritt Blvd.
         Nashville TN  37219
     David Padgett, Ph.D.
            OFFICE:  4407758747
       Oberlin College
     Mercedes Padilla
            OFFICE:  787-767-8181x2243
               FAX:  787-766-0150
            jcaemer@prtc. net
       PR Environemtal Quality Board
         P.O. Box 192785
         San Juan PR 00919
     Susan Pagano
            OFFICE:  610-696-4150
               FAX:  610-696-8608
            spagano@thcahill. com
       Cahill Associates
         104 South High Street
         West Chester PA  19382
     Harry Parrott
            OFFICE:  414-297-3342
               FAX:  414-297-3808
            hparrott/r9@fs.fed. us
       USDA-Forest Service
         310 W.Wisconsin Ave.
         Milwaukee Wl  53203
     Laszlo Pasztor, Ph.D.
            OFFICE:  361-356-3694
               FAX:  361-212-1891
            lacus@rissac. hu
       GIS Lab, Research Inst. for Soil Science & Agri. Chemistry
         Hungarian Academy of Sciences
         Herman Otto ut 15
         H-1022  Bud  Hungary
     Clair (Pat) Patterson
            OFFICE:  410-260-8512
               FAX:  410-260-8595
            ppatterson@dnr.state. md. us
       Maryland DNR-Forest Service
         Tawes State Office Building E1 580 Taylor Avenue
         Annapolis MD  21401
As of: Thursday, September 30, 1999

 Environmental Problem Solving with Geographic Information Systems Conference
 September 22-24,1999                 List Of Attendees                      Cincinnati, Ohio
     Michael J. Pavilonis
             OFFICE: 412-442-4066
               FAX:  412-442-4194
            pavilonis.michael@dep. state, pa.us
Pennsylvania DEP
  400 Waterfront Drive
  Pittsburgh PA  15222-4745
     Alan Peacock
             OFFICE: 317-337-3414
               FAX:  317-337-3235
             apeacock@dowagro. com
Dow AgroSciences
  9330 Zionsville Road
  Indianapolis IN  46268
     Norman E. Peters
             OFFICE: 770-903-9145
               FAX:  770-903-9199
U.S. Geological Survey
  3039 Amwiler Rd., Suite 130
  Atlanta GA  30360-2824
     Carl Petersen
             OFFICE: 502-564-8390
               FAX:  502-564-2088
             cpeteOO@pop. uky.edu
Unversity of Kentucky
  1854 Bellefonte Drive
  Lexington KY  40503
     Herb Petitjean
             OFFICE: 502-564-6717
               FAX:  502-564-5096
Kentucky Dept. for Environmental Protection
  14 Reilly Road
  Frankfort KY  40601
     Rebecca Petty
             OFFICE: 614-466-4801
               FAX:  614-466-4556
Ohio Department of Health
  246 N. High Street, 5th Floor
  Columbus OH  43266-0588
     Kenneth Pew
             OFFICE: 216-881-6600x814
               FAX:  216-881-2738
            pewk@neorsd. org
N.E. Ohio Regional Sewer District
  4415 Euclid Avenue
  Cleveland OH  44103
     Robert R. Pierce
             OFFICE: 770-903-9113
               FAX:  770-903-9199
  3039 Amwiler Road, Suite 130
  Atlanta GA  30360-2824
     Michael Plastino
             OFFICE: 202-260-0048
               FAX:  202-260-7926
EPA Office of Water
  401 M Street, SW(MC 4102)
  Washington  DC  20460
     Mark Porembka
             OFFICE: 412-442-4327
               FAX:  412-442-3428
            porembka.mark@dep. state, pa.us
PA Dept. of Environmental Protection
  400 Waterfront Drive
  Pittsburgh PA  15222
     Jacob Prater
            OFFICE:  765-759-6898
Ball State University
  6613 W.Jackson St.
  Muncie IN  47304
As of: Thursday, September 30, 1999

 Environmental Problem Solving with Geographic Information Systems Conference
 September 22-24,1999
     Garry Price
            OFFICE:  717-772-5630
              FAX:  717-787-5259
            price. garry@dep. state, pa. us
List of Attendees
       PA Dep.
         400 Market Street
         Harrisburg PA   17105
          Cincinnati, Ohio
     Eric R. Pruitt
            OFFICE:  812-323-8871
       Indiana University
         714 E Cottage Grove West
         Bloomington  IN  47408
     Tom Quinn
            OFFICE:  513-489-6611
              FAX:  513-489-6619
         7209 E KemperRd.
         Cincinnati  OH  45241
     Nancy J. Radle
            OFFICE:  651-779-5102
              FAX:  651-779-5109
            nancy, radle @dot. state, mn.us
       Minnesota Department of Transportation
         3485 Hadley Avenue, N - Mail Stop 620
         Oakdale MN  55128
     Charles Rairdan
            OFFICE:  916-557-7833
              FAX:  916-557-7848
            crairdan@spk. usace. army.mil
       USAGE - Sacramento District
         1325 "J" Street
         Sacramento CA 95814
     Alexandrine Randriamahefa
            OFFICE:  256-726-7436
              FAX:  256-726-7055
       Oakwood College
         7000 Adventist Blvd.
         Huntsville AL  35896
     Ravi Rao
            OFFICE:  404-562-8349
            rao. ravi@epamail. epa.gov
       USEPA  Region 4, OPM, PAB
         61 Forsyth Street
         Atlanta GA  30303
     Alan Rao, Ph.D.
            OFFICE:  6179241770
       Vanasse Hangen Brustlin, Inc.
         101 Walnut Street
     Andrew Rawnsley

            ronz@ravensfield. com
       Ravensfield Geographic Resources, Ltd.
         P.O. Box 410
         Granville OH  43023
     Alan Rea
            OFFICE:  208-387-1323
              FAX:  208-387-1372
       US Geological Survey
         230 Collins Road
         Boise ID 83702-4520
     Jeffrey Reese
            OFFICE:  765-213-1269
              FAX:  765-747-7744
            masine@iquest. net
       Delaware County, Indiana GIS
         100W. Main St., Room 204
         Muncie IN  47305
As of: Thursday, September 30, 1999

 Environmental Problem Solving with Geographic Information Systems Conference
 September 22-24,1999
     Jon Reid
            OFFICE: 513-558-1723
            jon. reid@uc.edu.
List of Attendees
          Cincinnati, Ohio
       University of Cincinnati
         3223 Eden Ave. ML 056
         Cincinnati OH  45267-0056
     Chad K. Rhodes
            OFFICE:  313-237-3093
               FAX:  313-224-1547
            rhodesc@EnvAfrs. ci. detroit.mi. us
       City of Detroit Dept. of Environmental Affairs
         660 Woodward Avenue, Suite 1590
         Detroit Ml  48226
     Carl Rich
            OFFICE:  303-202-4316
              FAX:  303-202-4354
            clrich@usgs. gov
       U.S. Geological Survey
         P.O. Box 25046, MS-516, Denver Federal Center
         Denver CO  80225
     Ian Richardson
            OFFICE:  519-884-0510
               FAX:  519-884-0525
            irichardson@rover. com
         651 Colby Dr.
     John Richardson
            OFFICE:  402-562-8290
               FAX:  404-562-8269
       US EPA Region 4
         61 Forsyth Street
         Atlanta GA 30303
     James M. Rine, Ph.D.
            OFFICE:  803-777-7792
              FAX:  803-777-6437
            jrine@esri. esri. sc.edu
       Earth Sciences & Resources Institute
         University of South Carolina
         901 Sumter Road, Room 401
         Columbia SC  29208
     Jim Rocco
            OFFICE:  330-562-9391
               FAX:  330-562-9391
            roccojl@worldnet. att. net
       Sage Risk Solutions LLC
         360 Heritage Road
         Aurora OH   44202
     Edward Rogers, Jr.
             OFFICE:  9085982600
            erogers@bemsys. com
       BEM Systems, Inc.
         100 PassaicAve.
         Chatham NJ  07928
     Cynthia Root
            OFFICE:  319-398-5678
               FAX:  319-398-1250
            croot@kirkwood. cc. ia. us
       Kirkwood Community College
         6301 Kirkwood Blvd. SW
         Cedar Rapids IA  52406
     Lloyd Ross
            OFFICE:  216-739-0555
               FAX:  216-739-0560
            lsr@haleyaldrich. com
       Haley & Aldrich, Inc.
         5755 Granger Road, Suite 100
         Independence OH  44131
     Randall Ross, Ph.D.
            OFFICE:  (580)436-8611
            ross. randall@epamail. epa.gov
       R.S. Kerr Environ. Research Center U.S. EPA/ORD/NRMRL PAPER
         919 Kerr Research Dr., P.O. Box 119
         Ada OK  74820
As of: Thursday, September 30, 1999

 Environmental Problem Solving with Geographic Information Systems Conference
 September 22-24,1999
     Holly Roten
            OFFICE: 419-353-1531
            rotenh@bgnet. bgsu.edu
                                   List of Attendees
                                          Bowling Green State University
                                            455 South Enterprise, Apartment H.
                                            Bowling Green OH  43402
                      Cincinnati, Ohio
     Chris Rudnick
            OFFICE:  502-564-6716
               FAX:  502-564-5096
                                          Kentucky Environmental Protection-State Superfund
                                            14 Reilly Road
                                            Frankfort KY  40601
     Nel Ruffin
            OFFICE:  606-257-1299
            nruffin@engr. uky.edu
                                          Kentucky Water Resources Research Institute
                                            233 Mining & Minerals Resource Building
                                            University of Kentucky
                                            Lexington KY 40506
     Steve Saines
            OFFICE:  614-644-2752
               FAX:  614-644-2909
                                          Ohio EPA
                                            P.O. Box1049
                                            Columbus OH
     Scott Samson
            OFFICE:  606-257-3767
               FAX:  606-257-4354
                                          University of Kentucky
                                            Rural Sociology Program, Garrigus 500
                                            Lexington  KY  40546
     Vicki Sandiford
            OFFICE:  919-541-2629
               FAX:  919-541-7690
            sandiford. vicki@epa.gov
                                          US EPA
                                            RTP NC  27711
     George Sarapa
            OFFICE:  412-262-5400
               FAX:  412-262-3036
            sarapg@lrkim ball, com
                                          L. Robert Kimball & Associates, Inc.
                                            415 Moon Clinton Road
                                            Moon Township PA  15108
     Gary Schaal
            OFFICE:  614-265-6769
               FAX:  614-267-2981
            gary.schaal@dnr. state, oh. us
                                          Ohio Department of Natural Resources
                                            1952 Belcher Drive C-4
                                            Columbus OH   43224
Devin Scheak
       OFFICE:  513-921-5124
          FAX:  513-921-5136
                                               IMAGO, Inc.
                                                 553 Enright Avenue
                                                 Cincinnati  OH  45205
     Tom Schneider
            OFFICE:  937-295-6466
               FAX:  937-285-6404
            tom.schneider@epa. state.oh. us
                                          Ohio EPA
                                            401 E. 5th Street
                                            Dayton OH   45402
     Theresa Schnorr
            OFFICE:  513-558-5984
                                            4676 Columbia Parkway
                                            Cincinnati OH  45220
           As of: Thursday, September 30, 1999

 Environmental Problem Solving with Geographic Information Systems Conference
 September 22-24,1999
     Susan Schock
            OFFICE: 513-569-7551
               FAX: 513-487-2513
List of Attendees
           Cincinnati, Ohio
         26 W. Martin Luther King Drive
         Cincinnati OH  45268
     Chris Schroeder
            OFFICE:  402-441-6272
               FAX:  402-441-8323
            cschroed@ci. lincoln. ne.us
       Lincoln-Lancaster County Health Dept.
         3140 "N" Street
         Lincoln  NE  68510
     Joe Schubauer-Berigan
            OFFICE:  513-569-7734
       U.S. EPA National Center for Envir. Assessment
         26 W. Martin Luther King Drive
         Cincinnati OH  45220
     Susan Schultz
            OFFICE:  513-861-7666
               FAX:  513-559-3155
            schultz23@fuse. net
       Mill Creek Restoration Project
         42 Calhoun Street
         Cincinnati OH  45219
     Robert Scott
             OFFICE:  404-675-1753
               FAX:  404-675-6246
       Georgia EPD
         4220 International Parkway, Suite 101
         Atlanta GA   30354
     Irena Scott, PhD
            OFFICE:  614-644-8020
               FAX:  614-644-7740
            iscott@gw.odh.state. us
       Ohio Department of Health
         246 North High Street
         Columbus OH  43266-0118
     Tom Seibert
            OFFICE:  502-564-6716
               FAX:  502-564-7484
            torn. seibert@mail. state, ky. us
       Kentucky Division for Waste Management
         14 Reilly Road
         Frankfort KY  40601
     Gabriel Senay, Ph.D.
            OFFICE:  5135697096
         26 W. Martin Luther King Dr.
         Cincinnati OH  45268
     Barbara Seymour
            OFFICE:  301-622-9696x108
               FAX:  301-622-9693
            bseymour@plexsc. com
       Plexus Scientific
         12501 Prosperity Drive, # 401
         Silver Spring MD  20904
     Jo Ann Shaw
            OFFICE:  573-751-9370
            nrshawj@mail. dnr.state.mo. us
       MO Dept. of Natural Resources
         205 Jefferson St.
         Jefferson City MO  65102
     Jerry Shi
            OFFICE:  614-292-6193
               FAX:  614-292-7688
            shi. 19@osu.edu
       The Ohio State University
         125 South Oval Mall
         Columbus OH   43210
As of: Thursday, September 30, 1999

 Environmental Problem Solving with Geographic Information Systems Conference
 September 22-24,1999
     Andy Shirmeyer
List of Attendees
          Cincinnati, Ohio
            OFFICE:  765-285-2327
               FAX:  765-285-2606
            jeflinl@gw. bsu. edu
       Department of Natural Resources and Environmental
         Ball State University
         Muncie IN  47306
     Massoud Shoa
            OFFICE:  502-564-6717x219
               FAX:  502-564-2705
       Kentucky Division of Waste Management
         14 Reilly Road
         Frankfort KY  40601
     Leila Shultz
            OFFICE:  617-327-4294
               FAX:  617-327-4294
            shultz@oeb. harvard, e du
       Harvard University
         22 Divinity Avenue
         Cambridge MA  02138
     Amardeep Singh
            OFFICE:  916-653-2726
               FAX:  916-653-6366
            asingh@fnd. csus.edu
         905 23rd Street #3
         Sacramento CA  95816
     Nathan Sloan
            OFFICE: 765-285-2327
               FAX:  765-285-2606
            jeflirt l@gw. bsu. edu
       Department of Natural Resources and Environmental
         Ball State University
         Muncie IN  47306
     Ryan Sloan
            OFFICE:  765-289-7816
            resloan(a)hotmail. com
       Dept. of Natural Resources , Ball State Univ.
         323 1/2
         Muncie IN  47304
     Mike Sloop
            OFFICE:  770-673-3624
               FAX:  770-396-9495
            msloop@brwncald. com
       Brown and Caldwell
         41 Perimeter Center East
         Suite 400
         Atlanta GA 30346
     Kathleen Smaluk-Nix
            OFFICE:  502-574-1358
               FAX:  502-574-1389
            ksmaluk@louky. org
       Office of Health and Environment
         600 W. Main St. , 4th Floor
         Louisville KY  40202
     Art Smith
            OFFICE:  502-574-2511
              FAX:  502-574-1389
            asmith@louky. org
       City of Louisville Office of Health and Environment
         600 W. Maine Street, 4th Floor
         Louisville KY  40202
     Christopher Smith
            OFFICE:  608-221-6330
               FAX:  608-221-6353
       Wisconsin DNR
         1350 Femrite Drive
         Monona Wl  53716
     Gerry Snyder
            OFFICE:  303-312-6623
               FAX:  303-312-6063
            snyder.gerry@epamail. epa.gov
       US EPA
         999 18th Street, Suite 500, 8IG
         Denver CO  80202-2466
As of: Thursday, September 30, 1999

 Environmental Problem Solving with Geographic Information Systems Conference
 September 22-24,1999
     Marilyn Sommer
List of Attendees
          Cincinnati, Ohio
            OFFICE:  502-564-3256
            maras425@ntr. net
       Kentucky Resource Cabinet
         1933 Deerwood
         Louisville KY  40210
     Puneet Srivastava
            OFFICE:  501-682-0018
              FAX:  501-682-0010
         800 National Drive
         P.O. Box 8913
         Little Rock AR  72205
     Greg Starkebaum
            OFFICE:  303-763-7188
              FAX:  303-763-4896
            gstarkebaum@techlawinc. com
       TechLaw, Inc.
         300 Union Blvd. #600
         Lakewood CO  80228
     Steven K. Starrett, Ph.D
            OFFICE:  785-532-1583
              FAX:  785-532-7717
            steveks@ksu. edu
       Kansas State University
         Manhattan KS  66506-2905
     Dr. Gerry  L.
            OFFICE:  301-415-5265
              FAX:  301-415-5348
       U.S. Nuclear Regulatory Commission
         TWFN-MS T7C6-11545 Rockville Pike
         Rockville MD  20852
     Joel Stocker
            OFFICE:  860-345-4511
              FAX:  860-345-3357
            jstocker@canr. uconn.edu
       UConn CES, NEMO Project
         1066 Saybrook Rd., Box 70
         Haddam CT  06438-0070
     Brad Stone
            OFFICE:  502-564-6716
              FAX:  502-564-4049
            brad, stone @mail. state, ky. us
       KY Div. of Waste Management
         14 Reilly Road
         Frankfort KY  40601
     William Story
            OFFICE:  775-687-4670
              FAX:  775-687-6396
            bstory@ndep. carson-city. nv. us
       Nevada Division of Environmental Protection
         333 W. Nye Lane
         Carson City NV  89706
     Steven V. Strausbauch
            OFFICE:  210-536-6134
              FAX:  210-536-1130
         2513 Kennedy CIR
         Brooks AFB TX  78235
     Ryan Stults
            OFFICE:  765-285-2327
              FAX:  765-285-2606
            jeflinl@gw. bsu. edu
       Department of Natural Resources and Environmental
         Ball State University
         Muncie IN  47306
     Prathiba Subramaniam
            OFFICE:  513-474-9780
            prathiba@gqc. com
       Global Quality Corporation
         3304 Jefferson Avenue #104
         Cincinnati OH  45220
As of: Thursday, September 30, 1999

 Environmental Problem Solving with Geographic Information Systems Conference
 September 22-24,1999                List Of Attendees                     Cincinnati, Ohio
     Bhagya Subramanian
            OFFICE:  5135697349
            subramanian. bhagya@epa.gov
  26 W. Martin Luther King Dr.
  Cincinnati OH  45268
     Parrish Swearingen
            OFFICE:  912-926-1197x114
              FAX:  912-926-9642
            parrish. swearingen@robins. af.mil
  455 Byron Street
  Robins AFB GA  31098
     Jane C. Thapa
            OFFICE: 518-402-7713
              FAX: 518-402-7599
            jct02@health. state, ny. us
  547 River Street, Room 400
  Troy NY  12180-2216
     Geoff Thompson
            OFFICE:  765-285-2327
              FAX:  765-285-2606
            jeflinl@gw. bsu. edu
Department of Natural Resources and Environmental Management
  Ball State University
  Muncie IN  47306
     Gordon Thompson
            OFFICE:  513-782-4556
            gwthompson@theitgroup. com
The IT Group
  312 Directors Drive
  Knoxville TN  37923
     Richard E. Thornburg
            OFFICE:  513-564-1785
              FAX:  513-564-1776
            Rick. Thornburg@cinhlthe.rcc. org
Cincinnati Health Dept.
  3845 Wm. P. Dooley By-Pass
  Cincinnati OH  45223
     Larry Tinney
            OFFICE:  702-897-3270
              FAX:  702-897-3285
            ltinney@lmepo. com
Lockheed Martin Environmental Services
  980 Kelly Johnson Dr.
  Las Vegas NV  89119
     Alejandro F. Tongco
            OFFICE: 405-744-8974
               FAX: 405-744-7008
            al@seic. lse.okstate.edu
Oklahoma Spatial & Environmental Information Clearinghouse,
  Stillwater OK   74076
     Oury Traore
            OFFICE: 606-986-2373
              FAX: 606-986-2619
            otraore@maced. org.
  433 Chestnut St.
  Berea KY  40403
     Michael E. Troyer, Ph.D.
            OFFICE:  (513)569-7399
  26 W. Martin Luther King Dr.
  Cincinnati OH  45268
     Amy Tucker
            OFFICE:  765-281-1026
            amyetucker@prodigy. net
Ball State University
  100S. CalvertAve.
  Muncie IN  47303
As of: Thursday, September 30, 1999

 Environmental Problem Solving with Geographic Information Systems Conference
                                        List Of Attendees                     Cincinnati, Ohio
                                              Hull and Associates, Inc.
                                                 4700 Duke Drive, Suite 172
                                                 Mason OH  45040
September 22-24,1999
    W. Lance Turley
            OFFICE: 513-459-9677
              FAX:  513-459-9869
            hurley @hullinc. com
     Rachel Tyler
            OFFICE:  352-378-6517
                                              Florida Rural Community Assistance Project
                                                6212 NW 43rd St. Suite A
                                                Gainesville FL  32697
     Tom Ungar
            OFFICE: 216-623-2738
              FAX: 216-621-4972
            thomas. ungar@mw.com
                                              Montgomery Watson
                                                1300 East 9th Street,  Suite 2000
                                                Cleveland  OH  44114
     Andre van Delft
            OFFICE:  31152695286
              FAX:  31152695335
                                              TNO Building and Construction Research
                                                P.O. Box 49
                                                Delft - Netherlands 2600 AA
     Laurens Van der Tak
            OFFICE:  703-471-6405
              FAX:  703-471-1508
                                              CH2M Hill
                                                13921 Park Center Rd., Suite 60
                                                Herndon VA  20171
     Robert J. van Waasbergen
            OFFICE: 408-737-7697
               FAX: 408-737-7978
            robertvw@aeds. com
                                              Applied Environmental Data Services
                                                201 W. California Ave. #206
                                                Sunnyvale CA  94086
     Dirk Vandervoort
            OFFICE:  360-475-6915
              FAX:  360-475-6901
            vandervoort@ctc. com
                                              Concurrent Technologies Corporation
                                                510 Washington Avenue, Suite 120
                                                Bremerton WA  93777-1844
     James M. Vanek
            OFFICE: 412-442-4031
              FAX: 412-442-4328
            vanek.james@dep. state.pa. us
                                              PA Dept. of Environmental Protection
                                                400 Waterfront Drive
                                                Pittsburgh PA  15222
     Paul Vermaaten
            OFFICE:  517-788-4075
              FAX:  517-788-4641
                                              City of Jackson
                                                161 W. Michigan Ave.
                                                Jackson Ml  49201
     Janet Vick
Gwinnett County
  75 Langley Drive
  Lawrenceville GA  30045
     Todd Vikan
            OFFICE:  937-259-9850
              FAX:  937-259-9869
            tvikan@kpmg. com
                                              KPMG LLP
                                                3139 Research Blvd. Suite 200
                                                Dayton OH  45420
                           As of: Thursday, September 30, 1999

 Environmental Problem Solving with Geographic Information Systems Conference
 September 22-24,1999                List Of Attendees                      Cincinnati, Ohio
     Jack Wachter
            OFFICE:  513-352-6992
               FAX:  513-352-4970
            jack.wachter@cinems. rcc. org
City of Cincinnati
  Office of Environmental Mgmt.
  805 Central Ave., Suite 610
  Cincinnati OH  45202-1947
     Scott Wade
            OFFICE:  734-332-1200
               FAX:  734-332-1212
            swade@limno. com
LIMNO-Tech, Inc.
  501 Avis Drive
  Ann Arbor Ml  48108
     Tim Wade
            OFFICE:  702-798-2117
               FAX:  702-798-2208
  944 E. Harmon Avenue
  Las Vegas NV  89119
     Jerry Wager
            OFFICE:  614-265-6619
               FAX:  614-262-2064
            jerry.wager@dnr.state.oh. us
Ohio Department of Natural Resources
  1939 Fountain Square Ct. E-2
  Columbus OH   43224
     James Wang
            OFFICE:  404-562-8280
            wang.james@epamail. epa.gov
USEPA Region 4
  61 Forsyth Street
  Atlanta GA  30303
     Lizhu Wang
            OFFICE:  608-221-6335
               FAX:  608-221-6353
            wangl@mailo. dnr. state, wi. us
Wisconsin DNR
  1350 Femrite Drive
  Monona WI  53716
     Xinhao Wang, Ph.D.
            OFFICE:  513-556-0497
               FAX:  513-556-1274
            xinhao. wang@uc.edu
School of Planning, University of Cincinnati
  621ODAAP Building
  Cincinnati OH  45221-0016
     Robin Wankum
            OFFICE:  913-458-6538
               FAX:  913-458-6633
            wankumrd@bv. com
Black and Veatch Special Projects
  6601 College Blvd.
  Overland Park KS  66211
     Stephen Want
            OFFICE:  301-951-4681
               FAX:  301-652-1273
  Congressional Information Service, Inc.
  4520 East-West Highway
  Bethesda MD  20814-3389
     Mark Warrell
            OFFICE:  502-564-6716
               FAX:  502-564-2705
            mark.warrell@mail.state.ky. us
Kentucky Division of Waste Management
  14 Reilly Road
  Frankfort KY  40601
     Steve Webb
            OFFICE:  405-702-8195
               FAX:  405-702-8101
            steve.-webb@deqmail.state.ok. us
Oklahoma Dept. of Environmental Quality
  707 N. Robinson Ave.
  Oklahoma City OK  73102
As of: Thursday, September 30, 1999

 Environmental Problem Solving with Geographic Information Systems Conference
 September 22-24,1999
     Kevin Wehrly
List of Attendees
          Cincinnati, Ohio
            OFFICE:  513-569-7301
               FAX:  513-569-7609
            wehrly. kevin@epa.gov
       USEPA-NERL MS-642
         26 W. Martin Luther King Drive
         Cincinnati OH  45268
     Tom Weir
            OFFICE:  215-685-9436
               FAX:  215-685-7593
            thomas. weir@phila. gov
       Air Management Services
         321 University Avenue / Spelman Building
         Philadelphia PA
     Joseph P. Wellner
            OFFICE:  513-251-2730
               FAX:  513-251-0200
            •wellnerj@ttnus. com
       Tetra Tech NUS, Inc.
         1930 Radcliff Drive
         Cincinnati OH  45204
     Russell Weniger
            OFFICE:  210-671-4844
               FAX:  210-671-2241
            russell.weniger@lackland. af.mil
       US Air Force
         2240 Walker Avenue
         Lackland AFB TX  78236-5637
     Bill Wheaton
            OFFICE:  919-541-6158
               FAX:  919-541-5929
       Research Triangle Industries
         3040 Cornwallis Road - P.O. Box 12194
         Research Triangle Park NC  27709-2194
     Brad White
            OFFICE:  937-384-4215
               FAX:  937-384-4201
       Roy F. Weston, Inc.
         2566 Kohnle Drive
         Miamisburg OH  45342
     Ron White
            OFFICE:  513-648-5920
               FAX:  513-648-4029
       Fluor Daniel Fernald
         P.O. Box 538704
         Cincinnati OH  45253-8704
     Charlotte White-Hull
            OFFICE:  513-627-1197
              FAX:  513-627-1208
            whitehull. ce@pg. com
       Procter & Gamble Company
         11810 East Miami River Road, Room 1A04T BTF
         Ross OH  45061
     D. Whitmire
       Ball State University
         6613 W.Jackson St.
         Muncie IN  47304
     Derik Whitmire
            OFFICE:  765-285-2327
               FAX:  765-285-2606
            jeflinl@gw. bsu. edu
       Department of Natural Resources and Environmental
         Ball State University
         Muncie IN  47306
     Duane Wilding
            OFFICE:  410-974-7276
               FAX:  410-974-7200
            dwild@menv. com
       Maryland Environmental Service
         2011 Commerce Park Drive
         Annapolis MD  21401
As of: Thursday, September 30, 1999

 Environmental Problem Solving with Geographic Information Systems Conference
 September 22-24,1999
     Frances Wildman
            OFFICE: 309-782-7907
              FAX: 309-782-5038
            wildmanf@ria. army.mil
                    List of Attendees
                                     Cincinnati, Ohio
                           Rock Island Arsenal
                             5IORI-SEV, Bldg210
                             Rock Island IL  61299
     Dave Williams
            OFFICE: 270-827-3414
              FAX: 270-827-1117
            williams@kgs.mm. uky. edu
                           Kentucky Geological Survey
                             P.O. Box 653
                             Henderson KY  42419
     Matt Williams
            OFFICE: 765-285-2327
              FAX: 765-285-2606
            jeflinl@gw. bsu. edu
                           Department of Natural Resources and Environmental Management
                             Ball State University
                             Muncie IN  47306
     Steve Williams
Ohio EPA
  P.O. Box1049
  Columbus OH
     Lesley Hay Wilson
            OFFICE: 512-367-2952
              FAX: 512-367-2953
            hay_wilson@mail. utexas. edu
                           University of Texas at Austin
                             10100 Burnet Road, PRC Building 119
                             Austin TX  78758
     James Wolfe
            OFFICE: 304-696-6042
              FAX: 304-696-5454
           jawolfe@marshall. edu
                           Marshall University
                             400 Hal Greer Blvd.
                             Huntington WV  25755-2585
     Tim T. Wright
            OFFICE: 513-732-7499
              FAX: 513-732-7969
                           Clermont County GHD
                             2275 Bauer Road  Suite 300
                             Batavia OH  45103
     Patricia Yaden
            OFFICE: 502-564-6716
              FAX: 502-564-4049
            patty.yaden@mail.state.ky. us
                             14 Reilly Road
                             Frankfort KY  40601
     Thomas K. Yeager
            OFFICE: 717-772-4018
              FAX: 717-772-3249
            yeager. thomas@dep.state.pa. us
                           Pennsylvania DEP-Water Supply Management
                             P.O. Box 8467, 11th Floor, RCSOB
                             Harrisburg PA  17105-8467
     Greg Young
            OFFICE: 513-576-0009
              FAX: 513-576-9756
            youngg@srwenvironmental. com
                           SRW Environmental Services
                             55 West TechneCenter, Suite C
                             Milford OH  45150
     Pao-Chiang Yuan
            OFFICE: 601-968-2466
              FAX:  601-968-2344
            pcyuan@yahoo. com
                           Jackson State University
                             P.O. Box18489
                             Jackson MS   39217
                           As of: Thursday, September 30, 1999

 Environmental Problem Solving with Geographic Information Systems Conference
 September 22-24,1999
     Donald Zelazny
            OFFICE:  703-916-7987
              FAX:  703-916-7984
            dzelazny@plaii. com
List of Attendees
          Cincinnati, Ohio
      Platinum International Inc.
        5350 Shawnee Road, Suite 200
        Alexandria VA  22312
     David Zellmer
            OFFICE:  540-552-8185
              FAX:  540-552-8189
            dzellmerCciiyt. edu
      Virginia Tech
        428 Winston Avenue
        Blacksburg VA  24060
     Guang Zhao, Ph.D.
            OFFICE:  8037344833
            zhaog@columb30. dhec.state.se. us
      Information Technology Section, SC Dept. of Health & Env. Control
        2600 Bull Street                                 PAPER
        Columbia SC 29201
     Peter Max Zimmerman
            OFFICE:  410-887-2188
              FAX:  410-887-3182
      People's Counsel for Baltimore County
        Courthouse 47, 400 Washington Ave.
        Towson MD   21204
As of: Thursday, September 30, 1999

         Environmental Problem Solving
     with Geographic Information Systems

               September 21 - 23, 1999
       This presentation was not made available.

Please contact the presenter with questions and comments.
                     Back to
                   1999 Agenda

                The Watershed Assessment Project: Tools for Regional
                                  Problem Area Identification
                                         Christine Adamus
                   St. Johns River Water Management District, Palatka, Florida
The St. Johns River Water Management District of Flor-
ida recently completed a major water resources plan-
ning effort. As part of this planning effort, the St. Johns
River Water Management District created a geographic
information systems (CIS)  project called the Watershed
Assessment, which included a nonpoint source pollution
load model. This paper introduces the planning project
and the Watershed Assessment, and describes how the
results of the model are  being  used to guide  water
management activities in northeast Florida.


The St. Johns River Water Management District (Dis-
trict), one of five water management districts in Florida,
covers 12,600 square miles  (see  Figure 1).  The St.
Johns River starts at the southern end of the District and
flows north; it enters the Atlantic Ocean east of the city
       District Boundary
       Major Drainage
       Basin Boundaries
Figure 1. St. Johns River Water Management District, Florida.
of Jacksonville. The cities of Orlando, Daytona Beach,
and Jacksonville are partially or entirely within the Dis-
trict boundaries. Ad valorem taxes provide primary fund-
ing for the District.

The  District  boundaries  are  somewhat  irregularly
shaped because Florida water management districts are
organized on hydrologic, not political, boundaries, which
greatly improves the District's  ability  to  manage  the
resources. On the north, the  District shares the St.
Mary's River with the state of Georgia and on the south,
shares the Indian River Lagoon with another water man-
agement district. Most of the water bodies the District
manages, however, have drainage basins that are en-
tirely contained within the District's boundaries.

Water management districts in  Florida have amassed
extensive CIS libraries, which they share with local and
statewide agencies. These libraries include basic data
layers such as detailed land use, soils, and drainage
basins.  Districts also  coordinate data  collection and
management to ensure data compatibility.

District Water Management Plan

All activities and  programs of the water management
districts  are related to one or  more of the following
responsibilities: water supply,  flood protection, water
quality management, and  natural systems management.

Each water management district recently  completed  a
district water management plan (Plan). The main pur-
pose of these Plans is to provide long-range guidance
for the resolution  of water management  issues. The
Florida Department of Environmental Protection will use
these five Plans as the basis for a state water manage-
ment  plan. Each water management district used the
same format, which comprised the following components:

• Resource assessment: What  are the problems and
  issues  related to each of the four responsibilities
  listed above?

• Options evaluation:  What options are available for
  addressing the problems?

• Water  management policies: What existing District
  policies influence the decisions that must be made?

• Implementation  strategy: What is the best plan for
  addressing the problems?

The Watershed Assessment Project

The District created the Watershed Assessment project
as part of is resource assessment. This  CIS project
examines the entire District to identify problems related
to flood protection, ecosystems protection, and surface
water quality.

The flood protection component is the only part of the
Watershed Assessment that is  not complete. It will in-
volve  simple overlays of floodplain boundaries with ex-
isting  and future land  use. Floodplain boundaries  are
defined as  Federal  Emergency Management Agency
(FEMA) flood insurance rate map 100-year flood hazard
areas. In many areas, these designations are not very
accurate, yet we  decided to  proceed with their use
because  they  are the best available  information for
many parts of the District. In areas where little hydrologic
information is available and where the District has not
conducted any  related studies, the FEMA  data are a
helpful starting point. This echoes a theme of the Water-
shed Assessment project: the assessment  is primarily
intended  to fill in gaps where we have  not performed
previous resource assessments, not to supplant existing

The ecosystems protection component of  the Water-
shed Assessment is based heavily on a project identify-
ing priority habitat in Florida, conducted by the Florida
Game and Fresh Water Fish Commission (1). It is similar
to gap analyses that the U.S. Fish and Wildlife Service
currently is conducting in many parts of the country. For
the Watershed Assessment, we modified the data some-
what and examined ways to protect the habitat in coop-
eration with local agencies.

The surface water quality component of the Watershed
Assessment has two main parts. The first uses water
quality data from stations that have been spatially refer-
enced so that we can map them and combine the infor-
mation with other information, such as the second part
of the water quality component. This second part is a
nonpoint source pollution  load  model,  which is dis-
cussed in more detail below.

The Pollution Load Screening Model

The nonpoint source pollution load model is the Pollu-
tion Load Screening Model (PLSM), a commonly used
screening tool in Florida.  It is an empirical  model that
estimates annual loads to  surface waters  from storm-
water runoff. Our goal in  designing this model was to
identify pollution load "problem areas" for examination in
the Plan.

In these types of models, annual pollutant loads are a
function of runoff volume and mean pollutant concentra-
tions commonly found in  runoff. Runoff volume varies
with soil and  land  use, while  pollutant concentrations
vary with land use. For the PLSM, pollutant concentra-
tions were derived from studies conducted solely in Flor-
ida. A report describing the model in detail is available (2).

Usually, this kind of model combines CIS with a spread-
sheet: the CIS supplies important spatial information
that is input into a spreadsheet where the actual calcu-
lations are made. The PLSM is different, however, be-
cause  we  programmed  it entirely within CIS.  The
District's  CIS software is  ARC/INFO, and the  model
employs an ARC/INFO module called GRID, which uses
cell-based processing and has analytical capabilities
(3). All the model calculations  are done in  the CIS
software, resulting in a more flexible model with useful
display capabilities.

Model input consists of grids, or data layers,  with a
relatively small cell size (less than 1/2 acre). We chose
this cell size based on the minimum mapping unit of the
most detailed input data layer (land use) and the need
to retain the major road features. The model has four
input  grids:  land use, soils,  rainfall,  and watershed
boundaries. For any given cell, the model first calculates
potential annual runoff based on the land use, soil, and
rainfall in that cell. It then calculates annual  loads by
applying  land-use-dependent  pollutant concentrations
to the runoff.

For this model:

• Land use is from 1:24,000-scale aerial photography
  flown in 1988 and 1989. The model incorporates 13
  land use categories.

• Soils  are  the  Soil  Conservation  Service  (SCS)
  SSURGO database, which corresponds to the  county
  soil  surveys. The PLSM uses  the hydrologic group
  designation of each soil type.

• Rainfall was taken from a network of long-term  rainfall
  stations  located throughout the District.

• Watersheds  were delineated  by the United  States
  Geological  Survey  (USGS)  on  1:24,000-scale,
  7.5-minute  maps and digitized.

Model output  consists  of a runoff grid and six pollutant
load grids. We  calculated  loads for total phosphorus,
total nitrogen,  suspended  solids,  biochemical oxygen
demand,  lead,  and zinc. We chose these pollutants
because reliable data were available and because they
characterize a broad range of nonpoint pollution-gener-
ating land uses,  from urban to agricultural. The  model

calculates runoff and loads for any point in space, allow-
ing the user to see the spatial distribution of loads. An
example of a total phosphorus load grid for one sub-
basin in  the Jacksonville, Florida vicinity is shown in
Figure 2.
The grids themselves  provide a detailed view of model
output. Model results can also be summarized by water-
shed, using the watershed boundary grid, and the infor-
mation can be examined from a basinwide perspective.
We have applied PLSM results in other useful ways at
the District. For example, District staff felt that previous
sediment sampling sites were not appropriately located,
so the  District water quality  network  manager used
model results to locate new sampling sites, focusing on
problem areas as well  as areas where we expect to see
little or no nonpoint impact.
Figure 2.  Distribution of total phosphorus loads, Ortega River
         subbasin (darker areas represent higher loads).

Application of Model Results in the Plan

Because the goal of the model was to identify potential
stormwater runoff problem areas, we needed to simplify,
or categorize, the model results for use in the Plan. We
calculated the per acre watershed load for each pollut-
ant and  defined "potential stormwater runoff problem
areas" as those individual watersheds with the highest
loads for all pollutants. Problem areas for one major
basin in the District, the lower St. Johns River basin, are
depicted  in Figure 3.

We also  ran the  model with future  land  use  data ob-
tained from county comprehensive plans. Because the

                                                          •^^ Subbasins
     h  1
                                                              Potential Stormwater
                                                              Runoff Problem Areas
                                                              Scale in Miles

                                                            0   4   8  12  16
                                                      Figure 3.  Potential stormwater runoff problem areas, lower St.
                                                               Johns River basin.
county maps are guides to future development, and not
predictions of actual development, we exercised caution
when using the results. Problem areas were defined as
those watersheds with projected loads greater than or
equal to existing problem areas. Also, District planners
combined model results with information about individ-
ual counties' regulations and policies to evaluate where
problems are most likely to occur.

Prior to compiling the Plan, the District conducted work-
shops in each county in the District,  in which problem
areas identified by the PLSM were discussed with local
agency staff, officials, and the public. We provided large,
hard copy  maps depicting stormwater runoff problem
areas combined with results of a separate water quality
analysis on county-based maps. These maps proved to
be powerful tools for initiating discussions and gathering

In the Plan, stormwater runoff problem areas were re-
ported for each of the 10 major drainage basins in the
District. The information was also repackaged  in  a
county-based format to create a quick reference for local
agencies. District planners recommended strategies for
addressing problems; these strategies vary as appropri-
ate for each county. Examples include the need to as-
sess  compliance with  existing stormwater  permits,
encourage stormwater reuse  during the stormwater and
consumptive use permitting processes, coordinate with
municipalities that are implementing stormwater man-
agement plans,  encourage and assist significantly af-
fected  municipalities to create stormwater utilities, and
improve monitoring in problem areas that do not have
sufficient water quality data.

In conclusion, the Watershed Assessment CIS project
has proved to be useful not only to the St. Johns River
Water Management District, but also to local govern-
ments. Large  projects such as this could not be com-
pleted in a reasonable time without the use of CIS. Also,
for ARC/INFO users who have been restricted to vector
processing, the cell-based processing available in GRID
is  a powerful modeling tool.


1.  Cox, J., R. Kautz, M. MacLaughlin, and T.  Gilbert. 1994. Closing
   the  gaps in  Florida's wildlife habitat conservation system. Florida
   Game and Fresh Water Fish Commission, Office of Environmental
   Services, Tallahassee, FL.
2.  Adamus, C.L., and M.J. Bergman. 1993. Development of a non-
   point source pollution load screening model. Technical Memoran-
   dum No. 1. Department of Surface  Water Programs, St. Johns
   River Water Management District, Palatka,  FL.
3.  ESRI. 1992. Cell-based modeling with GRID. Redlands, CA: En-
   vironmental  Systems Research Institute, Inc.

       Update of GIS land use attributes from land surface texture
                       information using SIR-C images

                                 Francisco J. Artigas1

The Meadowlands in northern New Jersey were used as dumping grounds for decades and
today dense canopies of common reed (Phragmites australis) cover most of the District's open
spaces. Our objective is to evaluate the utility of multi-frequency SAR in updating GIS land use
information by means of prospecting for anomalously high backscatter in open spaces that
could indicate the presence of building or metallic waste concentrations near the surface. We
combined a land use vector coverage with a co-registered SIR-C C-HV image acquired October
10, 1994 to isolate officially designated "open space" parcels. Groups of four or more pixels with
anomalously high backscatter values were converted to a vector coverage and draped over a
very fine spatial resolution color infrared digital orthophoto of the District. We discuss the
implications for the use of operational imaging radar for monitoring land use and management
of open spaces within a dynamic and complex urban environment.

The Hackensack Meadowlands Development Commission (HMDC) overseas the orderly
development of the Meadowlands District (District) which is a 82 square kilometer degraded
urban estuary four kilometers west of New York City in northern New Jersey. The unregulated
use of District lands as disposal sites for solid and industrial waste for more than 150 years has
turned these meadows and wetlands into one of the most environmentally assaulted areas in
the U.S. (Grossman 1992). The HMDC (a New Jersey State Agency) oversees the preservation
and development of more than 4,375 parcels of land including 1,012 hectares of landfill. This
study presents the results of an effort to use Geographical Information Systems (GIS) in
combination with Synthetic Aperture Radar (SAR) images to update the District's current land
use database. This was accomplished by detecting and documenting areas in open spaces that
showed unusual surface texture roughness as a consequence of land use change or
disturbance by disposal of solid waste.
1 Rutgers University, Center for Information Management, Integration and Connectivity. 180 University
Ave., Newark NJ 07102. artigas@cimic.rutgers.edu

Land use research based on SAR images is emerging as a promising new operational
technology for the applied earth sciences. Images of this area from the Shuttle Imaging Radar
(SIR) with a ground resolution of 12.5 m were first made available in 1994. In 1995, Canada
launched Radarsat-1 (ground resolution 8-100 meters) which some consider the starting point
for the true application and commercialization of radar images (Glackin 1997). By the year 2002,
NASA's LightSAR will be collecting radar images of the earth from a space platform on a
continuous basis.  In just a few years it is expected that there will be an abundance of high-
resolution radar imagery which have some notable advantages over traditional spectral

The main advantage of radar over spectral remote sensing sensors (e.g. Landsat, Spot, SPIN-
2) is that it can capture images of the earth's  surface under any weather conditions, day and
night.  Radar can also penetrate vegetation canopies and at certain wavelengths penetrate the
first centimeters of the soil (Xia et al 1997). The intensity of backscatter from radar microwave
pulses is highly correlated to surface texture.  A useful rule-of-thumb in analyzing radar images
is that the brighter the  backscatter on the image, the rougher the surface being imaged
(Freeman 1996).

Since SIR-C images became available there  has been great interest by scientists to document
radar backscatter  from diverse earth surfaces (Alpers and Holt 1995, Beaudoin et al 1994,
Cordey et al 1996, Freeman and van de Broek 1995). The areas of interest have predominantly
been large (10 to 5000 square kilometers), and focused on relatively homogeneous surfaces
(e.g. deserts, boreal forest, crops, ocean etc.). There have been a less number of studies
reported for urban areas (Taket et.  al, 1991; Xia  et al. 1997) and no reported studies that look at
spots of less than  0.1 hectares (0.25 acres) within a complex urban matrix.

Our research used an  image from an October 1994 Space Shuttle flight (SIR-C) that captured
the New York City and North New Jersey Metropolitan areas. As far as radar scattering is
concerned, such an urban scene is a target of considerable complexity. It provides a variety of
surfaces with sharp edges that go from scales of several hundred feet (buildings) to just a few
centimeters (building surfaces and fields with rubble).

Our specific objective was to evaluate the use of radar images in locating structures and debris
fields in the District and lay out an effective methodology for updating  GIS land use  attributes

based on landscape surface texture information. We discuss the implications of this technology
for monitoring land use and management of open spaces within a dynamic and complex urban

The HMDC maintains a GIS system to manage zoning, land use and block and lot information
of District properties (parcels) for the purpose of land use management and planning. A geo-
referenced parcel information coverage (Figure 1A.) was used to identify open space areas. The
areas selected included: landfill, parkland, riparian, vacant and open water. A binary vector
polygon coverage was created where open area parcel id's acquired the value of one while the
rest of District parcels acquired a value of zero. The binary coverage was converted to a raster
image (pixel size 12.5 meters on each side) where pixels representing open space areas (black)
have a value of one and all other pixels have a value of zero (Figure 1B.).

A set of images for the District were documented and manipulated with raster software. The
SIR-C sensor acquired images at three microwave wavelengths: L-band ~ 24 cm; C-band ~ 6
cm and X-band ~ 3 cm resolution (Freeman et al 1995). However, only L and C bands were
available for the study. Pixel size is 12.5 meters on each side. Incidence angle was 64 degrees
with illumination from the southwest. SAR transmits pulses of microwaves in either horizontal
(H) or vertical (V) polarization and receives in either H or V. The polarization available for each
band (L and C) were HH (horizontally transmitted  and horizontally received) and HV. A third
image for each band labeled total power (TP) is the combination of all polarizations in one
image. Images used in this study were C-band HH, C-band HV, C-band TP, L-band HH, L-band
HV and L-band TP (figure 3). The radar backscatter value for each pixel is measured in decibels
(dB). In our case values ranged from -40dB, very smooth surfaces in  light gray indicating open
water, to +5dB very rough surfaces in dark  gray indicating building structures (Figure 3).

Determining the accuracy of each radar pixel on the ground  is important since we used the x, y
coordinates of radar pixels to navigate to spots on the ground and verify anomalous returns.
The original radar images were geo-referenced to the New Jersey State Plane Coordinate
System  by using  11 control points from a geo-referenced GIS vector coverage of District parcels
(Figure 1C). The  average  RMS error of the re-sampling procedure was 1.3 m. However, the

     Figure 1. A.- GIS parcel vector coverage of the District. B.- Binary raster image of
     the District, black areas correspond to open spaces. C.- Geo-referenced radar
     image with an overlay of District parcels. D.- Image containing radar pixel values for
     open space areas.
parcel vector coverage also had digitizing and tolerance value errors. The original parcel map
was digitized from a 1": 200' scale map of the district. The resolution of the digitizing table was
0.0127 m (0.005 inches). The digitizing error on the ground was in the order of 30 cm. The
tolerance level for the coverage (minimum distance before vector lines snap) was originally set
to 4.2 meters. Therefore, the estimated error of parcel limits is 4.5 m. When the RMS error is

included, then the total error on the ground of radar pixels for this study is at least 6 meters or
half a pixel.

Once the images were geo-registered, the open area binary layer (Figure 1B) multiplied them
each. As a result,  all pixels not corresponding to open spaces were made zero. All other pixels
maintained their dB values (Figure 1D). To avoid working with negative numbers, pixel values
(dB) were re-scaled to fall within the range of 0-255. Pixel frequency distributions for all six radar
images were graphed (Figure 2.). Backscatter distributions of images were inspected for
speckle effects and contrast.

In order to select an image for field verification, two main criteria were used: 1- The image
should clearly discriminate between water and different vegetation types, and 2- The image
should have a minimum of speckle or background noise. The speckle or background noise is
usually associated with strong reflecting surfaces from structures in the vicinity of open

Backscatter  values from known power line towers in the middle of open spaces were used to
select a threshold pixel value for an anomalous return. An anomalous return in an open space
area with vegetation would be brighter than normal and most likely due to a corner reflector (i.e.
two surfaces in 90-degree angle). Known power line tower locations in the middle of open
spaces provided good examples of corner reflectors within a vegetation patch. High energy
double-bounce scatter mechanisms prevail from rectangular metallic surfaces compared to the
less energetic volumetric scatter mechanism that prevails  from a canopy of vegetation. A similar
reflectance mechanism (double bounce) would be expected from rubble (e.g.  concrete slabs or
metal artifacts) such as old cars and drums exposed or hidden under vegetation.

All pixels exceeding the threshold value were extracted and clusters of more than four  pixels
plotted. Clusters of at least four were selected to avoid selecting single pixels with unusual
brightness values located exactly on parcel boundaries that in reality do not correspond to true
open spaces. These clusters were converted to vector polygons and draped over aim
resolution geo-referenced digital orthophoto  (USGS 1995). The draped orthophoto images were
visually inspected for obvious structures on the ground that may have produced the bright
backscatter  return (Figure 6 and 7). Parcels classified as open space that contained structures
were re-classified in the GIS database. Spots were selected for ground verification when visual

inspection of the draped orthophoto images failed to reveal structures on the ground that could
explain the anomalous return. Moreover, ease of ground access and clusters size were
important considerations in selecting specific spots for field verification. Radar image
coordinates were used to navigate to the sites. Known reference points in the field were
selected and azimuth and distance from these points to the center of the anomalous spots were
used to navigate in the field. Sites were finally located in the field by using a 30-meter
measuring tape and a compass. Historical aerial  photography from 1966 was used to confirm
and explain some of the disturbances detected by radar.

Although several theoretical models have been developed to describe how ground objects
reflect radar energy (Taket 1991, Evans et al 1988),  most knowledge comes from practical
observations. There is no strong set of rules indicating what bands or polarizations work best on
different surfaces and under specific sensor conditions.  Given the time and resources available,
it was important to find one best image out of the six available to field verify.

An inspection of the pixel brightness frequency distribution values for all images (Figure 2)
shows that the distributions tend to be bimodal except for HH polarization. Cross-polarization
(HV) images function better to capture the difference between water and vegetation surface by
clearly separating water and vegetation into two distinct peaks. Of all images, C-band HV has
the best contrast (a greater valley between peaks) as well as a sharper break between the end
of the vegetation values (tall peak) and the tail of the extreme bright values.

The amount of speckle was another criteria for image selection. Speckle effects, or "noise", are
common in radar images capturing complex surfaces. Speckle can be caused by an object with
complex dielectric properties that behaves as  a very strong reflector at a particular alignment
between itself and the spacecraft. Structures such as metal bridges, elevated highways, and
power line towers behave as strong reflectors in urban areas. These structures influence the
value of neighboring pixels making them appear much brighter than the surfaces they actually
represent. In our case speckle effect is clear in C-band HH, C-band TP and L-Band HH (Figure
2B, 2C and 2E.). In all these cases there is an unusual high frequency of bright returns at the
high end of the distribution. Unfortunately, speckle effects affect open areas near strong
reflector objects. In our case this is very clear for C-band HH (Figure 2B). Speckle effect is also
clear in L-band TP. However, in this situation background noise manifests itself by cross and

                                                                            C-band IP
            28.40 SI .80 77./D lijifu li! LIL> 15i* 17880 JMi'D J2*6if 255.00  • I1 - •*!• HI  -  ' Jl   I! *  U -:-. £*« -''--'  I.OD 26.40 fl?D 77 iu IDiM liKn> 15j* '78.80 384.10 229.60 255,00
                                                                           L-band IP
          1.00 26.<0 51.80 77.20 102(0 12:0; 15242 ^I'SO 2^10 22V S5 255.00  1.00 26,-tO 51.80 77.20 10262 '2SOO '5':4Ii 173ai 2042: 22060 2
                                                                JB.-tt 51.30 77.20 1026i' -2ftLi[l '52: W 172 KO 204.20 22360 25500
       Figure 2. - Pixel brightness frequency distributions for C and L bands HH and HV
       polarizations. HV and TP polarizations for both bands, clearly separate open water
       and vegetation surface texture (bi-modal distribution). C-band HV shows the
       greatest contrast between surface textures (valley between peaks) and a sharp
       break between the vegetation peak and the upper tail of the brightness distribution.
star like patterns that have their origin at points with rough features and high dielectric constants
(Figure 3).
The final criteria used in selecting an image were based on radar wavelength. Shorter
wavelength (C) have better resolution than longer wavelength (L), thus improving the overall
spatial resolution and the ability to delimit detail and boundaries (Xia and Henderson 1997).
Based on these observations and since our objective is to determine surface texture differences
within small areas, of the two best images available (C-band HV and L-Band HV) we choose the
smaller wavelength image C-band HV.

 C-band HV
L-band HV
                                2 Km
 C-band TP
L-band TP
Figure 3. - Images of the Hackensack Medowlands (C and L bands, HH and HV
polarizations. Flat smooth surfaces in gray represent mainly open water bodies. The lower
right corner of the images shows the Hudson river and a part of Manhattan. West of the
Hudson is the Hackensack and Passaic rivers draining into Newark Bay. Other white-gray
surfaces represent wetlands. Dark gray and black areas represent developed urban areas.

                    Figure 4. -Pixel clusters from open spaces that
                    were classified as anomalous returns
                    (brightness value greater than 176 out of 255).
"vegetation pixels" from brighter backscatter returns. For this study, a value of 176 was selected
as the threshold value for an anomalous return. Figure 4 shows all pixels from open spaces that
exceeded the threshold value and were classified as anomalous returns for the District.
There were a total of 128 clusters of anomalous returns. These clusters varied from four to 53
pixels per cluster. The greatest number of clusters (61) were made out of only four pixels
(Figure 5). The most common clusters (82%) were made out of 10 pixels or less. The most
extensive cluster was made out of 53 pixels representing an area of 0.8 hectares.

                                    Pixel Cluster Size
         Figure 5. - Frequency distributions for clusters of pixels that exceeded the
         brightness threshold value of 176. More than 80% of the clusters were 10 pixels
         or less.
Figure 6 shows how a several clusters of radar anomalous returns overlay at least three distinct
parcels of land at the southwest corner of the District. These parcels were classified as vacant
in the GIS. Parcels range from 0.5 to 1.5 hectares. By zooming in the ortho-photo one can
actually see that these parcels contain structures (trailers and construction equipment) and are
not vacant.

Similarly, Figure 7, shows another anomalous return from  a border of a vacant lot (arrow) that
turned out to be a 0.1  hectares (0.25 acre), seven meter high bulldozed mound of earth mixed
with metal rubble, tires, old batteries, cloth, etc. It is impossible to tell by zooming in the ortho-
photo that this structure actually exists since it blends with the colors of the surrounding area. A
list of similar sites and their coordinates were detected and documented.

         Figure 6. - Anomalous radar clusters are shown as white polygons draped
         over a 1-meter resolution orthophoto. White lines represent parcel
         boundaries. Red circles mark parcels classified as vacant by the GIS yet
         they contain structures (metallic rigs and machinery) which offer corner
         reflector surfaces detected by radar.
The most interesting anomalous returns originated from what appeared to be a field with a
dense homogeneous canopy of Phragmatis australis. (Figure 7 circled area). After navigating to
the anomalous spots in the field, the only consistent difference in surface texture was a much
lower and sparse canopy of the common reed. These "open" spots seemed to be correlated to
differences in soil compaction. The effect is a sharp boundary between the tall reed  canopy and
the sparsely vegetated areas inside the selected spots. In some of these spots the soil was
littered with waste such as leather scraps, plastic wallpaper, roofing material, tiles and even
junked cars. Soil compaction, sharp canopy boundaries and scattered rubble provided the
surface features that favored double bounce reflections as opposed to volumetric scatter from
the immediate surrounding areas.

      Figure 7. - Radar anomalous returns (white polygons) for the South Hackensack
      part of the District. The white arrow shows a 40X40-meter area where there is a
      bulldozed mound of rubble. This mound, hidden under vegetation would be
      difficult to separate from the colors of the surrounding vegetation by using
      traditional spectral sensors. The anomalous returns included in the white circle
      indicate an area of what seems to be a homogenous canopy of the common
      reed. However, field inspection of these spots revealed differences in surface
      texture due to past disturbances
Recently declassified intelligence satellite photography from 1966 (3-m resolution) of the same

area (Figure 8) revealed a construction staging area exactly where the anomalous returns from

radar originate. What is today a "dense" stand of Phragmites australis, was in the late 1960's a

staging area for the construction of the New Jersey Turnpike. Roads and trails can be identified

in this area from the 1966 satellite photograph. This explains the differences in soil compaction

and the presence of debris in what seems today to be the middle of a common reed field.

        Figure 7. - Declassified spy satellite photograph from 1970 showing the
        same area of Figure 6. The brighter area inside the white circle shows a
        surface disturbance created by heavy machinery and rubble disposal that
        created differences in surface texture hidden under vegetation and
        detected by radar 24 years later
Discussion and Conclusions
The use of SAR images to detect disturbed or incorrectly classified land uses based on surface
texture proved to be a reliable and effective approach. Detection and navigation to the
previously unrecognized disturbed areas was done with less than 10 meters of error on the
ground. This in itself demonstrates the practicality and effectiveness of the method. One critical
part of the study was selecting the appropriate band and polarization to identify sites on the
ground since resource constraints prohibited the evaluation from both bands and all
polarizations in the field. Our criteria for image selection was in agreement with most research
to date which seems to prefer cross-polarized (HV) images for urban and land use cover
mapping over co-polarization (Bryan 1974, Henderson 1985). In our case the C-band HV

distribution better separates between "vegetation pixels" and a class of brighter pixel values
than L-band HV (Figure 2A and 2D). Contrary to findings reported in the literature (Xia 1997),
surface features appear smoother at shorter wavelengths (C-band HV) than at larger
wavelengths (L-band HV). Selecting C-band over L- band may have resulted in a loss of
information associated with surface texture within vegetation patches.  However, overall, we
found that shorter wavelength best separated known targets from the surrounding vegetation.
Selecting a threshold value, in other words, determining which pixels were brighter than normal
for a given surface was a critical decision of the study. There are many models for scatter
mechanisms (Evans 1988, Jacob 1993,) and they all tend to agree that the double bounce
mechanism carries the greatest amount of energy back to the receiver creating brighter pixels.
In this case the microwave bounces off one surface, hits another surface and returns back to
the receiver (corner reflector effect). We considered this mechanism to prevail in the case of a
target such as an electrical power tower, or slabs of concrete and metal rubble hidden under
vegetation or ground that has been compacted in open spaces surrounded by hard stems. Our
data suggests that volumetric scatter would be the dominant scatter mechanism from a canopy
of Phragmatis  australis. Incident radiation would bounce off stems, branches and leaves
scattering in all directions and less radiation would return to the receiver creating less bright
returns from these surfaces.

Incidence angle and "look" angle for each image vary according to the position of the sensor in
space in relation  to its ground target. In this case, our image had a fixed 64 degrees incidence
angle illuminating from the southwest. Different fly-over passes of the sensor will create images
with different incidence angles and therefore different backscatter patterns for the same area.
Images will clearly have to be selected according to the most favorable incidence angles and
illumination direction. Topography will also influence the backscatter pattern, as shadows from
mountains will hide surfaces that can not be imaged by radar. These factors emerged as
important limitations to surface texture detection using radar. In our specific case, the relatively
flat topography of the District would favor the use of radar.

Radar was able to identify specific parcels that were incorrectly classified in the GIS database.
Some parcels  contained  reflector surfaces from structures such as trailers, metallic rigs and
machinery which were easily recognizable from the ortho-photos.

In other cases (Figure 7 arrow), radar was able to detect rubble hidden under vegetation. The
40X40-meter mound could not have been separated from the surrounding vegetation by using
any other type of spectral sensor i.e. aerial photo or satellite image. In this case, the orientation
height (7 m) and the flat surfaces of metallic rubble associated with the mound  made it possible
for radar to detect and separate this particular structure from the surrounding vegetation.

Finally, radar was able to detect past disturbances within certain areas of normal looking reed
fields. In these sites, soil compaction by heavy machinery and dumping of construction rubble
and waste from  almost 40 years ago altered the natural hydrology and influenced community
plant development. Again, with radar it was possible to separate surface feature roughness and
detect these disturbances.

One mission of the HMDC is to balance development with preservation. This process involves
making informed decisions regarding what areas are to be developed and what areas are to be
set aside for preservation. To make these decisions, agency executives and technicians need to
be well aware of the location, quality and characteristics of each site. Currently, the land use
management team updates land use of District parcels by field recognizance every four to five
years. Under the current method, there may be enough time in between updates for an
undetected illegal land use practice to create considerable damage.

Our methodology proved that based on surface texture detection it is possible to  monitor and
update parcel information from areas as small as 0.1 hectare. Once periodic radar images
become available, standard backscatter returns for District parcels would be documented and
become a template for the current surface texture pattern. New images (with same incidence
angle and illumination direction) would be checked against the template. Changes in surface
texture due to constructions of structures, disturbance of soil surface, change in vegetation
cover or waste dumping would create a different backscatter pattern from the template that
would translate  into the identification of parcels where changes took place. The challenge ahead
is to integrate these systems and methodologies into an expert system with the ability to detect
change in surface texture, update the GIS database automatically, and continuously repeat this
process as  new images become available. This approach should be able to keep up with the
dynamic and complex land use changes in the District and help managers make  informed
decisions based on timely land use information.

This work was preformed with funding from a grant from the Hackensack Meadowlands
Development Commission and sponsored by the Rutgers University NASA Regional Application
Center and the Center for Information Management Integration and Connectivity (CIMIC). I
would like to thank Dr. Geoffrey Henebry for providing the images and for his comments on
radar backscatter mechanism. Finally, I would like to thank Matthew Ceberio and Alon Frumer
for their help with field verification.

Alpers, W., and B. Holt, 1995. Imaging of ocean features by SIR-C/X-SAR: An overview,
       Proceedings of the International Geoscience and Remote Sensing Symposium
       (IGARSS'95), July 10-14, Florence Italy, pp. 1588-1590.

Beaudoin, A., T. Le loan, S. Gozc. E. Nezry, A. Lopes, E. Mougin, C. C. Hsu, J. A. Kong and R.
       T. Shin. 1994.  Retrieval of forest biomass from SAR data. Int. J.  Rem. Sens. vol. 15, pp

Bryan M.L. 1974. Extraction of urban land cover data from multiplexed synthetic aperture radar
       imagery, in Proc. Ninth Int. Symp. Rem. Sens, of Environment. Environmental Research
       Institute of Michigan, Ann Arbor. Ml, pp. 271-288.

Cordey, R., J. R. Baker, S. Quegan, G. M. Foody, N.J. Veck, and A.  Wielogorska 1996. A study
       of the potential of  multi-component SAR imagery for agricultural  and forest studies, in
       Science results from the Spaceborne Imaging Radar C/X-Band Synthetic Aperture
       Radar (SIR-C/X-SAR): Progress Report, Jet Propulsion Laboratory  Publication No. 96-7,
       pp. 112-14.

Evans L. Diane, Tom G. Farr, Jakob J. Van Zyl and Howard A. Zebeker 1988. Radar
       Polarimetry: Analysis tools and applications. IEEE Transactions  on Geoscience and
       Remote Sensing,  Vol.26. No.  6.

Freeman, A., M. Alves, B. Chapman, J. Cruz, Y. Kim, S. Schaffer, J. Sun, E. Turner, and K.
       Sarabandi 1995. SIR-C data quality and calibration results, IEEE Transactions on
       Geoscience and Remote Sensing, Vol. 33, no. 4, pp. 848-857.

Freeman, A., and B. van den Broek, 1995. Mapping vegetation types using SIR-C data,
       Proceedings of the International Geoscience and Remote Sensing Symposium
       (IGARSS'95), July 10-14, Florence Italy, pp. 921-923.

Freeman A., 1996. What is imaging radar?. Spaceborne imaging Radar-C SIR-CED, Jet
       Propulsion Laboratory. Pasadena California.

Glackin L. David. 1997. Earth observation in transition: An international overview, Acta
       Astronomica Vol. 41, Nos. 4-10. pp. 413-420.

Grossman and Associates, Inc. 1992. Stage 1-A Archeological and Historical Sensitivity
       evaluation of the Hackensack Meadowlands. Grossman and Associates Inc. Technical
       Report, Lyndhurst. NJ.

Henderson  F. M. 1984. Confusion errors among urban land-cover types on SAR imagery.
       International Journal Remote Sensing, vol.6, no 10. pp. 1607-1622.

Jakob J. van Zyle, 1993. The effect of topography on radar scattering from vegetated areas.
       IEEE Trans, on Geoscience and Remote Sensing,  Vol. 31, No 1.

Richards J.A. 1986. Remote Sensing Image Analysis. Berling, Springer.

Taket N.D.,  S.M. Howarth, and R.E. Burge 1991. A model  for the imaging of urban areas by
       synthetic aperture radar. IEEE Transactions on Geoscience and Remote Sensing, Vol.
       29 No. 3.

USGS 1995. Digital Orthophoto Quadrangles . US Geological Survey. Reston, VA.

Xia Zong-Guo and F.M Henderson 1997. Understanding the relationship between radar
       response patterns and the Bio and Geophysical parameters of urban areas. IEEE Trans.
       on Geoscience and Remote Sensing, Vol.  35,  No.  1. pp 93-102.

                       GIS in the Confirmation Process
                     Dr. Raymond E. Bailey and Mr. Madhukar Mohan
The ultimate goal of remediation of the Department of Energy (DOE) Weldon Spring mixed
waste site is to release the site for unrestricted use to the extent possible. This dictates that an
accurate assessment of post cleanup activities is performed to confirm that contaminated soils
have been successfully treated or removed. The assessment begins by developing an accurate
3-dimensional picture of the  spatial distribution of contamination prior to the start of cleanup
activities. A geographic information system (GIS) was selected over traditional  manual methods
to map the initial spatial distribution of contamination  (pre-cleanup levels) and to manage the
confirmation database. The Weldon Spring site, consisting of approximately 217 acres, was
divided into eleven remedial  units (RU). Each RU was divided  into approximately 2,000 m2
areas known as confirmation units (CU). Upon completion of remedial activities within each RU,
Environmental Safety and Health technicians conducted a walkover survey with a 2-inch by 2-
inch sodium iodide (Nal) scintillation detector to establish removal of surface contamination to
levels at or below background levels of gamma-emitting radioactivity. Following the surface
scan,  soil samples were collected and analyzed for radiological and chemical contaminants of
concern listed in the Record  of Decision. Results of laboratory radiological  and chemical
analyses were used to populate the attribute portion of the GIS database. Geographic locations
in the database  were obtained from surveys of the sample locations. This established an
accurate location for each sample,  provided confirmation that the as-built excavation had
achieved the designed depths, identified areas that had achieved cleanup levels, and, if above
target cleanup levels,  identified locations of contamination requiring additional remediation.

The Department of Energy (DOE) Weldon Spring Site Remedial Action Project (WSSRAP) is a
mixed waste site located in St. Charles County, Missouri, approximately 48 km (30 mi.) west of
St. Louis (Figure 1).

                                                LOCATION OF THE
                                           WELDON SPRING CHEMICAL PLANT
                                                   AND THE
                                              WELDON SPRING QUARRY
                                                  FIGURE 1
                                                     GUI  |"" 7i7f-"i~
                             Figure 1

The WSSRAP consists of a 217-acre chemical plant area initially used by the U.S. Department
of the Army during the 1940s to produce the explosives trinitrotoluene (TNT) and dinitrotoluene.
After World War II, the structures were razed, decontaminated, and the site was re-graded. The
U. S. Atomic Energy Commission (predecessor to the DOE) built a chemical plant upon the
former Army site to process uranium and thorium ore concentrates. Production operations
proceeded throughout the 1950s and 1960s, resulting in the disposal of radioactively and
chemically contaminated waste on site.

Contaminated areas at the Weldon Spring Site included material from 40 building foundations,
four raffinate pits, two ponds, and two former dump areas (Figure 2).
The contaminants of concern requiring treatment are radioactive contaminants (primarily
radionuclides of the natural uranium and thorium-232 decay series) and chemical contaminants
(including naturally occurring metals and inorganic anions, as well as organic compounds such
as polychlorinated biphenyls and nitro-aromatic compounds). The remediation alternative
selected consists of removing material from contaminated areas, treatment as appropriate by
chemical stabilization and/or solidification, and disposal in an engineered disposal facility

                             Figure 2
constructed on site. The ultimate goal of site remediation activities is the release of the site for
unrestricted use to the extent possible. To achieve this goal, requires an accurate assessment
of post-cleanup activities (the confirmation process).

The site geographic information system, utilizing Arclnfo software operating on a Sun Unix
workstation, was selected to create an accurate 3-dimensional picture of the spatial distribution
of contamination prior to the initiation of remediation activities and to manage the confirmation
database. The geographic database was populated with Arclnfo coverages of the Weldon
Spring Site topography in 1954, aerial surveys of WSSRAP in 1993 and 1998, coordinates of
anomalies identified during geophysical surveys, and the location of characterization drilling and
sampling points. A complete picture of the spatial distribution of the pre-cleanup contamination
was provided by linking the analytical laboratory results of the characterization samples with
their geographic locations.
The confirmation database was created by dividing the 217-acres site into eleven remedial units
(RU). Each RU was further divided into approximately 2,000 m2 (0.5 acres) areas known as
confirmation units (CU) (Figure 3).

  763,000 E
754,000 E
755,000 E
756,000 E
 200   100   0
  P"H H r- -

                                                 Figure 3
The CU is the area for which a decision is made as to whether cleanup standards have been
attained. The size of the CU was selected to provide an area of approximately the same size as
that used in the risk assessment for a future residential lot. This size also provided manageable
areas capable of supporting the construction schedule when an excavated area needed to
remain open pending the confirmation that cleanup standards had been attained.
Upon completion of remedial activities within each RU, Environmental Safety and Health
technicians conducted a walkover survey with a 2-inch by 2-inch sodium iodide (Nal)
scintillation detector to establish removal of surface contamination to levels at or below
background levels of gamma-emitting radioactivity. Areas showing elevated readings greater
than 1.5 times background were designated as "hot spots", and additional material was

removed. After obtaining a surface scan at or below background readings, a 10 meter by 10
meter grid was surveyed and soil samples were collected and analyzed for radiological and
chemical contaminants of concern listed in the Record of Decision. Results of laboratory
radiological and chemical analyses were used to populate the attribute portion of the GIS
confirmation database. Geographic locations in the database were obtained from  surveys of the
sample locations and compared with design excavation limits. This established an accurate
location for each sample, provided confirmation that the as-built excavation had achieved the
designed depths, identified areas that had achieved cleanup levels, and, if above  target cleanup
levels, identified locations of contamination requiring additional remediation.

 Vulnerability Assessment of Missouri Drinking Water to Chemical Contamination
               Christopher J. Barnett, Steven J. Vance, and Christopher L. Fulcher
     Center for Agricultural, Resource, and Environmental Systems, University of Missouri,
                                       Columbus, Missouri

In 1991, the Missouri Department of Natural Resources
(MDNR) implemented the Vulnerability Assessment of
Missouri Drinking Water to Chemical Contamination pro-
ject. MDNR's Public Drinking Water Program (PDWP)
contracted with the Center for Agricultural, Resource,
and  Environmental Systems (CARES) to conduct this
assessment. They designed the project to determine
which,  if any, public water supplies are threatened by
chemicals being tested under the Safe Drinking Water Act.

Under  Phase II of the Safe Drinking Water Act, the
United  States Environmental Protection Agency (EPA)
required that all public drinking water systems be rou-
tinely monitored for 79 contaminants beginning January
1,1993. If a selected chemical parameter is not detected
in an area that would  affect a  water supply (where
"detected"  is defined as used, stored,  manufactured,
disposed of, or transported regardless of amount), then
the water supply need not be tested for that chemical.
Instead, that system would  be granted a use waiver,
meaning that the state would not  test for that chemical.
EPA grants use waivers for 43 of the 79 contaminants.
Use waivers can result in considerable cost savings.

Because use waivers are granted based on the spatial
relationship between drinking water sources and con-
taminant sources, accurate positional data needed to be
collected for those items. A geographic information sys-
tem  (CIS) was used to store and analyze this informa-
tion in a spatial context.

Water Sources

Water sources, as defined for this study, are the points
where water is drawn from a river, lake, or aquifer for
use in a public water supply. Our efforts focused primar-
ily on the development of the water source layers for the
CIS. These layers, containing wellheads, impoundment
intakes, and river intakes, were  created in house  or
obtained from  state and federal  agencies. MDNR
regional office personnel inspected these water source
layers in the  spring of 1993.  Since  these  personnel
routinely inspect Missouri public drinking water supplies,
their knowledge of these locations is  exceptional.
The updated water source information was mapped
on 1:24,000-scale USGS  topographic quadrangles at
the regional offices, then entered into the CIS. MDNR's
PDWP provided available attribute information, which
was associated with these layers. The layers offer the
most accurate and current information available. Only
the community (e.g., cities, subdivisions,  mobile home
parks)  and  the  nontransient, noncommunity  (e.g.,
schools, large businesses) water supply systems were
considered for water source mapping. This study did not
consider private wells.

The information is stored in  the CIS in the  form of
geographic data sets or layers.  The wellhead layer con-
tains 2,327  public wells and their attributes  (e.g., well
depth, casing  type). The majority of the wellheads are
located in the Ozarks and Southeast Lowlands. Natu-
rally poor ground-water quality prohibits  a  heavy reli-
ance on ground water for drinking water in other areas
of the state. The surface water impoundment layer con-
tains 105 points representing  the intake locations for
systems that rely on lake water. Additionally, the drain-
age basin and lake area are mapped for these systems.
The majority of the systems that rely on lake water are
located in northern and western Missouri. The final layer
represents the systems that use river water. The major-
ity of the 50 intakes are located on the Mississippi and
Missouri Rivers and on the major streams in the Grand
and Osage River basins.

Contaminant Sources

Contaminant sources, as defined  for this study, are the
points or areas where  existing databases indicate the
presence of a chemical contaminant. Incorporation of
contaminant data into the  CIS proved to be the  most
difficult task. These data usually contained very precise

information about what contaminants were found at a
site and  who was responsible, but the quality of the
locational information was often poor.

Ninety-three state and federal databases were reviewed
for contaminant information before performing the final
use waiver analysis. The contaminant information was
broken into two separate types, contaminant sites and
pesticide dealerships. The contaminant sites were loca-
tions at which certain chemicals were known to exist.
The pesticide dealerships were dealerships licensed to
distribute restricted  use pesticides.  Information  about
contaminant sites was extracted from the  databases
and entered into  Microsoft Excel, a spreadsheet pro-
gram. The  small amount of data with coordinate (lati-
tude/longitude)   or   map   information   was  readily
converted to the CIS. The majority of the contaminant
records, however, contained only address information,
often appearing as a rural  route address or post office
box number.

While the water source locations were being verified,
personnel at the MDNR regional offices reviewed the
contaminant site records. The regional office personnel
were familiar with their respective territories and could
assist CARES personnel in locating the  contaminant
sites. The Missouri Department of Agriculture pesticide
use investigators provided  additional information about
the locations  of  contaminant sites. All  contaminant
source information was also mapped on the  1:24,000-
scale  USGS topographic quadrangles and transferred
to the CIS.

Of more than 2,800 contaminant sites found in these
databases,  88 percent were geographically located and
used in the  study. At this time, the contaminant site layer
contains 2,493 points representing the information col-
lected on the  43 chemical contaminants required  by
MDNR. Each point contains a seven-digit chemical code
indicating the chemical it represents and serving as a
link to the chemical contaminant files. The contaminant
sites tend to be concentrated more in urban areas than
rural areas. Even though this  layer is being continually
updated, the basic distribution of contaminant sites re-
mains the same.

A second contaminant  source layer represents Mis-
souri's licensed pesticide dealers.  This information is
included to indicate potential contamination even though
specific chemicals  at dealership  locations are  not
known. At this time, we have been able to locate 1,344
dealerships out of 1,650. Two types of dealerships are
included in the layer, active dealers and inactive dealers.
Of the active dealerships in 1991, 91 percent were found
and entered into the CIS. Of the inactive dealerships, 79
percent were located.
Spatial Analysis

The final parameters for the use waiver analysis were
developed from EPA and MDNR guidelines and account
for the capabilities of the CIS. These parameters were
designed to present a conservative list of the systems
that needed to  be tested for the possible presence of
studied chemicals. Parameters forthe wellhead analysis
are as follows:

• A 1/4-, 1/2-, and 1-mile radius around each wellhead
  was searched for contaminant sites and pesticide
  dealerships (see Figure 1). Any contaminant sources
  found within  those radii were  reported  to PDWP.
  (PDWP requested that the results of the three radius
  analyses  be  reported,  but the 1/2-mile radius  was
  used to determine the issue  of the use waiver.)

• Any wellheads found within a contaminant area were
  denied a use waiver for that contaminant.

• Each highway and railroad within 500 feet of a well-
  head was recorded. This indicates the threat posed
  by the transport of chemicals near wellheads.

• Additionally, the percentage of the county planted in
  corn, soybeans, wheat, sorghum,  tobacco,  cotton,
  and rice was listed for each well to indicate the threat
  posed by agricultural chemical use within that county.

The parameters for the systems  relying on  lake water
are as follows:

• Any contaminant sources  found  within  a surface
  water impoundment drainage basin caused the asso-
  ciated intake(s) to fail use waiver analysis for those
                                    X = Contaminant
Figure 1.  Use waiver search radius distances.

• Any  area  of  contamination overlapping a drainage
  basin caused the associated intake to fail use waiver
  analysis for that contaminant.

• Transportation corridors passing through a drainage
  basin were noted  to  indicate the threat posed by
  transport of chemicals within the basin.

• The  percentage of the county planted  in the  seven
  crops mentioned above was listed  to indicate agricul-
  tural chemical use within the drainage basin.

Many of the rivers that supply water to systems in  Mis-
souri have their headwaters outside the state. To  fully
evaluate  the potential for contamination within those
drainage basins, we would have to collect data for large
areas outside of the state. For example, the Mississippi
and Missouri River drainage basins cover large portions
of the United States.  Because collecting data for those
areas would be impractical, we have recommended to
MDNR that use waivers not be granted to river supplies.

The following provides details on how the analysis  was
performed. The CIS searches around each wellhead for
each radius and  notes which contaminant sites affect
which wellheads. If a contaminant falls within that ra-
dius, we recommend that the wellhead be monitored. In
this example, the well is affected by one contaminant
within the 1/4-mile radius, two within the 1/2-mile radius,
and four within the 1-mile radius.


The results  of the use waiver analysis indicate which
systems may be affected by the use  of a chemical near
a water source. Several results show the substantial
savings realized from our analysis.  For example, the
analysis showed that only five wells  serving four public
drinking water systems were potentially affected by di-
oxin and should be monitored. By not testing the remain-
ing  systems for dioxin, the  state  can  realize  a
considerable cost savings, as the test for dioxin is the
most expensive test to perform.

The  final  wellhead  system  analysis shows that the
1/2-mile buffer analysis  affected a  total  of 447 well-
heads in 241 systems. That is, a chemical site or pesti-
cide dealership was found within 1/2  mile of 447 public
wellheads. A result form was generated for each of the
1,340 systems in the state listing each well or intake and
the  potential threat   posed  by nearby  contaminant

The cost of testing all wellhead systems for all 43 con-
taminants without issuing use waivers is more than $15
Table 1.  Estimated Cost Savings for Public Drinking Water
                Total Cost
 Mean Cost
per System
Total Cost
No use waiver    $15,533,100      $12,200             $0
With use waiver    $1,813,900       $1,400     $13,719,200

million (see Table 1). According to our analysis, CARES
estimates that only $1.8 million need be spent to monitor
vulnerable wells. Therefore, the state can save more
than $13.5 million in monitoring  costs.

Summary and Recommendations

To date, the investment the state made in the vulnerabil-
ity assessment project has provided many benefits. The
state saved several million  dollars in testing costs and
developed several spatial and nonspatial databases that
will have many uses. In addition, the  project established
a basic  framework for future assessments, which EPA
requires on a regular basis.

The basic data required for use waiver analysis are the
locations of water sources and the locations of potential
contamination sources.  CARES determined  that the
available data did not contain the information necessary
to map these  locations or that the data were of question-
able quality. Many layers required update and correc-
tion. Considerable  effort was  necessary to  improve
existing  locational information for both water source lay-
ers and  chemical contaminant files. Local knowledge of
an area was  heavily relied upon  to determine accurate
locations, particularly contaminant sites. The vast ma-
jority of these sites  contained only the address  as the
geographic reference. An address is not a  coordinate
system; it does not  indicate a fixed location on a map.
Because the  location of any chemical detection site is
of vital importance, state and federal agencies that col-
lect these data  need to record more  complete geo-
graphic  information. Ideally, a global  positioning system
could be employed to generate coordinates. Realisti-
cally,  the recording  of legal descriptions or directions
from an  easily located point would substantially improve
the quality of the current databases.

In many cases, data resided in digital format; however,
due to regulations or lack of agency cooperation, they
could only be distributed in paper format. Reentering
data from paper format into digital format required con-
siderable time and  expense. Interagency cooperation
should be emphasized to reduce  unnecessary data entry.

              Using GIS to Evaluate the Effects of Flood Risk
                        On Residential  Property Values
                 Alena Bartosova, David E. Clark, Vladimir Novotny, Kyra S. Taylor
                        Marquette University, Milwaukee, Wl 53201-1881
1. Introduction
Annually, flooding causes more property damage in the United States than any other type of
natural disaster. One of the consequences of continued urbanization is the tendency for
floodplains to expand, increasing flood risks in the areas around urban streams and rivers.
Hedonic modeling techniques can be used to estimate the relationship between residential
housing prices and flood risks. One weakness of hedonic modeling has been incomplete controls
for locational characteristics influencing a given property. In addition, relatively primitive
assumptions have been employed in modeling flood risk exposures.

We use GIS tools to provide more accurate measures of flood risks, and a more thorough
accounting of the locational features in the neighborhood. This has important policy implications.
Once a complete hedonic model is developed, the reduction in property value attributed to an
increase in flood risks can, under certain circumstances, be interpreted as the household's
willingness to pay for the reduction of flood risk. Willingness to pay estimates can in turn be
used to guide policymakers as they assess community-wide benefits from flood control projects.

2. Hedonic Theory and  Literature
The hedonic price model used in this study has its roots in the works of Lancaster (1966) and
Rosen (1974). It is based on the premise that individuals can choose consumption levels of
local public goods such as environmental quality through their residential location choice. The
model views the price of individual houses as dependent on a bundle of  housing characteristics.
These characteristics include those related to the structure (e.g., lot size, number of bathrooms,
etc.); the neighborhood (e.g., average commute time, median household income, etc.); the
environment (e.g., variables related to flood risk); and fiscal factors (e.g., property tax rates).

There are several underlying assumptions in this model. The model  assumes that the study
area is a single  market for housing services. It also assumes that all buyers and sellers have

perfect information on the alternatives that exist and that the housing market is in equilibrium.
This last assumption means that all households have made their utility maximizing choice in
terms of residential location given the prices of alternatives, all of which just clear the market.
The relationship outlined here can be linear only when repackaging of the house is possible,
and in general, this is not the case. When an individual makes a residential location decision,
they are accepting the entire bundle of housing characteristics. It is not possible to trade a
house with two full baths upstairs for the exact same house with one full bath upstairs and one
downstairs. Thus, the function is nonlinear.

Given the previous assumptions, the market clearing price of the house is treated as  parametric
and can be represented as p(Z), where Z = Zi,z2	zn is a vector of n structural, neighborhood,
and environmental characteristics. The housing market implicitly reveals the hedonic function,
p(Z), which relates prices and characteristics. This price function p(Z) is a reduced form
equation representing both supply and demand  influences in the housing market. The implicit
price of attribute n is given by the partial derivative of p(z) with respect  to attribute n, or pn(z) =
3p/3zn. That is to say, the partial derivative with respect to any of the aforementioned
characteristics in the function can be interpreted as a marginal implicit  price of that
characteristic. This marginal implicit price is the additional amount that must be paid by any
household to  move to a bundle of housing services with a higher level of that characteristic. For
example, the  coefficient on the number of rooms in a home may be interpreted as the price that
must be paid  by the household to move from a house with eight total rooms to the same house
with nine total rooms, all else constant. Since the function for housing is nonlinear, the marginal
implicit price depends on the quantity of the characteristic being purchased.

Several hedonic studies specifically address the issue of flooding including the effect of
floodplain regulations on residential property values (Schaefer 1990), the impact of subsidized
and non-subsidized flood insurance on property values (Shilling et al.,  1987), and the influence
of flood risk on property values (Barnard 1978; Park and Miller 1982; Thompson & Stoevener
1983; Donnelly  1989; Speyrerand Ragas 1991; Shabman and Stephenson 1996).  For the most
part, the results from these studies indicate that location in a floodplain, or proxies for flood risk,
negatively impacts residential property values. One study examined a major flood event
(Babcock and Mitchell 1980); however, this was done by a comparison of prices before and
after the event,  and  thus was vulnerable to bias due to omitted factors  in the analysis. None of

these studies measure flood risks directly, nor do they investigate the impact of a specific
flooding event in a hedonic framework.

3.  Definition of Flood Risks
A flood is defined for the purpose of this paper as a stream discharge greater than the capacity
flow of the channel. This is obviously a very simplistic definition. For example, Williams (1978)
presented 11 definitions of the channel bankfull flow, from which the flow that reaches the valley
active floodplain is the one accepted by most river morphologists. A flood of certain magnitude
occurs or is exceeded with a  certain frequency. The most common flow used for delineation of
floodplain is  the flow with the recurrence interval Tr = 100 years, i.e. the risk of flooding is r = 1 / Tr
= 1/100 = 0.01.

The delineation of the floodplain for a flow of given frequency is a tedious task. Such tasks usually
involve the development of a complex hydrologic/hydraulic model. Once calibrated, the model can
be used to simulate a wide range of flows and the flow-elevation relationship can be obtained.
Hydraulic models can be combined with GIS systems to delineate a floodplain for any recurrence
interval (e.g., McLin, 1993, Correia et al., 1998). However, this requires a considerable amount of
data and substantial effort. Thus, a simplifying alternative has been proposed in this study.

The extent of 100-year floodplain, often used for engineering and  flood insurance purposes, is
delineated by Federal Emergency Management Agency (FEMA).  The flood risk varies within the
floodplain and decreases with increased distance from the channel. The properties located within
the 100 years floodplain are under different risks of flooding and hence there is a need to express
a flood risk relation in the urban floodplain.

A schematic representation of the following concept is shown in Figure 1. The channel can
contain a flow with a certain recurrence interval. This flow is called a capacity flow, or bankfull
flow. As one moves away from the river's edge, the probability of flooding decreases, and at some
point at a distance x from the river the recurrence interval of flooding becomes 100 years, i.e., the
risk of flooding is r(x) = 0.01. This is the extent of the 100-year floodplain that is useful for many
engineering  and flood insurance purposes.

Channels of natural streams are in equilibrium with the flow. Leopold, Wolman, and Miller (1995)
document that channels of rivers in eastern and Midwestern US have a channel capacity that can
contain a flow that has an approximate recurrence interval of about 1 1/4 years. For example, if the
smallest flow that leaves the channel is about a 2-year flow before urbanization, then the risk of
flooding at the edge of the river is r(0) = 1 / 2 = 0.5.

                               Figure 1: Concept of flood risk
       Flood risk
The scale of the risk function r(x) should be logarithmic, i.e., a zero risk of flooding is expected to
occur at an infinitely large distance x from the river edge. The logarithmic form of the risk function
is selected for convenience and simply expresses the fact that floods on rare occasions may
extend further than the 100-year floodplain limits. The logarithmic risk function can be expressed
                                     rfc) =  CIO*'                                  Eq
The function parameters in Eq. 1 can be easily estimated from the knowledge of the risk of
exceeding the bankfull capacity flow and from the extent of the 100-year floodplain: C
corresponds to the risk of exceeding the bankfull flow, or, C = /tO). The risk function can be
integrated across the floodplain cross-section, as shown in the following equation, in which
subscripts L and R correspond to the left and right bank floodplains:

                         (x) dx+l ^ (x) dx= r(0) J [10"K^ + 10"K^ ] dx               Eq. 2
The magnitude of the floodplain shape coefficient, K, can be obtained from the extent of the
100-year floodplain at the point of interest on the river, denoted as X10o, and from the risk of
exceeding the bankfull discharge, /tO):
=  tog
=  -KX100                      Eq. 3

                                  K =  ~3"'"/J  '  "                               Eq. 4
                                            X 100

Finally, substituting for K in Eq. 2 from Eq. 4 yields the following expression for the floodplain risk

                         R =
                                /        ,
                             23(2+tog r(0)J
The dimension of the floodplain risk parameter R is length/time, and a possible unit is meter/day.
However, the unit does not have a physical meaning, as R is only a measure of the flood risk over
a floodplain. R increases with an increase in the size of the floodplain and with an increase in the
risk of overbank flow. This floodplain risk parameter changes along the stream. The integration of
the flood risk over the watershed represents an overall risk of flooding of the watershed, the flood
risk factor that can be used in comparing watershed management alternatives.

This characterization of flood risks will be used to assign unique values of flood risk to each
property within the floodplain. The flood risk measure, FRM, calculated in a GIS environment, is a
negative logarithm of the flood risk r(x). The anti-logarithm of the flood risk measure is basically a
recurrence  interval, i.e., FRM = 2 for Tr= 100 years.

4. Empirical Model
a. Study Area
The study area for this analysis is located approximately 11.5 miles (18.5 km) along the middle
to lower sections of the Menomonee River through the cities of Wauwatosa and Milwaukee,

Wisconsin. The Menomonee River is a 71.85 (15.5 km) mile river system and discharges into
the Milwaukee River about 0.9 mile upstream of where the Milwaukee River enters Lake
Michigan. This region was selected to encompass two significant areas, the city of Wauwatosa
and the Valley Park neighborhood in Milwaukee. Wauwatosa makes up a great portion of the
study area and lies within the Menomonee River watershed boundaries. Located west of
Milwaukee in northern Milwaukee County, Wauwatosa is just over 13 square miles (34 km2) with
a population of 49,300. Furthermore, it is a high density residential area, with more than 22.8
persons per net residential acre (55 persons/ha). Valley Park, the other area of concern, is the
smallest and most isolated neighborhood in Milwaukee. The study area is shown in Figure 2.

Figure 2: Menomonee River watershed. Location of properties in 100-year floodplain.
These two areas are significant for this study as a result of their susceptibility to flooding.
Specifically, the study examines the short and intermediate run impacts of a 100-year flood that
occurred in June of 1997. The flood was the worst rain for the Milwaukee  Metropolitan area
since August 6, 1986. After the first night of the rainfall, totals ranged as high as 9.78 inches (25

cm), indicating a flood recurrence interval exceeding 100 years. Roads were shut down and
many residents lost power. Damage for Milwaukee County alone was estimated to be $37
million, including $24 million to residential property. About 70 homes in the County incurred
major damage including collapsed basements and roofs forcing residents to evacuate their
homes. Approximately 2100 homes sustained lesser damage. As a result of the flood,
Wauwatosa submitted a Hazard Mitigation Grant Program application for the acquisition of a
number of structures located in the floodway on the Menomonee River. They used Community
Development Block Grant funds to acquire flood prone structures as a means of creating open
space in the riverfront floodway. Of the 20,289 structures in Wauwatosa, about 738 are located
in the special flood hazard area, 669 of which are residential. Due to its susceptibility to flood
disaster, Wauwatosa was invited by  FEMA in June of 1998 to participate in a nationwide effort
to become a "Project Impact" community. This program would develop efforts to minimize the
risk of damage from  natural disasters. Valley Park also suffered from the flood in terms of water
levels. However, there is a great sense of community in the neighborhood, which became
evident in the recovery period following the disaster. Both Wauwatosa and the city of
Milwaukee, in which  "Valley Park" resides, are participants in the National Flood Insurance
Program (NFIP); Wauwatosa entering in 1978 and Milwaukee in 1982. The NFIP implements
floodplain management regulations which ensure that development in flood-prone areas is
protected from flood  damages. However flood insurance is mandatory only for those properties
residing within the 100-year floodplain. This increase  in cost associated with location in the
floodplain may reduce property value for those houses.

b. CIS Analysis
ArcView, a Geographical Information System (GIS), was used in several aspects of this study.
First, it was used to spatially define flood risks. Second, properties were geocoded to the street
address, and finally location specific data were matched to each property. We  describe each of
these activities below.

The properties were  geocoded to the precise street address using the ArcView GIS package. A
key to the geocoding process is the accuracy of addresses, the geographic files, and matching
of the addresses to the geographic files.  The addresses and geographic files received from
outside sources (MLS and Wisconsin Department of Transportation) are believed to be accurate
given the sources' own incentive for accuracy of the files. ArcView assigns a score to each

match made for the properties. Of the 1475 observations, 1402 of them (or approximately 95%)
were given a score of 75 or above on a 100 point scale. The majority of these received a score
between 98-100.1 The resulting sample size is 1431, as 44 were unable to be geocoded and
eliminated from the sample. Once geocoding of properties was completed boundary files for
geographic areas were digitized if they were not already available as ArcView shape files. For
example, the 100-year floodplain was geocoded from FEMA maps and maps provided by the
Southeastern Wisconsin Regional Planning Commission (SEWRPC). Other spatial boundary
data (e.g., school district boundaries, historic preservation district data) were also manually

Once the geocoding was completed, properties were matched to locational attributes of the
neighborhood using one of three techniques. When a neighborhood characteristic was defined
by a point in space (e.g., proximity to air quality monitors), straight line distance calculations
between the property and the attribute were used. If the attribute was defined by a polygon
(e.g.,  school districts, census block groups), then individual  properties were mapped to the
underlying polygon, and attributes of the polygon were attached to the property. Finally, buffers
were defined for various types of line data (e.g., roads,  railroads) and properties falling within
the buffer zone were identified.

Turning to the calculation of property specific flood risks, two basic approaches were considered.
The first is a vector-based approach that employed a custom developed ArcView Avenue scripts
program. This approach permits estimation of risks only at specific points rather than for complete
areas. The second more general approach works in a grid (raster) environment, and makes use of
the Spatial Analyst Extension for ArcView. It permits flood risk to be calculated for the entire
watershed, and specified points can be assigned the corresponding value from the underlying
polygon. The second approach was selected because of its future applicability in watershed
management applications.
1 A possible reason for a score at the lower end of the spectrum would be misspellings. For example, if an
address appears  as  "Menomone  Pkwy"  and the correct spelling  would  be  "Menomonee  Pkwy," the
addresses may still be matched and assigned a lower score as a  result. For this reason, the matches
receiving a score  of less than 80 were interactively re-matched by the author to ensure accuracy and
minimize error.

When we refer to the floodplain in this paper, it should be understood as the 100-year floodplain.
The width of the floodplain is the  key parameter in calculation of the flood risk, when r(0) is kept
constant. The floodplain width for any specified point, both inside and outside the floodplain, is the
distance of the flood fringe from the river bank for the river cross-section on which this point is
located. The calculation of the floodplain width corresponding to the selected locations had to be
done separately for inside and  outside of the floodplain. The floodplain width is calculated  as
                                    X   = X  + X                                   Eq. 6
                                      100    W      F                                    '
                                    y   = y  - y                                   EQ  7
                                      100    W      F                                    ""
where Xw is the distance from the river channel and XF is the distance from the floodplain  (see
Figure 3)
                       river channel
Figure 3: Calculation of floodplain width for locations inside and outside the floodplain
The floodplain was digitized as a polygon and used as such in calculations for the areas outside
the floodplain. For the areas inside the floodplain, it had to be converted into a polyline and
divided into several reaches. The calculation of the floodplain width for points inside the floodplain
was calculated separately also for left and right banks, although the calculation followed the same
procedure. The data essential for risk calculations  include digitized maps of the river channel and
100-year floodplain, as well as the watershed boundaries. The risk associated with  the capacity
flow has been estimated separately using the information from USGS on capacity flow and the
annual maximum series for the gage station in Wauwatosa. This station is located in the same
area as the majority of the properties. The recurrence interval associated with the capacity flow is
approximately 1 year,  i.e., r(0) = 1.

                                                                /\/river channel
                                                                I   l 100-year floodplain
                                                                   -log R
                                                                   	1 10-20
                                                                   I	1 20 - 100
                                                                   I—1100 -1000
                                                                      0.001 -0.444
                                                                   I	1 0.444 - 0.889
                                                                   |	1 0.8B3 - 1 .333
                                                                      1-333- 1.778
                                  Figure 4: Flood risk measure

Figure 4 shows the flood risk measure, i.e., the negative logarithm of the flood risk, in the area
where the properties are located. Individual  properties were assigned a value corresponding to
the underlying cell. The higher is this value,  the lower is the likelihood of flooding for the specific
property. An increase in this variable of one implies that flood risks decrease by an order of
magnitude. For example, as you move from flood risk measure of 2 to 3 you move from a risk of
0.01 (i.e., once per 100 years) to 0.001 (i.e., once per 1000 years).

c. Description of the Data
Detailed house attribute data as well as the sales prices of the houses were obtained from the
Multiple Listing Service  (MLS) for the Milwaukee Metropolitan Area. Information was collected
for each transaction, listed through the MLS, for the time period January of 1995- July of 1998.
This time frame provides an adequate period for property value fluctuation to occur as a result

of the flooding event in June of 1997, if this is the case. A total of 1,965 properties were listed
through the MLS in the study area for the time period examined. From this total, properties were
eliminated as a result of missing data for: the lot size (290), age of the house (198) and taxes
(2). Furthermore, the MLS database only includes properties sold through realtors, and thus
leaves out of the sample properties sold directly by the owner. This may reduce the possibility of
including "non-market" transactions in the sample, assuming that properties sold to relatives  or
close friends may be transacted by this means. Finally, as noted above, 44 properties were lost
as a result of geocoding difficulties, yielding a total sample of 1431 properties.

The variables in the model are organized into six categories: Structural, Neighborhood, Fiscal,
Disequilibrium,  Time Related and Flood. Many influences are controlled within the neighborhood
category in order to avoid misspecification biases and to account for spatial influences. For
simplicity, the fiscal variable (tax rate) and the disequilibrium control (days on the market) are
included in the Neighborhood category for the specification. Following Cropper (et al.) a semi-
log specification is chosen, and the model is specified by Eq. 8.

       LnRPRICE = f (Structural,Neighborhood, Time Related,Fiscal,Disequilibrium,Flood) Eq. 8

The variable definitions and data sources are reported in Table 1, and descriptive statistics are
in Table 2. The dependent variable is the log of real sale price of housing and is deflated by the
housing component of the CPI (1982-84) for the month in which the property sold.

i.      Structural Variables
The structural characteristics include the number of bedrooms, bathrooms,  other rooms,
presence of an attached garage, as well as square footage of the lot and the property. It is
expected that an increase in any one of the previous characteristics will increase the  sale price,
assuming that these attributes increase the housing services a property provides. Measures  of
area are included in linear and quadratic form to account for non-linearity in these variables.
Finally, the age of the  house is included expecting a negative relationship between the age of
the house and the sale price. This is based on an assumption that older homes may have dated
technology lacking several beneficial features that would increase the housing service provided
by the property.

ii.     Locational Variables
Each property was matched to numerous locational variables, including those in the
Neighborhood category. To account for various demographic characteristics, census data was
attached accordingly to the appropriate property. The census block group data captures the
racial and ethnic mix of the neighborhood. The sign for these variables cannot be predicted
without knowledge of a home purchaser's cultural preferences. The characteristics also include
measures of income and poverty, home occupancy, age of the neighborhood. Also,  the model
controls for the travel time  to work and the population density of the neighborhood. The latter
variable is included to control for aspects of the neighborhood correlated  with density which are
not measured (e.g., crime, cultural amenities).

The property  tax is included to account for fiscal effects, expecting that increases in  taxes would
decrease the sale price. Also capturing fiscal impacts is the teacher student ratio for the high
school  district in which the property resides.  A dummy variable is included to account for
residence within Wauwatosa or Milwaukee, which may capture a submarket influence and
perceptions associated with living in Wauwatosa (versus Milwaukee). The number of days a
property was  on the market is used in the model as a disequilibrium control variable.

Past studies have found historical preservation districts to positively impact property values
(Clark and Herrin 1997; Coffin 1989). The coefficients may be positive in  the case that creation
of the district  provides people with additional information about the housing stock and revitalizes
the neighborhood, yet also may  be negative if the structural restrictions reduce housing
demand. There are a total  of six preservation districts in this study area, three in Milwaukee and
three in Wauwatosa. Dummy variables are included for each of the districts.

As indicated in the theoretical review of the hedonic price model,  one of the influences on the
property sale price is environmental quality. Several variables controlling  for environmental
quality  factors are included within the neighborhood category including measures of air quality,
and proximity to Toxic Release Inventory sites. Accounting for the impact of local annoyance
factors is the  proximity of a residence to both highways and rail lines, as well as being located
on a major road.  One would expect these factors to negatively  affect property sale price in most
cases.  A variable is also included to capture scenic benefits of residing along the river, a
positive environmental attribute. This is measured by a dummy variable for those properties

residing on the Menomonee River Parkway. While some of the properties along the
Menomonee River Parkway may also be susceptible to flooding, only 7 of the 13 properties
along the Parkway are also in the 100-year floodplain. Thus, the effect of this variable should
pick up the scenic benefits of the river, while holding constant the risk associated with flooding
(accounted for by variables in the Flood category).

iii.     Time Related Variables
The model also includes dummy variables in the Time Related category for both the year and
season in which the property was sold. Business cycles may affect property values, and the
year variables are incorporated to capture the possibility of that influence. Furthermore, the year
variables may capture an interest rate effect. Similarly, the season dummies control for trends
that may be associated with time. There are no expected signs for the variables relating to time.

iv.     Flood Variables
Finally, variables representing the focus of this study are included in the Flood category and
also capture environmental quality. Other studies (Speyer and Ragas 1991, Schaefer 1990,
Donnelly 1989,  Park and Miller 1982, Thompson and  Stoevener 1983) have used dummy
variables accounting for a property's location inside or outside of the 100-year floodplain. All,
with the exception of Schaefer, have found a significant negative relationship between location
in the floodplain and the sale price of a property. This study differs from the previous studies in
that a continuous measure of risk is derived.  This permits floodplains of any periodicity to be
defined. We investigate floodplains in  100-year increments from  100-500 year floodplains. Over
the 3-year period, 15 properties sold in the 100-year floodplain, and 32 sold within the 500-year
floodplain. In addition, we examine the rate at which property values change within each

A second objective is to analyze the short run and intermediate run effects of a specific flood
event that occurred  in June of 1997. To do so,  two different measures are used. First, to
measure the short run impact, the floodplain  dummy is interacted with a dummy variable for
whether the property was sold after the flood event. Of the 1431  properties in the sample, 512 of
them were sold after the flood event and 4 of these were within the 100-year floodplain whereas
12 were within the 500-year floodplain. Second, to measure intermediate run effects, the
floodplain  dummy is interacted  both with the  dummy for whether the property was sold after the

flooding event and the number of days between the flooding event and the sale of the house. If
present, one would expect short run effects to be stronger than intermediate impacts, assuming
that the consequences of the flood event will taper off in the minds of homeowners and buyers
as time passes.

The coefficients on control variables in the structural, neighborhood, fiscal, disequilibrium and
time related categories differ minimally among the tables. To conserve space, these variables
are reported only once, with subsequent regressions reporting only the flood category variables.
Heteroskedasticity, a non-constant variance in the model's error term, is expected in this sample
of data since variance in selling price is likely to differ between the low-end and high-end of the
market. To test for the presence of heteroskedasticity, White's test is used and the null
hypothesis of no heteroskedasticity is rejected at the 95% level of confidence for each
regression (Gujarati, 1995). White's correction is employed to generate consistent estimates of
the standard errors. All models estimated explained approximately 91% of the variation in the
real housing price.

/'.      Structural Variables
All structural variables are significant at the 99% level of confidence, except the dummy
accounting for whether the garage is attached.  The number of garage spaces is significant, with
each additional  space increasing the value of the home by 4.8%. The number of bedrooms,
other rooms, half baths, and full  baths all positively impact property sale price. One  additional
half bath, full bath, bedroom, and other room, increases the property value by 11.2%, 6.2%,
5.0%, and 5.8% respectively. The large magnitude of the coefficient on the half bath variable
suggests that it  may be serving as a proxy for other structural features of the house. Both
square footage variables, interior and lot, increase property value at a decreasing rate reflected
by positive linear terms and negative quadratic terms. The  partial derivative of sale price with
respect to the interior square footage (3Real Price/3Building area) is equal to [• AREA+ 2
*• AREAso*Building area]. Evaluated at the mean for interior square footage (705.7 sq.ft. or 0.65
m2), property value increases by 6.8% for an increment of 100 square feet (or 0.72%/m2).
Similarly, an increment of 1000 square feet for the lot size increases sale price by 1.7% (or
0.18%/m2 evaluated at the mean). Finally, other things equal, age has a negative effect on

property value (i.e., 1.6% for each additional 10 years). Inclusion of a quadratic term for age
made both the linear and quadratic terms insignificant.

/'/'.     Locational Variables
Evaluating the demographic variables taken from the block group data, many coefficients
appear to  be significant at the 99% confidence level. Exceptions include population density and
the percent of occupied housing units, and percent owner occupied units. Population density
has a negative relationship with property value suggesting that on the net, urban scale related
disamenities have a stronger influence than that of amenities, yet the variable is insignificant.
The racial variables reveal that higher concentrations of Asian (as compared to nonwhite other
race) populations in a neighborhood positively affect property values. Specifically, a 1%
increase in the Asian population increases property value by 3%. The impact of Hispanic
populations, on the other hand, decrease real home sale prices by 2.5%. Percent White is
positive and significant, raising prices  1.3% per 1% increase, whereas percent Black is not
significant. Note, that most of the neighborhoods in the study areas have relatively few minority
households. As expected, higher poverty rates in a neighborhood decrease home sale price, yet
the effect is not great. Median household income, also reflecting socioeconomic dimensions of
the neighborhood, positively impacts property values. Measured by the median year of houses
built in the neighborhood, older neighborhoods have significantly higher priced housing in the
study area. This is somewhat contrary to the sign on the age variable, yet it may suggest that
people prefer historic surroundings in  a neighborhood along with the benefits of a
technologically advanced home. Finally, in line with the existing theory, each additional 10
minutes of commute time decreases the home sale price by 9%.

The tax rate, incorporating fiscal effects into the model, negatively impacts property value.
Specifically, a 1% increase in the property tax rate (e.g. 4.3% to 5.3%) decreases the property
sale price  by 2.0%. The teacher student ratio included to proxy the quality of education does
have a positive effect, yet is insignificant.  Also insignificant  is the number of days a house was
on the market. The dummy variable accounting for city jurisdiction is significant indicating higher
sales  prices (by a magnitude of 19%)  in Wauwatosa than in Milwaukee. However, Valley Park is
only one small area in Milwaukee and the dummy accounting for location in Valley Park was

The effect of historic preservation districts was positive in all cases confirming that historic
preservation districts provide home buyers with additional information regarding the housing
stock and serve the purpose of revitalizing the neighborhood. The influence of five of the six
districts was significant. The most dramatic of all influences was that of The McKinley Boulevard
Historic District in Milwaukee, increasing property value by 49%. The Concordia Historic District,
also in Milwaukee, has a similar effect with 41% increase in property value as a result of
residing within the district. The one historic preservation district that did not have a significant
impact was The Wauwatosa Avenue Historic District. These districts were also interacted with
age,  yet the resulting variables were insignificant and doing so overwhelmed the significance of
the individual dummies. Therefore,  they were not included in the final regression.

Several other variables in the neighborhood category were indicative of the surrounding
environmental quality. The quality of the air measured by the sulfur dioxide reading negatively
impacts property sale price as we would expect, and this effect is significant at the 99% level of
confidence.  Furthermore, location within one mile of a Toxic Release Inventory site has the
effect of reducing home sale prices by 2.8%, all else constant. Two of the variables representing
local annoyance factors significantly reduce the sale price of a home. Specifically, residence on
a major road and residence within a quarter of a mile of rail lines reduce home sale prices by
5.7% and 6.0% respectively. On the other hand, residence within a quarter of a mile of
Interstate 94 increased sales prices for homes by 8.5%. It is possible that this variable is
controlling for non-work related travel accessibility  in addition to an annoyance factor. Finally,
residence along the scenic Menomonee River Parkway has the significant effect of increasing
property value by 7.1%, all else constant.

/'/'/'.     Time Related Variables
The seasonal dummy variables are insignificant indicating that the season in which a house is
sold has no  impact on the sales price.  The year dummy variables indicate that real housing
prices have  fallen over the time period 1995- July of 1998. The effect in 1996 is insignificant;
however, housing prices significantly decreased for both 1997 and 1998.

iv.     Flood Variables
There are two objectives in terms of flood risk for this study. The first objective is to determine
the effect that flood hazard in general has on property value. In the first regression reported in

Table 3, we proxy flood risk using the negative log (base 10) of the expected flood frequency ,
i.e., flood risk measure (see
Figure 4). The log  of the value is included due to the rapid rate at which flood risks fall as
distance from  the river increase, and elevation rise. The findings  indicate a clear relationship
between  reduced flooding risk, and increased property  values.  However,  the value of the
coefficient  is  extremely  low. This finding  is  not  surprising,  given  that  the vast majority of
properties are well  beyond even the 1000-year floodplain.  Hence  a  reduction of risk from say
10E-23 to 10E-24 is of negligible value to those residents.

To investigate the variation of flood risks within floodplains, we explore several different
specifications. First, we examine the 100-year floodplain. Although flood risk is continuously
defined, lenders only require that properties in the 100-year floodplain purchase flood insurance.
In Table 4, we report the findings on a regression that includes a dummy variable for whether
the property lies within the 100-year floodplain. In addition, we interact that  variable with  the
recurrence interval,  i.e., anti-log of the flood risk measure. The recurrence interval takes  on
values between 6.3  (i.e., a flood is expected with a probability of 1/6.3) for the property closest
to the river, and 100 for a property at the edge of the 100-year floodplain. Both the dummy
variable and the risk interaction term are statistically significant. The findings suggest that
properties at the edge of the river would sell for approximately 7.8% than those outside the
floodplain. However, as flood risk diminishes by 10 years (e.g., from a one-year flood frequency
to an 11-year frequency) property values would increase by 2.3%. This implies that the
detrimental effect of the flood risk is eliminated after the expected flood risk falls to once  every
33.3 years.

In Table 5, we add a second interaction term to consider the effect of a flooding event. The
variable Days  since is the number of days since the flood in June of 1997. Hence, it measures
the effect of the flooding event on the impact of the  100-year floodplain. The inclusion of  this
variable renders the floodplain dummy variable insignificant, although it remains negative. This
is due to multicollinearity between the two variables. Treating the coefficient on the dummy
variable as point estimate, it suggests that properties (at the edge of the river) selling in the
floodplain prior to the flood sold for 5.1% less than comparable properties outside the floodplain
prior to the flood.  Those selling a year after the flood would sell for 18.9% less than properties
outside the floodplain. The pattern did not appear to be nonlinear, although  note that it was not

possible to capture longer-term effects due to the fact that the sample did not extend further into
the future. Thus, it appears that at least over the short term, the flooding event did reduce
property values beyond what they were prior to the flood.

In the final model presented in Table 6, we explore whether wider floodplains generate
detrimental effects on properties within those areas. Thus, we define floodplains between 100
and 200 hundred years, 200 and 300 years, and so on.  Given that the detrimental effects of
flood risk appear to dissipate within the 100-year floodplain, it is not surprising that none of the
other floodplain categories are negative and significant. Indeed, the region between the 300 and
400-year floodplain sells at a premium over those outside the floodplains. We also explored
whether the flooding event negatively influenced any of the property values within the 200 year
and beyond areas, and found no evidence of detrimental impacts.

6. Conclusions
This study employed GIS tools to more accurately characterize flood risks in an urban
watershed. An interpolation  scheme to evaluate the level of flood risk in the watershed has been
developed and applied to the Menomonee River watershed. Together with a wide range of other
locational attributes, flood variables were matched to geocoded properties to investigate
impacts on housing prices. Our findings support the hypothesis that increases in flood risk
decrease values for residential properties within the 100-year floodplain. Unlike other studies
which conclude that there are uniform impacts within the floodplain, we find declining  effects
with reduced risk. Furthermore, there is evidence suggesting that flooding events heighten
sensitivity to such risks and  raise the property price premium associated with a given  level of
flood risk. Negative impacts beyond the 100-year floodplain are not found.

The use of GIS tools to complement statistical analyses of urban spatial  problems will continue
to grow as PC-based GIS software becomes more  powerful, and geographic data sources more
abundant. In addition, GIS tools  can serve as a conduit for interdisciplinary work as geographic
modeling in the physical sciences and engineering  is integrated with spatial modeling by  social

This research is sponsored by the EPA/NSF Watershed Management Program in a form of a
grant to Marquette University. Views and findings included in the presentation are those of the
authors and not of the funding agency.

              Table 1: Variable Definitions and Data Sources
        Dependent Variable and Variables in the Structural Category
Variable Name
Real Price
Age house
Full bath
Half bath
Other rooms
Building area
Garage spaces
Garage attached
Lot size
[mean, standard deviation]
Real sale price of the property
(1982-84 dollars)
Age of the house in years
Number of full baths in house
Number of half baths in house
Number of bedrooms in house
Total rooms minus number of bedrooms
Area of the master
in square feet
Note: Due to data limitations, all of the
square footage is not captured
Number of garage spaces
1 = garage attached, 0 = otherwise
Lot area in square feet
is the
Variables in the Neighborhood, Fiscal, and Disequilibrium Control Categories
Variable Name
Sulphur Dioxide
Major road
% mile I94
Commute time
% railroad
Toxic Release Inv.
[mean, standard deviation]
Distance weighted value of the nearest air
monitor, computed as sulfur
dioxide/distance of monitor to property
1 = property resides on a primary road,
0 = otherwise
1= property resides on the Menomonee
River Parkway,
0 = otherwise
1= property within a quarter of a mile of
Interstate 94, 0 = otherwise
Average household travel time to work for
the block group in minutes
1= property within a quarter of a mile of
railroad tracks, 0 = otherwise
1= property within a quarter of a mile of a
manufacturing facility on the Toxic Release
Inventory, 0 = otherwise
LandView III
Census of




Variable Name

Historic Preservation

TS ratio

Pop density

Median year built

Median HH income

% Asian

% Black

[mean, standard deviation]
HPDTOSA 1= resides within The
Wauwatosa Avenue Historic District, 0=
HPDCHURCH 1= resides within The
Church Street Historic District, 0=
HPDWASH-HIGH 1= resides within The
Washington Highlands Historic District,
0= otherwise
HPDCONCORD 1= resides within The
Concordia Historic District, 0=otherwise
HPDMCKINLEY 1=resides within The
McKinley Boulevard Historic District,
HPDHIMOUNT 1= resides within The
Washington-Hi Mount Boulevards Historic
District, 0=otherwise
Teacher student ratio for the school district
in which the property resides

Population density in the block group,
measured as people per square mile

Median year of houses built in the block

Median household income of the block

Percent of the block group population that
is Asian

Percent of the block group population that
is Black


(first three)
(last three)

Census of
Census of
Census of
Census of
Census of








Variable Name
%Occupied units
%Owner occupied
% Poverty
Tax rate
Valley Park
Days on market
[mean, standard deviation]
Percent of the block group population that
is Hispanic
Percent of block group population which
falls into the "other" category
Percent of the block group housing units
that are occupied
Percent of block group housing units that
are owner occupied
Percent of block group population that is
below the poverty line
Tax payment/ [sale price/1000]
1 = property resides in Wauwatosa,
0 = Milwaukee
1 = property resides in Valley Park,
0 = otherwise
Number of days the house was on the
Census of
Census of
Census of
Census of
Census of

Time Related Variables
Variable Name
Seasonal Dummy
[mean, standard deviation]
SPRING=1 (March-May), 0=otherwise
SUMMER=1 (June-Aug), 0=otherwise
FALL=1 (Sept-Nov), 0=otherwise
WINTER=1 (Dec-Feb), 0=otherwise
1= dwelling sold in ith year, 0=otherwise
i = 1995, 1996, 1997, 1998
Winter is
1995 is

Variables in the Flood Category
Variable Name
Flood Risk Measure
Recurrence Interval
Days since
[mean, standard deviation]
1= resides in the 100-year, 0=otherwise
1= resides in space beyond 100 year
flood and within 200 year flood,
1= resides in space beyond 200-year and
within 300 year flood, 0=otherwise
1= resides in space beyond 300-year and
within 400 year flood, 0=otherwise
1=resides in space beyond 400-year and
within 500 year flood, 0=otherwise
Minus log of flood risk
The expected number of years between
flooding events
1= after the June 1997 flood,
0 = otherwise
The number of days since the June 1997


                      Table 2: Descriptive Statistics
             Dependent Variable and Structural Characteristics:
Full bath
Half bath
Other rooms
Building area
Garage space
Garage attached
Lot size
Sulpher Dioxide
Major road
% mile 194
Commute time
% mile railroad
Toxic Release Inv.
HPD Church
HPD Concord
HPD McKinley
HPD Himount
TS ratio
Pop. Density
Median Year Built
Median HH income
in Neighborhood,


Fiscal, and Disequilibrium Control







Valley Park
Days on Market


Flood Risk Measure
Recurrence Interval10o
Recurrence Interval50o
Days since
Related Variables
Related Variables





Table 3 - Hedonic Regression with Log Flood Risk
Structural Characteristics
Full bath
Half bath
Other rooms
Garage space
Garage attached

Building area
Building area *
Building area




Neighborhood and Fiscal Characteristics
Sulpher Dioxide
Major road
% mile 194
% mile railroad
Commute time
Toxic Release Inv.
Teacher Student ratio
Population Density
Median HH Income
% Black
%Owner occupied
% Occupied units
% Poverty
Tax rate
Median year built
Mean dep. variable

Variable Coefficient t-score

Time Dummy Variables
Year 1996 -0.014904
Year 1997 -0.075591
Year 1998 -0.079498
Spring quarter -0.00728
Summer quarter -0.009696
Fall quarter -0.001184
Historic Preservation Districts and
HPD Church 0.063261
HPDConcordia 0.412596

HPD High Mount 0.141946
HPD McKinley 0.486035
HPDWauwatosa 0.069102
HPD Wash. Highlands 0.213099
Wauwatosa 0.198344
Valley Park -0.023755
Menomonee Pkwy 0.071265
Flood Risk Variables
Flood Risk Measure 0.000253
Disequilibrium Control
Days on market -8.17E-06

Adjusted R-squared 0.914996
S. E. of regression 0. 1 3761 1
Log likelihood 831.532







     Table 4: Model II—Flood Risk within thefloodplain

  LnRPRICE = f (Structure, Neighborhood, Time Sold, Flood),

 Variable                           Coefficient    t-statistic
Floodplain10o*Recurrence Interval
    Table 5: Model III—Flood Risk and a Flooding Event
  LnRPRICE = f (Structure, Neighborhood, Time Sold, Flood),
Coefficient    t-statistic
Floodplain10o*Recurrence Interval
Floodplain10o*Days Since Flood
  Table 6: Model III—Flood Risk in Expanded Flood Zones
  LnRPRICE = f (Structure, Neighborhood, Time Sold, Flood),
Coefficient    t-statistic
Floodplain10o*Recurrence Interval
Floodplain10o*Days Since Flood


Babcock, M., and Mitchell, B. 1980. "Impact of Flood Hazard on Residential Property
      Values in Gait (Cambridge), Ontario." Water Resources Bulletin, 16: 532-537.

Barnard, J.R. 1978. "Externalities from Urban Growth: The Case of Increased Storm
      Runoff and Flooding." Land Economics, 54: 298-315.

Clark, D.E., and Herrin, W.E. 1997. "Historical Preservation Districts and Home Sales
      Prices: Evidence from the Sacramento Housing Market."  The Review of Regional
      Studies, 27: 29-48.

Coffin, D.A. 1989. "The Impact of Historical Districts on Residential Property Values."
      Eastern Economic Journal, 15: 221-28.

Correia,  F.N., Rego, F.C., Da Graca Saraiva, M., and Ramos,  I. 1998.  "Coupling GIS
      with hydrologic and hydraulic flood modeling." Water Resources Management,

Cropper, M.L., L.B. Deck, and K.E. McConnell. 1988. "On the Choice of Functional Form
      for Hedonic Price Functions." Review of Economics and Statistics, 70: 668-675.

Damianos, D., and Shabman.  L.A. 1976. "Land Prices in Flood Hazard Areas: Applying
      Methods of Land Value Analysis." Offices of Water Resources  Research. Project

Donnelly, W.A. 1989. "Hedonic Price Analysis of the Effect of a Floodplain Location on
      Property Values." Water Resources Bulletin, 25: 581-586.

Gujarati, DamodarN. 1995. Basic Econometrics McGraw-Hill,  Inc., New York.

Kiel, K.A., and McClain,  K.T. 1995. "House Prices during Siting Decision Stages: The
      Case of an Incinerator from Rumor through Operation," Journal of Environmental
      Economics and Management, 28: 241-55.

Lancaster, K.J. 1966. "A New Approach to Consumer Theory," Journal of Political
      Economy, 74:132-157.

McLin, S.G. 1993.  "Combined GIS-HEC procedure for flood hazard evaluation." Report
      No. LAUR933008, Los Alamos National Lab., NM, USA

Park, W.M., and Miller, W.L. 1982. "Flood Risk Perceptions and Overdevelopment in the
      Floodplain." Water Resources Bulletin, 18: 89-94.

Rosen, S.  1974. "Hedonic prices and implicit markets: product differentiation in pure
      competition." Journal of Political Economy, 82: 132-57.

Shabman, L. and Stephenson, K. 1996. "Searching for the Correct Benefit Estimate:
      Empirical Evidence for an Alternative Perspective." Land Economics, 72: 433-39.

Shaefer, K.A. 1990. "The Effect of Floodplain Designation/Regulations on Residential
       Property Values: A Case Study in North York, Ontario." Canadian Water
       Resources Journal, 15: 319-333.

Shilling, J.D., Sirmans, C.F.,  Benjamin, J.D. 1987. "Flood Insurance, Wealth Distribution,
       and Urban Property Values." Journal of Urban Economics, 26: 43-53.

Speyrer, J.F., and Ragas, W.R. 1991. "Housing Prices and Flood Risk: An Examination
       Using Spline Regression." Journal of Real Estate Finance and Economics, 4:

Thompson, M.E., and Stoevener, H.H. 1983. "Estimating Residential Flood Control
       Benefits Using Implicit Price Equations." Water Resources Bulletin,  19: 889-95.

Tobin, G.A., and Newton, T.G. 1986. "A Theoretical Framework of Flood Induced
       Changes in Urban Land Values." Water Resources Bulletin, 22: 67-71.

US EPA (1998): http://www.epa.gov/ ...BASINS

Williams, G.P. (1978): Bank-full Discharge of Rivers. Water Resources Research
Zimmerman, R. 1979. "The Effect of Flood Plain Location on Property Values: Three
       Towns in Northeastern New Jersey." Water Resources Bulletin, 15: 1653-6

                 A GIS Demonstration for Greenbelt Land Use Analysis
                                          Joanna J. Becker
                     Environmental Planning Services, Santa Rosa, California

The goal of this project was to demonstrate what analy-
ses could  be undertaken with a GIS  program without
substantial GIS training ortime input. The demonstration
attempted to show how planning staff and decision-makers
could easily and usefully employ GIS. It was not intended
as a complete study of all possible variables. Only avail-
able  data  were used.  Diverse techniques were  pre-
sented while keeping the content as simple and relevant
as possible. The project was designed as  a demonstra-
tion using  regional scale data and was combined  with
another parcel-specific demonstration that showed
urban  GIS applications.

The  demonstration showed the following  modeling

• Buffer zones

• Combination of variables (overlays)

• Weighting  of values

• Absolute value variables

• Reclassification of final values

Site Location

The San Luis Obispo watershed comprises an area of
approximately  84  square  miles. The watershed drains
into the Pacific Ocean  at Avila, California.  The major
creek in the watershed, San Luis Obispo Creek,  is a
perennial  creek,  but many of its tributaries have  only
seasonal flow. Agriculture and grazing are the major
land uses  in the watershed, although a significant num-
ber of areas  have been  developed.  Growth  of these
areas is moderate to limited but has a pronounced effect
on the watershed. The watershed also supports a large
amount of riparian and other natural vegetation. Figure 1
demonstrates  the distribution  of  land cover/land  use
within the watershed.
                                      Study Perimeter
 City Greenbelt
           Pacific Ocean
Figure 1. A generalized map of the San Luis Obispo area show-
        ing the location of the two ARC/INFO GIS study areas:
        (1) the parcel analysis in the Dalideo area and (2) the
        regional analysis in the city greenbelt stippled area
        surrounding the city. All boundaries are approximate
        and are for schematic purposes only (1, 2).


The City Council approved the open space element of
the San Luis Obispo General Plan in January 1994 and
identified a greenbelt area that extended from the Urban
Reserve Line approximately to the boundaries of the
San  Luis Obispo Creek watershed. The intent of this
greenbelt area is to provide a buffer to the city and to
preserve the agricultural and natural resources of the

The data forthe watershed were already available through
the work of the Landscape Architecture Department of
California  Polytechnic State University  at  San  Luis
Obispo for a study of San Luis Obispo Creek (1), and
the city's greenbelt area lay approximately within those
boundaries (see Figure 1). Variables were selected that
could be extracted from the available data.

The data for the creek study were initially entered into
workstation ARC/INFO in a polygon format. They were
then transferred to MacGIS, a PC raster program, for
simplicity of use. The final product was then transferred
back to ARC/INFO as a grid format and viewed in a PC
version of ARC/VIEW using DOS files.

In interpreting the overlay of values, the assumption was
made that the occurrence of high  values for the most
variables would result in the most suitable land for that
land  use.  This was  presented as a  range of  values
derived from the total values divided into three approxi-
mately equal groups of high, medium, and low. In addi-
tion to providing a composite analysis, however, any one
of the data sets can be queried separately such that, for
example, slopes greater than 20 percent could be iden-
tified or two layers such as storie index and distance
from roads could be compared.


Note that the ratings of high, medium, and low are based
on available data, and the rating of low implies no suit-
able  use.  In addition, these ratings do not  imply that
categories rated low could not be used for a particular
land  use but rather that other land  uses might be more
appropriate. For example, open space use was rated
low for  flatter slopes  but only  because this category
would likely be more suitable for agriculture.

The demonstration used an existing cell size of the data
on the MacGIS program of 75 meters per side, which is
assigned during the initial conversion process.  There-
fore, buffer zones are  in aggregates of 75 meters. This
size cell does not allow for minute analysis but reduces
the size of the files,  which may become  extensive in
raster format.

In presenting the final analysis, land contained within the
urban reserve limit line has been excluded.


Initially,  eight variables from the  available data were
deemed suitable for this analysis:

• Slope

• Storie index (indicating soil fertility)

• Distance from major roads

• Distance from creeks
• Erosion hazard1

• Oak woodlands

• Land use compatibility

• Grasslands2
After selecting the six variables,  the  categories were
receded to conform to a rating of high, medium, or low.

Each land use was then evaluated separately for every
variable except for combining the variables of slope and
distance from creeks for rangeland  analysis. In this case,
composite values were assigned to the two variables, then
receded to produce a high, medium, or low rating.

After obtaining maps for each of the variables according
to land  use,  the  maps were  compiled to indicate the
density of overlays for each land  use category. In as-
sessing the suitability of land for the three land uses, the
values of all the variables, except land use, were aggre-
gated and a rating system developed. In  addition, a
double weight was assigned to the storie index in evalu-
ating agriculture  (because this  is a primary index for
considering prime agricultural land). If less than 75 me-
ters, the  distance from  creeks  also  received  added
weight in consideration of open space preservation (be-
cause this is likely to ensure  the least erosion and
pollution to the waterways). The weighting then altered
the scores as follows:
Land Use

Open space

Number of Values

After the values had been  assigned for each ranking,
further receding established three categories of high,
medium, and low for each land use.

The land use buffer was added to this receded  aggre-
gate map and resulted in an additional three values due
to the interaction of the buffer with each category. These
additional values were receded according to each land
use to produce a final map with three values.

The Urban  Reserve Area was overlaid on the final map
to exclude urban areas.


In determining what properties would  be most suitable for
each land use, the following assumptions were made.
1 After reviewing the material, erosion hazard was eliminated because
 it was similar to the Soil Conservation Service (SCS) storie index
 data while the identification of native grasslands fell only within the
 area currently designated as open space land, so it was not included.
2 See above note.

Open Space

This land is desirable to preserve as open space  be-
cause of the existence of scenic or significant natural
resources. It could also be land that is inappropriate for
other uses due to the presence of such factors as steep
slopes or poor  soils. A distinction is  made  in the final
map between land that is designated open space for
recreational uses, such as parks, and land preserved for
habitat or species protection. Open space adjacent to
an urban area would be rated high if public accessibility
was desirable but low if its purpose was resource pro-
tection or preservation. Separate maps  based on two
types of proposed uses present the contrast  in analysis.

The analysis of the variables thus was rated as follows:

A. Land with steep slopes and therefore less suited for
   other purposes.

B. Land that has oak woodland vegetation resources.

C. Riparian land.

D. Low storie index  indicating a  lack of suitability for
   agriculture or rangeland.

E. At least approximately one-eighth of a mile from a
   major road to avoid negative impact on wildlife.

F.  Either approximately one-quarter of a mile from ur-
   ban areas if designated to protect resources or adja-
   cent to urban areas if designated to serve as parks and


This land use includes all forms of agricultural activity;
obviously, its suitability for specific crops and practices
would vary. The determination of suitability would need
to be made on a site-specific basis.

For the purpose  of general agricultural suitability,  the
highest land suitability for agriculture was a rating of the
variables as follows:

A. Land that does not have oak woodland.

B. Land that is  not close to  perennial creeks  (to avoid
  fertilizer/pesticide runoff contamination).

C. The flattest slopes.

D. The highest storie index.

E. Proximity to a major road (considered an advantage
  for trucking and farm equipment access).


Some types of  livestock  can graze under most condi-
tions, but for purposes of this analysis, land more suited
for  either open space or agricultural designation was
rated above that of rangeland. The major limitation to
suitability of land for rangeland was  a  combination of
steep slopes and proximity to creeks. A rating of medium
for the other variables was considered  the most desir-
able for rangeland purposes.

Details of the Variables

Storie Index

In determining the most suitable uses according to soil
fertility, the SCS storie index rating was used, with a
modification of the categories to three to accommodate
the ratings of high, medium, and low that were used
throughout the  analysis.  Therefore, the first two SCS
categories  of excellent and good were combined into
Category 1. Categories fair and poor were combined
into Category 2, and categories very poor, nonagricul-
tural, urban,  and mines  were combined to compose
Category 3.

Subsequently, receding was undertaken to  prioritize
these categories according to land use:
Open space

High for Category 1
High for Category 2
High for Category 3
The five principal arterials of the watershed were used
in this analysis:

• Highway 1

• Highway 227

• Los Osos Valley Road

• U.S. 101

• Avila Valley/San Luis

A buffer on each side of the road was created, which
was then receded into the three categories of more than
75 meters, 75 meters to 150 meters, and greater than
150 meters. Each land use was then evaluated accord-
ing to these criteria, with agriculture deemed the most
suitable closest to the roads and open space the least


The slope categories in the San Luis Obispo watershed
data set were divided into a number of classes, which were
receded  into  the most  appropriate  grouping  of three
classes for the analysis of the three land uses. The existing
categories were not altered for the purpose of the demon-
stration, so they do not necessarily represent the most
ideal slopes for the particular uses. A separate category of
less than 10 percent slopes was provided for agriculture
because  most  agricultural practices  require flat land.

Slopes between 10 and  21.5 percent would be limited
to such  activities as orchards or vineyards.
Open space:
 < 10 percent
 10 to 21.5 percent
 > 21.5  percent

 < 10 percent
 10 to 46 percent
 > 46 percent
High (H)
Medium (M)
Low (L)

No  slope analysis was undertaken for rangeland  be-
cause this category was combined with that for streams
(see Streams section).

Urban Adjacency

Urban adjacency was treated as  an overlay of the  ag-
gregate map of the  other variables because it is an
absolute value. That is, this variable has no ranking.
Land is either within or outside the buffer. The suitability
of each land use  adjacent to urban areas was deter-
mined, then the aggregate map was adjusted according
to a comparison of the receded aggregate values with
the  designated ranking of land use suitability.

The first step was to recede the  existing data on land
uses  (interpreted  from 1989 aerial photography  ob-
tained from the United States Department of Agriculture
[USDA], Agricultural Stabilization  and  Conservation
Service, Atascadero, California) into urban/commercial
areas and  nonurban/noncommercial  areas. A  buffer
zone of approximately one-quarter of a mile was then
applied  around the  urban/commercial  areas and an
analysis undertaken of suitability for the three land uses
to be located within this buffer.

In making the analysis, the following assumptions were
Land Use
Agriculture    L

Rangeland    M

Open space   H

           Conflict with dust, noise,
           pesticides, and urban use
           Fire hazard of open
           grassland near buildings
           Most suitable if used as
           Least suitable if
           designated for
           habitat/wildlife protection
Therefore, the analysis provided for two planning alter-
natives for open space, with the scenarios presented as
separate maps overlaid on the aggregate map for the
other rangeland variables.

In interpreting this map, the combined values were rated
according to the above criteria, with any values lower
than the desired ranking receiving a low value no matter
what the aggregate value had been. This produced the
following results:

• Agriculture: A rating of low for any land within the
  buffer zone.

• Rangeland: A rating of medium for any land within
  the buffer zone except that rated low for the aggre-
  gate map.

• Open Space: A rating equivalent to the rating of the
  aggregate  map  if the land  was  to  be used for
  parks/recreation, or a rating of low for any land within
  the buffer zone if the land was to be used for habitat


The  original  data  for  streams (from 1:24,000-scale
USGS maps), which included both intermittent and per-
ennial creeks, were used for rangeland analysis. A buffer
of 150 meters was then applied to these streams, and
these data were combined with the slope  analysis. This
combination was important in evaluating the  erosion
hazard and resulting stream pollution caused by nitro-
gen waste and hoof disturbance. The complete stream
complex for the watershed area was therefore evaluated
using the following matrix:
                                       Stream Buffer
                                       < 75 meters
                                       75 to 150 meters
                                       > 300 meters


                              21.5 to 46


                The stream data were then modified for agricultural and
                open space land use analysis to indicate only the per-
                ennial creeks as defined by the California Department
                of Fish  and Game. Buffers were created for this as
                Stream Buffer

                < 150 meters
                150 to 300 meters
                > 300 meters

                Oak Woodlands
                      Open Space    Agriculture
                                       The receding of oak woodland data for agriculture was
                                       different than for the other two land uses  because
                                       the presence of oak woodlands is  not conducive to
                                       No oak woodlands
                                       < 10 percent
                                       > 10 percent

The suitability of oak woodlands for open space and    References
rangeland was ranked as follows:                         ^  Ha||ock  B G  L s Bowker WD  Bremer  and D N  Long  19g4

                       Open Space        Rangeland       Nutrient objectives and best management  practices for San Luis
                       —*	*	        	"	       Obispo Creek. San Luis Obispo, CA: Coastal Resources Institute,
                                                             California Polytechnic State University.
< 33 percent          L                   M
                                                          2.  Community Development Department of San Luis Obispo. 1994.
33 to 75 percent      M                  H                Open space element for the city of San Luis Obispo general plan.
> 75 percent - high    H                  L                san Luis Obispo, CA.

          Small Is Beautiful: GIS and Small Native American Reservations—
                      Approach, Problems, Pitfalls, and Advantages
                                          Jeff Besougloff
          Upper Sioux and Lower Sioux Indian Communities, Redwood Falls, Minnesota

The Lower Sioux Indian Reservation
The Lower Sioux Indian Reservation covers 1,743 acres
in  southwestern  Minnesota  bordering the Minnesota
River. The land base consists of several hundred acres
of prime, flat agricultural land, a large wetlands slough
complex, prairie  pothole wetlands, bottom land wet-
lands, small lakes, and approximately 250 acres of tim-
ber and brush. The  elevation ranges from Minnesota
River level to the adjacent bluffs several  hundred  feet
This rural reservation contains a moderate amount of
infrastructure, including paved and dirt roads, 12-acre
sewage lagoon serving a moderately sized casino, com-
munity water system  composed of a  tower and small
treatment plant for the 90 mostly single-family dwelling
homes,  convenience  store/gas station/gift shop, com-
munity center, small two-room schoolhouse, pottery works
with  gift  shop, warehouse, and  church.  The  casino-
fueled economic  boost to the community recently re-
sulted in improvements to infrastructure and plans for
additional projects.

Office of the Environment
The  tribal government was  formed under the Indian
Reorganization Act of 1934. The governing body is an
elected five-person tribal community council that admin-
isters several government departments and is responsi-
ble for all government activities.
Under a  U.S. Environmental Protection Agency (EPA)
Region 5 multimedia grant, the Upper Sioux and Lower
Sioux Office of the Environment (OE) was formed in late
1992. This unique joint venture between two tribes  and
EPA envisioned moving the tribal governments into com-
pliance with major federal environmental legislation. At
the present time,  only the Lower Sioux are developing
a tribal geographic information system (GIS). Therefore,
this article is solely  applicable to this community, al-
though adoption  of an Upper Sioux Community  GIS
would likely follow a similar lifeline.

Environmental Regulation in Indian Country

Reservations are subject to a bewildering array of envi-
ronmental regulations. Numerous meetings, publications,
projects, and court decisions are devoted to determining
what law does  or does  not  apply  on  any particular
reservation. In very general terms, the following can be
stated: state environmental regulations do not apply,
federal regulations do apply, and tribal regulations may
apply. From a tribal environmental employee's point of
view, numerous environmental regulations (whether fed-
eral or tribal) do exist that apply to reservation activities
and land, and they require compliance.

The Problem and the Solution

The OE's responsibility is to bring the reservation into
compliance with the  14 major pieces of environmental
legislation administered through EPA and directly appli-
cable to tribes. The OE finds itself responsible for any
and all other applicable environmental regulations and
all other less-regulated environmental media. The OE
currently has a staff of one.

In addition to the  responsibility of moving the tribes into
compliance with federal environmental regulations, the
OE also develops environmental infrastructure, insti-
tutes environmental programs, and performs grant writ-
ing. Lower Sioux programs  currently  include Clean
Water Act (CWA) and Safe Drinking Water Act (SDWA)
compliance,  solid waste  planning  including  develop-
ment and institution of a household recycling program,
wetlands regulations compliance, wetlands  mapping
and  restoration,  National Environmental  Policy Act
(NEPA) compliance and site assessments, basic hydro-
logical data gathering and mapping, radon testing and
mitigation, environmental education as necessary, SARA

Title III compliance and planning, and a variety of related

Contracts  or grants are currently being administered
under several Bureau of Indian Affairs (BIA) programs,
U.S. Geological Survey (USGS) and U.S. Army Corps
of Engineers (COE) matching funds programs, two EPA
programs,  one  Federal  Emergency  Management
Agency (FEMA) program, one Administration of Native
Americans (ANA) program, one Great Lakes Intertribal
Council  (GLITC) program,  and a cooperative  project
with the National Tribal Environmental Council (NTEC).

Needless to say, responsibilities of the OE are limited by
staff hours rather than need. As distressing as the res-
ervation's  unaddressed environmental  needs  are,
equally distressing (prior to CIS development) was the
helter skelter manner in which the OE digested the data
and information flowing into the office.  Because of its
broad responsibilities and the administrative problems
being encountered, the OE began to investigate devel-
oping a tribal CIS.

The Lower Sioux GIS

System Selection

The Lower Sioux GIS system is a networked PC station
through the Bureau  of Indian Affairs Geographic Data
Service  Center's  (BIA  GDSC's)  two  Sun   MP690
SparcServers in Golden, Colorado.  The GDSC is the
hub of BIA's GIS and remote sensing program, known
as the Indian  Integrated Resources Information  Pro-
gram (IIRIP). The purpose of the IIRIP is twofold: first,
make GIS and remote sensing technology available to
tribes and BIA personnel; second, transfer these tech-
nologies to tribal organizations.

Database  development  and management functions,
technical support, development of simplified user inter-
faces, remote sensing interpretation, and  implementa-
tion   of  equipment  directives   are  performed   by
approximately 30 GDSC employees for approximately
230 GDSC users.  User technical support  is also avail-
able through  BIA field offices, each of which has a
designated GIS coordinator. Simplified  user interfaces
for specialized programs have  been developed  includ-
ing  the Lightening Display System and the Land Title
Mapping System.  Quality control  is provided for non-
BIA-produced data that will be inputted.

The GDSC has standardized on the ARC/INFO family
of software produced  by Environmental Systems Re-
search Institute (ESRI). GDSC has developed a number
of hardware/software configuration options depending
on tribal needs and financial resources and based upon
GDSC experience. The OE happily relies on this expe-
rience to avoid the familiar horror stories related to
equipment and software incompatibility.
Based upon GDSC configuration advice, the initial GIS
setup will be on the OE's existing Compaq PC using
Tektronix Terminal Emulation software (EM4105) and a
Multitech modem (MT932BA). The system can use the
OE's Hewlett Packard (HP) DeskJet 500,  although a
significant upgrade, possibly to an HP Paint or HP Excel
Paint, is soon  expected. Initial startup hardware and
software costs  are minimal in this configuration. Costs
for the  above equipment and introductory training  are
less than $5,000.

GIS Users at Lower Sioux

Initial setup and data loading will be in the OE, and the
OE employee will receive introductory training on  the
system. Because the OE is formed through a coopera-
tive agreement between two tribes, the Upper Sioux and
the Lower Sioux, the  OE is centrally located between
the reservations. The system will probably be relocated
to the Lower Sioux Community Center within 1 year. A
tribal government employee will receive advanced GIS
training and be available for all tribal government depart-
ments and businesses.


In addition  to tribal contributions, funding has come
through several sources and joint agreements  with  the
tribe  and BIA, EPA, and ANA.


The GDSC supplies no-cost training to tribes. The Geo-
graphic Data Service Center 1995 Training Catalog (no
federal document number available) offers eight formal
courses repeatedly throughout the year, a 5-week intern
program, and a cooperative student program. Courses
are held at the  GDSC or by request at BIA field offices
and tribal locations.

The GDSC also produces the monthly The Service Cen-
ter Review (ISSN 1073-6190), a helpful compilation of
current issues, available resources, system  bugs, and
other items of interest to GDSC users.

Data Collection and Input

Data collection can be divided  into three  categories:
aerial photography, portable global positioning system
(GPS) data, and ARC/INFO export files created under
contracted studies.

Aerial Photography

Surface features  and  topography will be obtained  us-
ing  aerial photography reduced to GIS format, then
downloaded to the GDSC. Coverages will consist of 62
categories of features on a scale of 1  inch = 100 feet
with 2-foot contour intervals.

Global Positioning Systems

Use of a portable Trimble, Inc., GPS Pathfinder Pro XL
submeter GPS mapping system purchased with assis-
tance from an ANA grant will allow updating of surface
features and addition of nonsurface features as neces-
sary. It will also facilitate input of attribute data.

The GPS will also be used during field work by USGS
on the Lower Sioux hydrological  mapping project to
obtain  data that  otherwise would  not  be  put  into
ARC/INFO export file for any reason (i.e., it might not be
directly related to the project at hand or outside the
agreed upon data to be converted to ARC/INFO export
file form but nevertheless is of importance to the OE).
The alternative  is that this  type of information  never
makes it into the CIS and is lost.

ARC/INFO  Export  Files

Fortunately, most federal agencies that supply funding
to tribes for environmental work are well versed in CIS
applications and the need for CIS-ready data. The OE
now requires all information and mapping to be deliv-
ered as an ARC/INFO export file with  data registered to
a real world coordinate  system. Downloading of this
data to the GDSC mainframes allows for direct input of
the data. The OE has contemplated,  but not acted on,
conversion  of existing data for the CIS. This is an ex-
pensive and time  consuming process that must be
weighed in comparison with  recollecting the data. Ironi-
cally, the lack of reservation data therefore becomes a
benefit because time  consuming  and expensive data
conversion is unnecessary.

The Intertribal GIS Council

Information  gathering, networking,  and  addressing
uniquely tribal problems were some of the accomplish-
ments at the first annual meeting of the Intertribal GIS
Council (IGC) held in June 1994. Vendors as well as BIA
regional office and  GDSC  representatives answered
questions  and presented  panels. This annual confer-
ence is likely to become a major benefit to the tribe as
it continually develops the GIS.

The Future

As the tribal government becomes more familiarwith the
GIS, its uses, and advantages, recognized governmen-
tal needs  will  likely drive the development of further
coverages. The  OE  also expects to access existing
governmental data of importance to the tribe in an effort
to expand the GIS database and is  actively seeking
sources of such information.

Philosophical Caveat

Albert Einstein stated that, "The significant problems we
face cannot be solved at the same level of thinking we
were at when we created them." Some assume that GIS
is the next level of reasoning in the environmental pro-
fession because we can accomplish tasks more quickly,
more efficiently, with more variables accounted for, and
beyond what we  could have hoped to accomplish prior
to GIS.

Essentially, what we have  gained is speed  and the
capacity to include additional data, which is not what
Einstein was referring to when he spoke of  the next
level. Wisdom,  in the sense  of a  higher  level  of
understanding,  is  the  necessary ingredient to  the
solution of current environmental  problems;  in other
words, movement beyond  the paradigm  that created
the problem. GIS may be the tool that pushes the envi-
ronmental professional to the next level of wisdom by
presenting the data and information  in a  manner that
allows the user to stand back and see more clearly on
a higher plane. But that level can be  found only within
the environmental professional  himself or herself and
not within  GIS.

Reach File 3 Hydrologic Network and the Development of GIS Water Quality Tools
                                        Stephen Bevington
  Water Quality Section, Division of Environmental Management, North Carolina Department of
              Environment, Health, and Natural Resources, Raleigh, North Carolina

The application of geographic information system (GIS)
tools to water quality management is limited by the lack
of geographically referenced data describing the surface
water environment. Ongoing efforts at the local, state,
and federal level are producing a multitude of GIS data
coverages describing land use/cover and relevant water
quality data files. As these data coverages become
available, water quality managers will need to develop
new analysis techniques to take advantage of the  vast
amount of geographically referenced data. A key step in
the development of analytical tools for  water quality
management will be the development and maintenance
of a coverage describing the structure and hydrology of
surface waters.

Reach File 3 (RF3) is one potential source of surface
water maps and topology  for the development of a
CIS-based  water quality  analysis tool. This paper de-
scribes a pilot project designed to examine the suitability
of RF3 as a network system for the collection, integra-
tion, and analysis of water quality data.

To be considered an appropriate water quality analysis
tool, RF3 should provide the following functions:

• Present  a working environment that allows users  to
  explore geographic relationships between surface water
  features, landmark features, and data coverages.

• Allow  users to select specific stream segments, in-
  cluding all points  upstream and  downstream of a
  given  point.

• Provide tools to assist users in partitioning water quality
  databases into hydrologically meaningful subsets.

Reach File 3

RF3  is a hydrographic database of the surface waters
of the United States. The database contains 3 million
river reaches mapped at 1:100,000 scale. The source
for RF3 arcs were digital line graphs (DLGs).

Attribute data for RF3 arcs include the major-minor DIG
pairs, stream name, water-body type, stream order, and
a unique identifying reach number. The  unique reach
numbers are structured in such  a way as to provide a
logical  hydrologic framework. Reach numbers can be
used to sort the database for all reaches in any specified
watershed  or  locate  all upstream  or downstream

The U.S. Environmental Protection Agency (EPA) origi-
nally designed RF3 as a tabular data set. It evolved into
a GIS data coverage, and EPA and the U.S. Geological
Survey (USGS) will  likely maintain it as a surface water
mapping standard. At present, RF3 as a GIS data layer
is not widely used for water quality applications.

RF3 Pilot Study: Uppeffadkin River Basin

The Upper Yadkin River basin (USGS h03040101) was
selected  to test  RF3 water quality applications (see
Figures 1 and 2). The Upper Yadkin  was chosen be-
cause of the availability of water quality and stream flow
data layers in that area. Also, the Upper Yadkin RF3 file
contained arcs depicting lakes and double-line rivers as
well as simple stream networks. These two-dimensional
water features present interesting complications to net-
work routing and  path-finding.
Figuie 1.  The Upper Yadkin River watershed, North Carolina
        and Virginia.

Figure 2  RF3 hydrography for the UppeYadkin River basin.

Two forms of point source data were used in the study:
National   Pollutant  Discharge   Elimination  System
(NPDES)  wastewater discharge  points and  USGS
gages. The NPDES coverage includes data  on the per-
mit limits  such as daily flow, dissolved  oxygen, bio-
chemical oxygen demand  (BOD), and ammonia. The
USGS gage coverage includes data  on several flow
statistics for each USGS gage in the basin. Both data
layers contain information about the location of the site
and stream with which it is associated.

Coverages of counties and cities were also made avail-
able for geographic orientation.

Preparing the Network

The original RF3 file received from the USGS  had sev-
eral topological issues that needed to be  addressed
before RF3 could function as  a stream network. First,
not all arcs were connected to each other (see Figure 3).
The ARC/INFO command TRACE was  used to  select all
connected arcs.  This  revealed three  major blocks of
connected arcs and many isolated arcs. The three major
blocks were easily connected in ARCEDIT by extending
the main tributary links between the blocks.  Processing
of the isolated  arcs was not pursued for  this study.
Complete  processing of arcs for this RF3 basin would
not be difficult or time consuming, with  the  possible
exception  of the many arcs surrounding  the  lake.  A
functional  network encompassing a high percentage  of
the arcs was not difficult to achieve, however.

The second network issue concerned the direction of the
arcs. RF3 has all arcs oriented toward the top of the
watershed, with the exception of one side of double-line
streams. Arcs  that make up double-line streams are
oriented up one side of the double-line section and down
the other (see  Figure 3). Clearly, this complicates rout-
ing. To allow for accurate downstream routing,  arcs on
the downward-facing side  of the stream were flipped
using ARCEDIT. With all arcs in the  network facing
upstream,  most hydrologic routes can be traced. Given
the network system alone, upstream routing from dou-
ble-line streams does not function properly, ignoring all
tributaries  on one side of the double-line stream.
Double-Line Stream Routing

Many possible solutions  exist for the problems caused
by double-line streams.  Some involve  improving the
network (e.g., by adding center-line arcs down the mid-
dle of double-line streams). This would involve not only
adding arcs but establishing conductivity with all tribu-
taries.  This option will involve significant topological
changes to RF3.  To  maintain compatibility with other

Figure 3  Original conductivity of RF3 hydrograph

RF3 and DIG sources, this option should be considered
only as part of a major RF3 upgrade.

At the other  end  of the technological spectrum, one
could simply instruct users  to  watch for double-line
streams and  select arcs from both sides of the river.
Users may have trouble with this  option, however, if they
are not working at an appropriate scale to easily differ-
entiate between double- and single-line streams.

A third option is to program an arc  macro language
(AMI) to  check for double-line  streams  and  run  up-
stream traces from both sides of the stream. The diffi-
culty in this method is  to find the appropriate starting
place on both banks. The algorithm developed to do this
goes as follows:

• Select stream segment and trace upstream. (Results
  in incomplete trace.)

• Find the minimum segment and mile of selected dou-
  ble-line streams.

• Unselect all double-line streams below minimum seg-
  ment and mile.

• Add to selection  all non-double-line streams.

• Trace from original point both upstream and down-
  stream.  (Results  in completed upstream trace.)
Results and Conclusions

AMLs and menus were written that can  perform  up-
stream and downstream traces on the RF3 stream net-
work  and select  data  points within  500 feet of  the
stream. Lists of attributes can be returned to the screen.
This system is easy to use and can be used to quickly
identify  general watersheds  and water  quality data
points. An AMI can be used to trace upstream from a
double-line stream given only one point on the stream
(see Figure 4). The success of these methods suggests
that two-dimensional surface water features can be suc-
cessfully integrated into RF3 water quality analyses.

This system could be further developed to support poly-
gon analysis using the ARC command BUFFER. Other
developments could include the procedures to write se-
lected attributes to files and increased flexibility for the
screen environment and outputs.

This pilot project demonstrates only a few of the poten-
tial applications of RF3 to water quality. Success in this
pilot project suggests that RF3 is a potentially valuable
water quality analysis tool. It may also be a valuable tool
for demonstrating the results of water quality analyses
to managers or the public.

Because RF3 will require some processing before network
algorithms can  be  run, it is important to plan for the inte-
gration of RF3  into other CIS tools and data coverages.

                                                        ,, , Mj',< ,   ',  V . ^
                                                         *> , V ^ • '  "   \ j
                                                          i. -'-  <-; v  \  '"%
                                                          ' Jj 'f "f    *
Figure 4  Upstream and downstream traces of RF3 hydrogrgph

Ongoing efforts to  update RF3 may address some of   proceed in a way that is compatible with ongoing efforts
these problems. If RF3 is to be developed into a produc-   to update  RF3  and the  development of  new data
tive water quality management tool, it is  important to   sources.

   Assessing the Long-Term Hydrologic Impact of Land Use Change Using a GIS-NPS

                            Model and the World Wide Web*
                                                                          JON HARBOR2
                                                                         BERNIE ENGELS
                                                                         KYOUNG.J. LiM3
                                                                            DON JONES3
 This paper covers the contents of two presentations in the Diffuse Source session: "Assessing Long-
 Term Impact of Land Use Change on Runoff and Non-Point Source Pollution Using a GIS-NPS Model"
 and "A Web-based CIS Model for Assessing the Long-Term Impact of Land Use Change (L-THIA CIS
 WWW): Motivation and Development".
 Oak Ridge National Laboratory, PO Box 2008, MS 6237, Oak Ridge, TN 37831-6237; Phone: (423) 241
 9272; Email: bhaduribl@ornl.gov
: Department of  Earth  & Atmospheric Sciences, Purdue University, West Lafayette, IN 47906-1397;
 Phone: (765) 494 9610; Email: jharbor@purdue.edu
1 Department of Agricultural & Biological Engineering, Purdue University, West Lafayette, IN 47907-1146;
 Phone: (765) 494 1198; Email: engelb@ecn.purdue.edu. kjlim@ecn.purdue.edu,


Assessment of the long-term hydrologic impacts of land use change is important for optimizing
management practices to control runoff and non-point source (NPS) pollution associated with
watershed development. Land use change, dominated by an increase in urban/impervious
areas, can have a significant impact on water resources. Non-point source (NPS) pollution is the
leading cause of degraded water quality in the US and urban areas are an important source of
NPS pollution. Despite widespread concern over the environmental impacts of land use
changes such as urban sprawl, most planners, government agencies and consultants lack
access to simple impact-assessment tools that can be used with readily available data. Before
investing  in sophisticated analyses and customized data collection, it is desirable to be able to
run initial  screening analyses using  data that are already available. In response to this need, we
developed a long-term hydrologic impact assessment technique (L-THIA) using the popular
Curve Number (CN) method that makes use of basic land use, soils and long-term rainfall data.
Initially developed as a spreadsheet application, the  technique allows a  user to compare the
hydrologic impacts of past, present  and any future land use change. Consequently, a NPS
pollution module was incorporated to develop the L-THIA/NPS model.

Long-term daily rainfall records are  used in combination with soils and land use information to
calculate  average annual runoff and NPS pollution at a watershed scale. Because of the
geospatial nature of land use and soils data, and the increasingly widespread use of GIS by
planners, government agencies and consultants, the model is linked to a Geographic
Information System (GIS) that allows convenient generation and management of model input
and output data, and provides advanced  visualization of the model results. Manipulation of the
land use layer, or provision of multiple  land use layers  (for different scenarios), allows for rapid
and simple comparison of impacts. To  increase access to L-THIA, we have begun development
of a WWW-accessible version of the method. Using databases housed on our computers, the
user can select any location in the US and perform L-THIA/NPS analyses.

In this paper we present applications of the WWW-based L-THIA/NPS and L-THIA/NPS GIS
model on the Little Eagle Creek (LEG)  watershed near Indianapolis, Indiana. Three historical
land use scenarios for 1973, 1984, and 1991 were analyzed to track land use change in the
watershed and to assess the impacts of land use change on annual average runoff and NPS
pollution from the watershed and its five sub-basins.  Comparison of the  two methods highlights
the effectiveness of the L-THIA approach in assessing the long-term hydrologic impact of urban


sprawl. The L-THIA/NPS GIS model is a powerful tool for identifying environmentally sensitive
areas in terms of NPS pollution potential and for evaluating alternative land use scenarios to
enhance NPS pollution management. Access to the model via the WWW enhances the  usability
and effectiveness of the technique significantly. Recommendations can be made to community
decision makers, based on this analysis, concerning how development can be controlled within
the watershed to minimize the long-term impacts of increased stormwater runoff and NPS
pollution for better management of water resources.

For decision makers, such as land use planners and watershed managers,  it is important to
assess the effects of land use changes on watershed hydrology. At present numerous
hydrologic models are available that focus on event-specific assessment and management of
hydrologic impacts of land use change. Traditionally the focus of these urban surface water
management models has been on the control of peak discharges from individual, high
magnitude storm events that cause flooding. Models such as those developed by the US Army
Corps of engineers (HEC-1, 1974), the US Department of Agriculture (TR-20, 1983; TR-55,
1986), and the US Environmental Protection Agency (Huber and Dickinson, 1988) are routinely
used in assessing how proposed land use changes will affect runoff quantity. Although
hydrologic impact assessment based on individual, high magnitude storm events is an
appropriate approach for designing runoff control facilities for reducing local flooding and
improving water quality, it is of limited use for attempts to understand the long-term hydrologic
impacts of land use change. However, it has been increasingly realized that there is a long-term
hydrologic impact associated  with land use change, and that this is dominated by runoff
generated from frequently occurring, smaller storm events rather than extreme,  high magnitude
storms (Harbor,  1994; McClinktocket al., 1995).

Realizing the importance of NPS pollution, over the last 25 years, models including SWMM
(Huber and Dickinson,  1988), AGNPS (Young et al., 1989), and WEPP (Nearing et al., 1989,
Flanagan and Nearing, 1995) have been created with capabilities to assess the impacts of NPS
pollution on runoff quality in addition to standard assessments of peak discharges. Because
NPS pollution from agricultural areas was originally identified as the major cause of water
quality degradation, most NPS pollution models focus on typical agricultural pollutants such a
sediment, nutrients (nitrogen and phosphorus), and organic compounds (pesticides and
herbicides). However, heavy metal  pollution from urban areas has recently been identified as a


leading cause of NPS pollution problems (Novotny and Olem, 1994) but estimation of heavy
metal pollution from urban areas with existing hydrologic/NPS pollution models is quite limited.

In assessing the long-term hydrologic impacts of land use change planners, developers, and
community decision makers usually avoid using the existing hydrologic models because these
models are too complex, data intensive, time consuming, expensive, and requires considerable
user-expertise (Harbor, 1994). To overcome the limitations of traditional hydrologic models, the
Long-Term Hydrologic Impact Assessment (L-THIA) model was developed as a user-friendly
tool for long-term runoff estimation (Harbor, 1994). L-THIA is built around the Natural Resources
Conservation Service's Curve Number  (CN) technique that is the core component of many
sophisticated hydrologic models (Williams et al., 1984; Young et al., 1989). Curve numbers or
CN values represent surface characteristics of a soil-land use complex. In L-THIA a long-term
(typically 30 years) daily precipitation record is used along with soil and land use information to
compute daily runoff for estimating annual average runoff. The model was initially developed as
a simple spreadsheet application (Harbor, 1994; Bhaduri et al., 1997). Subsequently a C
program was developed for the model to facilitate input data  handling and model application.
The L-THIA model was further expanded to L-THIA/NPS by adding a NPS pollution assessment
module. To enhance spatial data management, spatial analyses, and advanced visualization of
model results Geographic Information Systems (GIS) have been utilized. L-THIA GIS (Grove,
1997) and L-THIA/NPS GIS have been developed as customized applications of commercial
GIS software. Recently, a WWW-based version of the L-THIA/NPS model has been developed.
In the WWW-based implementation of the L-THIA/NPS model, the user provides land use and
hydrologic soil group information and L-THIA/NPS is run using long-term daily precipitation data
queried  from an ORACLE database. By determining and comparing the average annual runoff
depths and NPS pollutant loads for land use scenarios from different time periods, it is possible
to assess the absolute and relative changes in runoff and NPS pollution due to land use

Structure of L-THIA and L-THIA/NPS Model
The L-THIA model was originally developed as a preliminary hydrologic impact assessment tool
that focused  on predicting the percent increase in annual average runoff from a watershed due
to some land use change represented by a change in the CN value for the watershed. The
model utilizes a lumped parameter design to minimize model complexity and to reduce the


expense and time involved in data collection. For a watershed with multiple land use categories
and/or sub-watersheds, the model can be applied as a lumped (composite CN) as well as a
distributed (distributed CN) approach (Grove, 1997, Grove et al., 1998).

Runoff Calculation:
Daily runoff is calculated using the USDA NRCS Curve Number (CN) method for a daily
precipitation data set spanning many years (typically 30 years). The  CN method is an empirical
set of relationships between  rainfall, land use characteristics, and runoff depth. CN values,
ranging from 25 to 98, represent land-surface conditions and are a function of land use,
hydrologic soil group (or soil  permeability), and antecedent moisture condition (USDA SCS,
1986). The basic equations used in the CN method for standard or average conditions are:
       R=(P   0.                R = OforP<0.2S                       (1)
            (P + 0.8S)
           (CN )

       R = runoff depth (inches)
       P = precipitation depth (inches)
       S = potential maximum retention (inches)
       CN = Curve Number

Antecedent Moisture Condition (AMC) and CN Variation
The effect of antecedent rainfall and associated soil moisture conditions has long been
recognized as a primary source of variability in runoff determination. To account for this, the
Natural Resources Conservation  Services (NRCS) introduced the concept of an antecedent
moisture condition (AMC), also referred to as antecedent runoff condition (ARC).

Three AMCs are defined as a step function of 5-day antecedent rainfall, and an AMC remains
constant for the specific range of antecedent rainfall values.  Definitions of growing and dormant
seasons are not easily available and to keep calculations simple and consistent, growing and
dormant seasons were assumed  to begin on April 15 and on October 15 of any year,


respectively. CN values for AMC 1 and 3 are determined by the following relationship as
described in NEH-4 (USDA SCS, 1985).
CN, = •
CN, = -
                   10 + 0.13CAT,
[where, CN!, CN2, and CN3 represent CN values for AMC 1, 2, and 3 respectively.]
Runoff analyses were performed with the CN values for AMC 1 , 2, and 3 where AMC is a step
function of 5-day antecedent rainfall (Table 1).

5-day Antecedent Rainfall (
Dormant Season
< 13
Growing Season
    Table 1: Criteria for determination of Antecedent Moisture Conditions (SCS, 1972).

NPS Pollution Calculation:
L-THIA/NPS GIS: Pollutant Build-up and Washoff
The most common urban NPS pollutant estimation technique in current deterministic water
quality models including STORM and SWMM is the pollutant "buildup-washoff" function (Huber,
1986; Barbe et al., 1996). "Buildup" refers to  all dry-weather processes that lead to
accumulation of solids and associated pollutants on a watershed surface which are "washed off"
during subsequent storm events. In developing a NPS pollution sub-model for L-THIA, it was
assumed that pollutants accumulate on a land surface as a linear function of time.
           j = (number of days) x
                              [U = accumulation rate for pollutant i (mass/area/day);
                              Mi = Ultimate pollutant accumulation (mass/area);]
For this study, daily accumulation rates of solid particles (dust and dirt) for urban land uses (low
and high density residential, industrial, and commercial) were adopted from the SWMM manual.
Daily dust and dirt accumulation values are reported as a function of curb (road) density, and


thus road densities for the urban land uses were required to produce dust and dirt accumulation
values as mass/area. Although road density values for different urban areas have been reported
in the literature, for this study values of road density for Tulsa, Oklahoma (Heany et al., 1977)
were chosen as representative of the Indianapolis, Indiana area where the model was applied.
Daily build up values of pollutants on non-urban land uses (agricultural, grass/pasture, and
forest) could not be found in literature and were not included in the daily simulations of NPS
pollution analyses. However, annual average loading rates for non-urban land uses were taken
from literature and used to calculate NPS pollution in the GIS analysis. The NPS pollutant
loading values are reported in Table 2.
Total N
Total P
Annual average loading rate (kg/ha/year)
  Table 2. Annual average pollutant loading values used in L-THIA/NPS GIS simulations.

For the washoff function, a non-linear washoff equation was used. The washoff relationship is
an exponential function of the runoff depth. This approach has been successfully used in
numerous studies (Haith and Shoemaker, 1987; Dikshit and Loucks, 1996) and was utilized in
the NPS simulation for the L-THIA model because daily runoff depths are calculated in the
runoff sub-model which then can be used in the washoff function. The washoff function is
expressed as:

       wk,,= 1  -exp(-1.81 a,,,)
               M = fraction of the pollutant mass removed from the land use k on day t;
             Qkt = runoff from land use k on day t (cm);
WWW-based L-THIA/NPS: Event Mean Concentration (EMC)
In the Web-based version of L-THIA/NPS, Event Mean Concentration (EMC) data were
introduced to predict NPS pollutants for non-urban areas as well as urban areas (Baird and


Jennings, 1996). The EMC data used were compiled by the Texas Natural Resource
Conservation Commission (Baird and Jennings, 1996). Numerous literature and existing water
quality data were reviewed by Baird and Jennings (1996) with respect to eight categories of land
use and several parameters.  Land use categories defined were (1) industrial; (2) transportation;
(3) commercial; (4) residential; (5) agricultural cropland (dry land and irrigated); (6) range land;
(7) undeveloped/open; and (8) marinas. The total pollutant load for a NPS pollutant divided by
runoff volume during a runoff event yielded the Event Mean Concentration for that pollutant.
EMCs should be reliable for determining average concentrations and calculating constituent
loads (Table 3).
NPS Pollutant
Total Nitrogen (mg/L)
Total Phosphorus
Total Lead ( g/L)
Total Copper ( g/L)
Total Zinc ( g/L)
Land use classification
           Table 3. Event Mean Concentration for each land use classification
                               (Baird and Jennings, 1996)
The L-THIA model has been linked with Arc/INFO® GIS software as a GIS application (Grove,
1997). The ArcView® GIS software was chosen for L-THIA/NPS GIS application because
ArcView® is the dominant desktop GIS, and it has a friendlier graphical user interface than
Arc/INFO®. The GIS application is implemented through a linked-model approach that utilizes
both the graphical and spatial data handling capabilities of a GIS as well as the speed and
flexibility offered by a standard executable program. The required input data is initially selected
in the GIS before the L-THIA/NPS executable is called by the GIS. The executable calculates
the annual average runoff depths for all land uses and annual average dust and dirt amounts
(kg/km2) for all CN values for urban land uses (low density residential, high density residential,
industrial,  and commercial). These calculations are based on daily rainfall data spanning many
years. The output file created by the L-THIA/NPS executable is then read back into the GIS and
used to produce final results.


WWW-Based L-THIA/NPS Setup
A user-friendly L-THIA WWW interface was developed using Java/Java Script, HTML, and CGI
scripts (http://pasture.ecn.purdue.edu/~sprawl/lthia2). This interface provides easy access to the
model and potentially improves understanding of the results through graphical representation.
Figure 1 shows the L-THIA/NPS WWW interface.
                                                            Help |
               Home Search Netscape
                            c*  >*    m
                            Print  Security    Stop
  . jj* Bookmarks .j^. Location: Ihttp; //danpatch. ecn.purdue. edu/"sprawl/LTHIA2/
                                      i\ flgi" What's Related
   J TV 8, Magasine |_J CLASS :_J FVS i_j Computer _j[ Models fj GSH i_j KWNU _J TECH ,J[ Cool_Pages
   ' Home
  |l> Input
  I/ Output
   • Case Studie
   • Related Site
                                  Urrteil States
                                  Enviranmental Protection
(Long Term Hydroloqic Impact Assessment) WWW
               • Scenario
               Name :
               • Area Units
• State :

• County:
• Hydrologic
Soil Group
                              Tippecanoe  -• I
                     LAND USE
                                          VEAR1  YEAR2 YEARS
                                          Iteo   : Ibo    ! Eo
                    HD Residential  -

           i Mtp :/A^ww. epa.gov/region5/
                                                   •^ a
          Figure 1. LTHIA/NPS WWW Interface.
Depending on the location the user selects, weather data for the nearest weather station are
queried from the database and reformatted for the L-THIA run. The user selects one of the eight
land use classifications, hydrologic soil group information and provides the area for this
combination for each time step the L-THIA/NPS WWW system is to be run. Tables, bar charts,
and pie charts for runoff and NPS pollution are generated on the fly for display in the user's
WWW browser. The tabular output provides all  information the user provided in the input


interface, the Curve Number, runoff depth, and runoff volume for each time step. Bar graphs
provide runoff depth, runoff volume, total volume, average runoff depth, and NPS pollution
information. Pie charts provide land use and runoff volume for each time of interest. LTHIA/NPS
WWW has several advantages over the traditional model and decision support system
approach: 1) It is accessible through the Internet using only a WWW browser, 2) Database and
GIS data are maintained at a single location, 3) All model users access the same version of the
model, and 4) All data are verified by the model  maintainer so errors due to input data can be

The L-THIA/NPS model was applied to the Little Eagle Creek (LEG) watershed, a rapidly
urbanizing watershed in the northwest section of Indianapolis, Indiana and its suburbs. The LEG
watershed is 70.5 km2 in size and consists of five smaller sub-basins (Figure 2). This watershed
has experienced extensive  urbanization over the past three decades. Land uses ranging from
non-urban natural grass and forested areas and agricultural areas to typical urban residential,
commercial,  and industrial categories exist in the LEG watershed.

As part of a long-term hydrologic impact assessment study (Grove, 1997), digital land use data
were generated from LANDSAT satellite imagery (80m resolution Landsat Multi-Spectral
Scanner imagery) for 1973, 1984, and 1991 and these three images represented temporal
changes in land use in the watershed. In this study, these  land use coverages along with the
Soil Survey Geographic (SSURGO) soils data (1:20,000) were used to analyze the long-term
impact of land use change on runoff and non-point source pollution. Only hydrologic soil groups
B and C are present  in the watershed. The watershed and sub-watershed boundaries were
delineated from a Digital Elevation Model (DEM) using the Arc/INFO GRID module (Grove,
1997). Curve Numbers ranged from approximately 60 to 97 for all sub-basins in the watershed.
Six land use  categories were delineated using ERDAS Imagine software and were used in L-
THIA/NPS GIS simulations. These areas of these land use categories and hydrologic soil B and
C groups were used  in the WWW L-THIA/NPS simulations.

   Figure 2. Location of the Little Eagle Creek watershed.
Land Use Change
There is a significant increase in urban land uses between 1973 and 1991 with the majority of
the changes taking place between 1973 and 1984 (Table 4). Grouping agricultural, forest, and
grass/pasture as non-urban and low density residential, high density residential/industrial, and
commercial as urban land uses, 49.3%, 63.5%, and 68.1% area of LEG watershed was urban in
1973, 1984, and 1991 respectively. Thus, there was a 14.2% increase in urban land uses
between 1973 and 1984 and a 4.6% increase in urban areas between 1984 and 1991. The
increase in urban land uses is not uniformly reflected in all the urban land use categories.

Land Use
HD Residential /
LD Residential
Area (km2)
% Change in Individual Category
Table 4. Land use distributions in Little Eagle Creek watershed for 1973, 1984, and 1991.

For individual land use categories, high density residential and commercial areas show
tremendous increase in the watershed while low density residential areas show a 28.4%
decrease between 1973 and 1984 and a 2.6% increase between 1984 and 1991. The initial
decrease in low density residential areas is possible conversion of low density residential to high
density residential areas. The increase in urban areas is followed by an equivalent decrease in
non-urban areas. However, all the non-urban land uses decrease at the same rate. Forested
areas show the greatest loss with a 62.6% decrease, followed by grass/pasture with a 28.8%
decrease between 1973 and 1991. Agricultural areas show minimum change (14.7% decrease)
during the same time period.

Impact of Urbanization on Annual Average Runoff and NPS Pollution
L-THIA/NPS analyses were performed to assess the impact of land use change on average
annual runoff and NPS pollution for the LEC watershed. There are significant changes in
average annual runoff volumes and NPS pollution loads from the LEC watershed as a result of
land use change. The results from L-THIA/NPS GIS and L-THIA/NPS web-versions are
presented in Table 5 and Table 6 respectively. However, changes in runoff volume or NPS
pollution do not have a simple or linear relationship with land  use change.

Change in runoff and NPS pollution using L-THIA/NPS GIS simulation:
Runoff (m3)
Nitrogen (kg)
Lead (kg)
Copper (kg)
Zinc (kg)
% Change
1973 to
1984 to
1973 to
Table 5. Average annual runoff volume and NPS pollution from LEG watershed using L-
        THIA/NPS GIS that uses daily pollutant build-up and washoff functions for
        pollution calculation.
Change in Runoff and NPS pollution using WWW L-THIA/NPS simulation:
Runoff (m3)
Nitrogen (kg)
Lead (kg)
Copper (kg)
Zinc (kg)
% Change
1973 to
1984 to
1973 to
Table 6. Average annual runoff volume and NPS pollution from LEG watershed using
        web-based L-THIA/NPS that uses Event Mean Concentrations (EMC) for pollution
The annual average runoff volumes predicted by L-THIA/NPS GIS are approximately half of
those predicted by the web-version of the model. This is primarily because two different sets of
daily precipitation data were used for the two simulations and the one used for the web version
had several large storm events that produced significantly more runoff.  However, considering
the relative change in runoff volume, very similar results were obtained  from both simulations.
The amounts of urban or impervious areas dominantly control the volume of runoff produced
from a watershed. For example, in  LEG watershed, 87% of the total runoff volume (81% with
web-based L-THIA/NPS) was produced from urban areas that occupied only 49% of the total
watershed area in 1973. In 1984 and 1991, urban areas occupied less than 70% of the total
watershed area but contributed over 93% of the annual average runoff volume (over 90% with


web-based L-THIA/NPS). Percent increase in average annual runoff volume is greater between
1973 and 1984 than between 1984 and 1991 because a greater percentage of non-urban land
use is changed to urban (more impervious) land use during the former time interval (Figure 3).
     60.00% -
     40.00% -
     20.00% -
                             L-THIA/NPS GIS
                      WWW L-THIA/NPS
Figure 3. Changes in annual average runoff and NPS pollution from the Little Eagle Creek

For the NPS pollutants, the relative change in annual average NPS pollution from LEG
watershed is not only controlled by the nature of land use change but also by the nature of the
pollutants. The time period between 1973 and 1984 experienced a much greater amount of
urbanization than the time period between 1984 and 1991. This pattern of land use change is
also reflected in changes in average annual runoff volume and metal pollution from the LEG
watershed. However, total pollutant loads predicted by web-based L-THIA/NPS are roughly


higher by an order of magnitude than those predicted by the L-THIA/NPS GIS model. This
difference can be attributed primarily to the different methods of pollution calculations by the two
simulations and also the different concentration values of the pollutants used. Using EMC
values in the web-version of the model, two different days with the same amount of runoff will
produce the same pollutant loads. In the GIS version, that uses pollutant build-up and washoff
functions, two similar runoff events can produce significantly different pollutant loads depending
upon masses of pollutants that accumulated before those two runoff events. Moreover, Bhaduri
(1998) showed that more than 90% of the days in the study area are dry (AMC 1), and thus
before any runoff event there will be a significant amount of pollutant accumulated on the

One significant difference between the two approaches of pollution calculations can be
observed in the predicted changes in nutrient pollution. L-THIA/NPS GIS predicts decreasing
nutrient pollution with increasing urbanization in the watershed. Nitrogen and phosphorus are
dominantly  produced from agricultural areas. Moreover, the other non-urban land uses (forest
and grass/pasture) show significant decrease between 1973 and 1984. Thus, this small change
is nutrient loading between 1973 and 1984 is most plausibly related to the small reduction in
agricultural  area in the watershed. On the contrary, the web-version of L-THIA/NPS predicts
changes in  nitrogen and phosphorus loads that conform to the increasing trend in urbanization.
Because nitrogen and  phosphorus are typically identified  as non-urban pollutants, it might be
assumed that conversion of non-urban land uses to urban areas would significantly reduce
nutrient pollution from a watershed. Our analyses on LEG watershed indicate that, between
1973 and 1991, a conversion of 19%  areas from non-urban to urban land uses results in annual
average nitrogen and phosphorus loads being increased  by about 60%. This is primarily
because there is only a small reduction of agricultural area and a large increase in urban areas
in the watershed. Although urban areas produce nutrients at a much lower rate than non-urban
areas, but increases in urban land uses produce runoff at a significantly higher rate and thus,
the web-version predicts very high  nutrient loads.

Heavy metals, such as lead, copper, and zinc, are considered "urban" pollutants because urban
land uses contribute a  major portion of the metal pollution from a watershed. For the LEG
watershed using L-THIA/NPS GIS, we found that only 49%  of the area had urban land uses but
they contributed 98% of total lead load, 92% of total copper load, and 93% of the total  zinc load
from the watershed in 1973. However, for individual metal pollutants, this 18% increase in urban


areas (or an equivalent decrease in non-urban areas) between 1973 and 1991 results in 76.5%,
56.2%, 67.8% increase in lead, copper, and zinc loads from the watershed respectively.
Predictions of changes in metal pollution from web-based L-THIA/NPS simulation were similar
to those from the GIS version (Figure 3).

Assessment of the long-term hydrologic impacts of land use change is important for optimizing
management  practices to control runoff and non-point source (NPS) pollution from urban
sprawl. The L-THIA/NPS model uses the popular curve number technique and empirical
relationships between land uses and pollutant accumulation and wash off processes to estimate
the relative impacts of land use change on annual average runoff and NPS pollution. L-
THIA/NPS uses readily available data to overcome the difficulties of long-term modeling by
existing hydrologic models because of their complexity and extensive input data requirements.
Moreover, most traditional hydrologic/NPS pollution models do not emphasize the changes in
loads of typical urban pollutants such as heavy metals, which can be addressed by L-
THIA/NPS. The model is linked to a GIS to  enhance input data generation, data management,
and advanced visualization of model results. The GIS version computes NPS pollution using
daily pollutant build-up and washoff functions. A World Wide Web based version of the model
has also been developed that provides easy access to the model through the Internet. This
web-based version of the model uses Event Mean Concentrations (EMC) of pollutants for
predicting NPS pollution.

L-THIA/NPS was applied to the Little  Eagle Creek (LEG) watershed, an urbanizing watershed
near Indianapolis, Indiana, to provide estimates of changes in annual average runoff volumes
and NPS pollution loads for three time periods: 1973, 1984, and 1991. Increases in urban land
uses were much higher between  1973 and  1984 than between 1984 and 1991. Non-urban land
uses, particularly agricultural areas, are the dominant sources of nutrient (nitrogen and
phosphorus) pollution but the majority of the metal pollution is contributed from urban areas.
Overall, increasing urbanization resulted in  increases in annual average runoff volume and
metal loads. The L-THIA/NPS GIS simulations predicted decreases in nitrogen and phosphorus
loads from the LEG watershed. However, the web-based version of the model indicated
increases in nutrient pollution with increasing  urbanization. This difference can be explained by
the two different methods of pollution calculations by the GIS and web-based versions of the
model. This difference in pollution calculation is also reflected in the absolute values of pollution


predicted by the two versions. However, considering relative change in runoff and NPS pollution
from urbanization, both simulations indicate a very similar trend and direction of changes in NPS
pollutants for the Little Eagle Creek watershed.

L-THIA/NPS GIS is a simple and user-friendly model that makes it attractive for applications to
other watersheds. Although L-THIA/NPS GIS is designed to run with easily available data, such
data is often not readily available for most of the watersheds. Thus,  compilation of model input
data sets through field experiments for a variety of watersheds characterized by different
geography, climate, and land uses will greatly enhance model applications and performance in
a wider range of watersheds. These field-measured data sets should be used to calibrate the L-
THIA/NPS model and validate the model predictions. Future work should also explore the
sensitivity of the L-THIA/NPS model to the spatial and temporal scales of input data. Work in
progress is aimed at allowing a user to access a modified web-based L-THIA/NPS model that
will take advantage of GIS functionality in the analysis. In this modified web-version, the users
will not only be able to access the model through a web-browser, but will also be able to select
or define a watershed using system-supplied  maps, and then run L-THIA/NPS analyses run
using land use, soils and climate databases stored on our server. The user will then be able to
manipulate the GIS land use data in the browser environment or on a remote computer, and run
multiple L-THIA analyses to compare hydrologic impacts from different land use scenarios.

The authors would like to thank Dr. Darrell Leap and Dr. Darrell Norton of Purdue University for
their valuable suggestions, comments, and review of the manuscript. Daniel Pack of Oak Ridge
National Laboratory provided valuable help with some of the graphics. Funding for this work was
provided in part by the Purdue Research Foundation and the United States Environmental
Protection Agency.


Baird, C., and M. Jennings. 1996. Characterization of Nonpoint Sources and Loadings to the
       Corpus Christi Bay National Estuary Program Study Area, Texas Natural Resource
       Conservation Commission.

Barbe, D. E., J. F. Cruise, and X. Mo.  1996. Modeling the buildup and washoff of pollutants on
       urban watersheds. Water Resources Bulletin, Vol. 32, No. 3, p. 511-519.

Bhaduri, B. 1998. A geographic information system based model of the long-term impact of land
       use change on non-point source pollution at a watershed scale. Unpublished Ph.D.
       dissertation, Purdue University, West Lafayette, Indiana 47907-1397.

Bhaduri, B., M. Grove, C. Lowry, and J. Harbor.  1997. Assessing the long-term hydrologic
       impact of land use change. Journal of the American Water Works Association. Vol. 89,
       No.  11, pp. 94-106.

Flanagan, D. C. and Nearing, M. A., 1995. USDA-Water Erosion Prediction Project: Hillslope
       profile and watershed model documentation. NSERL Report No. 10. West Lafayette,
       Indiana: National Soil Erosion Research laboratory, USDA-ARS.

Grove, M. 1997. Development and application of a GIS-based model for assessing the long-
       term hydrologic impacts of land-use change. Unpublished MS Thesis, Purdue University,
       West Lafayette, IN 47907-1397.

Grove M. and J. Harbor, and B. Engel. 1998. Composite versus distributed curve numbers:
       effects on estimates of storm runoff depths. Journal of the American Water Resources
       Association. Vol. 34, No. 5, pp. 1015-1023.

Haith, D. A. and L. L. Shoemaker. 1987. Generalized watershed loading functions for stream
       flow nutrients. Water Resources Bulletin, Vol. 23, No. 3, pp. 471-478.


Harbor, J. 1994. A practical method for estimating the impact of land use change on surface
      runoff, groundwater recharge and wetland hydrology. Journal of American Planning
      Association, Vol. 60, pp.91-104.

Heany, J. P., W. C. Huber, M. A. Medina, Jr., M. P. Murphy, S. J. Nix, and S. M. Haasan. 1977.
      Nationwide evaluation of combined sewer overflows and urban stormwater discharges -
      Vol.  II: Cost assessment and impacts. EPA-600/2-77-064b (NTIS PB-266005),
      Environmental Protection Agency, Cincinnati, Ohio.

Huber, W. 1986. Deterministic modeling  of urban runoff quality. In Torno, H. C., Marsalek, J,
      and  Desbordes, M. (Eds.)  Urban  Runoff Quality. Springer-Verlag. 893 p.

Huber, W. C., and R. E. Dickinson. 1988. Storm Water Management Model, Version 4: User's
      Manual. US Environmental Protection Agency, Report No. EPA-600/3-88-001a. 569 pp.

McClintock, K., J. Harbor, and T. Wilson. 1995.  Assessing the hydrologic impact of land use
      change  in wetland watersheds, a case study from northern Ohio, USA. In: McGregor, D.
      and  Thompson, (eds.) Gee/morphology and Land Management in a Changing
      Environment, pp. 107-119.

Nearing, M. A., Foster, G. R., Lane, L. J., and Finkner, S. C.,  (1989). A process-based soil
      erosion  model for USDA-water erosion prediction project technology. Transactions of the
      ASAE, Vol. 32, No. 6, pp.  1587-1593.

Novotny, V. and H.  Olem. 1994. Water Quality.  Van Nostrand Reinold. New York. 1054  pp.
U.S. Army Corps of Engineers, 1985. Flood  hydrograph package HEC-1. The Hydrologic
      Engineering Center, Davis, CA.

U.S. Department of Agriculture, Soil Conservation Service, 1983. Computer programs for
      project formulation - hydrology. Technical Release 20.


U.S. Department of Agriculture, Soil Conservation Service, 1986. Urban hydrology for small
      watersheds. Technical Release 55.

USEPA, 1983. Results of the National Urban Runoff Program. Vol. 1. USEPA, Washington,

Williams, J. R., Jones, C. A., and Dyke, P. T., 1984. The EPIC model and its application. Proc.
      ICRISAT-IBSNAT-SYSS Symposium on Minimum Data Sets for Agrotechnology
      Transfer, March 1983,  Hydrabad, India, p. 111-121.

Young, R. A., C. A. Onstad, D. D. Bosch, and W. P. Anderson.  1989. AGNPS: A non-point
      source pollution model for evaluating agricultural watersheds. Journal of Soil and Water
      Conservation Society, Vol. 44, No. 2, p. 164-172.

                    Ecological Land Units, GIS, and Remote Sensing:
                          Gap Analysis in the Central Appalachians
                          Ree Brannon, Charles B. Yuill, and Sue A. Perry
                        West Virginia University, Morgantown, West Virginia

The gap analysis team in West Virginia is assessing the
state's natural communities  as part  of a nationwide,
comprehensive planning  effort. Underrepresented  or
unrepresented habitats represent gaps  in the present
network of conservation lands and conservation activi-
ties.  After identifying  these gaps,  we can assess
whether our current management direction will maintain
natural diversity and will prevent additional species from
being classified as threatened or endangered.

The  relationship  between vegetation and  ecological
variables serves as the basis for classifying  ecological
land  units. To characterize the ecological  land  units,
many layers  of physical data  can  be integrated in a
geographic information system  (GIS). Satellite imagery
and videography map existing conditions over the state.
The existing vegetation is classified to reflect physiog-
nomic and floristic elements to correlate with  vertebrate
and butterfly  habitat requirements.  This correlation of
vegetation and wildlife  habitat creates mappable habitat
types.  Analysis  of these  habitat types with  land-
ownership data indicates where the species-rich  areas
occur in the landscape and whether the most species-
rich areas are protected.


To respond to the urgency of habitat loss and its effect
on species diversity, scientists must implement a  meth-
odology for  rapid assessment and  documentation of
natural communities at a scale pertinent for  regional
management activities.

Geographic  information systems  (GIS) and remote
sensing support the development  of ecological land
classifications over large regions. CIS-based mapping
of ecological land classes allows users to combine and
display environmental variables for spatial modeling and
refinement of ecological land  units (1).
The Gap Analysis Project is a comprehensive planning
effort  for land  conservation in the United States. The
objective of the Gap Analysis Project is to identify spe-
cies, species-rich areas, and vegetation types underrep-
resented  or  unrepresented  in  existing biodiversity
management areas. Unprotected communities are the
gaps  in the  conservation strategy. The Gap Analysis
Project is not merely identifying communities with the
largest number of species; its ultimate goal is to identify
clusters  of habitats that link the greatest  variety of
unique species.

Local  areas with  considerable diversity of  habitat or
topography usually have richer faunas  and  floras (2).
Nature reserves, which incorporate a variety of habitats,
may be the best guarantee for long-term protection of
biodiversity. By protecting species-rich regions, we can
reduce the enormous financial and scientific  resources
needed to recover species on the brink of extinction.

The West Virginia Gap Analysis Project began in 1991.
The objective is to map existing vegetation and to use
that as the foundation to model potential distribution of
vertebrate and  butterfly  species. High cost  precludes
intensive  field  inventory and  monitoring of  wildlife.
Therefore, habitat modeling is critical to predict wildlife
species composition and potential ranges over the vari-
ous landscape types of West Virginia. Lastly, the vege-
tation  map will provide a record of the existing habitat to
use in monitoring changes due to human activities and
natural disturbances.

Pilot Study Area

Distribution of wildlife and  plant communities will be
modeled for the entire state. Initially, we will focus on  a
smaller pilot study area. This region includes approxi-
mately 50,000  hectares in  the  central Appalachian
Mountains and spans several physiographic provinces
and vegetative communities. Generally, soils  in the pilot
study  area are of two kinds: acidic soils that  develop a

clay horizon  from extensive leaching  over time and
younger soils that are found on steep slopes and where
environmental conditions, such as cold climate, limit soil
development (3, 4).

The vegetation  types include spruce-fir, oak-pine, high
elevation bogs,  northern hardwoods, Appalachian oak,
mixed-mesophytic,  open  heath barrens,  and   grass
balds. The mixed-mesophytic, the most diverse in West
Virginia, lies  primarily west  of this area,  but localized
stands do occur in the lower elevations. Cover types
within the Appalachian oak and mixed-mesophytic types
are not discrete and will be difficult to delineate.

The pilot study area includes  a variety of land uses, such
as residential, commercial, industrial, mining, and agri-
culture. Portions of the Monongahela National Forest in
this area are  the Fernow  Experimental Forest, and the
Otter Creek and Laurel Forks Wilderness Areas.


The following discussion describes the methods  formu-
lated and data compiled for the West Virginia Gap Analy-
sis  Project.

Describe Ecological Land Units With Existing

Davis and  Dozier (1) note  that a  landscape can be
partitioned by ecological variables, which contributes to
an ecological land classification. This process is applied
frequently to analysis and mapping of natural resources.
Davis and  Dozier classified vegetation  in California
based on the documented associations of vegetation
with terrain variables. They based this approach on the
assumptions that the arrangement of natural landscape
features is  spatially ordered by an  ecological interde-
pendence among terrain variables and that actual vege-
tation  is  a  reliable  indicator  of these ecological
conditions. Similar documentation exists for the distribu-
tion of vegetation types in West Virginia, and the gap
analysis team is proceeding  along a similar course.

West Virginia lies in two  major provinces, the Eastern
Broadleaf Forest and the Central Appalachian Broadleaf
Forest-Coniferous Forest-Meadow Provinces (5). Within
these provinces are several broad vegetation types. The
gradient diagram in Figure 1  (6) illustrates the range of
these types.  The  vertical  axis represents elevation  in
feet. Three vegetation types emerge distinctly along the
elevation gradient. The horizontal axis  is  not quite as
explicit. This gradient spans  moist, protected slopes to
dry, exposed ridgetops, and the vegetation types are
much less distinct. Drier  oak and pine types occur al-
most exclusively on exposed ridgetops.  The vegetation
types along the  horizontal  axis are the mixed mesophytic
forest association  of the  Allegheny and Cumberland
Mountains and can have 20 to 25 overstory and under-
story species per hectare in North America (7).

The distribution of the vegetation along gradients such
as elevation  and soil moisture  lends  itself  to  a  CIS
analysis. Physical data such as elevation and soil mois-
ture regime can be incorporated into a CIS. These lay-
ers of information can be manipulated graphically  or
mathematically to model the spatial distribution of vege-
tation types or to provide useful ancillary data for clas-
sification of satellite imagery (see Figure 2). Much of this
data is digital and can be used to substantially reduce
the time required to develop a database. To standardize
output,  members  of the  national  gap  team  have
specified ARC/INFO as the software to generate final

Classify Satellite Imagery To Create a Map of
Current Distribution

Remote sensing  provides an effective means to classify
forests, and the Gap Analysis Project has successfully
used it in the western United States (1,  8, 9).

The Gap Analysis Project is  using Landsat Thematic
Mapper imagery  in all states to standardize the baseline
information. The hypothesis  is  that the spectral data
from the imagery is related  to  the distribution of the
ecological land units and land use across the landscape.
The data include all spectral bands, except thermal, for
the entire state. The West Virginia project is  using two
seasons of data, spring  and fall. Temporal changes,
which record phenologic variation in the deciduous spe-
cies, enhance classification accuracy. The spectral reso-
lution   is  30-meter  pixels.  This  is   equivalent   to
approximately 1/6 hectare (1/2 acre). Our final product
will be a series of 1:100,000 maps. The minimum map-
ping unit is 40 acres.

The mountainous terrain in the Appalachian Mountains
offers disadvantages and advantages for using remotely
sensed data.  Irregular topography can cause inconsis-
tencies in the spectral data that diminish the classifica-
tion accuracy. Similar cover types may have different
spectral signatures; for instance,  if one  stand is in sun-
light and the other is shaded.  Also, phenology can vary
due  to  microclimatic influences.  Conversely, topo-
graphic features  influence the distribution of vegetation
types, and ancillary data, such as digital elevation mod-
els (OEMs), enhance classification  results of the im-
agery. The West Virginia gap analysis team selected the
following strategy for image classification.

1. Stratify the imagery using ecological units based on
   a hierarchical scheme. Bauer et al.  (10) found that
   an initial stratification of physiographic regions was
   necessary  to reduce  the  effect  of  broadscale
   environmental factors  caused  by  changes over
   latitude.  Therefore,  stratification  enhanced  the










Red Spruce
Forest ._. .,

Red Spruce -
Yellow Birch
Northern Hardwoods
--------- , Oak
._--"' " ', ' Forest /
x' Hemlock I . ,'' /
and ,' ,-' /' '
Hemlock- ' -' / ,-'*
Cove / Hardwoods ,' / ,-'
Hardwoods ' ''
White /'
Oak /
1 / ' '
, / ; i
/ Red Oak - / chestnut/
/ / White Oak / oak ''
Forest / ,'
,' \ ,' I Black Oak
Coves Flats Sheltered Open Slope
Canyons Draws/Ravines Slopes NE, E, S, W, NV





Figure 1.  Environmental gradients for vegetation.

   efficiency of  the training  data. An  interagency
   committee, including ecologists from the Monongahela
   National Forest, West Virginia Division of Forestry,
   and  geologists  from the  State  Geologic  Survey,
   generated  a draft map of physiographic regions.
   They delineated sections based on geomorphology
   and climate. Sections were divided into subsections:
   those most  typical of the section or those that are
   transitional,  or irregular, to the section. Figure 3 is a
   draft map of these sections and subsections in West

2. Classify stratified imagery using  the ancillary data.
   High resolution imagery has not been used widely in
   the eastern United  States, where  forests  are not
   homogeneous stands of relatively few tree species
   as they are  in the West. Researchers who classified
   eastern forests from satellite  imagery attempted to
   find the most distinctive spectral band combinations
   for discriminating cover types (10-13). One recent
   technique (14) uses  a nonparametric approach that
   combines all spectral and informational categories to
   classify imagery. We are testing a variety of methods
   such   as   nonparametric  processes,  traditional
   clustering techniques, and use of derived vegetation
   indexes to find the most successful method.

3. Assess   accuracy   with   random   plots   from
   videography. Videography will be  acquired in the
   spring  of 1995.  Aerial transects, which extend the
   length  of the state, will  be flown with approximately
   30-kilometer spacing. By regulating flight altitudes,
   the resolution  per frame can  be  captured at  1
   kilometer  per frame. About 7,000  frames will be
   collected, which make up a 3 percent sample of the
   state. About 2 to 3 percent of the videography frames
   will be field verified. With this strategy, we  will test
   the effectiveness of using videography, instead of
   intensive field plots, to verify classification of satellite
   imagery.  Areas  of special  interest,  which  the
   systematic transects may not capture, will require
   separate flights. The bulk of the videography  will
   provide  training   data  for  supervised   image
   classification. The  remaining frames will be  used to
   assess the accuracy of the classification.
                      ;>,  Streams
                      ."I;;-- Ecological Subregions
                      ";:>- Digital Elevation Models
                      i-* Soils
                              Agricultural, Tilled
                              High Elevation, Red
                              Low Elevation, Coniferous
                              Forest (Hemlock)
                              Open Grassland, Deep
                              Soil (Grass Bald)
                              High-Density Urban
Landscape Unit
Figure 2.  GIS and the development of ecological land units.

Figure 3.

                                  g]  Pilot Study Area
Physiographic regions of West Virginia and pilot
study area.
   In summary, 100 percent of the state will be classified
   using the  Thematic Mapper imagery.  Aerial video-
   graphy,  covering  approximately 3 percent  of the
   state, will help to verify the image classification, and
   2 to 3 percent of the videography will be verified from
   transects on the ground.

4. Determine sources of data for image classification.
   Due to the increasing interest in CIS, digital data are
   more readily available from a  variety of sources,
   such as the federal government, state agencies, and
   private companies. Acquisition of available data sets
   can  substantially  reduce the  time  and  cost  of
   database development.  Users  must bear  in  mind,
   however, that databases are developed with differing
   objectives and techniques, so  one must consider
   scale and standards  of production when  deciding
   which data sets are appropriate for project design.
   The  West Virginia gap  team  determined  that the
   following CIS coverages are  important for image

The U.S. Geological Survey's (USGS's) graphic infor-
mation  retrieval  and analysis  system (GIRAS) ear-
marked land use/land cover data. The classification was
done several years ago, and although these data are not
current, they  provide excellent information concerning
urban and agricultural land  use. Land-cover categories
represent Level II classifications from Anderson's (15)
system. The maps are produced at a 1:250,000 scale,
so they require few CIS operations to piece together a
regional coverage of land use.

The Southern Forest Experiment Station mapped U.S.
forestland using advanced very high resolution radiome-
ter (AVHRR) satellite imagery (16). This imagery is rela-
tively current (1991 to 1992), but the resolution is coarse
at 1 kilometer per pixel (100 hectares or 247 acres). The
classes are based on Forest  Inventory and Analysis
plots established by the U.S. Department of Agriculture
(USDA) Forest Service and Kuchler's  (17)  potential
natural vegetation types.  We  are  using the  maps  to
depict broad changes in forest type over a region, such
as the state of West Virginia. This coverage does not
show land use.

The eastern region of the Nature Conservancy has com-
pleted a draft of the classification of the terrestrial com-
munity alliances (18). The classification hierarchy is that
prescribed  by the national gap team, and as such, re-
flects physiognomic and floristic characteristics neces-
sary for  correlating  vegetation structure and floristic
composition with vertebrate  habitat requirements. The
descriptions include the range of alliances and charac-
teristic species of the overstory, understory,  and herba-
ceous  layer.  This  provides information  on  associated
species not detectable  by image classification.

The National Wetlands Inventory data are available digi-
tally. Maps have been digitized  at a scale of 1:24,000,
and the classification scheme  is from Cowardin  et al.
(19). Coverages come with attribute data for each poly-
gon, arc, or point as needed. A labor-intensive effort is
required to join the maps in a CIS for an area the size
of West Virginia, but the information will be invaluable
for masking water and forested wetlands  on the satellite
imagery.  The U.S. Fish and Wildlife Service  includes
detailed instructions for converting the data to coverages.

Field data,  much of it already digital, has  been  acquired
from many sources. Commercial timber companies pro-
vided data for timber stand  composition and age groups.
The USDA Forest Service ecologist conducted transects
throughout the Monongahela National Forest to charac-
terize ecological land units. Forest inventory and analy-
sis plots are also  available. We acquired these data to
verify videography classification.

The  USGS has digital  data  of  terrain elevations. The
West Virginia gap analysis team acquired 3  arc-second
OEMs as an additional band in the satellite imagery. We
will use these data  to generate coverages of slope,
aspect,  and  elevation  classes to  further  stratify the
physiographic regions of the state. This will increase the
accuracy of the classification.

The Soil Conservation Service created a statewide da-
tabase called STATSGO, produced  at  a scale  of
1:250,000.  For West Virginia, the map of the soil map-
ping units consists of approximately 450 polygons. Each
mapping  unit is an aggregation of soil components that
occupy a certain percentage of the mapping unit area.
The database is extensive and includes  information on
soil attributes such as soil taxonomy, soil chemistry, soil
structure, and interpretations for erodability  and wildlife
habitat. For an attribute such as soil temperature, each
component  has  an individual  value   so  that  each
mapping  unit may have several different values for soil

temperature. Attribute information is difficult to query in
ARC/INFO, where there is a one-to-many relationship
between polygons and database entries (for instance,
each mapping unit, or polygon, has several soil compo-
nents). We found that exporting the attribute information
from ARC/INFO to another software package such as
Excel was easier. The values can  be aggregated by
attribute and then imported into ARC/INFO to produce
individual  coverages such  as soil texture, soil depth, or
soil group. STATSGO data can provide useful informa-
tion for the physical variables that influence vegetation,
such as soil moisture and  nutrient availability.

To review, the project researchers will first identify the
physical parameters  that  govern the distribution of
ecological land units and the existing vegetation in the
state. Then, the team will gather applicable ancillary
data of  physical data in CIS to support the image clas-
sification.  Once the imagery has been classified,  the
wildlife models can be incorporated.

Integrate Terrestrial Vertebrate Models

Once classification is  completed for the state, the gap
analysis team will integrate  terrestrial vertebrate mod-
els. Concurrent with the image classification, the team
will develop a species profile for each vertebrate species
known to occur in the state. These profiles, when com-
pleted,  will be  condensed into  rule-based models for
associated species that can be linked to the ecological
land units (see Figure 4).  This step will create habitat
types. After integration, the team will generate maps that
display species richness of vertebrates for each habitat
type (see Figure 5). These maps link spatial data to the
species database. This enables users to identify areas
in the landscape that combine habitats with the greatest
number of unique species. When a coverage  of land-
ownership is  overlaid on this map,  land managers or
conservation groups can take a proactive stance to seek
protection of  critical habitats. Additional analyses that
Figure 4.  Informational flow chart for wildlife data.
users can perform are displays of the potential distribu-
tion of vertebrate groups, such as upland salamanders.
Another analysis  would be to report the species that
occur in  the fewest habitats  and that would be most
vulnerable to landscape changes. Clearly, CIS provides
a powerful environment for quick and efficient retrieval
of spatial data for management decisions.


To summarize, the West Virginia gap analysis team is
assessing the  natural communities in the state as a part
of the national comprehensive planning effort. We need
to conduct the assessment rapidly, compiling existing
information and  integrating these  data  with CIS and
remotely sensed data.  Ecological land units are classi-
fied according to the  relationship between vegetation
and ecological variables. Satellite  imagery is used to
map existing  conditions  over the  state. The existing
vegetation is classified to reflect physiognomic and flor-
istic elements  to correlate with vertebrate and butterfly
habitat requirements.

The gap analysis team is using many widely available
data sets such as OEMs, land use/land cover, wetlands
inventory, and soils data. While these can reduce  the
time and cost of developing  an ecological  database,
they do present implications for project design and  ac-
curacy.  When the user  combines  maps of different
scales, accuracy  is constrained by the  map with  the
smallest  scale. Additionally,  data  sets  may be con-
structed with objectives for an intended use that is  not
compatible with  project  needs. The classification of
AVHRR data  reflects forest types but not land use, so
another source may be required for these data.

The Gap Analysis Project is not a substitute for intensive
biological studies at a  fine scale. It is merely a quick
assessment at a broad scale  to provide information on
existing conditions. While accumulating data and mod-
eling  potential wildlife  distributions, we will  identify
where inventory  data may  be lacking. Additional work
must be done  to verify wildlife models and the classifi-
cation of vegetation, but this preliminary analysis will be
a valuable framework that  will direct future  studies of
biological diversity. Finally, this effort will  provide a data
set that can be used to monitor changes to land cover
and land use.


This study is being funded by the Cooperative Research
Units  Center,  National Biological Survey, United  States
Department of the Interior. This manuscript is published
as Scientific Article No. 2499  of the West Virginia Agri-
culture and Forestry Experiment Station.

                                                                                                   • Very High
                                                                                                   n High
                                                                                                   E3 Moderately High
                                                                                                   m Moderate
                                                                                                   H Moderately Low
                                                                                                   n Low
                                                                                                   B Very Low
                                                                                                   H Comm. Timber Lands
                                                                                                   H Public Lands
Figure 5.  Relative species richness.


 1.  Davis, F.W., and J. Dozier. 1990. Information analysis of a spatial
    database for ecological land classification. Photogramm.  Eng.
    and Remote Sensing 56:605-613.
 2.  Scott, J.M., B. Csuti, K. Smith, J.E. Estes, and S. Caicco. 1991.
    Gap analysis of species richness and vegetation cover: an inte-
    grated biodiversity conservation strategy. In: Kohm, K., ed. Bal-
    ancing on  the brink of extinction: the Endangered Species Act
    and  lessons for the  future. Washington,  DC:  Island Press.
    pp. 282-297.
 3.  Stephenson, S.  1993. Upland forests of West Virginia. Parsons,
    WV: McClain Printing Co.
 4.  Buol, S.W, F.D. Hole,  and R.J. McCracken. 1980. Soil genesis
    and classification, 2nd ed. Ames, IA: Iowa State University Press.
 5.  Bailey, R.G. 1994. Ecoregions and subecoregions of the United
    States (map). U.S. Department of Agriculture, Forest Service.
    Revised  February 23.
 6.  Fortney,  R. 1994.  Vegetation pattern  of  central Appalachian
    Mountains. (Unpublished).
 7.  Barbour, M.G.,  J.H. Burk,  and WD. Pitts. 1987. Terrestrial  plant
    ecology, 2nd ed. Menlo Park, CA:  Benjamin/Cummings Publishing
 8.  Davis, F.W, D.M. Stoms,  J.E. Estes, J. Scepah, and  M.  Scott.
    1990. An information  systems approach to the  preservation of
    biological diversity. Int.  J. Geogr.  Info. Syst. 4:55-78.
 9.  Scott, J.M., F. Davis, B. Csuti, R. Noss, B. Butterfield, C. Groves,
    H. Anderson, S. Caicco, F.  D'Erchia, T.C. Edwards, Jr., J. Ulliman,
    and R.G. Wright. 1993. Gap analysis: A geographic approach to
    protection of biological diversity. Wildl. Monogr. 123:1-41.
10.  Bauer, M.E., T.E. Burk, A.R. Ek,  PR. Coppin, S.D. Lime, T.A.
    Walsh, O.K. Walters, W Befort, and D.F. Heinzen. 1994. Satellite
    inventory of Minnesota forest resources. Photogramm. Eng. and
    Remote Sensing 60:287-298.
                                                                                               0          10
11.  Benson, A.S., and S.D. DeGloria. 1985. Interpretation of Land-
    sat-4 thematic mapper and multispectral scanner data for forest
    surveys. Photogramm. Eng. and Remote Sensing 51:1,281-1,289.

12.  Hopkins, PR, A.L. Maclean, and T.M. Lillesand. 1988. Assess-
    ment of thematic mapper imagery for forestry application under
    lake states conditions.  Photogramm. Eng. and Remote Sensing
13.  Schriever, J.R., and R.G. Congalton. 1993. Mapping forest cover-
    types in New Hampshire using multitemporal Landsat Thematic
    Mapper data.  Papers presented at ASCM/ASPRS Annual  Con-
    vention and Exposition, New Orleans, LA (February), pp. 333-342.

14.  Cetin, H., T.A. Warner, and D.W  Levandowski.  1993. Data clas-
    sification, visualization, and enhancement using n-dimensional
    probability density functions (nPDF): AVIRIS, TIMS, TM, and geo-
    physical applications. Photogramm. Eng. and Remote  Sensing
15.  Anderson, J.R.,  E.E. Hardy, J.T Roach, and R.E. Witmer. 1976.
    A land  use  and  land cover classification  system  for use with
    remote sensor data.  U.S. Geological Survey Professional Paper
    965. Washington, DC: U.S. Geological Survey.
16.  Zhu, Z., and D.L. Evans. 1992. Mapping midsouth forest distri-
    butions: AVHRR satellite data and CIS help meet RPA mandate.
    J. Forestry 27-30.
17.  Kuchler, A.W. 1964. Potential natural vegetation of the contermi-
    nous United States (map and manual). American Geographical
    Society  Special  Publication 36.  New  York,  NY:  American
    Geographical Society.

18.  Sneddon, L., M. Anderson,  and K. Metzler. 1994. A classification
    and description of terrestrial community alliances in the Nature
    Conservancy's eastern region: First approximation. Washington,
    DC: U.S. Department of the Interior Fish and Wildlife Service.
19.  Cowardin,  L.M.,  V  Carter, F.C. Golte,  and T.E. LaRoe. 1979.
    Classification of wetlands and deepwater habitats of the United
    States.  Report  No.  FWS/OBS-79/31.  Washington,  DC:  U.S.
    Department of the Interior  Fish and Wildlife Service.

      GIS Watershed Applications in the Analysis of Nonpoint Source Pollution
                     Thomas H. Cahill, Wesley R. Horner, and Joel S. McGuire
                          Cahill Associates, West Chester, Pennsylvania

Geographic information systems (GIS) have been used
to evaluate the impact of nonpoint source (NPS) pollu-
tion in a variety of watersheds and drainage systems
over the past 20  years (1-6). During that period, our
understanding of the sources and hydrologic transport
mechanisms of NPS pollutants, both in particulate and
soluble forms, has greatly increased (7-9). Our ability to
create and manipulate land resource data, however, has
advanced at a far more dramatic rate. Whereas 20 years
ago, both computer system capabilities and peripheral
hardware limited the process of encoding, storing, and
displaying spatial data, today we  can encode land re-
source data, analyze  it,  and produce stunning visual
displays at a relatively low cost.

The question is: what has this experience told us regard-
ing the yet unresolved problem of water quality degra-
dation from NPS  pollution in our streams,  lakes, and
coastal waters (10, 11)?

The purpose of this paper is to report on several recent
studies of this nature that created a GIS as a tool to
analyze  NPS pollution. This paper will not cover all
aspects of these studies; detailed  reports on each pro-
ject are available from the authors or respective clients.
The objects of these studies were:

• A medium-sized lake draining  a fairly small watershed

• A riverine system with  multiple use impoundments

• A 100-mile stretch of Atlantic coastal estuary

These water bodies all have one common  ingredient:
NPS pollution significantly affects them. While the pri-
mary  focus of these studies was  to understand the
dynamics of surface water quality, and specifically the
NPS component, the further objective was to document
the causal link between identified water resource prob-
lems  and the watershed-wide management  actions
needed for their remediation. Thus, GIS serves not only
as a mechanism for analysis of NPS pollution sources
but also  as the tool by which  to evaluate  alternative
methods that would reduce or prevent this pollution.
Study Concepts

These three studies illustrate different approaches to
both aspects of this problem.  In the 93-square-mile Up-
per Perkiomen Creek watershed (UPW) study, the ob-
jective was to develop a management program that
would reduce nutrient load in a system of reservoirs at
the base of the watershed. An essential element in the
analysis underlying GIS design (ARC/CAD) was to be
able to differentiate and evaluate pollution sources in the
watershed, while  providing the technical basis for an
innovative and far-reaching management program on all
levels of government; that is,  GIS was used not only to
analyze the problem but to help formulate the solution.

In the more focused Neshaminy Creek study,  Cahill
Associates (CA) designed a detailed pixel/raster format
for GIS to support detailed hydrologic modeling (12) and
NPS  loading  analysis. This  study,  carried out  under
Pennsylvania's Act 167 stormwater management pro-
gram, was under a legal  requirement to translate tech-
nical  findings  into subdivision regulations that all  30
watershed municipalities would  adopt. This mandate
required much more geographically specific rigor in the
GIS approach and in the management recommenda-
tions the law stipulated.

These two projects (see Figure 1), when taken together,
illustrate the critical relationship between understanding
the appropriate level of detail in GIS system design, GIS
development with modeling and other analytical require-
ments, and ultimately, the proposed management ac-
tions for watershed-wide  implementation.

In the New  Jersey  Atlantic  Coastal Drainage (ACD)
study, the objective was to document more completely
the magnitude and sources  of NPS pollutants,  espe-
cially nutrients, entering New Jersey estuarine coastal
waters. The GIS design placed special attention on the
role of urban or developed land uses situated along the
coastal fringe, particularly the  maintained or landscaped
portions of developed sites. Most previous studies have
largely ignored this factor. Instead, they have focused
water quality  analysis typically on  NPS loadings  as a

    Upper Perkiomen
                            N. HAMPTON

                        ^x  >Vi

         Perfiomen Creek Watershed^^y
Figure 1.  Regional location of Upper Perkiomen and Neshaminy
function of impervious area coverage, with the assump-
tion that loadings increase as imperviousness increases.

On the contrary, the CA thesis states that certain pollut-
ant loadings, such as nutrients, maximize in areas with
relatively moderate densities  (1/2- to  1-acre lots) and
percentage impervious cover but with large maintained
lawnscapes. Because sandy  soils allow soluble NPS
pollutants to pass as interflow to points of surface dis-
charge with surprising ease, they exacerbate the prob-
lem of nutrient applications in typical coastal drainage
areas. CIS application in this case enabled estimation
of the nutrient loading to coastal waters. Existing fertil-
ized lawn areas  were  calculated  to  be  a significant
source of nutrient pollution, with loadings from new land
development posited as  an even more serious problem
for New Jersey's coastal waters. CIS was then applied
to evaluate the suitability of various best management
practices (BMPs), based on the physical and chemical
properties of the soil mantle and the existing and antici-
pated land use.

The Upper Perkiomen Creek
Watershed Study


The UPWin southeastern Pennsylvania is a tributary of
the Schuylkill River in the Delaware River basin  (see
Figure 2). Serious eutrophication  problems occurring in
the system of reservoirs lying at the  base of this rela-
tively  rural watershed prompted  the  study.  The study
                                                                                               New Jersey


  Upper Perkiomen Creek Water:
                                                                                            Scale in Miles
                                                      Figure 2. The Perkiomen Creek watershed in the Delaware River
effort evolved from concerns on the part of the Delaware
Riverkeeper,  a private nonprofit environmental organi-
zation dedicated to promoting the environmental well-
being of the Delaware  River watershed.  The  Upper
Perkiomen Creek has experienced various water quality
problems, especially the eutrophication of Green Lane
Reservoir, a  large raw water supply storage reservoir
(see Figure 3). Green Lane's highly eutrophic condition
has been a constant since shortly after initial construc-
tion over 35 years ago,  but the relative importance of
NPS inputs  has dramatically increased. Whereas 10
years ago point source input  was the major source of
phosphorus, elimination of some point sources and ad-
vanced waste treatment for others has greatly reduced
that component of pollutant loading, while NPS sources
have remained constant or increased. Current analysis
indicates that NPS pollution constitutes over 80 percent
of the annual load of phosphorus (see Figure 4) into the
Green  Lane  Reservoir and  is well  in excess  of the
desired loading to restore water quality (see Figure 5).

Nonpoint Source Analysis

Calculating the NPS load was an essential ingredient in
the study and relied on developing accurate measure-
ment of NPS transport during stormwater runoff periods.
Certain  pollutants, specifically those associated  with
sediment and particulate transport such as phosphorus,
have produced a  "chemograph" that parallels but does
not exactly follow the traditional form of the hydrograph
(see Figure 6). The pollutant mass transport associated
with this runoff flux frequently constitutes the major frac-
tion of NPS discharge in a given watershed (8, 13). In



Figure 3.  Green Lane Reservoir in the Upper Perkiomen watershed, 814 acres, 4.3 BG.

                  500 Pounds
                   per Year
                  500 Pounds
                   per Year
  Nonpoint Sources
     Dry Flow
  5,604 Pounds
    perYear^ 30%
    Nonpoint Source
     Total = 84%
                           Point Sources
                            Direct 2,486
                          Pounds per Year
                                        Point Sources
                                      to Tributaries 592
                                       Pounds per Year

                                       Direct Drainage
                                        1,172 Pounds
                                         Per Year
                         Nonpoint Sources
                         Storm Flow 8,036
                         Pounds per Year
   Total Load = 18,889 Pounds per Year
Figure 4.
Sources of total phosphorus mass transport into the
Green Lane Reservoir from the Upper Perkiomen wa-
tershed (71 square miles) in an average flow year—in
pounds per year.
                            Average Year, Riverkeeper-
                              1993 = 18,889 Pounds per Year
                            Eutrophic = 8,416 Pounds per Year
                            Oligotrophic = 4,208 Pounds per Year
        Riverkeeper -1993   Eutrophic
the UPW study, operating continuous sampling stations
at two key gage  locations above the reservoir allowed
the measurement of stormwater chemistry of this type
and produced estimates of wet weather transport of phos-
phorus and sediment. Surprisingly,  the  NPS transport
during dry  weather,  calculated by subtracting the point
sources, was  also significant and is attributed to live-
stock discharges and septage drainage.

But the wet weather  proportion of NPS  pollution still
dominates  lake water quality. Many have said that water
quality in a given watershed is a function of land use,
but that statement  is as  unsatisfying as saying  that
runoff is a function of rainfall. Experience has taught us
that  neither process is quite that simplistic, nor does
either follow a direct  linear  relationship  of  cause and
effect. The causal mechanisms that generate a certain
mass  load  of  pollutant in a  drainage basin certainly
result from how much mass  of that  pollutant is  applied
to the landscape  within the drainage, which in turn  is
scoured from the landscape during  periods of surface
saturation,  transported in, and diluted by runoff. The end
result is a  concentration of pollutant in the stormwater
that might  be several orders  of magnitude greater than
during dry weather flow, the hydrologic period tradition-
ally used to measure and define water quality.

Developing NPS analysis or algorithms  for stormwater
quality modeling requires replicating the  specific hydro-
graph and  its associated chemograph, as well as defin-
ing the mechanisms by which pollutants are scoured
from the land surface, transported in runoff, and pass
through the river system. Total phosphorus (TP),  for
example, is transported with the colloidal soil particles
(see Figure 7), so sediment transport and deposition
constitute a key mechanism.
Figure 5.  Reduction in annual phosphorus load required  to
         achieve improved trophic level.
  3/4/930:00  3/4/9312:00  3/5/930:00  3/5/9312:00  3/6/930:00  3/6/9312:00
Figure 6.  Storm hydrograph in the Upper Perkiomen watershed
         illustrating the dramatic increase in total phosphorus
         and suspended sediment during runoff.
                                               Adding to these complications is the question of whether
                                               to model single or multiple events. Is the chemodynamic
                                               process  one in which the transport takes place over a
                                               series of storm events, so that each storm moves the
                                               pollutant mass a given distance in the drainage and then
                                               allows it to settle in the channel only to resuspend it with
                                               the next  peak of flow? Or does the total mass transport
                                               occur in one single dynamic, from corn field or suburban
                                               lawn to lake, estuary, or other sink, that is hours or days
                                               downstream in the  drainage? The issue of how storm-
                                               water transport of pollutants takes place is of paramount
                                               importance in  current planning  and regulatory  imple-
                                               mentation (11) because many of our current BMPs are
                                               relatively ineffective in removing NPS pollutants.  This
                                               understanding is critical even as we attempt to intervene
                                               in the pollutant generation process by changing the way
                                               we cultivate the land, fertilize our landscapes, or for that
                                               matter, how we alter the land  surface during growth.

3 000
2 500
Q- 9 nnn
o_ ^,uuu -

ouu •
n _

.* V."**" *
^*r" **

^"'" •


'. « "


, Main E
" Regre;

ranch rA2 =
3 65 Samples
ssion ~
                                500           1,000          1,500

                                          Suspended Solids (PPM)
Figure 7.  The relationship between total phosphorus and suspended sediment concentrations during runoff is strong but varies
         with different watersheds.
GIS Evaluation

The  GIS data files  on land use/land cover that were
created for the UPW show that the bulk of the area is
still quite undeveloped and rural (see Figure 8), with the
steeply sloped and igneous rock areas  in the  headwa-
ters in forest cover (38 percent) and the valleys in mixed
agriculture (44 percent). The urban/suburban land com-
poses the remaining 18 percent and largely consists of
several older, historic  boroughs linked together in a
lineal pattern with widely scattered, low-density residen-
tial areas. Much of the existing housing is turn-of-the-
century at quite high densities,  mixed with a variety of
commercial  and  other uses.  This  pattern  contrasts
sharply with typical large-lot suburban subdivisions. In
fact,  these watershed boroughs resemble the "village"
concepts that innovative planning theorists advocate in
a variety of important ways.

The  watershed (see Figure 9)  is blessed, or cursed,
depending upon one's perspective, with a multiplicity of
local governments including four different counties and
18 different municipalities. This arrangement poses spe-
cial challenges for management program implementa-
tion.  Population  projections  indicate  that additional
development will occur at moderate rates throughout the
watershed, reflecting recent trends.

Farming, both crop cultivation and dairying, is a major
existing land use in the watershed, although agriculture
is not especially robust and appears to be declining. This
lack  of agricultural vibrancy becomes a major factor in
determining  how to impose additional  management
measures on agricultural pollution sources. GIS tabula-
tion of agricultural land totals some 19,000 acres above
the reservoir, which can be compared with the estimated
TP and suspended solids  (SS) mass transport reaching
the lake. Considering only the agricultural land to be
the source of this NPS input (not quite true) suggests
an  average annual  yield of 180#/acre/year-SS  and

This sediment/phosphorus yield is more than sufficient
to maintain a eutrophic condition in the reservoir system.
The problem with this yield, however, is that it is two
orders of  magnitude less than  commonly accepted
methodologies of soil erosion, such as the universal soil
loss equation (14), would suggest might come from such
a watershed. Analysis of the cultivation practices taking
place on farmland in the watershed estimates soil ero-
sion to be approximately 5 to  10 tons per acre or more
per year, far more than is observed  passing out of the
basin into the reservoir. The phosphorus applications on
both cultivated  and  maintained residential landscapes
also appear much greater than the mass transport actu-
ally measured in the flowing streams, which represent
perhaps 7 percent or less of the annual land application.

The implication for NPS  analysis is that the standard
shopping list of either agricultural or urban BMPs might
only reduce the mass  transport by a  relatively small
fraction,  even  if successfully applied  throughout the
drainage. As Figure 7 illustrates, most of the phosphorus
transport occurs on  the colloidal fraction of sediment
particles, which tend to remain in  suspension as storm-
waters pass through  conventional detention structures,
terraces, or grassed swales.

To consider more radical measures, GIS was used  to
determine  possible  alternatives, such  as  creating  a
stream buffer system (see Figure 10) with  various set-
back distances  from the perennial stream network, and
to evaluate how great an impact this might have on
agricultural land use and  urban development. Land


Figure 8.  GIS data files showing land use/land cover characteristics for the Upper Perkomien watershed.

Figure 9.  Existing land use/land cover GIS file for the Upper Perkiomen watershed. The 95-square-mile basin includes portions of
          four counties and 18 municipalities.

                                                                                          x' X  -1"-" s
                                                                                         /  S  i
                                                                                       (Hydrologic Subgroup)
             /'           '
                                                                                         Key Map
•]  11
Figure 10.  GIS analysis of stream corridors allows evaluation of riparian buffer systems, potential agricultural land loss, potential
          septic system discharges, and related NFS reduction with selected best management practices.

use at varying distances (100 feet, 200 feet, and 1,000
feet) from streams was tabulated,  including all land area
in the "active" agriculture categories. This CIS documen-
tation allowed estimation of the significant NPS reduc-
tion  in loadings that a  riparian corridor  management
program could achieve.

In the same way, CIS analysis helped estimate pollutant
loadings  from malfunctioning onsite  septic systems.
Counts  of structures in nonpublicly  sewered  areas
within varying distances from the stream system were
developed using CIS data files. The nearly 300 potential
systems within a 200-foot radius of those streams drain-
ing into the Green Lane Reservoir identified in this man-
ner, with pollutant generation factors applied, became
the basis of a dry weather pollutant estimation. Although
this approach was dependent on  a variety of assump-
tions, alternative approaches of evaluating the  problem,
such as field visits to actual onsite systems throughout
the watershed, would not have been feasible.

For urban and suburban development, the management
focus was to  estimate NPS loadings from future land
development. CIS was used to demonstrate NPS pollut-
ant load implications of future growth envisioned in the
watershed's  keystone  municipality, Upper  Hanover
Township. Here, an increase of 15,000 residents would
convert  1,772 acres into residential, commercial, and
industrial uses. Nonpoint pollutant loadings generated
by this new land  development constituted significant
increases in  phosphorus,  suspended solids, metals,
oil/grease, and other pollutants, and would reverse any
improvements in Green  Lane Reservoir water quality
that  recent  wastewater  treatment  plant  upgrades

From a water quality perspective, future alternative land
use configurations that  concentrate development and
minimize ultimate  disturbance  of the  land surface
yielded would substantially reduce NPS pollutant load-
ings  into the reservoirs. This entire process of testing
land  use  implications  of  different management  ap-
proaches for their water quality impacts indicated that
pollutant loads could be minimized far more cost effec-
tively through  management actions, both structural and
nonstructural, which varied from  the areawide to the

Neshaminy Creek Watershed Stormwater
Management Study


The Neshaminy Creek watershed, including 237 square
miles of mixed urban and rural land uses, lies primarily
in Bucks County, Pennsylvania, and flows directly into
the Delaware  River (see Figure 1). The 1978  Pennsyl-
vania 167 Stormwater Management Act, which required
that counties prepare Stormwater management plans for
all 353 designated watersheds in the state, mandated
the Neshaminy study. This act further stipulated  that
municipalities then needed to implement the watershed
plans through  adopting the necessary  municipal ordi-
nances and  regulations. In fact, the Neshaminy study
had three water resource management objectives:

• Prevent worsened  flooding  downstream caused by
  increased  volumes of runoff from land development.

• Increase ground-water recharge.

• Reduce  NPS pollutant loadings from new development.

In the initial study design, water quality and NPS issues
were secondary to flooding concerns. When Pennsylva-
nia's Stormwater management program was conceived,
the state focused  on  preventing watershed-wide flood-
ing. Clearly, detention basins have become the primary
mode of managing peak rates of Stormwater discharge
site-by-site in most communities. Because detention ba-
sins only control peak rates of runoff and allow signifi-
cantly increased total volumes of water discharged from
sites,  however, the increased Stormwater volumes can
theoretically  combine and create worsened  flooding
downstream. Consequently, most Act 167 planning has
focused on elaborate hydrologic modeling designed to
assess the seriousness of potential cumulative flooding
in watersheds under study.

In the case of the Neshaminy, however, the record sug-
gested that although  localized flooding could be an is-
sue, an existing network of eight multipurpose flood
control structures constructed during the 1960s served
to prevent significant flooding. Water quality  certainly
was a serious Stormwater concern, however, especially
in the areas  flowing into the reservoirs where recrea-
tional use  had become intense. Several of the existing
impoundments  were  multipurpose,  their permanent
pools providing critical recreational functions for a  bur-
geoning Bucks County population. At the same time, the
proliferation  of development in the watershed, with its
increased  point and nonpoint sources, had degraded
streams and  seriously affected the reservoirs. While the
total stream  system  in the watershed was of  concern,
the future  of the reservoirs came to be particularly im-
portant in developing the total Stormwater management
program for the Neshaminy watershed.
The Neshaminy lies at the heart of Bucks County, Penn-
sylvania's  primary population and employment growth
county (see Figure 11). Although the Neshaminy water-
shed has already experienced heavy development, es-
pecially in the lower  or southern portions, farmsteads
and large  areas of undeveloped land still exist, espe-
cially in headwater areas. Agriculture has  been a major
land use in the past, but farms rapidly are converting to
urban uses as the wave of urbanization moves outward
from Philadelphia  and from the Princeton/Trenton met-
ropolitan  areas. Growth projections  indicate  continu-

                                                                       Nl SM 1MIM rK'-.FK V, ATlTttSHLO
                                                                    S'lORWAlflH MANAGEMENT PI AN
Figure 11.  Land use/land cover in the Neshaminy basin of Bucks County, Pennsylvania.  The watershed covers 237 square miles
           in southeast Pennsylvania.

ation of this rapid growth and a continuing change  in
existing land use/land cover, together with  projected
development with in the required 10-year planning horizon.

Physiographically, the watershed spans both the Pied-
mont and Atlantic coastal plain provinces, with  rolling
topography and relatively steep slopes underlain by Tri-
assic formation rock,  including the Lockatong, Bruns-
wick, and Stockton formations.  This bedrock ranges
from being a poor aquifer (Lockatong) to an excellent
aquifer (Stockton) where the many rock fractures allow
for  considerable ground-water yields. Soils  are quite
variable, ranging from good loam (hydrologic soil group  B)
to clays  and other  types with poor drainage charac-
teristics (e.g., high water table, shallow depth to bed-
rock). A large proportion of the soils in the watershed are
categorized  as hydrologic soil group C, which is mar-
ginal for many stormwater management infiltration tech-
niques (see  Figure 12) and produces a relatively large
proportion of direct runoff. With an annual rainfall of 45
inches, base flow  accounts for  about 12 inches and
direct runoff accounts  for 10 inches.

The system  of  eight stormwater control structures,
which were built over the past three decades  under the
federal PL 566 program, have altered the hydrology  of
the watershed (15). In addition, in heavily developed
portions  of the watershed,  impervious  surfaces com-
bined with numerous detention basins prevent the bulk
of the precipitation from being recharged, and the vol-
ume of total runoff proportionally increases. An elabo-
rate system of municipal and nonmunicipal wastewater
treatment plants also  adds  to this alteration of the hy-
drologic cycle. These plants discharge wastewater efflu-
ent that,  in  some  cases,  constitutes the bulk of the
stream flow during dry periods. While the impact of NPS
was evident throughout the drainage, it was of special
interest in the impoundment network, especially those
impoundments that were conceived as multipurpose  in
function and constitute major recreational resources  in
the watershed.

CIS Design

Act 167 requirements and the needs of the hydrologic
and other modeling  used in planning both heavily influ-
enced  the CIS developed for the Neshaminy. Spatial
data files, including  existing land use, future  land use,
and soil  series aggregated by hydrologic soil groups,
were created by digitizing at a 1-hectare (2.5-acre) cell
resolution. The encoding process that helped design the
CIS used a stratified random  point sampling technique
that similar studies  had developed and applied (1, 3).
The encoding process used a metric grid of 5-kilometer
sections, subdivided into 2,500 1-hectare cells (100 me-
ters on a side), aligned with the Universal Transverse
Mercator (UTM) Grid System. This grid appears in blue
on  U.S. Geological Survey  (USGS) topographic maps.
These maps served as the framework of reference for
all data compilation. Within each 100-meter cell, a ran-
domly located point was  chosen (see  Figure  13)  at
which the specific factor was encoded as  representative
for the cell, using a digitizer tablet.  This approach al-
lowed extraction  of the data from the respective source
documents with some rectification necessary for many
types of source maps and  photographs.

The  combination of soil series and cover in  each cell
helped to calculate the curve number and unit runoff per
cell. The 45,000-cell data file was then used to calculate
total runoff for a range of events in each of 100 subbas-
ins that averaged 1.95 square miles each. The resultant
hydrographs, used in combination with a separate linear
data file in CIS describing  the hydrographic network of
stream geometry, routed and calibrated the hydrologic
model (TR-20). NPS mass transport loadings were es-
timated on an annual basis by cell, again  using the land
use/land  cover data file, and total loads summed by
groups of subbasins above critical locations. This issue
was  particularly important  with  respect to the drainage
areas above the  impoundments, where NPS  pollutants
were of greatest  concern.
The  soil  properties  data file was especially useful  in
evaluating certain management objectives, such as the
opportunity for recharging  ground-water aquifers. The
spatial variation  in relative effectiveness of  infiltration
BMPs was considered for both quantity and quality miti-
gation because the best  methods for NPS  reduction
usually include recharge where possible. The soil series
corresponding  with  new growth  areas were  classified
regarding  their suitability  for these  BMPs,  which are
most efficient on well-drained or moderately well-drained
soil. Thus, the alternative impacts of future growth could
be considered  in terms  of potential generation (or man-
agement) of NPS loads. A BMP selection methodology
(see Figure 14), which was developed forthe 30 munici-
palities within the watershed, focused on new land de-
velopment  applications and  considered both  water
quantity and quality management objectives. BMP se-
lection is a function of several factors, including:
• The need for further  peak rate reduction.

• The recharge sensitivity of the project site (defined as
  a function of headwaters  stream location, areawide re-
  liance on ground water for water supply, or presence
  of effluent limited streams).

• The need for priority  NPS pollution controls (location
  within reservoir drainage).

Development of two "performance" levels of BMP selection
techniques gave municipalities some degree of flexibility
in  developing their new stormwater  management pro-
grams. This system required only the minimally acceptable
techniques  but recommended the more fully effective
ones, hoping that municipalities would strive to incorporate

           Region With Labeled Zones


46 47 48 49
            Zone With Labeled Cells



50 462 464 466 468 470
Figure 13.  Raster/pixel design of GIS for Neshaminy modeling
           study. Each pixel is 1 hectare (2.47 acres).

recommended management measures wherever possi-
ble. The BMP selection methodology also was sensitive
to type of land use or proposed  development, assigning
typical  single-family  residential  subdivisions  different
BMPs than, for  example,  multifamily and  other nonresi-
dential  proposals  (including commercial  and  industrial
proposals). The  selection  process also determined size
of site to be  a factor, differentiating between sites of 5
acres or more  because  of the varying degrees of cost
and  effectiveness  of different BMP  approaches.  The
methodology, if properly and fully implemented, should
achieve the  necessary  stormwater-related  objectives—
both  quantity and quality—that the analysis had deemed
necessary  (16).

GIS was especially important in its  ability to test  how
reasonable the BMP selection methodology was. Such
tests included the ability to evaluate, for each municipal-
ity, the following factors:

•  The nature and extent of the projected development.

•  The size of development/size of site assumptions.

•  Other vital BMP feasibility factors such as soils and
   their appropriateness  for different  BMP techniques.

GIS  also  enabled  analysis  of the water quantity and
quality impacts of projected growth on a  baseline basis,
assuming continuation of existing stormwater manage-
ment practices. Water quality loadings to individual res-
ervoirs  and  to  the  stream  system  could  be  readily
demonstrated.  Because overenrichment of  the  reser-
voirs was so crucial, researchers could  estimate phos-
phorus   and    nitrogen   loadings   from    projected
development assuming  existing  stormwater practices,
even on a  municipality by municipality basis.

New Jersey Atlantic Coastal
Drainage  Study


The third study considered a much larger coastal water-
shed in New Jersey (see Figure 15). The New Jersey
                                         1' Impoundment I
                                          INon Impoundment!.
                                          Drainage      F
                               Non Rechargel
                               Sensitive    r
                                          I Impoundment I
                                          \ Drainage    |
                                          INon Impoundment!
                                          '(Drainage      \
                                         J Impoundment
                                         1 Drainage
                                          INon Impoundment!.
                                          Drainage	I
                                          I Impoundment
                                          1 Drainage
                               INon Recharge|_
                                          INon Impoundment!
                                          "[Drainage	T
         Not Applicable
         Required: Multi-Resi and Non-Resi Overs Acres, Porous Pave. With
         Underground Recharge Beds for Paved Areas and Infiltration Devices for
         Non-Paved Areas, Sized for Peak; Other Uses, Infiltration Devices for
         Paved and Nonpaved Areas, Sized for Peak

         Not Applicable
         Required: Multi-Resi and Non-Resi Overs Acres, Dual Purpose Detention
         Basins for Paved/Nonpaved Areas, Sized for Peak; Other Uses, Detention
         Basins Sized for Peak
         Recommended: All Uses/Sizes, Porous Pave. With Underground Recharge
         Beds for Paved Areas; Minimum Disturbance or Wet Ponds/Artificial
         Wetlands for Nonpaved Areas, All Sized for Peak
         Required: All Uses and Sizes, Porous Pave. With Underground Recharge
         for Paved Areas; Minimum Disturbance for Nonpaved Areas

         Required: Multi-Resi and Non-Resi Overs Acres, Porous Pave. With
         Underground Recharge Beds for Paved Areas and Infiltration Devices for
         Nonpaved Areas
         Recommended: All Uses/Sizes, Porous Pave. With Underground Recharge
         for Paved Areas; Minimum Disturbance and/or Infiltration Devices After
         Site Stabilization

         Required: All Uses/Sizes, First-Flush Settling Basins for Paved Areas; for
         Nonpaved Areas,  Minimum Disturbance/Wet Ponds/Artificial Wetlands
         Recommended: for All Uses/Sizes, Porous Pave. With Underground
         Recharge Beds for Paved Areas; Minimum Disturbance for Nonpaved
         Required: Multi-Resi and Non-Resi Overs Acres, First-Flush Settling
         Basin; Other Uses, Detention Basins (No Change)
         Recommended: Porous Pave. With Underground Recharge Beds for Paved
         Areas; for Nonpaved areas, Minimum Disturbance/Wet Ponds/Artificial
Figure 14.  BMP selection methodology used with the GIS database in the Neshaminy basin modeling study.

Atlantic Coastal Drainage  (ACD) includes an area  of
2,086  square miles,  with  barrier islands  (50  square
miles), wetlands/bays/estuaries (285 square miles), and
a unique scrubby pitch pine-cedar forest, known  as the
Pine Barrens, largely covering  the 1,750 square miles
of mainland interior (see  Figure  16). This flat coastal
plain comprises a series of unconsolidated sedimentary
deposits of sand, marl, and clay, which increase in thick-
ness toward the coastline.  Over the past 16,000  years,
as the  ocean level has risen, the water's edge  has
progressed inland to its present position.  Ocean  cur-
rents and upland erosion and deposition have created a
long, narrow series of barrier  islands  that absorb the
energy of ocean storms and buffer the  estuary habitats
from the scour  of waves and currents. Between the
mainland and barrier islands are embayments and es-
tuaries of different sizes and configurations. Inland erosion
and  marine sediments  have  gradually filled many  of
these areas, creating extensive wetlands (17).

In this ACD region, new land development and  popula-
tion growth have caused significant degradation of water
quality from an increase in both point source and NPS
pollution. Although the array of pollutants is ominously
broad, increased nitrogen  and phosphorus loadings
have resulted in enrichment of back bays, estuaries, and
nearshore waters, contributing to algal blooms, declining
finfish and  shellfish populations, diminished  recreational
                                                    	Atlantic Coastal Drainage

                                                    	Cafra Management Area

                                                     —    Pine Barren Region Including
                                                          Cedar-Pine Fringe

Figure 15.  The ACD of New Jersey includes approximately 2,000 square miles of land area from the Manasquan River to Cape May.

Figure 16.  Aerial photograph of New Jersey illustrating the Barrier Islands and estuary system situated along the Atlantic coast.

opportunities, and a variety of other problems (18). A
major source of these nutrients is point source sewage
treatment plants (STPs), but the effluent outfalls of almost
all these STPs discharge into nearshore ocean waters
beyond the barrier islands. Thus, NPS pollutants almost
totally dominate the water quality in the estuaries and
back bays (19, 20).

These  NPS pollutants, which rain scours from the land
surface and flushes into coastal waters  with each rain-
fall, comprise a  largely  unmeasured and unmanaged
flux of contaminants.  Prior research on coastal water
quality has given considerable attention to NPS pollution
generated from paved or impervious surfaces, particu-
larly roadways and  parking lots where hydrocarbons,
metals, suspended  solids, biologic oxygen  demand
(BOD), and other pollutants have been measured.

Although these NPS pollutants are certainly of concern
in  New Jersey's coastal waters, the enrichment  issue
has led to  a focus  on NPS pollution produced when
creating large areas of pervious and  heavily maintained
landscape,  such as lawns and other  landscaped areas,
in  the sandy soil context of the coastal  area. Typically,
significant quantities of  fertilizer and other  chemicals,
which are applied on these new pervious surfaces, are
naturally low in nutrients. Although a modest portion of
the applied  fertilizer runs off directly into surface waters,
larger quantities of soluble pollutants, such  as nitrates
and  herbicides,  quickly percolate  down through the
sandy soil, then move rapidly as interflow to the estuary

In  this coastal drainage of unconsolidated sediments,
the hydrologic cycle differs from inland watersheds. Of
the 45-inch  average annual rainfall, only a small fraction
(2.5 inches  per year) becomes direct runoff,  with the
balance rapidly infiltrating into the  sand strata  (21).
Most of the infiltration that reaches the ground water
(20 inches per year) discharges to surface streams
(17 inches per year) within  a few hours following rainfall,
producing  a lagging and attenuated hydrograph. This
rapid infiltration, combined with the sand texture of the
soil, has a  major bearing on the water quality implica-
tions of new land development. Thus, urbanization  of
coastal regions  has  dramatically altered  hydrologic
response,  with  every square foot of new impervious
surface converting what had been approximately 41.5
inches of infiltration into  direct runoff to  bays and  estu-
aries, with a turbid soup of NPS pollutants.

Even  in areas  that  have  maintained  infiltration, the
coastal soils do not remove NPS pollutants as efficiently
as other areas of New Jersey that overlie consolidated
formations with heavier clay soils. These soils provide a
much more thorough removal of NPS pollutants through
physical, chemical, and  biologic processes, as rainfall
percolates through the soil mantle.
With development of coastal areas, increased impervi-
ous areas and changing flow pathways (inlets and storm
sewers) convey nonpoint pollutants introduced by devel-
opment (from both pervious and impervious  surfaces)
directly to the coastal waters. In addition, freshwater
recharge to the underlying  aquifer decreases with the
increase in   impervious  surfaces, with  resulting  in-
creases in saltwater intrusion into the sand aquifers and
contamination of ground-water supply wells along the
coast.  Further compounding the loss of the stormwater
for ground-water recharge are increased ground-water
withdrawals necessary for new watersupply. In sum, urban
growth within the ACD, with its  1.13 million permanent
residents (and still growing) and an additional 1.5 million
summertourists, has dramatically altered the natural drain-
age system (and landscape) in a way that significantly
increases the  discharge of NPS pollutants (22).

GIS Approach

New Jersey's Department of Environmental Protection
already had  developed  a  computerized GIS  system
(ARC/INFO)  for environmental analysis and resource
planning, so this study aimed to use existing  GIS work
and to refine this GIS system. Although data files for
municipal boundaries, watershed  areas, and a variety of
other factors  already existed, land use/land cover data
had not been developed and constituted a major work
task. The subsequent land use/land cover file included
the entire 2,000 square  miles of the ACD, but this fo-
cused  on the urbanized area (212 square miles) that
occupied about  11 percent of the coastal fringe. The end
product was  a polygon file that described about 2,500
polygons of urban/suburban land, each averaging about
0.1 square miles (see Figure 17).

Using  aerial  photographs combined with USGS  base
maps and extensive field reconnaissance, each polygon
was classified by:

• Land use type.

• Percentage of impervious cover and  maintained areas.

• Degree of  maintenance (fertilization)  being provided
  to these maintained areas.

Although classifying land use type and extent of imper-
vious cover/maintained areas was a relatively straight-
forward evaluation  process (rated within one of 11
categories by percentage, 0 to 5 percent, and so forth),
the third variable, degree of  maintenance, required spe-
cial treatment and  data  development procedures. De-
gree of maintenance was translated into high, medium,
and low categories, with high maintenance exemplified
by golf courses or other intensively maintained areas.
Medium maintenance assumed  chemical application
rates comparable with those recommended by Rutgers
University state agronomists. Finally,  low maintenance
was typified by a wooded or otherwise naturally vegetated

Figure 17.  Urban land use polygons digitized for the New Jersey coastal drainage. The 2,500 polygons shown cover approximately
           212 square miles (11 percent) of the ACD area of 2,000 square miles.

lot and assumed little or no regular chemical application.
Research staff executed considerable field reconnais-
sance to objectify this judgment-based rating technique
(see Figure 18).

Nonpoint Source Analysis
Because the  drainage is almost entirely estuarine, the
hydrologic aspects in this study were almost irrelevant
except as a tool to describe the pollutant transport proc-
ess. Such coastal drainage  systems do not allow the
measurement of hydrographs and chemographs  (see
Figure 6), except on inland riverine segments or se-
lected  infrastructure  points of discharge (storm sewer
outfalls). Thus, the NPS analysis focused on the pollut-
ant production and  transport  process, especially the
nutrients applied to the  maintained  landscapes, which
are a major part of coastal urbanization.
This study required a great deal of effort to produce an
indextable relating urban cover characteristics (percent-
age impervious, amount of chemical application) to NPS
production potential.  For each of the 2,500 urban land
polygons CIS described, estimates of the NPS loading
for a number of pollutants were generated. Potential
loadings were then aggregated by subwatershed. Total
NPS loadings could then be compared with  point source
loadings for the entire coastal drainage. Interestingly,
the NPS loading dominated water quality  in the estu-
arine drainage while the point sources, discharged by
ocean  outfalls to nearshore  waters  beyond the barrier
islands, were the major source of nutrient pollution in
this portion of the coastal environment (see Figure 19).
Given the estimates of NPS pollution, the major ques-
tion involves how to control or reduce these loads. The
suitability of selected BMPs for the reduction/prevention
of pollutant generation was then evaluated and spatially
identified within the drainage (see Figure 20). This figure
evaluated the use of constructed wetland systems as a
structural measure. That is, CIS allowed state regulators
to identify not only what works best in terms of water
quality protection measures, but also where these meth-
ods could work successfully.  This analysis was driven
by a detailed evaluation of the combinations of natural
conditions CIS identified within the study area. For ex-
ample, certain BMPs can be applied on soils that have
a certain set of characteristics  (permeability, depth to
seasonal high water table) and that are presently in a
given land cover and planned for urbanization.

CIS  also  aided  in  evaluating  alternative BMP tech-
niques,  including  reduction in nutrient applications and
land  management BMPs such as elimination of artificial
landscapes, again using  a  combination of natural fea-
tures and land use patterns (see Figure 21). The result
of this analysis considered the relative proximity of ur-
ban  land  uses to the coastal  waters as significantly
increasing the potential for NPS transport. State regula-
tory  programs establish  minimum setback criteria for
development in sensitive areas, and these criteria may
be modified to consider  pollutant production potential
based on CIS delineation of pollutant production.
New Jersey has been  striving to develop NPS manage-
ment programs for coastal areas to  reduce  existing
sources of pollution as well as  prevent the creation of

                                                                             R = Residential Use
                                                                             C = Commercial Use

                                                                             # = Percent Impervious

                                                                             H = High Maintenance
                                                                             M = Medium Maintenance
                                                                             L = Low Maintenance
Figure 18.  Classification of urban polygons by land use, percentage impervious cover, and degree of land fertilization.

                               Maryland  s
  N = 224 Tons'
  P = 52 Tons   eve"Me   Ocean City
              N = 142 Tons N = 151Ton;
              P = 33 Tons
                         P = 33 Tons
                                  Atlantic County
                                  N = 1,183 Tons
                                  P = 276 Tons
                             Ocean County Monmouth, Bayshore
rvpanrn,,niv     Ocean County  North         Monmouth, NE
Ocean County     Centra|       N = 859 Tons  Long Branch, Deal
South            N = 871 Tons  P = 201 Tons  ocean, Asbury Park
N = 267 Tons     P = ,n, T_               South Monmouth
                                          N = 1,966 Tons
                                          P = 458 Tons
      P = 62 Tons

Atlantic Ocean
                      P = 203 Tons
Figure 19.  Point and NFS discharges to the ACD. Data shown are in tons of TP and NOs-N per year.
                                                                                             BMP Suitability
                                                                                            • -  (1) Suitable
                                                                                            g-  (2) Generally Suitable
                                                                                            M~  (3) Limited Suitability
                                                                                             Unsuitable Areas
                                                                                            D-  (4) Unsuitable Soils
                                                                                            H-  Open Water
                                                                                            E3~  Urbanized Areas
Figure 20.  BMP analysis  using GIS.   Files consider soil suitability, current vegetative cover, and BMP criteria on vacant and
           developed land parcels.

pollution. As most regulatory agencies have discovered,
NPS management programs can  be difficult to imple-
ment, especially when confronting issues of land  use
management. To substantiate the need for new manage-
ment programs amidst these controversies, the ability to
document causal linkages (i.e., to generate data  and
statistics that make the case for NPS pollutant gener-
ators and resultant water quality degradation)  is very
important. The need for documentation of various types
is especially great  given the  less than perfect data re-
cord of water quality in coastal and other waters. All of
these factors come together to make the value of a  CIS
system for water quality management very real.


This CIS-driven analysis indicates that NPS pollutants,
especially the nutrients phosphorus and nitrogen, gen-
erated  from fertilized  fields or maintained  landscapes
surrounding new  residential, commercial,  and other
types of development in drainage systems, contribute
significantly to water quality degradation. In effect, the
particulate-associated  phosphorus and the soluble ni-
trates serve as surrogates for the full spectrum of NPS
pollutants that each rainfall washes  from the land. A
comprehensive water quality  management program
must include structural measures to remove pollutants
this runoff conveys, as well as management of the con-
tributing  landscape to reduce (and perhaps eliminate)
the application of these chemicals  within the drainage.
In planning  new development, management actions
should occur on a variety of levels or tiers.  On  an
areawide basis, growth should proceed (with guidance
and management) in a manner that would reduce total
pollutant discharges;  therefore,  the  total  amount of
maintained area being created should  be as concen-
trated  as  possible. On the  more site-specific level,
measures and construction techniques that reduce the
quantity of pollutants generated are essential. Required
development guidelines must include, but not be limited to:

• Prevention of excessive site disturbance and ongoing
  site maintenance (described as a policy of minimum
  disturbance and minimum maintenance).

• Use of special materials  for reduction  of storm-
  water runoff (porous pavement and ground-water

• Use of stormwater treatment systems (water quality
  detention basins, artificial wetlands).

In sum, the regulatory framework must contain both
"how to build" guidelines, as well as "where not to build"
guidelines. CIS can be a powerful tool in both of these

While inland lakes serve as nutrient traps for these NPS
pollutants, perhaps the greatest potential impact is the
gradual process of excessively enriching our coastal
waters. As population continues to migrate  to coastal
areas, the importance of protecting this fragile ecosystem
Figure 21. For certain regulatory criteria, the proximity of land uses to the water's edge was a consideration in BMP selection.

increases. The  pollution  that new  land  development
generates,  including  the  discharge  of  point  source
wastes, should not be allowed to enter coastal  waters;
it should not be  allowed to destroy the natural balance
that exists  between  land  and  water.  The  concept of
stormwater management takes on  an  entirely different
meaning when viewed as one of the basic mechanisms
of this NPS pollution transport. For centuries, engineer-
ing of the shoreline has intensively focused on protect-
ing human  developments  from  the ravages  of ocean
storms.  Now,  however,  the  converse seems  to  be
emerging: ocean waters need  protection  from  the im-
pacts  of human development.


 1.  Bliss,  N., T.H. Cahill,  E.B. MacDougall,  and C.A. Staub.  1975.
    Land resource measurement for water quality analysis. Chadds
    Ford, PA: Tri-County Conservancy of the Brandywine.
 2.  Cahill, T.H., R.W. Pierson, Jr., and B.R. Cohen. 1978. The evalu-
    ation of best management practices for the reduction of diffuse
    pollutants in an agricultural watershed. In: Lohr, R.C., ed. Best
    management practices for agriculture and silviculture. Ann Arbor,
    Ml: Ann Arbor Science.
 3.  Cahill, T.H., R.W. Pierson, Jr., and B.R. Cohen. 1979.  Nonpoint
    source model calibration in  the Honey Creek  watershed. #R-
    805421-01. Athens, GA:  U.S.  EPA  Environmental Research
 4.  Cahill, T.H. 1980. The analysis of relationships between land use
    and water quality in the Lake Erie basin.  Burlington,  Ontario:
    International Association of Great Lakes Research.
 5.  Cahill, T.H., J. McGuire, and C. Smith. 1993. Hydrologic and
    water quality modeling with geographic information systems. Pro-
    ceedings of the Symposium on Geographic Information Systems
    and Water Resources, AWRA, Mobile, AL (March).
6.  U.S. EPA. 1991. Remote sensing and GIS applications to nonpoint
    source planning. Proceedings of the U.S. EPA Workshop for Re-
    gion 5 and Northeast Illinois Planning Commission,  Chicago, IL
    (October 1990).
 7.  Baker, D.B., and J.W Kramer. 1973. Phosphorus sources and
    transport in an agricultural basin of Lake Erie. Proceedings of the
    16th Conference, Great Lakes Research, Ann Arbor, Ml (Septem-
 8.  Cahill, T.H., and T.R. Hammer. 1976. Phosphate  transport in river
    basins. Proceedings of the International Joint Committee on Flu-
    vial Transport Workshop, Kitchener, Ontario (October).
 9.  Cahill  Associates. 1989. Stormwater management  in  the  New
    Jersey coastal zone. Trenton, NJ: State of New Jersey, Depart-
    ment of Environmental Protection, Division of Coastal Resources.
10.  Delaware Riverkeeper. 1993. Upper Perkiomen Creek watershed
    water  quality management  plan. Lambertville, NJ: Delaware
    Riverkeeper/Watershed Association of the Delaware River.

11.  U.S. EPA. 1993. Guidance specifying  management measures for
    sources of nonpoint  pollution in coastal waters.  EPA/840/B-
    92/002. Washington, DC.

12.  Soil Conservation Commission. 1982. TR-20,  project formula-
    tion—Hydrology. Technical Release No. 20. PB83-223768. Land-
    ham, MD: Soil Conservation  Service.

13.  Cahill,  T.H., M.C. Adams, and W.R.  Horner. 1990.  The  use  of
    porous paving for groundwater recharge in stormwater manage-
    ment systems.  Presented at the 1988 Floodplain/Stormwater
    Management  Symposium, State College, PA (October).

14.  Soil Conservation Commission. 1974. Universal soil loss equa-
    tion. Technical notes,  Conservation Agronomy No. 32. Portland,
    OR: West Technical Service  Center.

15.  Cahill,  T.H., M.  Adams, S. Remalie, and C. Smith. 1988. The
    hydrology of flood flow in the Neshaminy Creek basin, Pennsyl-
    vania.  Jamison,  PA: The Neshaminy  Water Resources Authority

16.  BCPC. 1992.  Neshaminy Creek watershed stormwater manage-
    ment plan, Vol. 1: Policy document, and Vol. II: Plan implemen-
    tation.  Doylestown, PA:  Bucks County Planning  Commission

17.  Clark, J. 1977. Coastal ecosystems: Ecological considerations for
    management  of the coastal zone. Washington, DC: The Conser-
    vation  Foundation.

18.  New Jersey Department of Environmental Protection. 1988. The
    state of the ocean: A report  by the blue ribbon panel in ocean
    incidents. Trenton, NJ.

19.  Cahill,  T.H., M. Adams, C.L. Smith, and J.S. McGuire. 1991. GIS
    analysis of nonpoint source pollution in the  New Jersey coastal
    zone. With Whitney, S., and  S. Halsey, New Jersey  Department
    of Environmental Protection,  Division  of Coastal Resources. Pre-
    sented at the  National Conference on Integrated Water Informa-
    tion Management, Atlantic City, NJ (August).

20.  Cahill,  T.H.,  M.  Adams, C.L. Smith, and  J.S. McGuire. 1991.
    Living  on the edge: Environmental quality in the coastal zone.
    With Whitney, S., and S. Halsey, New Jersey Department  of
    Environmental Protection,  Division of Coastal  Resources.  Pre-
    sented at the  International  Conference on Integrated  Stormwater
    Management, Singapore (July).

21.  Martin, M. 1989. Ground-water flow  in the  New Jersey coastal
    plain. Open File  Report 87-528. West Trenton, NJ: U.S. Geologi-
    cal Survey.

22.  National Oceanic  and Atmospheric  Administration. 1989. Se-
    lected characteristics in coastal states, 1980-2000. Rockville, MD:
    U.S. Department of Commerce,  National Oceanic and Atmos-
    pheric  Administration, Strategic Assessment Branch, Ocean As-
    sessment Division (October).

     GIS Standards for Environmental Restoration and Compliance

                              Bobby G. Carpenter, P.E.
                          The CADD/GIS Technology Center
                            Waterways Experiment Station
                   Army Engineer Research and Development Center
                                Vicksburg, Mississippi

The CADD/GIS Technology Center (formerly Tri-Service CADD/GIS Technology Center) was
established at the Information Technology Laboratory (ITL), U.S. Army Engineer Waterways
Experiment Station (WES), Vicksburg, Mississippi in October 1992. The CADD/GIS Center's
primary mission is to serve as a multi-service  vehicle to set computer-aided design and drafting
(CADD) and geographic information system (GIS) standards; coordinate CADD/GIS facilities
systems within the Department of Defense (DoD); promote CADD/GIS system integration; support
centralized CADD/GIS hardware and software acquisition; and provide assistance for the
installation, training, operation, and maintenance of CADD and GIS systems.

The term  geospatial data refers to data that can be referenced to a specific geographic location on
the Earth. The specific geographic location can be depicted by a graphic feature on a map or
drawing (e.g., a building, monitoring well,  road, location where an environmental sample is
collected, etc.).

Environmental restoration activities involve the investigations and cleanup efforts associated with
the identification and removal of chemical or radioactive contaminants present in the soil,
groundwater, surface water, or sediment.  The most common environmental restoration programs
within the DoD include the Installation Restoration Program (IRP) (involves all Air Force, Army, and
Navy installations) and the Defense  Environmental Restoration Program - Formerly Used Defense
Sites (DERP-FUDS) (administered by USAGE).

Environmental compliance activities involve conducting everyday business practices in a manner
which complies with U.S.  Environmental Protection Agency (EPA) and state/local environmental
regulatory agency laws and regulations. Environmental compliance activities include the
management or removal of "toxic substances" (asbestos containing materials, lead  paint, PCBs),

proper handling of hazardous materiels, proper handling and disposal of hazardous wastes,
monitoring and management of regulated storage tanks, monitoring and management of permitted
surface water discharges and permitted air emissions.

Why Do We Need Data Standards? The collection, storage, management, and analysis of
geospatial data are critical components of environmental restoration and compliance activities.
Geospatial data can be stored in a number of ways (i.e., paper,  microfilm, and/or electronically)
which may not be readily accessible and usable, or easily shared with, or reported to others. CADD
and GIS technology can provide cost-effective and efficient tools to apply and manage such data.
However, careful planning and the use of consistent data storage and CADD/GIS system
implementation standards are necessary to achieve the full potential offered by CADD and GIS

Tri-Service Spatial Data Standards (TSSDS) and Tri-Service Facility Management Standards
(TSFMS) Development. One of the CADD/GIS Center's (http://tsc.wes.army.mil) major initiatives
has been development of the TSSDS and TSFMS. This project  involves the development of
graphic and non-graphic standards for GIS and facility management implementations at Air Force,
Army, and Navy installations, and Army Corps of Engineers Civil Works activities. The
TSSDS/TSFMS are the only "nonproprietary" standard designed for use with the predominant
commercially available "off-the-shelf" GIS (e.g., ESRI ARC/INFO & ArcView; Intergraph  MGE &
GeoMedia; AutoDesk AutoCAD, Map and World; and Bentley MicroStation  & GeoGraphics) and
relational database software (e.g., Oracle & Microsoft Access). This design, in conjunction with it's
universal coverage, have propelled the TSSDS into the  standard for GIS implementations
throughout the DoD, and well as the De Facto standard for GIS  implementations in other Federal,
State, and Local Government organizations; public utilities; and  private industry throughout the
United States and the World.

The TSSDS/TSFMS are distributed via CD-ROM and the Internet (http://tsc.wes.army.mil). A "User
Friendly" interactive Microsoft Windows based software application installs the TSSDS on desktop
computers and networks, provides viewing and printing  capability, and generates SQL code for
construction of the GIS database. The CADD/GIS Center annually updates and  expands the
TSSDS and TSFMS data coverages. Release 1.80 of the TSSDS/TSFMS was completed in
February 1999.

Contributors and Coordination. The TSSDS and TSFMS have been developed based on input
from various technical experts; review and analysis of existing working DoD and state GIS; review
and analysis of various existing database management systems used throughout DoD and the
federal government; and content contributions from federal, state, local, and private sector sources.
The CADD/GIS Center has organized Field Working Groups (FWGs) whose membership is
composed of subject matter, CAD, and GIS technical experts to assist in the development of the
Tri-Service Standards and other CADD/GIS projects. The CADD/GIS Center's Environmental FWG
has been very active is assisting  in the development of the TSSDS and TSFMS.

The CADD/GIS Center is coordinating development of the TSSDS/TSFMS with other DoD and
Federal standards initiatives such as the Defense Environmental Security Corporate Information
Management (DESCIM)  program, the Federal Geographic Data Committee (FGDC), the Defense
Information Standards Agency (DISA), the National Imagery and Mapping Agency (NIMA), and the
Environmental Protection Agency (EPA).

Some of the specific DoD and  Federal initiatives contributing to the environmental restoration and
compliance content  of the TSSDS/TSFMS include: (1) Air Force "Environmental Restoration
Program Information Management System", (ERPIMS) (formerly called IRPIMS); (2) USAGE
Alaska District "Environmental  Data Management System" (EDMS); (3) Army Environmental Center
(AEC) "Installation Restoration Data Management Information System" (IRDMIS); (4) Southwest
Division Naval Facilities Engineering Command, "Navy Environmental Data Transfer Standard"
(NEDTS); (5) USAGE "Formerly Used Defense Site (FUDS) Database"; (6) Air Force Aeronautical
Systems Center (ASC) and USAGE District, Louisville "Draft System Specification for the Technical
Data Management System"; (7) DESCIM Cleanup, Explosive Safety, and other data modeling work
groups; (8) Edwards AFB, Patuxent River Naval Air Station, and other DoD GIS; (9) CADD/GIS
Center's Environmental FWG,  (10) Environmental Protection Agency; (11) FGDC Facilities
Standards Working  Group,  and (12) EPA's Environmental Data Registry.

TSSDS/TSFMS Data Model Structure. Both graphic (i.e., symbols, text fonts, line styles/types,
and level/layer schemas) and nongraphic (e.g., database attribute tables and domains) geospatial
data requirements are addressed in the TSSDS/TSFMS. The TSSDS/TSFMS data model consists
of five basic levels of hierarchy: Entity Sets, Entity Classes, Entity Types (includes Entities) (TSSDS
only), Attribute Tables, and Domain Tables.

Entity Sets (or Themes) are broad groupings of features and related data. The TSSDS/TSFMS
structure currently includes the following twenty-five themes: (1) Auditory, (2) Boundary, (3)
Buildings, (4) Cadastre, (5) Climate, (6) Common, (7) Communications, (8) Cultural, (9)
Demographics, (10) Environmental Hazards, (11) Ecology, (12) Fauna, (13) Flora, (14) Geodesy,
(15) Geology, (16) Hydrography, (17) Improvements, (18) Landform, (19) Land Status, (20) Military
Operations, (21) Olfactory, (22) Soil, (23) Transportation, (24) Utilities,  (25) and Visual.

Entity Classes are logical groupings of features and data within an Entity Set for data management

The TSSDS Entity Classes contain logical groupings of "real-world", geographically referenced
(geospatial) features (entity types & entities) with related (graphic) database attribute tables. Each
Entity Class consists of a separate map or drawing file (i.e., category or design file in MGE;
workspace in ARC/INFO; design file in MicroStation; drawing file in AutoCAD). The current TSSDS
Entity Classes in the Environmental Hazards Entity Set include: (1) characterization, (2) surface
water pollution, (3) munitions  remediation, (4) emergency preparedness (spills, etc.), (5) general,
(6) groundwater pollution,  (7)  hazardous materiels/hazardous waste, (8) munitions
materiel/munitions waste,  (9)  pollution remediation,  (10) regulated tanks, (11) sediment pollution,
(12) sites, (13) building environmental concerns, (14) solid waste, (15) air pollution, and (16) soil

The TSFMS Entity Classes contain logical groupings of (non-graphic) database attribute tables
which contain temporal or event data for specific "business" activities or functions. The TSFMS
Classes in the Environmental Hazards Entity Set include: (1) hazardous materiel management, (2)
munitions waste management, (3) asbestos containing  materiel  management, (4) surface water
discharge management, (5) hazardous waste management, (6) regulated tank management, (7)
PCB (polychlorinated biphenyl)  management, (8) lead paint management, (9) indoor air
management, (10) field measurements management, (11) remediation management, (12)
environmental management,  (13) munitions materiel management, and (14) air quality

Entity Types are a grouping or collection of like,  or similar,  features (entities) which appear
graphically on a map or drawing. Each entity type has an associated attribute table. Entities can be
represented as one of the following three categories:

   •   Boundary (Polygon) - A line string (or group of arcs) which forms the perimeter of an area.
       An example would be the boundary of a lake.

   •   Point - A single point representing the geographical location of a feature; e.g., a well. Points
       are normally represented on a map by a symbol. The TSSDS provide symbols in the native
       formats of AutoCAD, MicroStation, and ARC/INFO.

   •    String/Chain - A line or group of arcs.

An Attribute Table is a relational database table containing non-graphic, or attribute, information
about an entity. Attribute Tables which are linked directly to a graphic entity and contain data
directly related to that entity can be classified as "graphic" (i.e., TSSDS) attribute tables. Attribute
Tables not directly linked to an entity but which contain data required for a "business process" or
function, along with  data and relationships linked through specific data field ids which may be
queried for geospatial and  relational analysis, can be classified as "nongraphic" (i.e., TSFMS)
attribute tables.

Domain Tables contain lists of codes (i.e., permissible or valid values) for populating specific fields
in  the Attribute Tables; e.g., units of measure, material types, etc.

Join relationships are mechanisms by which relational databases link multiple records of a common
attribute or item and provide access to the records through the use of queries. Join relationships
are established in the TSSDS/TSFMS through the use of "Primary  Key" attribute fields in a "parent"
attribute table and "Foreign Key" attribute fields in related "child"  attribute tables.

Integration of Approved FGDC Geospatial Data Standards into the TSSDS. Executive Order
12906, "Coordinating Data Acquisition and Access: The National Spatial Data Infrastructure"
(NSDI), which was signed by the President on 11 April 1994, requires that all Federal agencies use
the FGDC Metadata Standard to document new geospatial data  and  make them electronically
accessible through the use of a National Geospatial Data Clearinghouse. Executive Order 12906
also assigned authority for the development of national geospatial data standards to the FGDC.
The FGDC standards development program ensures that standards are created under an open
consensus, with participation by non-federal and federal communities.

The FGDC geospatial data standards provides a "Logical Data Model" consisting of descriptive
feature names (entity), attribute names, and domain names. However, this data model must be fully
developed into a "Physical Data Model" before it can be implemented in a GIS. That is, all
symbology (e.g., symbols, colors, fonts, line types); level/layer schemas; coverages; file table,
attribute, and domain names which are compatible with commercially available GIS and relational
database management systems must be developed. The TSSDS provides the "Physical Data
Model" for implementation of the approved FGDC geospatial data standards in a GIS. The TSSDS
has been designed to comply with the Spatial Data Transfer Standard (SDTS) data model, and has
been updated to permit compliance with the recently revised FGDC Metadata Standard. Provisions
of the FGDC Bathymetric Geospatial Standard (International Hydrographic Standard (IHO S-57))
were incorporated into the TSSDS Release 1.6. The FGDC Vegetation, Wetlands,  and Soils
standards have been incorporated into the CADD/GIS Center's TSSDS/TSFMS Release 1.8. In
addition, two of the standards currently under development by the FGDC Facilities  Working Group
(Environmental Hazards  Geospatial Standard and Utilities Geospatial Standard) originated from the

      Planning Strategies for Siting Animal Confinement Facilities:
         The Integrated Use of Geographic Information Systems
                  and Landscape Simulation Technologies

                  T. L Cartlidge1, B. Chamberlain2, D. G. Pitt1, M. Olson3,
                           B. Halverson2 and T. Harikrishnan1

             1 Department of Landscape Architecture, University of Minnesota, Minneapolis, MN
                          2Hoisington-Koegler Group, Minneapolis, MN
        Department of Landscape Architecture, The Pennsylvania State University, University Park, PA
Through presentation of a case study, a series of planning strategies for siting animal
confinement facilities in the rural landscape are compared and contrasted. The strategies use
geographic information systems (GIS) technology to develop an environmental protection
framework for a 31,000-hectare watershed in west central Minnesota. The framework is based
on desires to maintain landscape integrity as reflected in enhanced biological diversity,
conserved soil and improved water quality as well as to maintain neighborhood cohesion among
farm and non-farm neighbors in the rural landscape. Design of the production landscape is
compared and contrasted under three alternative scenarios: a) the use of Euclidean zoning; b)
the use of overlay zoning; and c) the use of technical assistance to small-scale operators to
enhance adoption of whole farm planning. Low elevation aerial oblique renderings are
presented as a means of communicating to stakeholder groups the spatial organization and
visual appearance of agriculture in the rural landscape following implementation of the

The location and operation of animal confinement facilities in rural landscapes is an  issue of
large environmental concern in states having economic bases that include livestock production.
In the recent gubernatorial elections in Minnesota, for example, all candidates were  expected to
voice their position on a legislatively proposed moratorium on animal confinement facility siting.
The State's Environmental Quality Board is preparing a Generic Environmental Impact
Statement (GEIS) on animal  confinement facilities in Minnesota. Among other concerns, the

GEIS is examining the effects of animal confinement facilities on water quality, air quality, the
structure of local economies, animal health, human health and the changing characteristics of
rural communities. In addition, the GEIS will also examine the effectiveness and capability of
local land use planning strategies in considering animal confinement facility siting.

The diversity of issues being considered in the GEIS is symptomatic of the complexity of the
issues surrounding animal agriculture in the contemporary rural landscape. No longer can the
focus rest solely on the impacts of production facilities on physical and biological characteristics
of the environment. Evaluation of animal agricultural issues in the twenty-first century also
requires assessment of social, economic and political concerns. The changing structure of
American agriculture has affected the manner in which farmers conduct their business as well
as the definition of who is a farmer.  Scale of operation has increased significantly. Corporate
structures of farm enterprises often  break the traditional land-based agrarian ties of operators as
land becomes little more than another factor of production. Rural community structure
deteriorates as seemingly strangers operate land once cared for by trusted neighbors. Small
township governments sometimes find themselves overpowered by the corporate structures
with which they must often interact in issues related to animal agriculture.

The physical, biological,  social economic and political issues associated with animal agriculture
must be examined at the scale of the individual production facility, the immediate landscape
context of the facility as well as the  region within which the facility is located. Examining the
spatial dimensions of these issues at the landscape and regional levels lends itself directly to
the use of geographic information systems (GIS) technology.  Planners attempting to resolve the
complexities of animal agriculture issues can use GIS technology to integrate the diverse issues
across space and spatial scales, and they can use the technology to develop sophisticated map
representations of their findings and recommendations.  The map representations can serve as
a basis for creating realistic image simulations to present recommendations to various
stakeholder groups.

The University of Minnesota Extension Service sponsored a Rural Landscape Project to
demonstrate how local units of government could use geographic information systems (GIS)
technology and image simulation technologies to enhance their abilities to plan for animal
confinement facilities. The demonstration project was located in the 31,000-hectare Sacred

Heart drainage basin, a tributary of the Minnesota River in west central Minnesota. Figure 1
illustrates existing land use and cultural settlement patterns in the basin.
            Figure 1. Land Use and Cultural Settlement
                            Sacred Heart Basin
                           Renville and Redwood Counties, MN
         Land Use
d] Cropland
•• Pasture
'.':.;,.-'.I Farmstead
•nil Forest
m Other rural development
1.11 Urban
• Open water
• Wetland
                                   and Drainage
                                   /y State highway
                                   /^County state-aid highway
                                     County road
                                   V City street or township road
* Farmer
" Non-farmer
• Retired farmer
Objectives of the demonstration project were:
     •    To evaluate at the local level - in rural communities- the land management
          alternatives that can be used to sustain animal confinement agriculture in the rural
          landscape. This objective sought the identification of strategies to conserve soil,
          maintain water quality, enhance biological diversity, contribute to regional economic
          health, maintain the viability of individual farm enterprises and enhance the well
          being of people living and working in the landscape. The objective sought means of
          building and sustaining healthy ecological systems, healthy economies and healthy
          communities in the Sacred Heart basin.
     •    To broaden input and further discussion of critical issues related to animal agriculture
          and the rural landscape between producers and their neighbors, policy-makers and
          communities, and state and local governments.

     •    To help foster consensus on principles for sustaining the rural landscape of west
          central Minnesota.

The case study is proceeding in three phases. The first phase involved a series of in-depth
interviews and focus group discussions with selected animal agriculture operators in west
central Minnesota. Along with an examination of agricultural statistics for the region and a
review of relevant literature, the interviews and discussions enabled the demonstration team to
better understand the issues associated with animal agriculture operation in the region. The
second phase of the case study, and the subject of this paper, developed a series of planning
strategies to enable continuation of sustainable animal agriculture in one of the region's
watersheds. In developing the strategies, plan view mapped representations of the designs
were created. To better communicate the design strategies to different sets of stakeholders
during  a subsequent round of workshops, the strategies were also represented as hand drawn,
low-elevation aerial oblique renderings. The renderings were prepared to offer stakeholders a
sense of how the design strategies would affect spatial organization of agriculture in the rural
landscape as well as engender a sense of the landscape's visual appearance following
implementation of each strategy. The third phase,  to be conducted in the fall of 1999, will
engage a  series of stakeholder groups in the region in conversation about the planning
strategies. The workshops will involve use of the GIS mapped information as well as the aerial
perspective renderings of the various landscape design alternatives.

Designing an Environmental  Protection Framework
The first step in creating a planning strategy for siting animal confinement facilities involved
development of an environmental protection framework. The purpose of the framework was to
identify those components of the landscape in the Sacred Heart basin that warrant special
protection in meeting the objectives of conserving  soil, protecting water quality, enhancing
biological  diversity and promoting community values. The framework was defined  in terms of
two dimensions. Landscape integrity,  a measure of the ability of landscape structure to maintain
ecological function, defined a balance between environmental quality and agricultural production
values. Neighborhood  cohesion defined relationships between farm and non-farm residents of
the rural landscape.

Defining Landscape Integrity
Landscape dimensions critical to defining integrity included:
      a)  slopes exceeding 12% slope;
      b)  highly erodible soils;
      c)  existing surface hydrology;
      d)  soils that are occasionally or regularly flooded;
      e)  hydric soils as defined by USDA-NRCS criteria.
      f)   existing forest, wetland, open water and grassland communities;
      g)  National Wetland Inventory sites;
      h)  land enrolled in the Conservation Reserve Program;
      i)   that one-fourth of the watershed's total area whose soils contain the lowest potential
          productivity for corn and soybean;
      j)   locations of existing animal confinement facilities; and
      k)  existing field pattens.

Data related to these dimensions were gathered from the Soil Survey Manual of Renville
County, Minnesota, digital aerial ortho-photographs of the watershed, digital raster graphic
images of the 7-1/2 minute Topographic Quadrangles for the watershed and various existing
data sources maintained by agencies of state government (e.g. National Wetland Inventory
sites, existing animal confinement facilities). The data were  compiled into a GIS data base using
Arc-lnfo™technology. Figure 2 illustrates the data used in defining landscape integrity.

Determinants of Soil Conservation and Water Quality. Soils that were steeply sloping, highly
erodible, occasionally or regularly flooded or contained hydric conditions, especially where they
were located in close proximity to surface hydrologic features, were defined as being critical to
maintaining surface and ground water quality.

Determinants of Biological Diversity. The original pre-settlement vegetation of the watershed
contained upland conditions of xeric, mesic and hydric prairie and bottomland conditions
predominated  by cottonwood, elm, silver maple and other floodplain species, oak forest and oak
openings existed on the south-facing bluffs along the Minnesota River. One hundred and twenty
years of agriculture in the watershed has essentially eliminated the prairie vegetation from the
landscape, and over 80% of the wetlands have been  drained. The cultural landscape of farming
introduced upland forest vegetation to the late nineteenth century landscape in the form of

                   Figure 2. Components of Landscape Integrity
                  Hydro log i
                 •*V* Perennial stream
                 .•'-..• Intermittent stream
                 /V Drainage channel
                 ^B Open water
                 I  I Wetland
                 HI National Wetlands Inventory site
                 ^^ Frequently or occassionally flooded land
                    Hydric soil
                    of Soil
 I   I Open water
 EBI Highly erodible land
 HI3I Slopes exceeding 12 percent
                  ^| Open water
                  j-.'.."i Not prime agr land
                  ill Lowest soil productivity
 I  I Wetland
 ^H Open water
 l;:;:;:-l Grassland
    Forest windbreak
 •1 Forest patch
 • CRP I and holdings
shelterbelt patches and strips of hedgerow and fencerow vegetation. As economic scale of farm
operations increased in the last thirty years, many of the hedgerows and fencerows were
removed. Similarly, many shelterbelt patches associated with abandoned farmsteads were
removed. The result of these cultural influences was the creation of a mono-typical landscape
matrix characterized by increasingly larger fields of either soybean or corn. This pattern was
broken only where farmers enrolled less productive  land in the Conservation Reserve Program
(CRP) during the 1980's. Identification of remnant wetland systems, forest patches and strips,
grassland communities and CRP land holdings became a priority step in defining  patterns of
biological diversity in the watershed.  Forest patches were differentiated from strips using a 200-
meter forest patch width criterion.

Defining Neighborhood Cohesion
Assessing neighborhood cohesion involved mapping the locations of farmers, retired farmers
and rural non-farm residents in the landscape. The density of non-farm residents in each land
section was mapped. This strategy assumed that non-farm rural residents have greater aversion
to the externalities of animal confinement operation (e.g. odor, impacts of transportation,  etc.).
Figure 3 illustrates the density of non-farm residents in each  land section.
              Figure 3. Density of Non-farm Rural Residents Per Section
                                   Sacred Heart Basin
                                Renville and Redwood Counties, MN
                                              /\/ State highway
                                                   County state-aid road
                                                \/ County road
                                                \ / Township road or city street

                                                •  Non-farm resident
                                               |i:i:i|||iii|:| One non-farm resident per section
                                                   Two non-farm residents per section
                                                   Three non-farm residents per section
Design of Environmental Protection Framework
The dimensions of landscape integrity and neighborhood cohesion provided the raw material for
creation of the environmental protection framework (see Figure 4). Within the context of these
dimensions, design of the framework pursued five principles:
          Define and protect riparian corridors within the watershed. Accomplishment of
          this design principle ranged from protection of the extensive floodplain systems that
          exist along the bottomlands of the Minnesota River and its tributaries to delineation
          of ten meter vegetative buffer strips along the constructed upland drainage systems.
          Definition of the riparian corridors included bluff and steeply sloped landscapes
          adjacent to the floodplain and drainage systems. It also included integration of any of

     the dimensions of either landscape integrity or neighborhood cohesion that were
     adjacent to the riparian system. For example, corridor definition incorporated areas
     that were adjacent to the riparian corridor and contained soils with low productivity

•    Provide a network of upland connections among riparian corridors. This
     objective established an upland network of connectivity among the riparian corridors.
     Components of the network were defined initially on the basis of remnant wetland,
     woodland and grassland communities as well as existing CRP holdings. The location
     of such factors as the presence of low productivity or hydric soil conditions, steep
     slopes, highly erodible soils, or National Wetland Inventory sites connected remnant
     communities to themselves and to the riparian corridors.

•    Provide buffering around non-farm residents living  in the rural landscape.
     Areas containing high concentrations of rural non-farm residents became
     components of the environmental protection matrix.

•    Use existing patterns of cultural settlement to establish the boundaries of the
     framework. Within the context of maintaining the integrity of the spatial pattern of the
     environmental protection framework, existing field patterns provided a basis of
     defining framework edges.

•    Define uses appropriate to landscape characteristics. Within the environmental
     protection framework, land uses would be permitted based on their appropriateness
     to the resource characteristics of the landscape. For example, low input perennial or
     agro-forestry crops were considered appropriate in vegetative buffers  of upland
     drainage riparian systems.

    Figure 4. Environmental Protection Framework
                        Sacred Heart Basin
                     Renville and Redwood Counties, MN

                               /%/" State highway
                               /\/ County state-aid road
                                / \ , County road
                                   / Township road or city street
                                    Open water
                                    Environmental Protection
                                               6 Miles

Defining Farm Production Units in the Landscape
The environmental protection framework established an armature within which commodity
production values in the landscape can be realized.  Farm production units involved in pursuit of
these values were classified along three continua: crop diversity; farm scale; and operation
distribution. Crop diversity varied from highly specialized production of one or two cash crop
commodities (e.g. corn and soybean) to farm units that combine dairy and pork production with
forage and pick-your-own truck crop production. Farm scale referred to the capital  intensity of
the farm enterprise, and it generally correlated with geographic size of the operation. Finally,
operation distribution referred to the geographic dispersion of the enterprise across the

 landscape. Distribution varied from farm units operating in one concentrated locus to units
operating across several different locations. As illustrated in Figure 5, these three criteria were
used to describe farm units operating in the Sacred Heart watershed.
                      Figure 5. Farm Production Units Can Be
                       Characterized Along Three Continua.
                                               Crop Diversity
                     Operation Distribution

Designing the Production Landscape
Landscape not included in the environmental protection framework was defined as production
landscape. Three principle functions occurred within the production landscape: animal
production; crop production; and residential uses associated with both farm operators as well as
non-farm inhabitants. The case study investigated three scenarios by which these uses might
be managed and spatially organized within the production landscape to establish a sustainable
rural environment. Each scenario described a prototypical set of conditions that might be found
in areas wherein the policies inherent in the scenario were adopted.  A low-elevation, oblique
aerial perspective rendering accompanies the presentation of each scenario to depict its spatial
structure and appearance in the landscape.

Scenario One: Large Scale Zoning Approach
The first design scenario (see Figure 6) involved establishment of four exclusive use zoning
     a)    an environmental protection zone;
     b)    an animal agriculture enterprise district;
     c)    an intensive cropping district;  and
     d)    a rural residential district.

Characteristics of the environmental protection zone were previously defined, and the
characteristics of the remaining districts are described below.

Animal Agriculture Enterprise District
An animal agriculture enterprise district consisted of large-scale confinement facilities clustered
within the same geographic proximity. Animal agriculture was the permitted use within the
district.  Location of the considered such criteria as: a) proximity to components of the
environmental protection zone;  b) maintaining a 1.5 mile buffer between the enterprise district
and components of the rural residential district; and c) existing transportation infrastructure. A
set of performance criteria relating to air and water quality, resource recycling and the
enhancement of biological diversity regulated operation of the animal  agriculture uses within the
zone. The enterprise districts were large enough to permit establishment of centralized
treatment facilities for manure management and treatment.

A manure management cooperative,  structured similarly to a rural electric cooperative,
managed the waste products of several enterprise districts. The scale of the cooperative's
operation permitted it to become involved bye-product. The cooperative established and
maintained integrated relationships between generators of manure sludge within the enterprise
district and users of manure sludge in the adjacent cropping district. The cooperative also
managed integrated relationships between forage producers in the cropping district and forage
consumers in the enterprise district. These cooperative relationships allowed the creation of
closed production systems  in which size of the enterprise district was  determined by the forage
production and sludge handling capabilities of producers in the adjacent cropping district.

Cropping District
Within a cropping district, large scale crop production, intended in part to supply forage for and
process manure sludge from adjacent animal enterprise districts, was the preferred use. The
location of the cropping districts was based on three criteria: a) soil productivity, b) constraints
and opportunities defined by the environmental protection framework; and c) provision of a 1.5
mile buffer between the enterprise districts and the rural residential districts. The transportation
infrastructure within a cropping district was maintained primarily to foster delivery of products to
market and interaction between individual producers in the cropping district and animal
producers in the enterprise district. Management of this infrastructure involved the  conversion of
redundant township roads into field roads or the transformation of these rights-of-way into
components of the environmental protection framework. Performance standards and incentives
operating within  the cropping district encouraged adoption of best management practices,
establishment of a routine fallow program for fields and creation of the environmental protection
framework. Since lands within  the cropping district would be down-zoned to crop production
uses, a transfer of development rights program (TDR) or a purchase of development rights
program (PDR) was established to permit landowners to participate in the development
windfalls that accrued to landowners in the rural residential district.  Either of these  programs
allowed development rights transfers into the rural residential district were such non-farm uses
were permitted and encouraged.

Rural Residential District
The fourth zoning district was a rural residential district. Within this district, rural residential uses,
hobby farms and small-scale farm operations were primary uses. Pre-existing residential uses in
the cropping or enterprise districts remained in their current locations, although they would be
eligible for relocation into a rural residential district. All future rural residential development
would occur within a rural residential district. Future development was encouraged to follow
cluster strategies in implementation, and  density  bonuses became available to  developers
pursuing cluster strategies. Location of the rural residential districts were based on: a)
constraints and opportunities defined by the environmental protection framework; b) provision of
a 1.5 mile buffer between the enterprise districts  and the rural residential districts; and c) the
suitability of soils for development and domestic waste management (either as  septic tank
drainfield effluent or alternative technologies). Within the rural residential districts, the existing
pattern of transportation infrastructure was maintained.

Policy Environment of Large Scale Zoning Approach
The policy environment of the large scale zoning approach involved an intensive regulatory
framework. Having defined the environmental protection framework, a significant amount of
centralized planning established the boundaries of the other three districts. An administrative
framework managed the performance standards and incentive systems operating in the
districts, and the framework established the operating procedures of the development rights
              Figure 6. Large Scale Zoning Approach
transfer program. A forage production and manure sludge-handling cooperative managed
integrated relationships between forage production and sludge handling.

Scenario Two: Mixed Scale Overlay District Approach
Rather than relying on the rigidity of an Euclidean approach to land use separation through
exclusive use districts, scenario two adopted the flexibility of overlay district zoning. In this
scenario, an application for development or expansion of an animal confinement facility
triggered a request to implement an overlay district. The district consisted of all lands within a
1.5 mile radius from the confinement facility (see Figure 7). The application was reviewed in a
site selection process that examined the overlay district's location relative to: a) constraints and
opportunities defined by the environmental protection framework; b) existing transportation
infrastructure; and c) proximity to rural non-farm residential uses.

Applicants were required to negotiate and secure operating agreements with all landowners
within the confines of the overlay district. The requirement to negotiate operating agreements
with all landowners within a 1.5 mile radius of the led to the agglomeration of such facilities
within the rural landscape. However, rather than  legislating these concentrations as was true in
the zoning approach, the agglomerations evolved through market activity. Performance criteria
relating to air and water quality, resource recycling and the enhancement of biological diversity
governed operation of the livestock facility within the overlay district. To the extent practical,
operators were encouraged to pursue bye-product recovery (e.g. capture of methane production
for small-scale energy generation). The existing pattern of transportation infrastructure within
the overlay district was maintained.

The overlay district concept required a less intensive regulatory framework than did the large
scale zoning approach. Administration of the site selection  and overlay district creation process
was required along with administration of performance standards. Otherwise,  administrative
functions occurred in the private sector through landowner  to landowner transactions or through
adjudicatory processes initiated to resolve compensatory claims issues.
                Figure 7. Mixed Scale Overlay District Approach

Scenario Three: Small Scale Diversified Farming Approach
The third scenario (see Figure 8) assumed a livestock commodity price and economic structure
that favored a return to smaller scale more diversified farm operations. In this scenario,
individual farms pursued both crop and livestock production, and alternative and low-input
agricultural systems became more widely accepted. Individual operators became both forage
producer and forage consumer;  both manure producer and manure consumer. Farming became
a closed system wherein operating scale was based on potential productivity and manure
assimilative capacity of soils. Farmers pursued whole farm planning with technical assistance
from USDA-NRCS personnel. Implementation of this planning activity, coupled with incentives
to encourage adoption of best management practices, enabled establishment and maintenance
of the environmental protection framework. Performance criteria relating to air and water quality,
resource recycling and the enhancement of biological diversity governed farming operations.
The policy environment of scenario three required regulatory activities associated with
administration of performance standards. The technical assistance involved in whole farm
planning coupled with educational efforts and incentives associated with adoption of best
management practice adoption also required administrative activity.
                         Figure 8. Small Scale Diversified
                                 Farming Approach

None of the three scenarios, by itself, represents the single planning approach to
accommodating animal confinement facilities in the rural landscape. Collectively, the scenarios
represent a catalog of strategies that can be applied in different physiographic and cultural
contexts. In many instances, the most appropriate solution may involve a combination of
strategies. Regardless of how the production landscape is designed and managed, it is
important to frame the designs around the establishment and maintenance of the environmental
protection framework. GIS technology is well suited for design of the framework. Execution of
the alternative production landscape designs is also well suited to the application of GIS
technology. Maps produced from GIS technology lend themselves readily to the production of
illustrative graphics, which can be used to further explain the spatial structure and appearance
of a sustainable rural landscape for animal production.

    A GIS-Based Approach to Characterizing Chemical Compounds in Soil and
                           Modeling of Remedial System Design
                    Leslie L. Chau, Charles R. Comstock, and R. Frank Keyser
                          ICF Kaiser Engineers, Inc., Oakland, California
Introduction: The Problem

The cost-effectiveness of implementing a computerized
geographic information system (CIS) for environmental
subsurface characterization should be based on long-
term remedial objectives. A CIS project was developed
to characterize soil contamination and to provide design
parameters for a soil vapor extraction remedial system,
as part of a  $120-million  remediation and "land sale"
project in California. The primary purposes of the CIS
were to efficiently combine and evaluate (model) dispa-
rate data sets, provide "new" and more useful informa-
tion to aid in short-term engineering decisions, and
support the development of long-term cleanup goals.

The project had a major change in scope early on, and
the schedule was expedited to allow for the develop-
ment of "land  sale" options and for actual site redevel-
opment at the earliest opportunity.  Characterization of
chemically affected soil would have been compromised
given the above  circumstances without an  ambitious
undertaking of concurrently developing and implement-
ing a CIS with three-dimensional  (3-D) geostatistical
and predictive modeling capabilities.

The GIS Approach

Computer solutions included the use of a cross-platform
(DOS and UNIX) GIS to quickly and systematically in-
corporate spatial and chemical data sets and  to provide
a distributed data processing and analysis environment
(see Figure 1). Networked, DOS-based relational data-
bases were used to compile and disseminate data for
the numerous investigatory  and   engineering  tasks.
UNIX-based computer aided  design (CAD) and model-
ing applications received data from databases,  per-
formed  quantitative  analyses,  and   provided  3-D
computer graphics. Given the aggressive project sched-
ule, exclusive  use of one platform would not be realistic
due mainly to the limited data modeling capacity and 3-D
graphics in DOS systems. On the other hand, the high
startup and operating costs  of several UNIX worksta-
tions would render their exclusive use much less cost-

The hardware  and  software configurations were inte-
grated in a client/server Intergraph  InterPro 6400 with
48 megabytes of memory. It is largely a 3-D CAD system
with add-on modules of geologic mapping and 3-D vis-
ual models capable  of consolidating  both environmental
and engineering parameters for analysis (see Figure 1).
Textual environmental and geologic data were extracted
by SQL queries  from  relational databases and were
transferred to  mapping  and modeling  modules  via
PC/TCP cross-platform linkage.

The GIS assisted in making short- and long-term deci-
sions  regarding health-risk-based  regulatory strategy
and engineering feasibility. Use of spatial statistical and
predictive  models was part  of a CIS-based decision-
making loop (see Figure 2). The  process  supported
concurrent activities in:

• Data collection: field program.

• Numerical models of remedial system configurations.

• Development of cleanup goals from health  risk

• Remedial design  with CAD capability.

Site Background

In early 1993, the remedial investigation of the operable
unit for soil at a former aircraft manufacturing facility in
southern California was thought to be ready for remedial
alternatives feasibility study. ICF Kaiser Engineers, Inc.,
was awarded the contract to perform feasibility studies
on applicable soil cleanup technologies and to sub-
sequently design and manage the installation and early
operation of the selected technologies. After $700,000
was spent evaluating data collected by previous consult-
ants, it was decided that an additional $5 million worth

Token Ring Network



Data Visualization and Computer-Aided Engineering



                               Health Risk
                                                  Modeling Module
                                       3-D Kriging Spatial
                                        3-D Vapor Flow
                                       3-D Chemical Mass
                                                             High Impact 3-D
                                                              Solid Models
               Analysis of
               Static and
              Dynamic Mass
  3-D Piping


System Housing
Accelerated Implementation To Establish Cle
Require Low Overhead Upfront Cost and Off-the
Data Visualization
Report Generation
anup Goals
Shelf Software

Regulatory and

Extended Engineering Phase
Specialized CAD Software and Analytical Staff

Figure 1.  Multiplatform GIS project.
                   Ongoing Site Investigation
                   Soil/Soil Vapor Sampling
                      Data Processing
                     1. Storage (RDBMS)
                      2. Dissemination
                     1. Spatial Correlation
                     2. Fate and Transport
                                Treatment Alternatives
   Health Risk
                                Environmental Planning
                                 Site Redevelopment
Establishment of
 Cleanup Goals
                       Final Remedial
Figure 2.  GIS-assisted decision tree.
of field activities were required to more definitively esti-
mate the volume of chemically affected  soil and the
nature and extent of contamination at the facility. Be-
cause of the data gaps, the selection and  design of
alternatives could not be addressed with a high degree
of certainty. Hence, computer assisted  data processing
was crucial to speed up the feasibility study, accelerate
downstream work, and  reduce the overall project sched-
ule to the minimum.

The site is environmentally complex, covering an area
of approximately 120 acres. As a result of nearly  half a
century of aircraft production and development, soil be-
neath the facility is affected by fuel and  heavy oil hydro-
carbons (TPH) commingled with volatile  compounds,
mainly  perchloroethylene (PCE) and trichloroethylene
(TCE) (see Figure 3). Ground  water at 170 feet below
ground surface is  affected by TCE and PCE, but it is not
part of the drinking water aquifer. The facility has  been
demolished, and  shallow contaminated soil  has   been
excavated and  back-filled to an interim  grade.


Health-Risk-Based Cleanup Goals

Central to determining  the volume and  kinds  of data to
be collected was  the question  of whether chemicals in
soil represented potentially unacceptable risks to human

                                                            PCE+TCE TOTAL CONCENTRATION

                                                             • •1.0.
                 VAPOR DISTRIBUTION
Figure 3.  Aircraft manufacturing facility in California. Outline of demolished buildings located at the 120-acre site are shown as
         surface features for reference. A geostatistical model of a 3-D kriged VOC soil vapor cloud in the subsurface was simulated
         with Intergraph's MGVA. Views displayed are: 3-D isometric, vertical section of chemical isoplaths, and a nearly plan view.
         Digital simulation also illustrates VOCs affecting ground water in a dispersive nature at a depth of nearly 170 feet bgs
         (shown at bottom of isometric view).
health and to the environment, with the former being of
particular concern to construction workers onsite during

Because site redevelopment was scheduled to begin in
the near term, data collection and CIS analysis concen-
trated on shallow depths (top 20 feet), with decreasing
sample  density at greater depths. A health-risk-based
cleanup goal (HBCG) approach to collecting more data
was to establish cleanup goals for near-term  remedia-
tion of the shallow soils as well as for long-term remedial
measures of contaminated soils at greater depths. Fur-
ther, various regulatory agencies had  to  approve  the
estimated cleanup goals in a short  time. Ongoing  site
demolition and  excavation  schedules encouraged  the
aggressive regulatory negotiations. The shallow cleanup
goals for volatile organic compounds (VOCs) and TPH
determined the volume  of contaminated soil to be re-
moved. At greater depths, data gaps were minimized to
more definitively characterize  the nature  of TPH and
VOC contaminations and to facilitate the implementation
of long-term remedial objectives (i.e., in situ soil vapor

In situ soil vapor and soil sampling composed the field
program, which provided data to map  the  subsurface
distribution of volatile organic compounds, including
TCE and PCE. Only in situ soil sampling was used for
characterizing TPH. The ratio of soil vapor to soil sam-
ples was 4:1. No previous soil vapor information was
available. ICF Kaiser has been refining the technique of
comparing results from paired soil vapor and soil sam-
ples in past and similar projects. Hydraulic probes were
used instead  of drilling to acquire soil vapor samples at
shallow  depths. This minimized waste  and cost in the
field program significantly.

Risk Assessment and Spatial Analysis

Human  health risk analyses were conducted for the
entire site, and risk factors were contoured and overlaid

on maps of past usage and known soil contamination
areas. Before the risk modeling could proceed, chemical
and lithological data gathered in the past 7 years and
those acquired  by ICF Kaiser  populated the environ-
mental relational databases. Approximately 522 soil va-
por probes were  located  in 100-square-foot spacings
with additional probes in areas requiring better plume
definitions.  The  database  contains  approximately
15,000 xyz-records of soil and soil gas laboratory ana-
lytical results. This information in text and graphics form,
combined with site infrastructures and building outlines
with  attributes of "past usage," were  stored as map
layers,  making up the CIS nucleus. Accuracy  of site
maps was verified with aerial photographs when avail-
able. Data types combined for computerized evaluation
included known locations of contaminated soil, contami-
nated ground water, soil types, and site features. Com-
posite risk maps of the above data were analyzed for
data gaps at discrete depth intervals. This analysis was
performed  while the field program  was in progress
and hence gave guidance to optimize the locations of
additional data points and to  minimize the number
of samples taken.

The MGLA/MGLM mapping module and the MSM ter-
rain modeling module tracked the earth excavation and
removal of contaminated  soil. Excavation was  largely
part of site demolition. It also expedited the removal of
TPH contaminated soils,  however, because no other
short-term means of remediation are available for these
substances. Tracking of removed soils was essential
because concurrent field activities were occurring in site
demolition, data gathering, and  risk modeling.

The CIS coordinated all three. Geologists and surveyors
provided  terrain data from daily  excavation activities,
which were transcribed into  database formats. Maps
illustrated the locations of excavated soil and removed
chemicals in soil at various depths. Although TCE and
PCE were of foremost concern as health risks, all com-
pounds and  some  metals  identified  in  soil  were
screened for unacceptable risk. Terrain modeling (map-
ping) as part of health risk assessment may seem un-
usual,  but  results of estimated cleanup  levels  and
accurate locations of left-in-place contamination, mostly
soils at greater depths, were critical to the cost-effective-
ness and proper design of long-term remedial systems.

Characterization of Subsurface VOCs

In situ soil vapor extraction (SVE) of total volatile com-
pounds in dense nonaqueous, liquid, gaseous, and ad-
sorbed solid forms in the subsurface produced favorable
results that have been well documented in recent years.
ICF Kaiser proposed a very large-scale SVE system
(see Figure 4),  perhaps the  largest  yet, as long-term
remedial technology for this former aircraft manufactur-
ing site. The primary design problem was speculating on
air flow capacity and operating time of the  complex
system components. The SVE system comprises three
fundamental elements:

• Front-end,  in situ  subsurface vents  (totaling 193

• Applied vacuum  and air transport manifolds  linking
  the subsurface vents to the treatment compound (dis-
  tance of one-quarter mile  with over 100 manifolds).

• A multivessel activated carbon treatment system.

To size the pipes, carbon vessels, and vacuum required
to achieve a certain rate of VOC removal, the total mass
and  nature of sorption had to  be  known. Due  to the
schedule-driven nature of this project, the SVE design
accounted  for  the time  needed  to accomplish the
cleanup goals.

To estimate the extent and  total mass of VOCs in the
subsurface, soil vapor data were input to a 3-D kriging
algorithm (1) to produce a concentration  continuum
model (see  Figure 3).  This solid model of predicted total
VOC concentrations took the form of a uniformly spaced
3-D  grid-block that completely encased the site. Cell
sizes ranged from 10 to 20 cubic feet, depending on the
model  run,  number of data clusters, density of data
points  in areas of clustered data, and  the standard
deviation of variances for estimated values in all cells.
The  Fortran program estimated a  concentration  value
for each cell based  on the nearest field sample(s).

The validity of such "block kriging" models can be judged
by the  size of the variances, smoothness, and agree-
ment with nearby field data. Because volume is a known
quantity in kriging, the total  mass can be calculated by
incorporating soil bulk density or porosity, both of which
were less than abundant for this investigation. Render-
ings  of kriged  results in 2-D plan view contour maps,
cross-sectional maps, and 3-D "vapor cloud" (see Figure
3) were included in client reports and used in regulatory
presentations and public forums.

Remedial Design Layout

Final Extension of a Fully Integrated CIS

With the total mass and extent of  VOCs derived from
3-D kriged results, the applied vacuum at individual vent
heads and  the cumulative pressure (negative) neces-
sary to extract and transport VOC vapors from the sub-
surface to the treatment system can be estimated. We
performed 3-D air flow analysis by use  of finite differ-
ence fluid flow models and chemical transport models.
The  Fortran codes  used to  approximate  compressible
flow  and chemical transport  were AIR3D (2) and VT3D
(3), respectively. Air flow simulations focused on  maxi-
mizing vacuums at the shallow depths down to 20 feet
to expedite  remediation of contaminated soils that were

Figure 4.  A rendering by the Intergraph 3-D Plant Design System of an in situ soil vapor extraction and treatment system. The
         cutaway section located near the upper left portion of the figure exposes some of the 193 subsurface extraction vents
         bottoming at 120 feet (bgs). These vents are located in a cluster for long-term extraction of the VOC vapor cloud presented
         in Figure 3. Vents are connected to a system of parallel airflow manifolds (right side of figure), which runs one-quarter of
         a mile to the treatment compound (foreground of figure).
not removed during site demolition and excavation. The
lower depths were  also  included in each simulation.
Transient mass transport models incorporate flow fields,
given by flow models,  and predicted cleanup times
based on  established HBCG cleanup goals. As VOC
concentrations in an operating SVE system fall below
cleanup levels in the top 20 feet, thus minimizing human
risk,  available vacuums thereafter will be diverted to
vents at lower depths to be part of long-term extraction
scenarios. Models suggested that cleanup for the top 20
feet can be accomplished within 1 year.

Numerical models prescribed  vacuum levels at each
vent head, which is the aboveground segment of a
subsurface SVE vent. The 193 vents are connected to
a system of parallel manifolds (see Figure 4) that trans-
port vapor to the treatment system.  With the vacuums
known at vent heads, the size of manifolds and capacity
of vacuum  blowers can  be determined and integrated
into the overall system design. With 3-D Plant Design
module as part of the Intergraph CAD/GIS,  manifold
layouts and treatment compound can be modeled in 3-D
and easily checked for pipe routing  interferences. The
final layout of the SVE system was overlaid onto contour
maps of total VOC concentrations to check on accuracy
and completeness of vent locations and manifold layouts.


Maximized Visual and Analytical Responses

One goal of this project was to expedite regulatory
negotiations and gain early acceptance  of cleanup
goals. The computerized data processing and visualiza-
tion contributed generously to the rapid understanding
of modeling results by expert regulators and the  lay
public. Likewise,  the  CIS facilitated the response to
regulatory comments.  Positive  comments  first came

from the client's in-house review of model  results and
the high impact 3-D color rendering of kriged VOC dis-
tributions in the subsurface (see Figure 3).

Analytically, benefits were derived from the efficiency of
electronic data access and the ability to "predict" the
presence of contaminant in areas with sparse field data.
The process of kriging involves the linear interpolation
and extrapolation  of existing  data. The resultant con-
taminant distribution is a "conservative" model that pro-
vided  the   best   fit  with  field  data  and validated
conceptualized subsurface conditions. Further, models
provided conservative  estimates of mass and extent of
PCE and TCE contaminations. Kriging also  provided
information on the uncertainty of the predicted chemical
distribution, which  is extremely useful for regulatory dis-
cussion and system design. The efficiency of computer
models allowed  investigators to  perform numerous
model  runs with varied boundary parameters, such as
cell size and search radii, in the kriging process.

Accurate mapping of excavated soil and the removal of
most TPH source  areas provided the  incentive to criti-
cally assess the feasibility of a no-action remedial sce-
nario  for these substances  at greater  depths.  With
removal of many TPH source areas, 1 -D finite difference
models (4) were used  to assess the mobility of TPH in
NAPL and adsorbed residual phase. Specifically, mod-
els assessed the likelihood of largely residual-phase
TPH  affecting ground water and  migrating upward to
affect indoor air volumes via gaseous diffusion. Results
were  extremely favorable; models predicted negligible
likelihood of TPH affecting  ground water or indoor air
volumes.  Combined with CIS graphic evidence of spe-
cific areas of excavated soil and  the absence of TPH
sources,  regulatory agencies accepted the  model  re-
sults,  and the  no-action remedial alternative for TPH
was approved.


1.  Deutsch, C.V., and A.G.  Journal.  1992. Geostatistical software
   library and user's guide. New York, NY: Oxford University Press.

2.  U.S. Department of the Interior Geological Survey. 1993. AIR3D:
   An adaptation of the ground-water flow  code MODFLOWto simu-
   late three-dimensional air flow in the unsaturated zone. Books and
   open file reports. Denver,  CO.

3.  Zheng, C. 1994. VT3D: Numerical model for VOC removal from
   unsaturated  soil (draft). Bethesda, MD: S.S. Papadopulos and
   Associates, Inc.

4.  Rosenbloom, J., P. Mock, P. Lawson,  J. Brown, and H.J. Turin.
   1993. Application of VLEACH to vadose zone transport of VOCs
   at an Arizona Superfund site. Groundwater monitoring and reme-
   diation (summer),  pp. 159-169.

     Using GIS/GPS in the Design and Operation of Minnesota's Ground Water
                           Monitoring and Assessment Program
                Tom Clark, Yuan-Ming Hsu, Jennifer Schlotthauer, and Don Jakes
                     Minnesota Pollution Control Agency, St. Paul, Minnesota

                                        Georgianna Myers
                        Water Management Consultants, Denver, Colorado

Minnesota's Ground Water Monitoring and Assessment
Program (GWMAP) is administered  by the Minnesota
Pollution Control Agency (MPCA) to evaluate baseline
ground-water quality conditions regionally and state-
wide. The  program uses a systematic sampling design
to maintain uniform geographic distribution of randomly
selected monitoring stations (wells) for  ground-water
sampling and data analysis. In  1993, geographic infor-
mation system (CIS) and  global positioning  system
(GPS) technologies were integrated into GWMAP, auto-
mating the selection of wells and the field determination
of well locations.

GWMAP consists of three components: the statewide
baseline network, regional monitoring  cooperatives,
and a trends analysis component. In the statewide
baseline network, Minnesota is divided into over 700
121-square-mile grid cells,  each with a centralized,
9-square-mile sampling region.  Wthin each target area,
single-aquifer, cased and grouted wells are sampled for
about 125  metals, organic compounds, and major cat-
ions and anions. We are currently finishing the second
year of a 5-year program to establish the statewide grid.
When complete, the statewide baseline component will
consist of  about 1,600 wells representing Minnesota's
14 major aquifers.

In 1993, approximately 4,000 well construction records
were selected for geologic and hydrologic review, using
a CIS overlay, from a database of 200,000 water well
records maintained in the state's County Well Index
(CWI). Using GPS, 364 wells were sampled and field
located. The  semiautomatic well selection process uses
existing electronic coverage of public land survey (PLS)
data maintained in CWI in conjunction with the digitized
systematic sampling grid. CIS has greatly reduced the
time needed for selecting sampling stations. With the
combination of CIS and GPS, program costs have de-
creased, allowing more resources to be applied toward sam-
pling, while efficiency and quality of data have improved.


Quantitative assessment of ground-water quality condi-
tions requires a highly organized  data collection pro-
gram that includes statistical  evaluation of monitoring
results (1, 2). States have difficulty providing the staff
and financial resources necessary to generate state-
wide quantitative  ground-water information. Wth the
use of geographic information system (CIS) and global
positioning system (GPS) technologies, however, states
have the potential to improve the quality of environ-
mental monitoring programs and to reduce the amount
of staff time needed to collect and evaluate data, thus
decreasing costs. The degree to which states realize
these potential benefits depends largely on how effec-
tively the technology can be incorporated into the design
of the monitoring  program. This paper describes how
CIS and GPS technologies are being integrated into the
design and operation  of Minnesota's Ground Water
Monitoring and Assessment Program (GWMAP) to im-
prove overall effectiveness.

The Minnesota Pollution  Control Agency (MPCA) has
sampled and analyzed ambient ground-water quality in
the state's 14 principal aquifers since 1978. In 1990, the
MPCA began a redesign of its ground-water monitoring
program to better assess water quality conditions state-
wide (3). Three program components resulted from the
redesign: a statewide  baseline network  for complete
geographic coverage, a trends analysis component for
intensive studies of how ground-water quality in specific
areas changes with time, and  a  regional monitoring
cooperative link to governmental units such as counties

to meet specific local ground-water assessment needs.
This paper describes the design and operation of the
statewide baseline network.

The design of the statewide network is geographically
and statistically based to automate well selection and
data interpretation. In 1993, the MPCA began integrat-
ing  CIS and GPS technologies into this  part of the
program. The implementation  of CIS  and GPS sur-
passed our expectations by reducing staff time re-
quired to select wells and evaluate analytical results
(see Table 1). In addition,  through  the elimination of
previously uncontrollable variables, the use of CIS and
GPS has increased the accuracy of GWMAP data.

Monitoring Program Description

Since 1992, GWMAP has selected 150 to 250 existing
water supplies yearly for ground-water sampling and
analysis of about 125 parameters, including major cat-
ions and  anions,  metals, and  volatile organic com-
pounds. Well selection  is a fundamental  element of
GWMAP that, if efficiently performed, supports the pro-
gram objectives by upholding the quality of the monitor-
ing  data and minimizing the operating costs.

A key  to the interpretation of monitoring  data is  the
technique  used to select wells for sampling (2, 4, 5).
Minnesota  has  over 200,000 active water wells with
approximately 10,000 new installations annually.  For
each well selected for GWMAP monitoring, a hydrologist
must individually review many well construction records.
An automated prescreening mechanism to facilitate well
selection can result in considerable time (and therefore
cost) savings. GWMAP chose CIS as the best tool for
this task. CIS enables the program to  combine a sys-
tematic sampling technique with hydrogeologic criteria
to ensure an efficient and consistent selection process.
As  Table 1 shows, CIS allowed us to more than triple
our geographic coverage and  wells initially selected,
while  dramatically  reducing the records that must be
individually reviewed. We realized a time savings of 2
months compared with  the time required  before CIS

In general, systematic sampling techniques use a ran-
domly  generated uniform grid  to  determine sampling
locations in space and/or time (5). Systematic sampling
was initially implemented in  GWMAP in 1991  using  a
manually generated spatial grid defined by the public
land survey (PLS) (3). Although the PLS is not 100

Table 1.  Well Selection in 1992 and 1993
percent geographically uniform, it was selected for the
grid to expedite well selection from existing digital data-
bases in which wells are organized by PLS location.

Systematic Sample Site Selection
Using GIS
Systematic sample site selection is a three-step process.
First, a database search of Minnesota's County Well Index
(CWI) (6), containing nearly 200,000 driller's records, is
conducted to include  all available water wells in the
region of interest. Second, the candidate pool is reduced
to those wells located within regularly spaced grid cells.
Third, further wells  are eliminated from the candidate
pool by applying geologic and well construction criteria
mandated in the GWMAP design (7).

Generating the Sampling Grid
The statewide sampling grid was generated from a randomly
selected origin (8). This grid consists of approximately 700
square cells, 11  miles on a  side  (see Figure 1). The
centroid of each cell is consecutively numbered and was
extracted to produce the origin of the sampling zone.
Figure 1.  Statewide baseline network sampling grid.
9 counties
26 counties
PLS Sections
Well Logs
Well Logs
6 months
4 months

Establishing the Sampling Zone

Each sampling zone consists of a 3- by 3-mile box from
which potential sampling sites are selected. It is gener-
ated by computing the coordinates of the four corners of
the box using the grid cell's centroid as the origin. To link
the sampling zone and grid cell, both are identified with
the same numerical code.

These sampling "target" zones,  a series of regularly
spaced, 9-square-mile boxes, are then made into a CIS
coverage and overlaid on top of the PLS coverage to
extract those sections that are associated with each of
the sampling zones. Ideally, each sampling zone should
cover exactly nine PLS sections (3). Due to irregularities
in the PLS system, however, portions of 16 to 20 sec-
tions usually fall within the sampling zone of each  cell
(see Figure 2).

   / \ /  PLS Boundary
Sample Zone
                     Sample Grid
                              County Border
Figure 2.  PLS and the sampling grid, Watonwan County.

Selection of PLS Sections

The  PLS  coverage was derived from the Minnesota
Land Management Information Center (LMIC) "GISMO"
file. It was originally created in 1979 by digitizing every
section corner in  Minnesota from the U.S. Geological
Survey (USGS) 7.5-minute quadrangle map series.

The  PLS section  information is necessary in the well
selection process because the original well construction
logs, maintained by the Minnesota Geological Survey
(MGS), are organized by PLS. Although most of the well
selection  process can  be  automated,  manual  file
searches for well records are still necessary and require
the PLS information.

Well Selection

After identifying the PLS sections within the sampling
grid, the statewide well database is imported as a point
coverage and  overlaid with the selected  PLS section
coverage. Thus, all wells that fall within the 16 to 20
sections are selected as potential candidates. The ac-
curacy of the well locations in CWI varies; most of the
point  locations  are  approximated  to  four  quarters
(2.5 acres). The CWI does not contain all well construc-
tion  information,  however, requiring that copies  of
driller's logs be made for GWMAP files.

The final well  selection  is done after applying the
9-square-mile sampling zone over the potential pool of
candidates. For wells that fall within the zone, the well
construction records are pulled from MGS  files, copied,
and submitted for hydrologist review.  Depending on the
target cell location, the number of candidate wells requir-
ing review may range from a few to more than 100. For
newly installed water wells whose records have not yet
been digitized by LMIC, the PLS locations of the wells
are manually plotted onto a map to confirm whether they
fall into a sampling grid cell. Typically, from 5 percent to
as many as 20  percent of selected wells that meet the
location criteria are sampled. This accounts for the hydro-
geologic and well construction criteria and the coopera-
tion  of  well  owners participating  in  the program.
Currently, interest in ground-water protection programs
runs high in rural Minnesota, with an acceptance rate of
up to 80 percent.

The  implementation  of CIS  in  well selection helped
GWMAP excel in two major areas. First, the develop-
ment  of the statewide CIS grid eliminated previously
uncontrolled variables by removing the  PLS spatial in-
consistencies from the systematic grid. Second, the CIS
reduced the manual workload with the automation of two
important steps in the well selection  process: the  gen-
eration of PLS section information to facilitate the data-
base search, and the identification of wells that meet the
geographic location criteria. The success of GWMAP
relies largely on the ability to use existing CIS cover-
ages. In using coverages created by  other entities, this
program  identified the need for a uniform standard for
data conversion and transfer.

Application of Global Positioning
Systems in  Ground-Water Sampling

In 1991, the U.S. Environmental Protection Agency
(EPA) established a policy that all new data collected
after 1992 should meet an accuracy goal of 25 meters
or better (9). The purpose of EPA's Locational Data
Policy (LDP) is to establish principles for collecting and

documenting consistently formatted locational data to
facilitate cross-programmatic, multimedia analyses. Ac-
curate geographic information is important to the spatial
analysis of  well sampling results. Any uncertainty in
sample location can compromise  hydrogeologic analy-
sis (10). GPS is an easy, cost-effective solution.

Global Positioning System Field Application

Beginning in October 1992, GWMAP employed GPS in
the field to assist in locating sample sites. Applying GPS
in the field has proven to be quite easy. The program
uses a multichannel C/Acode receiver with internal data
logging capability. Typically,  the receiver is placed di-
rectly on top of the wellhead and logs 100 to 150 GPS
readings into the receiver's internal memory in approxi-
mately 5 minutes.

The GPS is also used for navigation in the field to locate
sampling sites. Because sampling sites are predeter-
mined, their locations can  be extracted from a  topo-
graphic map. The approximate coordinates can then be
loaded into a GPS receiver. In most cases, the receiver
successfully led the field team within visual range of the
sampling site.

Because of the inherent selective availability (SA) of the
GPS, raw field data must go through a differential cor-
rection process to  achieve the goal of 25-meter accu-
racy (9,  11).

Data Management and Processing

Once the GPS receiver  is brought back from the field,
data are downloaded to a personal computer (I486 proc-
essor at a speed of 50 MHz) and differentially corrected
(11). The average or mean of the 100 or more readings
collected onsite is  calculated and  reported  as the site

The MPCAdoes not operate a GPS base station for the
purpose of differential  correction. The base station data
are obtained through a computer network (Internet) from
the Minnesota Department of Health (MDH) base station
located in  Minneapolis.

To facilitate  future data integration and  document data
accuracy for secondary application, GWMAP proposed
quality assurance codes for GPS data collected by the
MPCA. The value of the accuracy proposed is a nominal
value rather than an  absolute number (see Table 2).
Each of the seven processing methods is assigned a
separate code.

In the field experience of GWMAP, a nominal  accuracy
of 2  to 5 meters has been  consistently achieved after
the postdifferential correction and averaging have been
applied to the data. This technology is suitable for any
program that is designed to conduct either large-area or
intensive monitoring activities. It helps  to cut costs by
Table 2.  Proposed Nominal Accuracy Reference Table
Type of GPS
Receiver Used
Processing Method
Used To Correct Data
quality C/A code
quality with carrier
aid receiver
Survey quality
receiver (dual or
single frequency)
Postdifferential corrected        2-5

Real-time differential
corrected (RTCM)             2-5

Autonomous mode (no
correction)                15-100

Postdifferential corrected        < 1

Real-time differential
corrected (RTCM)             < 1

Autonomous mode (no       15-100

Postdifferential corrected       <0.1
increasing efficiency and accuracy of the data. The data
collected by GWMAP can be used not only in a regional
study but could be used directly in a site-specific inves-
tigation as well.

GWMAP also found that GPS can be  used most effi-
ciently by separating the two roles of field operator and
data manager. The field operators receive only the brief
instructions necessary to operate a GPS receiver before
going into the field. The data manager handles the data
processing details. The field operators can then concen-
trate their efforts on obtaining  ground-water samples
and conducting the hydrogeologic investigation.
CIS  and GPS  technologies made  it  possible for the
MPCA to implement the statewide GWMAP project by
optimizing the available funding and staff time. CIS mini-
mized  staff time spent on identifying  sampling areas,
manipulating the sampling grid, and selecting monitor-
ing sites. In addition, CIS enabled GWMAP to integrate
a variety of databases and maps of different scales.

Using GPS to locate sampling sites enabled GWMAP to
efficiently  obtain accurate geographic locational  data
with  relative ease. This eliminated the  degree  of uncer-
tainty that previously might have compromised the sta-
tistical evaluation of the hydrogeologic data.

GWMAP's success in integrating existing digital data to
automate  the  well  selection process  clearly  demon-
strated the importance of the ability to share information
with  others and the great need for a  broadly applied
standard for data conversion and transfer.


The authors wish to thank Renee Johnson  of the Min-
nesota Department of Natural Resources for her work
to convert the PLS data layerto CIS coverage. Susanne
Maeder of LMIC supplied the statewide CWI coverage,
and Susan Schreifels of MPCA conducted research on
the LDP and  made valuable suggestions on  implement-
ing GPS.

 1.  Nelson, J.D., and R.C. Ward. 1981. Statistical considerations and
    sampling techniques for ground-water quality monitoring. Ground
    Water 19(6):617-625.
 2.  Ward, R.C. 1989. Water quality monitoring—A systems approach
    to design. Presented at the International Symposium on the De-
    sign of Water Quality Information Systems, Colorado State  Uni-
    versity, Ft. Collins, CO.
 3.  Myers, G., S. Magdalene, D. Jakes,  and E. Porcher. 1992.  The
    redesign of the  ambient ground  water monitoring program. St.
    Paul, MN: Minnesota Pollution Control Agency.
 4.  Olea, R.A. 1984. Systematic sampling of spatial functions. Series
    on Spatial Analysis,  No. 7. Kansas Geological Survey, University
    of Kansas, Lawrence, KS.
 5.  Gilbert, R.0.1987. Statistical methods for environmental pollution
    monitoring.  New York, NY: Van Nostrand Reinhold.

 6.  Wahl, I.E., and R.G. Tipping. 1991. Ground-water data manage-
    ment—The county well index. Report to the Legislative Commis-
    sion on Minnesota  Resources. Minnesota  Geological  Survey,
    University of Minnesota, St. Paul, MN.

 7.  Clark, T, Y Hsu, J.  Schlotthauer, and D. Jakes. 1994. Ground-
    water monitoring and  assessment program—Annual  report.  St.
    Paul, MN: Minnesota Pollution Control Agency.

 8.  ESRI. 1993. ARC/INFO version 6.1, ARCPLOT command refer-
    ences. Redlands, CA: Environmental Systems Research Insti-
    tute, Inc.

 9.  U.S. EPA.  1992. Global positioning systems technology and its
    application in environmental programs. CIS Technical  Memoran-
    dum 3. EPA/600/R-92/036. Washington, DC.

10.  Mitchell, J.E. 1993.  A characterization  of the influence  of (x,y)
    uncertainty on predicting the form of three-dimensional surfaces.
    Proceedings of the  AWRA Spring Symposium on Geographic
    Information  Systems and Water Resources, Mobile, AL. pp. 559-

11.  Trimble Navigation, Ltd. 1993. GPS Pathfinder System,  general
    reference. Sunnyvale, CA: Trimble Navigation, Ltd.

 EPA's Reach Indexing Project—Using GIS To Improve Water Quality Assessment
                                           Jack Clifford
      Office of Wetlands, Oceans, and Watersheds, U.S. Environmental Protection Agency,
                                         Washington, DC

                              William D. Wheaton and Ross J. Curry
               Research Triangle Institute, Research Triangle Park, North Carolina

The Waterbody System (WBS), which the U.S. Environ-
mental Protection Agency (EPA) originally developed to
support preparation of the report to Congress that Sec-
tion 305(b) of the Clean Water Act requires, is a poten-
tially significant source of information on the use support
status and the causes and sources of impairment of U.S.
waters. Demand is growing for geographically refer-
enced water quality assessment data for use in inter-
agency data integration, joint analysis of environmental
problems, establishing program priorities, and planning
and management of water quality on an ecosystem or
watershed basis.

Because  location of the waterbody assessment units is
key to analyzing their spatial  relationships, EPA has
particularly emphasized anchoring water bodies to the
River Reach File (RF3). The reach file provides a nation-
wide database of hydrologically linked stream reaches and
unique reach identifiers, based on the 1:100,000 U.S.
Geological Survey (USGS) hydrography layer.

EPA began the reach indexing project to give states an
incentive  to link their water bodies to RF3 and to ensure
increased consistency in the approaches to reach index-
ing. After  a successful 1992 pilot effort in South Carolina,
an expanded program began this year. Working with
Virginia, a route system data model was developed and
proved successful in conjunction with  state use of PC
Reach File (PCRF), a PC program that relates water
bodies to  the reach file. ARC/INFO provides an extensive
set of commands and tools for developing and analyzing
route systems and for using dynamic segmentation.

One important advantage of the route system is that it
avoids the necessity of breaking arcs; this is an impor-
tant consideration in using RF3 as the base coverage in
a geographic information system (GIS). Using dynamic
segmentation to organize, display, and analyze water
quality assessment information also simplifies use of the
existing waterbody system data. Because of the variabil-
ity in delineation of water bodies, however, other states
used a number of different approaches.  Working with
these states has defined a range of issues that must be
addressed in developing a consistent set of locational
features for geospatial analysis.

Wider use of these data also depends upon increased
consistency in waterbody assessments within and be-
tween states. Several  factors complicate the goal of
attaining this consistency in assessment data:

• The choice of beneficial use as the base for assess-
  ment of water quality condition.

• The historical emphasis on providing  flexible tools
  to states.

• The lack of robust standards for assessment of water
  quality condition.

This paper explores possible resolutions to the problem
of building a national database from data collected by
independent entities.

Section 305(b) of the Clean Water Act and
the Waterbody System

Background of Section 305(b)

Since 1975, Section 305(b) of the Federal Water Pollution
Act, commonly known as the Clean Water Act (CWA),
has required states to submit a report on  the quality of
their waters to the U.S. Environmental Protection Agency
(EPA) administrator every 2 years.  The  administrator
must transmit these reports, along with an  analysis of
them, to Congress.

State assessments are based on the extent to which the
waters meet state water quality standards as measured

against the state's designated beneficial uses. For each
use, the state establishes a set of water quality criteria
or requirements that must be met if the  use  is to be
realized.  The CWA provides the  primary authority to
states to set their own standards  but requires that all
state beneficial  uses and their criteria comply with the
'fishable and swimmable' goals of the CWA.

Assessments and the Role of Guidelines

EPA issues guidelines to coordinate state assessments,
standardize assessment methods and terminology, and
encourage states to assess support of specific benefi-
cial uses (e.g., aquatic life support, drinking water sup-
ply, primary contact recreation, fish consumption). For
each use, EPA asks that the state categorize its assess-
ment of use support into five classes:

• Fully supporting: meets designated use criteria.

• Threatened: may not support uses in the future unless
  action is taken.

• Partially supporting: fails to meet designated use cri-
  teria at times.

• Not supporting: frequently  fails  to  meet designated
  use criteria.

• Not attainable: use support not achievable.

In the  preferred assessment method, the state com-
pares monitoring data  with  numeric criteria for  each
designated  use. If monitoring data  are not available,
however, the state may use  qualitative information to
determine use support levels.

In cases  of impaired use support  (partially or  not sup-
porting), the state lists the sources  (e.g., municipal point
source, agriculture, combined sewer overflows)  and
causes (e.g., nutrients, pesticides, metals) of the use
support problems. Not all impaired waters are charac-
terized. Determining specific sources and causes  re-
quires data that frequently are not  available.

States generally do not assess all of their waters each
biennium. Most states  assess a  subset  of their total
waters every 2 years. A state's perception of its greatest
water quality problems frequently  determines this sub-
set.  To this extent, assessments  are skewed toward
waters with the most pollution and  may,  if viewed as
representative of overall water quality, overstate pollu-
tion problems.

Assessment Data Characteristics

Each state determines use support for its own set of
beneficial  uses. Despite EPA's encouragement to use
standardized use categories, the wide variation in state-
designated beneficial uses makes comparing state uses
an inherent problem. This affects the validity of aggre-
gation and use  of data across state boundaries. Com-
parably categorizing waters into use support categories
also poses a problem; different states apply the qualita-
tive criteria for use support levels in very different ways.
Further limiting the utility of Section 305(b) data  is that
data are aggregated at the state  level and questions
about the use support status of individual streams can-
not be resolved without additional information.  While
some states report on individual waters in their Section
305(b) reports,  EPA's Waterbody System (WBS) is
the primary database for assessment information on
specific waters.

State  monitoring  and assessment activities are also
highly variable. States base assessments on monitoring
data or more subjective evaluation. The evaluation cate-
gory particularly differs among states.

Waterbody System

The WBS is a database and a set of analytical tools for
collecting, querying, and reporting on state 305(b) infor-
mation. It includes information  on  use support and the
causes  and sources of impairment for water bodies,
identification and locational information, and a variety of
other program status information.

As pointed out earlier, although some states discuss the
status of specific waters in their 305(b) reports, many do
not. The WBS is generally much more specific than the
305(b) reports. It  provides  the  basic assessment infor-
mation to track the status  of individual waters in time
and, if georeferenced, to locate assessment information
in  space. By allowing the  integration of water quality
data with other related data, the WBS provides a frame-
work for improving assessments.

WBS has significant potential for management planning
and priority setting and can serve as the foundation for
watershed- and  ecosystem-based analysis, planning,
and management. In this respect, it can play a vital role
in  setting  up watershed-based  permitting of  point
sources. The primary function of WBS is to define  where
our water quality problems  do and  do not exist. WBS is
increasingly used to meet the identification requirement
for waters requiring a total maximum daily load (TMDL)
allocation. It can serve as the initial step in the detailed
allocation analysis included in the TMDL process. In
addition, WBS is  an important  component of EPA par-
ticipation in joint  studies and  analyses.  For instance,
EPA is currently participating with the Soil Conservation
Service (SCS) in  a joint project to identify waters that
are impaired due  to agricultural nonpoint source  (NPS)
pollution. WBS  can also anchor efforts to provide im-
proved public access at the state and national levels to
information on the status of their waters.

It is important to recognize that  use of WBS is voluntary.
Of the 54 states, territories,  river basin commissions, and
Indian tribes that submitted 305(b) reports, approximately

30 used the WBS in the 1992 cycle. While submissions
for the 1994 cycle are not complete, we anticipate about
the same level of participation. This represents about a
60-percent rate of participation in WBS, which may be
the limit for a voluntary system. This severely limits use
of WBS assessment data for regional and national level
analysis. If data at the national level are required, man-
datory data elements, formats, and standards may be

EPA is  currently attempting  to  achieve  consistency
through agreement with other state and federal agen-
cies. The recent work of the Interagency Task  Force on
Monitoring offers hope for eventual consensus on the
need for nationally consistent assessment data and mu-
tually agreed upon standards for collection, storage, and
transfer. The Spatial  Data Transfer Standards already
govern spatial data, allowing movement of data between
dissimilar  platforms.  The Federal Geographic Data
Committee provides leadership in coalescing data inte-
gration at the federal  level; it provides a model for gov-
ernment and  private sector  efforts. This  level  of
cooperation, however, has not always been present in
water assessment  data  management. Assuming  that
national and regional assessment data are needed, if
consistency is elusive through cooperative efforts, regu-
lations may be necessary. Developing a national data-
base may  not be feasible without a mutual commitment
by EPA and the states to using common assessment

WBS was  originally developed as a dBASE program in
1987. It has undergone several revisions since then, and
the current Version 3.1 is written in  Foxpro  2.0.  The
WBS software provides standard data entry, edit, query,
and report generation functions. WBS has grown sub-
stantially in the years since  its inception, primarily in
response to the expressed needs of WBS users and
EPA program offices. The program's memory require-
ments and the size of the program and data files, how-
ever, are of growing concern to state WBS users and
the WBS program manager. Because of the wide range
of WBS user capabilities and equipment, users must be
equipped to support  an  array  of hardware from high
capacity Pentium computers to rudimentary  286  ma-
chines with 640 Kb of memory and small hard disks. This
range makes memory problems inevitable for some users.

While WBS contains over 208 fields, exclusive of those
in  lookup  tables, approximately 30 fields in  four  files
provide  the core data needed to  comply with 305(b)
requirements. These fields contain:

• Identification information for the water body.

• The date the assessment was completed.

• The status of use support for beneficial uses.
• The causes and sources of any use impairment in
  the water body.

The  uses WBS  considers are both state-designated
uses and a set of nationally consistent uses (e.g., overall
use,  aquatic  life support, recreation) specified in  the
305(b) guidelines. The other essential piece of informa-
tion is the geographic location of the water body, which
the remainder of this paper discusses in detail.

Significant differences exist in the analytical base as well
as in assessments. EPA provided  little initial guidance
on defining water bodies; therefore, states vary widely
in their configurations of water bodies. Water bodies are
supposed to represent waters of relatively homogene-
ous water quality conditions,  but state interpretation of
this guidance has resulted in major differences in water-
body definition.

Initially, many states developed linear water bodies, and
these were often very small. The large number of water
bodies delineated, however, created significant difficul-
ties in managing the assessment workload and were not
ideal in the context of the growing need for watershed
information. Some states, such as Ohio, developed their
own river mile systems.

As discussed below, some states indexed their water
bodies to earlier versions of the reach file, and therefore,
the density of the streams these water bodies include is
fairly sparse. Recently,  many states have  redefined
their water bodies on the basis of small  watersheds
(SCS basins, either 11-digit or 14-digit hydrologic unit
codes [HUCs]).

Locating  water bodies geographically  is a necessary
prerequisite to assessing water quality on a watershed
or ecosystem basis. The  WBS has always included
several  locational fields,  including county name and
FIPS, river basin, and ecoregion. These fields have not
been uniformly populated, however. One  of the WBS
files  includes fields for the  River Reach  File (RF3)
reaches included in the water body. While a few states
had indexed their water bodies to older versions of the
reach file (RF1 and RF2), however, no state had indexed
to RF3 until 1992.

In  1992,  EPA initiated a demonstration of geographic
information system (CIS) technology in conjunction with
the South Carolina Department of  Health and Environ-
mental Control. This project involved:

• Indexing South Carolina's water bodies to RF3.

• Developing a set of arc macro languages (AMLs) for
  query and analysis.

• Producing coverages of water quality monitoring sta-
  tions and discharge  points.

• Using CIS tools in exploring ways to improve water
  quality assessments.

South Carolina has defined  its water bodies as SCS

The results have  been very  encouraging. First, South
Carolina  took  the initial coverages and  decided they
needed much  more specificity  in their use support de-
terminations and  their mapping of  the causes  and
sources of impairment. As a result, they mapped these
features down to the reach level. Next, they decided that
they needed better locational information, so they used
global positioning  satellite receivers to identify accurate
locations for discharges and  monitoring stations. They
then used CIS query and analysis techniques to relate
their monitoring and discharge data to their water quality
criteria. South  Carolina is using CIS to actively identify
water quality problems and improve their assessments.

In 1993,  EPA worked cooperatively with several states
to index their water bodies to the reach file. Virginia, the
next state to be indexed,  demonstrated the successful
use of PC Reach File (PCRF) software (described later
in this paper) for indexing water bodies to the reach file.
Ohio and Kansas also are essentially complete. Each of
these  states required a somewhat different approach
than Virginia.  The need  for flexibility in dealing  with
states on reach indexing issues is essential. Existing
waterbody delineations often  represent considerable in-
vestment; therefore, EPA must  provide the capability to
link the state's existing assessment data to the reach file
in order to encourage state buy-in.

Figure 1 shows the results of Ohio's indexing of a typical
cataloging unit (CD). Figure 2 reflects part of the output
of the Kansas work. We can link use support, cause, and
source data to each of these water bodies now. In the
future, we hope to map these attributes at a higher level
of resolution, down to the reach segment level. CIS has
proven to be a useful assessment tool. With higher reso-
lution, it should prove to be even more helpful in identify-
ing water quality problems, picking up data anomalies,
and assessing management actions, strategies, and poli-
cies. This entire process has taught  us much and  has
strengthened enthusiasm for place-based management.

The Reach Indexing Project—
Georeferencing the Waterbody System

Purpose and Overview

The reach indexing project is designed to locate water
bodies using RF3  as an electronic base map of hydrog-
raphy and to code RF3 reaches with the specific water-
body identifier (WBID). After linking water bodies to their
spatial representation,  they can  be  queried and dis-
played with assessment data located in WBS files.

Reach indexing includes several steps. First, the state
must supply waterbody locations and WBIDs. The next
step entails developing a set of procedures for indexing.
Finally, the coded  RF3 data must be produced.

Input data to the indexing process includes:

• A list of valid WBIDs. In most cases, the state  has
  already input these identification numbers to the WBS.

• Information about the location of each water body. Lo-
  cational information may  be found in  marked-up
  paper maps showing waterbody locations or electronic
  files from  WBS  containing waterbody indexing
	 = 66011
— 66012
<*"*•»« 66013
oc*xx> 66014
=, 66015
•=a=r 66016
fifim 7
	 DDU 1 /
= 66019
__-. 66020
Figure 1.  State of Ohio water bodies in CU 04100008.

Figure 2. State of Kansas water bodies in CU 11070202.

  expressions (discussed later), orit may be embedded
  in the WBID itself.

• A complete set of RF3 data for the state being indexed.

Depending on the type of information the state supplies,
procedures used to index water bodies can  be almost
fully automated, semiautomated, or completely manual.

The final result of the indexing processes is a set of RF3
coverages that contain a WBID attribute. This  product
allows querying and displaying of assessment  data,
which is collected and stored  by water body, in a CIS

The Reach File Database

The reach file is a hydrographic database of the surface
waters of the continental United States. Elements within
the database represent stream segments. The elements
were created for several purposes:

• To perform hydrologic routing for modeling  programs.

• To identify upstream and downstream connectivity.

• To provide a method to uniquely identify any particu-
  lar point associated with surface waters.

The unique reach identifier has succeeded in associat-
ing other EPA national databases, such as STORET, to
surface waters. Any point within these databases can be
associated with and identified by a specific location on
any surface water element, such as a reservoir, lake,
stream, wide river, or coastline.

There are three versions of the reach file. The first was
created  in 1982 and contained 68,000 reaches. The
second version, released in 1988, doubled the size of
Version   1.  The third  version  (RF3)  includes  over
3,000,000 individual reach components.

The base geography of RF3 is derived from U.S. Geo-
logical Survey (USGS) hydrographic data (1:100,000
scale) stored in digital line graph (DIG) format. Unlike
DIG data, which are partitioned by quad sheet bounda-
ries, RF3 data are partitioned by CU. A CU is a geo-
graphic area that represents part or all of a surface
drainage basin, a combination of drainage basins, or
a distinct hydrologic feature. The USGS uses CUs
for  cataloging  and indexing water-data acquisition

The  continental United States comprises over 2,100
CUs.  CUs are  fairly small; for example, 45 units fall
partially or completely within the state of Virginia (see
Figure 3).

RF3 is a  powerful data source used in hydrologic appli-
cations for many reasons, including the following:

• RF3 has spatial network connectivity that topological
  upstream/downstream modeling tools use.

Figure 3. Cataloging units in Virginia.

• RF3  has attributes that describe connectivity, which
  offers the ability to accomplish upstream/downstream
  navigation analytically (without topological networking).

• RF3  has a simple and consistent unique numbering
  system for every stream reach in the United States.

• RF3  has built-in river mileage attributes that describe
  upstream/downstream distances along river reaches.

Use ofRF3 in the Indexing Process

When importing Reach File data from EPA's mainframe
computer,  an arc attribute table (AAT) is  automatically
built for each  RF3  coverage. The AAT  contains the
standard AAT fields, plus the items found in Table 1.

The CD item stores the USGS CD number of this piece
of RF3. Every arc in the coverage has the same value
for CU.

The SEG  item stores the number  of the stream seg-
ment to which the particular arc is assigned. SEG num-
bers start at 1 and increase incrementally by 1 to 'N' for
each CU.  A SEG  could  represent all the arcs of  a
mainstream, the arcs of a tributary, or piece of a main-
stream or  tributary. SEG numbers were defined in the
production of RF3.

Ml stores the marker index for each particular arc. The
Ml resembles a mile posting along  a stream. In reality,
the  Ml  field does not truly measure mileage along the
RF3 stream network. It does, however,  represent  a
method of producing a unique identifier (in combination
Table 1.  Fields Found in Arc Attribute Table

12070104-ID      CU      SEG     Ml
1 20701 04
1 20701 04
1 20701 04
1 20701 04
1 20701 04
1 20701 04
1 20701 04
with the CU number and the SEG  number) for every
reach in the United States (see Figure 4).

Together CU, SEG, and  Ml uniquely identify every arc
in RF3 nationwide. These three items are combined in
the  redefined item called RF3RCHID. This  provides a
powerful scheme  for consistently identifying locations
along streams everywhere in the country.

Along with the AAT file, a second attribute file is auto-
matically created for RF3 coverages. This file is always
named COVER.DS3. The DS3 file  stores a wealth of
information about  arcs in the coverage. Some of the
important fields in  the DS3 file contain:

• Upstream and downstream connectivity for navigat-
  ing  along reaches.

• Codes to describe the type of reach (e.g., stream,
  lake boundary, wide river).

• DIG major and  minor  attributes.

                     KS- KR-04-R001
                  WBID         WBNAME

                  KS-KR-04-R0001   Mainstem

                  KS-KR-04-W020   Tributaries

                  KS-KR-04-W030   Lakes
Figure 4. RF3, SEG, and Ml data elements.

Waterbody Locations

Because states define water bodies, they provide the
only information on waterbody location. South Carolina
was indexed to RF3 in 1992, followed by Virginia. Vir-
ginia indexed its water bodies using the PCRF program
instead of indexing in a CIS environment with ARC/INFO.

PCRF is a PC-based system that indexes water bodies
and locates other assessment data from WBS. PCRF
stores  the definitions of water bodies (including their
location) in a file that is linked to other WBS database
files that  contain information about the  assessment
status and quality of the waters.

A water body is a set of one or more hydrologic features,
such as streams, lakes, or shorelines, that have similar
hydrologic characteristics. Water bodies are the  basic
units that  states use to  report water quality  for CWA
305(b)  requirements. Depending on the state's assess-
ment goals and resources, water bodies can be defined
in several ways, including (see Figure 5):

• All streams within a watershed

• All lakes and ponds within a watershed

• Sets of streams with similar water quality conditions

PCRF  stores locational data for a water  body with a
unique WBID. WBS uses this WBID as a common field
to relate the  water body's definition and  location to de-
scriptive data about the water body's assessment status
and quality. The two most important files used in PCRF
are the SCRF1 and SCRF2  files.

The SCRF1 file simply lists the unique water bodies by
state. Table 2 offers an example. The most relevant data
for reach indexing in this file are the WBID, WBNAME,
and WBTYPE, as defined by the state. The  WBID, as
stated, is  a  unique  identifier for each water  body the
state has  defined. The  WBNAME  stores a verbal de-
Figure 5. Potential definitions of water bodies.

scription of the water body. Finally, the WBTYPE de-
fines the type of water body; for example, R is for river,
L is for lake.

The SCRF2 file contains an explicit definition of each
water body. Because of the complexity involved in de-
fining water bodies, this file may include more than one
record for each water body. The SCRF2 file can be
considered a waterbody definition language because it
contains specific codes, attributes, and keys that can be
converted  into specific reaches on the RF3 data, if read
properly (see Table 3). The  WBBEGIN and WBEND
fields contain explicit CD, SEG,  and Ml  attributes to
define the  location of the starting point and  ending point
for the water body. The WBDIR field contains an attrib-
ute  that describes  whether to go upstream or down-
stream from the WBBEGIN to the WBEND. In addition,
a blank WBEND field denotes that the  water  body
should include all  upstream or downstream reaches
(depending on the WBDIR) of the WBBEGIN reach.

Virginia used  PCRF to create an  SCRF2 file that con-
tains reach indexing expressions for all of their defined
water bodies. ARC/INFO macros were then written to
process this file and expand the expressions into the set
of specific arcs that compose  each water body. The
macros will be described in more detail later.
Table 2.  Example of SCRF1 File Data

WBID            WBNAME      WBTYPE

Table 3.  Example of SCRF2 File Data

WBID               WBDIR      WBBEGIN
10270104001 0.00
10270104005 10.80
10270104059 12.05
10270104038 0.00

States that have not already generated indexing expres-
sions in PCRF  must provide locations in some other
way. The most basic method is for the state to supply a
set of 1:100,000 USGS quad sheets that they have
marked up with locations of each water body. The
maps can be used  in conjunction with a digitizer to
manually select the appropriate  RF3  reaches and
code them with  the WBID.

The  state of Ohio created a CIS database of its river
reaches several years ago. The CIS coverage is rep-
resentational in  nature. The stream reaches are 'stick-
figures' only. Generally, they fall along the paths of the
actual streams, but they are schematic in nature and do
not show the true shape of  streams.  The CIS layer,
however, contains the attributes of Ohio's stream reach
numbering system, which is used to identify water bod-
ies as well. Ohio's river reach coverage contains infor-
mation on the locations of water  bodies and is being
manually conflated to transfer the WBIDs to RF3. The
conflation process will be covered  later in this paper.

The  state of Kansas had previously defined its water
bodies on RF2,  the precursor to RF3.  Kansas defined
some indexes by a set of RF3 SEG numbers in a CD
and  some by the RF3  reaches in a small watershed
polygon within a CD. The locations were, in effect, de-
fined within the WBID itself.

Indexing  Procedures

Procedures developed for performing waterbody index-
ing include automated, semiautomated,  and  manual

Automated Indexing Procedures

As stated, Virginia used PCRF to perform the indexing
operation. The state delivered an SCRF2 file containing
indexing expressions for all of its water bodies. AMI
programs were  created  to read the SCRF2 file and
select the reaches specified by each indexing expres-
sion. The selected sets of reaches were then coded with
the appropriate WBID. The macros were designed to run
on one RF3 CD at a time, so the operator specified runs
of up to 10  CDs at a time. The macros had to accom-
modate indexing expressions  that included:

• Select reaches upstream of a specified location.
• Select reaches on a reach-by-reach basis.

• Select reaches within a given polygon area.

• Select shorelines of lakes or ponds given latitude and
  longitude coordinates.

• Select reach downstream from a specified location.

Kansas water bodies  were  also  indexed through an
automated  process. Kansas supplied  an ARC/INFO
coverage of small watershed polygons (sub-CD  poly-
gons)  containing a watershed identifier. The state's
WBID contained all other information  necessary to de-
termine the RF3 CD and the set of reaches  making up
each water body. An example of a  Kansas WBID is
KS-KR-02-W030. This  is explained by the following:

• KS refers to the  state. All  WBIDs  in Kansas begin
  with  KS.

• The  second component (in this case KR) is an ab-
  breviation of the basin in which the water  body falls.
  KR indicates that this  water body is in the Kansas-
  Lower Republican River basin.

• The  third component contains the  last two digits of
  the eight-digit CD number. Although basins comprise
  several CDs, the last two digits of each CD in a basin
  are unique; therefore,  between the basin  (e.g., KR)
  designation and the last two digits of the CD (e.g.,
  02), the complete eight-digit CD number in which the
  water body falls is defined.

• The  next letter (in this  case W) denotes whether the
  water body is defined  by a watershed polygon (W),
  an RF2 SEG (R), or a lake or pond shoreline (L).

• Finally, the WBID ends with the number of the  poly-
  gon  (in this case 030) that contains the reaches for
  the water body in the watershed coverage.

The completed macros could index the entire state in  a
single run  provided that all the WBIDs were contained
in single file.

In all cases, Kansas has  indexed to RF2 reaches. Only
RF3 reaches originally created in RF2 production, there-
fore,  are coded with a WBID.

Manual Indexing Procedures

Because Ohio already  has a coverage of river reach
codes, WBIDs from this coverage had to be transferred
to the RF3 reaches they represent. This entailed using a
manual conflation process. The operator displayed a CD
of RF3 along with the Ohio river reach system for the
same area. In a simple process of 'pointing and clicking,'
the operator first selected an Ohio river reach arc, then
the RF3 arcs that seemed to coincide. As each RF3 arc
was selected, it was  coded with the WBID of the pre-
viously selected  Ohio  river reach arc.

Other states that  have no  means of describing water
bodies in electronic files may have to mark up paper maps
to show waterbody locations. These maps  can  then  be
used  in a  manual process  of selecting RF3  reach and
coding them with WBIDs either in ARC/INFO or in PCRF

Using the Route  System Data Model To
Store Water Bodies

Because water bodies can be defined as noncontiguous
sets of arcs and portions of arcs, a robust linear data-
base  model  is necessary to model these  entities.
ARC/INFO's route system data model seems well suited
for this application. The route system data model allows
one to group any set of arcs or portions of arcs into
routes. Each route is managed  as a feature in itself.
Attributes of water bodies are stored in a route attribute
table (RAT) and relate to all the arcs defined as the water
body. Figure 6 helps illustrate the route system model.

Each route comprises one  or more  arcs or sections of
arcs. ARC/INFO manages the relationship between arcs
and routes in the section table (SEC). The structure of
the SEC, which  is an  INFO table, is defined in Table 4.
Tables reflects how the sections that make up the above
routes would appear.
                1     1


                        Table 4.  Definition of Structure of SEC INFO Table

                        ROUTELINK*   The route upon which the section falls
                        ARCLINK*     The arc upon which the section falls
                        F-MEAS       The measurement value at the beginning of the
                        T-MEAS       The measurement value at the end of the section
                        F-POS        The percentage of the distance along the arc at
                                     which the section begins
                        T-POS        The percentage of the distance along the arc at
                                     which the section ends
                        SEC#         The internal identifier of the section
                        SEC-ID       The user identifier of the section
                        Table 5.  How Sections Appear in SEC INFO Table

                        LINKS   ARCLINK8  F-MEAS T-MEAS  F-POS T-POS SEC# SEC-ID
Figure 6. The route attribute table containing waterbody data.
Representing Water Bodies as Routes

ARC/INFO offers several ways of grouping sets of arcs
into discrete routes. One can use ARCEDIT to select a
set of arcs to group them into a route, or ARCSECTION
or MEASUREROUTE in ARC to  group arcs into routes.
The method described here uses  the MEASUREROUTE
command. This method requires that the AAT or a re-
lated table has an attribute containing the identifier of
the route to which an arc should be assigned. In the
application the authors employed, they converted the
SCRF2 file into an INFO table containing, for each arc
in the coverage, the RF3RCHID of the arc and the WBID
to which the arc should be assigned. The WBID item is
used to group arcs into routes. One route exists for each
unique WBID. Table 6 illustrates the table used in the
MEASUREROUTE command method. This table is re-
lated to the AAT of the RF3 coverage by the RF3RCHID.

An RAT is automatically created for the coverage, which
now can  be  related to other WBS assessment files for
display and query. Figure 4 illustrates the RAT. The most
important characteristic of the file is that it has only one
record for each water body. This simplifies the display
and query of water bodies based on water quality data.

Using EVENTS for Subwaterbody Attributes

Water bodies, as states define them, often constitute a
gross aggregation of the water in an area. States often
have  more specific data  about  particular stretches  of
streams within a  water body. A system is needed  to

Table 6.  Table Used in MEASUREROUTE Command Method
        To Group Arcs Into Routes
1 02701 04
1 02701 04
1 02701 04
1 0.00
1 1.30
1 2.10
2 0.00
3 0.00
3 1.15
5 0.00
6 0.00
                                                                         KS- KR-04-R001
query and  display data at the subwaterbody level.
ARC/INFO's dynamic segmentation tools and event ta-
bles are useful for this application. Once water bodies
have been defined and reporting methods have been set
up based on those water bodies, the task of redefining
them is cumbersome.

Event tables can help to keep these waterbody defini-
tions yet still offer the ability to store, manage, and track
data at the subwatershed level. Event tables are simple
INFO files that relate to route systems on coverages.
This concept and data structure can work in conjunction
with the predefined waterbody system. We  have already
seen how a route system called WBS is created in RF3
to group arcs into waterbody routes. This works quite
well when displaying  water bodies and querying their
attributes. A route system based on the WBID cannot,
however, act as an underlying base for subwaterbody
events because the measures used to  create the WBS
route system are not  unique for a particular route. For
example, in the route depicted in Figure 7, three loca-
tions are defined as being on WBID KS-KR-04-W020 and
having measure 1.0.

The mileage measurements along  SEG,  however, are
always unique (see Figure 8). To use EVENTS, therefore,
a second route system must be created based on the RF3
SEG attribute, which provides a unique code for each CD.

The ARCSECTION command, instead of the MEASURE-
ROUTE  command, is used to create the SEG route
system. This is because the measurement items (Ml on
the  AAT  and  UPMI  on  the  DS3)  already  store  the
summed measures along particular SEGs. Table 7  lists
the  contents of the resulting RAT table.

Because  the name of the route system  is  SEG, the
SEG# and SEG-ID are the names of  the internal  and
user IDs. The SEG item contains the actual SEG num-
ber in the RF3 coverage. Because the SEG numbers for
each RF3 CU coverage start at 1 and increase incre-
                                  Figure 7.  Measurements along SEGs.
                                  Figure 8.  Events located on RF3 data.

                                  Table 7.  Route Attribute Table

                                    SEG-ID              SEG#
                                  mentally by 1, the SEG item looks much like the SEG-ID
                                  and SEG#.

                                  Event tables contain a key item, the WBID or SEG, to
                                  relate them to the appropriate route system (see Figure 8).
                                  They also contain locational information on where to lo-
                                  cate the events on the route (either WBID to indicate the
                                  water body on the WBS route or SEG to identify the
                                  route in the SEG route system). Separate event tables
                                  can then relate use support, causes,  and sources as
                                  linear events.  FROM and  TO store the starting  and

ending measures for each event. Using event tables
allows us to apply many useful cartographic effects
(e.g., hatching, offsets, text, strip maps). Events can be
queried both in INFO and graphically. Event data can
help in producing overlays of two or more event tables.
An event table can display use support information (see
Table 8). WBS users can update their event tables using
RF3 maps supplied by EPA without having proficiency
in ARC/INFO. ARC/VIEW2 is expected to support
events and route systems. This will give users powerful
tools for spatial query of assessment data. Developing
event tables would also display and query data on the
causes and sources of use impairment. These events
can be offset and displayed to show the areas of inter-
action. More permanently, preparing line-on-line over-
lays can show intersections and unions.
An alternative approach is to use an EVENT-ID as a
unique identifier for each event. The SEG field stores the
number of the route (SEG) upon which the event occurs.
FROM and TO store the beginning and end measures
along the route upon which the event occurs. WBID
contains the identifier of the water body upon which the
event occurs (see Table 9). An event can occur within
a single SEG, across two or more SEGs, within a single
water body, or across two or more water bodies.
Additional attribute tables can be created to store de-
scriptive attributes for each event. These tables would
resemble the SCRF5 and SCRF6 files except that in-
stead of using the WBID to relate to a water body, a field
called 'EVENT-ID' would link the use, cause, and source
data to a particular event (see Table 10).
Both approaches offer some advantages. In either case,
they allow us to map our water quality assessment data
and communicate it in a meaninaful and useful wav.
Table 8. Event Table That Reflects Use Support Information
1 0.80 1.30 KS-KR-04-R0001
1 1.30 2.10 KS-KR-04-R0001
1 2.10 2.31 KS-KR-04-R0001
1 0.50 1.30 KS-KR-04-R0001
3 0.00 1.15 KS-KR-04-W030
4 0.00 2.5 KS-KR-04-W040
21 Fully
21 Partial
21 Not supported
40 Threatened
21 Fully
40 Not supported
Table 9. Using EVENT-ID as a Unique Event Identifier
1 1 0.80 1.30
1 1 1.30 2.10
2 4 0.00 2.5
Table 10. Using EVENT-ID To Link Use,
Data to an Event
1 900
1 -9
1 0500
2 1200
2 0900
Cause, and Source

         Nonpoint Source Water Quality Impacts in an Urbanizing Watershed
                         Peter Coffin, Andrea Dorlester, and Julius Fabos
                University of Massachusetts at Amherst, Amherst, Massachusetts

As part of the larger Narragansett Bay Estuary Project,
the University of Massachusetts Cooperative Extension
Service contracted with the university's METLAND re-
search team to develop a geographic information sys-
tem (CIS) database,  generate watershed-wide maps,
perform analyses, and develop a modeling procedure.
The objective was to educate local officials about the
impacts of development on water quality and to help
local boards minimize the effect of nonpoint sources of

Because  the receiving waters of the Narragansett Bay
are located far downstream in  Rhode Island, the up-
stream communities in Massachusetts are reluctant to
enact measures to improve water resources outside of
their jurisdictions. A CIS was used to create awareness
of existing downstream problems and to show the up-
stream communities  how development will ultimately
affect water resources in their own backyards.

To nurture this awareness, a  "buildout" analysis was
conducted for an entire upstream subwatershed, the
Mumford  River watershed,  containing  parts of four
towns, and roughly 50 square miles. This buildout was
coupled with a loading model using Schueler's Simple
Method to illustrate the potential impacts of future devel-
opment, and encourage local boards to minimize future
nonpoint  sources of pollution.

CIS proved  its usefulness by developing  customized
maps for each town, by generating several "what if
scenarios showing the  impacts of different zoning
changes,  by facilitating long-range planning  for small
towns without professional staff,  and by encouraging a
regional perspective on development issues. The entire
planning  process was most successful  in creating  a
series of  partnerships that will continue after the grant
expires. The university shared coverages with the state
CIS  agency,  creating  new coverages not previously
available, specifically  soils,  ownership,  and zoning.
Small towns  learned about the potential  of the new
technology. Students gained from hands-on experience
with real-world problems. State agencies saw their ef-
forts understood at the local level, especially as they
reorganize on a basin approach and begin to implement
a total mass daily loading (TMDL) procedure to coordi-
nate permitted discharges and withdrawals.

As greater emphasis is placed on controlling nonpoint
sources of pollution, more attention needs to be focused
on local boards, who control land use decisions in New


Project Description

Narragansett Bay is a  vital  resource for southern New
England.  The health of its  waters is  critical to the re-
gional economy, supporting  fisheries, tourism, and qual-
ity of  life.  Increased  development  along  the bay's
shorelines and throughout its drainage basin threatens
the quality of these waters,  however. The U.S.  Environ-
mental Protection Agency (EPA) recognized the threats
to this important water  body and designated the Narra-
gansett Bay under its National Estuary Program in 1985.

Completing a Comprehensive Conservation and Man-
agement  Plan (CCMP) for Narragansett Bay took  7
years.  The CCMP  identified seven priority areas for
source reduction or control, including the reduction of
agricultural and other nonpoint sources of pollution. The
nonpoint  source strategy identified United States  De-
partment  of Agriculture (USDA) agencies, conservation
districts, and other public and private organizations as
having principal roles in nonpoint source  management.

Whereas  the vast  majority of Narragansett Bay  lies
within the boundaries  of Rhode  Island, a  significant
portion of its pollution load originates in Massachusetts.
Recognizing that the watershed extends beyond state
boundaries, the USDA provided 3 years of funding to
Cooperative Extension  and  the Soil Conservation Serv-
ice (SCS) in both Massachusetts and Rhode Island to

coordinate their efforts in an innovative attempt to re-
duce the impact  of nonpoint sources of pollution on
Narragansett Bay. While water quality is a relatively new
focus for Cooperative  Extension, it fits well with the
historic mission of extending the knowledge base of the
land-grant colleges out into the community, and provid-
ing training and capacity building for local officials and
community organizations.

With such a large area of concern, the management
team decided to focus on a smaller subwatershed area
in  each state for the first 2 years. The strategy was first
to  develop a  program for the  mitigation of nonpoint
source pollution on the smaller scale of a watershed of
roughly 50 square miles, then take the lessons learned
and apply the  most appropriate efforts throughout the
larger watershed. By using similar strategies in Rhode
Island and Massachusetts, but choosing subwatersheds
that differ in terms of location relative to the receiving
water, size, staffing, and sophistication, the two states
gained from each other's experience, sharing the  suc-
cessful techniques and avoiding each other's mistakes.

For its pilot study, Rhode Island chose Aquidneck Island,
home of Newport, Portsmouth, and Middletown, with a
special focus on  protecting surface water supply reser-
voirs. Massachusetts chose an upstream watershed in
the Blackstone Valley, somewhat rural in character, but
rapidly undergoing a transformation to suburbia.

Watershed Description

The Blackstone River drops 451 feet in its 48-mile jour-
ney  from Worcester,  Massachusetts,  to  Pawtucket,
Rhode Island.  In the 19th century, this drop of roughly
10 feet per mile was ideally suited to  providing power to
mills during the early years of the industrial revolution.
By the Civil War, every available mill site was developed,
earning the Blackstone River the  name "The Hardest
Working River."

The Blackstone has a long history of pollution. First, the
textile industry, then steel, wire, and metal finishing in-
dustries used the  river for power, in their manufacturing
process, and for waste disposal.

In  Massachusetts, the Blackstone River is the  major
source of many pollutants to Narragansett  Bay. Based
on total  precipitation  event  loading  calculations, the
Blackstone River  is the principal source of solids, cad-
mium, copper,  lead,  nitrate, orthophosphate, and PCBs
to the bay (1). The Blackstone River has an average flow
of  577 million  gallons per day or 23.2 percent of the
freshwater input to the bay.

The  watershed  area  in  Massachusetts  equals  335
square miles; with a population of 255,682, this results
in  a density of 763 people per square mile. The Black-
stone Valley has  9,000 acres in agricultural use,  with
more land in hay (4,500 acres) than crops (3,700 acres)
to support its 4,400 animals.

Based on aerial mapping flown in 1987, the Blackstone
Valley has lost 5 percent of its cropland, 9 percent of its
pasture, and 21 percent of its orchards since 1971. The
valley remains more than 60 percent forested, but that
represents  a decrease  of 5 percent. The forest and
farmlands were lost to  development as low density
housing grew by 45 percent, commercial use grew by
15  percent, and transportation grew by 54  percent.
Waste disposal grew 52  percent to 582 acres, and min-
ing, which in this region represents gravel pits, grew 22
percent to 1,100 acres.

Watershed  soils consist mainly of compact glacial  till on
rolling topography, with 3 to 15 percent slopes. The river
and stream valleys are underlain by glacial-derived sand
and gravel  outwash, which provide drinking water to all
towns in the area  except Worcester and support the
large gravel pits. The high clay content in the till soils of
the uplands makes for  a high water table, which is
beneficial for growing corn  but causes problems for
septic systems.

Following a preliminary study of the subwatersheds, the
Mumford River in the Blackstone Valley was selected as
the focus watershed based on its size, location, land
use, and existing water quality (see Figure 1). The Mum-
ford  River watershed has an area of 57 square miles,
with  a length of 13 miles, and lies within the towns of
Douglas, Northbridge,  Button,  and  Uxbridge. These
towns share the attributes of small, rural communities
undergoing  rapid development,  with  no  professional
planning  staff (see  Figure 2).  According  to  the  1990
Census, Douglas grew 46 percent in 10 years to 5,438;
Uxbridge experienced 24 percent growth to 10,415; Sut-
ton increased 17 percent to 6,824; and Northbridge grew
9 percent to 13,371.

Project Strategy

Because the generation of nonpoint sources of pollution
is so closely tied to land  use, and because local boards
composed  of citizen volunteers have principal control
over land use in New England, the  key focus of this
program  is  to train local  boards to recognize and begin
managing the threat that nonpoint sources of pollution
pose to water quality. Local planning boards, conserva-
tion  commissions, and boards of health address land
use  issues and  can  regulate and shape  existing and
proposed development.  By developing a  program to
train local officials, Cooperative Extension can focus its
outreach where it will have the greatest impact in both
the short and long term. Local boards have the strongest
opportunity to comment on how land is to be used as it
undergoes development. Therefore, this project focused
on preventing future deterioration as  opposed to  fixing

Figure 1.  Map of Mumford River watershed study area.
existing problems. This is especially appropriate in a
rapidly urbanizing setting.

Both Massachusetts and Rhode Island chose to utilize
CIS technology because of its ability to store, analyze,
transform, and display geographic, or spatial informa-
tion. Its database management and analytical capabili-
ties make it a useful tool for pollution load modeling and
buildout scenario development, while its mapping capa-
bilities make it an excellent tool for sharing information
with local officials. This paper documents a case study
on how CIS technology was used to apply a watershed-
wide pollution loading model and to  develop buildout
scenarios for demonstrating to local officials the poten-
tial  impacts of future development  on water quality.

This project used CIS in four different applications:

• Printing customized, large-scale maps: This most ba-
  sic application of a CIS proved the most  useful for
  local officials. It was a revelation for some officials to
  see how their current zoning related to  actual land
  use. In  one town,  these maps inspired a change in
  zoning to  protect the area of a future water supply
  reservoir. These maps helped officials see how their
  towns fit into the regional picture and how their zoning
  and land use affected the adjoining towns, and vice-

• Performing  "buildout" analysis: A "buildout" analysis
  demonstrates the  consequences of existing zoning.
  It assumes that all land that can  be developed will be
  developed  at some future  date. In essence,  it is a
  spreadsheet that divides the land available for develop-
  ment in each zone  by the required lot size, subtracting

                                                                        I  I  Agriculture/Open Land
                                                                        HU  Developed Land
                                                                        V/\  Forest
                                                                        ggj  Unforested Wetlands
                                                                        PH  Lakes and Ponds
                                                                         Scale = 1:90,000

Figure 2.  Land use/land cover map of Mumford River watershed.

  a certain percentage for the road network and steep
  slopes. It is best used to evaluate different develop-
  ment scenarios, substituting different zoning require-

• Applying a watershed-wide pollutant loading model:
  CIS provided the input needed to apply the "Simple
  Method" for estimating existing and potential pollutant
  loads. Future pollution loading was estimated using
  a buildout  with existing zoning  and  again assuming
  the  implementation  of cluster  zoning. The  Simple
  Method was  compared in one subwatershed with the
  Galveston  Bay Method, which accounts for the hy-
  drologic class of the soils.

• Promoting planning for a greenway: Land use maps
  were  overlaid  with  parcel ownership  to show the
  existing  network of preserved  open space  and to
  identify those parcels of land having significant wild-
  life habitat and recreational value. In one town, these
  maps were used to gain funding for planning a river

Database Development

The most daunting aspect of using a CIS is the prospect
of spending a great deal of time and money creating a
useful database. Fortunately for Massachusetts, many
of the basic coverages needed for regional planning are
housed in a state agency, MASS CIS, and are available
for a small processing fee. These coverages  include
most of what appears on the standard United States
Geological Survey (USGS) map: roads, streams, town
boundaries, as well as watershed boundaries and land
use data generated from the interpretation of aerial pho-
tographs. The university entered into an agreement
whereby we gained access to this data at no charge, in
return for sharing the  new coverages that the project
would generate.

New coverages needed for the study included: zoning,
soils,  sewer and water lines, and land ownership, or
parcels taken from the assessor's maps. The soils maps
were obtained from the SCS, digitized by hand, then the
scale was converted with a computer program, "rubber-
sheeting," to achieve a uniform scale of 1:25,000.  All
other  new coverages were  transferred onto a USGS
topographical map at a scale of 1:25,000,  then digitized
directly into the computer. We obtained elevation data,
but the triangulation process used to convert elevation
data to slopes would require so much time and memory
that, for our purpose, deriving a slope map from the four
classes identified on the soils map was sufficient.

While CIS computer programs are powerful enough to
perform most overlay and analysis functions necessary
in  nonpoint source pollution load modeling, database
development and accuracy  issues can limit the effec-
tiveness of such modeling.  The choice of which model
to use is a function of which data are available for input.
Physics-based distributed models are more precise but
require detailed input parameters, beyond the scope of
this project.  The extent of our database limited us to
lumped-parameter  empirical  models. We chose  two
such models, the Simple Method and the Galveston Bay

GIS Applications

The Simple Method

Schueler (2) developed the Simple Method, one of the
simplest lumped-parameter empirical models. The input
data necessary to  compute pollutant loading with the
Simple Method are land use, land area,  and  mean an-
nual rainfall. Land  use determines which event mean
concentration (EMC) values and percentage of impervi-
ousness to use in the computation. The amount of rain-
fall  runoff is assumed  to   be a  function  of  the
imperviousness of various land uses. More densely de-
veloped areas have more impervious surfaces, such as
rooftops and paving, which cause stormwaterto run off
the  land  rather than be absorbed  into the soil. The
Simple Method can generate  rough figures for annual
pollutant loading within a watershed  and  can effectively
show relative  increases  in pollutant levels as  land is

The formula  used in the Simple Method is as follows:
                          (C)  * (A)  *  (2.72)
       (load) =  (runoff)  *  (EMC) *  (area)


L   = pounds of pollutant load per year
P   = rainfall depth (inches) over the desired time
     interval (1 year)
Pj  = percentage of storms that are large enough to
     produce runoff (90 percent)
Rv  = fraction of rainfall that is converted into runoff
     (Rv = 0.05 + 0.009 (I), where I represents the
     percentage of site imperviousness)
C   = flow-weighted mean concentration (EMC) of
     the targeted pollutant in  runoff (milligrams per
A   = area (in acres) of the study region

The Simple Method can be applied using a hand-held
calculator or a computer spreadsheet program. For this
project, the calculations were performed entirely within
the  ARC/INFO GIS environment,  where the input data
were stored. Results were exported to the Excel spread-
sheet program for presentation purposes.

The application of the Simple Method consists of three
major steps.

Step 1: Aggregate Land Uses and Obtain Area Fig-
ures for Land Use Categories Within Each Subbasin

The land use coverage in our database has 21 catego-
ries.  For the purpose of applying the Simple Method,
these were aggregated into the following six major cate-
gories, based  on development density: undeveloped
forest and other open land, large-lot single-family resi-
dential, medium-density residential, high-density  resi-
dential, commercial, and industrial. The aggregated land
use categories were matched with study basins from the
Nationwide Urban Runoff Program (NURP) for the pur-
pose of assigning EMC values.

Step 2: Enter Percentage Imperviousness and Event
Mean Concentrations for Each Land Use Type

The TABLES module of ARC/INFO was  used to assign
percentage of imperviousness and EMC values to indi-
vidual land use polygons within the watershed's subbas-
ins. The estimated percentage of imperviousness was
obtained  from  Schueler's guide to  using  the Simple
Method (2). EMC values for three pollutants—phospho-
rous,  nitrogen, and  lead—were taken  from selected
NURP study basins and were assigned to the aggre-
gated land uses within the watershed.

Step 3: Input the Simple Method's  Mathematical
Loading  Formula, Calculate  Loading Results  for
Each Distinct  Land Use Area, and Sum Results by
Watershed Subbasin

Finally, the pollutant load was  calculated for each dis-
tinct land use area within the Mumford River watershed
by inputting the loading formula through the TABLES
module of ARC/INFO. The mean  annual rainfall figure
was assumed to  be that of Worcester, Massachusetts,
or 47.6 inches. After calculating loading figures for phos-
phorous,  nitrogen, and lead for each distinct land use
area, these numbers were summed for each watershed
subbasin, using the ARC/INFO frequency table report-
ing capability.

The  Galveston Bay Method

As an experiment,  we  applied  the Galveston  Bay
Method to one of the subbasins to compare results with
the  Simple Method. The  slightly more sophisticated
Galveston Bay model considers soil drainage charac-
teristics in addition to land use/imperviousness to deter-
mine rainfall runoff. This method is similar to the Simple
Method, in that amount of rainfall  runoff and EMCs for
particular land uses are multiplied by land area to deter-
mine total pollutant  load (3).  Runoff in this method,
however,  is calculated  using the USDA SCS's TR 55
runoff curve model. The SCS model calculates runoff as
a function of both land use and soil type. Runoff equals
total  rainfall minus interception by vegetation,  depres-
sion storage, infiltration before runoff begins, and  con-
tinued infiltration after runoff begins (4).
The formula used with the Galveston Bay Method has a
structure similar to that of the Simple Method and is as

          P + 0.8[(100(yCN)-10]
   (load) =

L    = milligrams of pollutant load per year
P    = mean annual rainfall amount
CN  = runoff curve number, which is a function of
       soil type and land use
EMC = event mean concentration
A    = area (in acres) of the study region

The application of the Galveston Bay Method consists
of four major steps.

Step 1: Aggregate Land Uses and Obtain Area Fig-
ures for Land Uses Categories. Aggregate Soils Ac-
cording to Drainage Classes

Land use types were aggregated into the same six major
categories as the  Simple Method in orderto match EMC
values and to allow for later  comparison  of the two
pollution loading  methods. Soils were aggregated ac-
cording to drainage classes for use with the  USDA SCS
TR 55 runoff formula. The SCS  identifies four classes of
soils according to their drainage capacity:

Class A = excessively to well-drained sands or
          gravelly sands.
Class B = well to  moderately drained, moderately
          coarse soils.
Class C = moderately to poorly drained fine soils.
Class D = very poorly drained  clays or soils with a
          high water table.

Step 2: Overlay Soils Data With Land Use Data and
Clip This New Coverage Within the Subbasin

The ARC/INFO CIS overlay capability was used to over-
lay land use  and soils maps  for the Mumford  River
watershed on top of each other. This created new, dis-
tinct areas of different land use and soils combinations.
Because we were only applying this model  in one sub-
basin, the subbasin boundary was used in  conjunction
with the ARC/INFO "clip" command to  cut out (like a
cookie cutter) that portion of the watershed within the

Step  3: Assign  Runoff  Curve Numbers and EMC
Values to Each New Land Use/Soils Polygon Within
the Subbasin

EMC  values  were  assigned  to each  distinct  land
use/soils area in  the same  manner as they were as-
signed to  land use areas using the Simple Method.

Runoff curve  numbers were assigned to each distinct
land use/soils area  within the  subbasin according to
values established by the USDA SCS.

Step 4: Calculate Loading Results for Each Distinct
Land Use/Soils Area and Sum Results for the Entire

Finally, the pollutant load was calculated for each dis-
tinct land use/soils area within the subbasin by inputting
the loading formula through the  TABLES  module of
ARC/INFO. After calculating  loading figures for phos-
phorous,  nitrogen,  and  lead for  each  distinct land
use/soils  area, these figures were then summed for the
subbasin using the ARC/INFO frequency table reporting
capability. Results of this  modeling were converted from
milligrams per acre peryearto pounds per acre peryear
to facilitate later comparisons.

Buildout Scenarios

For planning purposes, CIS is most useful in its ability
to quickly generate  alternative scenarios. When these
development scenarios are coupled with a pollutant load
model as described above, alternative scenarios can be
evaluated according to their impact on water quality.
This project generated two different scenarios for each
of the four towns in the watershed: a maximum buildout
with existing zoning  and a maximum buildout with clus-
tered  development.

Maximum Buildout

A maximum buildout scenario was used to  show the
worst case for development according to current zoning
regulations (see Figure 3). The  result of this buildout is
expressed both in the number of new residential units to
be built and in the area of land to be converted from
undeveloped to residential and other urban uses.

Step 1: Eliminate From Consideration All Land That
Is Already Developed

Step 2: Eliminate From Consideration All Land That
Is Under Water

Step 3: Eliminate From Consideration All Land That
Is Protected From Development

These  protected lands included cemeteries, parks, and
all land permanently restricted from development.

Step 4: Reduce the Remaining Amount of Land by
20 Percent To Account for New Roadways and Ex-
tremely Steep Slopes

The remaining land was considered to have "developa-
ble" status. Wetlands  were included  in this category
because while a house probably would not be built on a
wetland, wetlands can and often do constitute portions
of the required lot size of large residential lots.

Step 5: Overlay the Land Use Coverage With Zoning
and Minimum Lot Size Information

This created new land use areas as a function of zoning.
All forests and fields were converted to a developed

Step 6: Divide Net Developable Land  Area  Within
Each Zone by Minimum Lot Size Allowed To Obtain
the Number of New Units

Results from the buildout are expressed  in the number
of new units.  Results can also be shown spatially by
shading in areas on the map according to future density

                                                                      Already Developed

                                                                      <1/4-Acre Lot Size

                                                                      1/4-to 1/2-Acre Lot Size

                                                                      1/2-to 1-Acre Lot Size


                                                                  !Z3 >2-Acre Lot Size

                                                                  !	I Protected/Public Land

                                                                      Scale = 1:90,000
Figure 3.  Maximum buildout scenario within the Mumford River watershed.

of development (darker shades for higher density, lighter
shades for lower density).
opment can reduce future levels of water pollution, es-
pecially from nutrients (see Figures 4 and 5).
Clustered Buildout

Another alternative development scenario was gener-
ated assuming the implementation of clustered develop-
ment. All areas zoned for lots larger than 1  acre were
changed to cluster zones, where three-fourths of the
land area remains undeveloped, and the remaining one-
fourth of the land area is developed at a  density of
1/2-acre  lot size.  With clustering, an area  zoned for
2-acre house lots still supports the same number of new
units, but three-quarters of the land  area remains open
space for passive recreation, protected wildlife habitat,
and as a buffer zone to filter runoff.

Step 1: Select All Land  Available for Development
Zoned for  1-Acre Lots or Larger

Step 2: Multiply Selected Land  by 0.75 and Add to
the Category of Protected Land

Step 3: Multiply Selected Land by 0.25 and Change
the Minimum Lot Size to One-Half an Acre

Step 4: Divide Step 3 by 20 Percent To Allow for New
Roads and Steep Slopes

Step 5: Divide Step 4 by 21,780 (One-Half an Acre)
To Determine Number of New Housing Units


Lumped-parameter empirical models were chosen for
this project and were applied to watershed  subbasins
ranging in size from 1 to 20 square miles and having an
average of  4 square miles. The application of the Simple
Method to  existing land use conditions allowed for a
comparison of the Mumford River watershed's subbas-
ins  for the purpose of identifying the subbasins that
contribute the highest  levels of pollutants per acre per
year. The development of a maximum buildout scenario
identified those areas within  the watershed that will
sustain the greatest amount of new growth.  The appli-
cation of the  Simple Method to this maximum buildout
scenario revealed that pollutant levels in surface water
runoff would  increase substantially for all subbasins in
the watershed. This finding supports the theory of  a
positive  relationship  between  development  and in-
creased pollutant levels from surface water runoff.

The development of a customized buildout scenario for
future development identified those  areas that are cur-
rently zoned  for large-lot  residential "sprawl" and that
can support higher development density under cluster
zoning, while protecting a significant amount of open
space that  can support a variety of beneficial uses. The
application  of the Simple Method  to the customized
buildout scenario revealed that the use of cluster devel-
Results determined  by applying the  Galveston Bay
Method to one subbasin  were  compared  with  those
obtained using the Simple Method. The predicted pollut-
ant loading from current conditions differed significantly
              Nitrogen Loading Estimates
  Whitin Reservoir
    West Sutton
    Wallis Pond
   Tuckers Pond
   Stevens Pond
    Rivulet Pond
    Morse Pond
  Manchaug Pond
    Lackey Pond
    East Douglas Jl
    Cross Street
    Badluck Pond
                     50       100      150
                         Percent Change
               n  1985-Cluster
   1985 - Maximum
Figure 4.  Chart showing difference in simple method results
         for nitrogen loading between maximum and custom-
         ized  buildout scenarios.
             Phosphorus Loading Estimates
      Whitin ._-
    West Sutton
    Wallis Pond
   Tuckers Pond
   Stevens Pond
    Rivulet Pond
    Morse Pond
  Manchaug Pond
    Lackey Pond
    East Douglas
    Cross Street
   Badluck Pond
         -10  0
                      20   30   40
                       Percent Change
1985 - Maximum
Figure 5.  Chart showing difference in simple method results
         for phosphorus loading between maximum and cus-
         tomized buildout scenarios.

between the two methods; the Simple Method consis-
tently predicted five times the amounts generated by the
Galveston Bay Method.

When the two methods were applied to both the maxi-
mum and customized buildout scenarios, however, the
percentage  growth of predicted pollutant loadings was
remarkably similar for both methods; the Simple Method
consistently predicted loadings  10 to 15 percent greater
than the Galveston Bay Method.  This  indicates that
while the Galveston  Bay Method  may provide more
accurate results  in predicting actual pollutant loading,
the Simple Method is adequate enough for evaluating
and  comparing  different  development scenarios  (see
Figures 6 and 7).


As states begin to implement a TMDL approach to regu-
lating water quality, they  face the quandary of how to
determine the extent of nonpoint source pollution in the
rivers. The crudest method is to subtract from the total
load those quantities generated by point sources and
call all the rest nonpoint source. While this is appropriate
in some settings, it is unacceptable in a watershed with
a long history of pollution because  a significant source
of pollution is the resuspension of historical sediments
stirred up by storms. The situation demands the devel-
opment of a model to predict the loading from nonpoint
sources. Only a computer can  handle the  multiple fac-
tors that interact to generate nonpoint sources of pollution.

As greater emphasis is placed  on watershed  planning,
the abilities of a CIS to input, store, manipulate, analyze,
and display geographic  information become indispensa-
ble. As the scientific community improves its knowledge
base for determining the critical factors influencing non-
point source pollution,  CIS technology is  improving in
its ability to store and handle large amounts of data.

While a detailed, physics-based distributed model would
be more accurate than the lumped-parameter models
used for this project, they are difficult to  apply at the
watershed scale. The real limiting factor is the provision
of all the data coverages needed to apply complex mod-
els.  Lumped-parameter models, such  as the Simple
Method and the Galveston Bay  Method, are ineffective
for accurately predicting  pollutant  loads, but they are
suitable for comparing and evaluating alternative devel-
opment scenarios.

Time, and the development community, will not wait until
all the answers are known.  Local  officials continue to
approve development with no thought to the impacts on
water quality. These officials need to be informed about
the implications of haphazard  growth. A CIS, with  its
ability to generate customized maps and quickly evalu-
ate alternative development scenarios, is a powerful tool
to  help  local officials visualize  how the decisions they
                                                                       Nitrogen Loading
                 50     100      150
                    Percent Change
 D Galveston Method Percent
   Change 1985 to Cluster

 ED Simple Method Percent
   Change 1985 to Cluster
Galveston Method Percent
Change 1985 to Maximum Buildout

Simple Method Percent
Change 1985 to Maximum Buildout
Figure 6.  Chart showing difference between Simple Method re-
         sults and Galveston Bay Method results for nitrogen
                Lead Loading
                  50     100      150
                     Percent Change
 D  Galveston Method Percent
    Change 1985 to Cluster

 E3  Simple Method Percent
    Change 1985 to Cluster
Galveston Method Percent
Change 1985 to Maximum Buildout

Simple Method Percent
Change 1985 to Maximum Buildout
Figure 7.  Chart showing difference between Simple Method re-
         sults and Galveston Bay  Method  results for  lead

make on paper today will have an impact on the land

1. Metcalf & Eddy, Inc. 1991. Assessment of pollutions in Narragan-
  sett Bay. Draft report to the Narragansett Bay Project.

2. Schueler, T. 1987. Controlling urban runoff: A practical manual for
  planning and designing urban BMPs. Washington, DC: Metropoli-
  tan Washington Council of Governments.

3.  Newell, C., et al. 1992. Characterization of nonpoint sources and
   loadings to Galveston Bay, Vol. I, Technical Report: Galveston Bay
   National Estuary Program (March).
4.  U.S. Department of Agriculture (USDA) Soil Conservation Service.
   1975.  Urban hydrology for small watersheds. Technical Release
   55. Springfield,  VA: U.S. Department of Agriculture.

Additional Reading

Arnold, C.L.,  et al.  1993. The use of Geographic Information System
images as a  tool to educate local officials about the land use/ water
quality connection. Proceedings of Watershed Conference, Alexan-
dria, VA.

Chesebrough, E. 1993. Massachusetts nonpoint source management
plan. Massachusetts Department of Environmental Protection, Office
of Watershed Management (October).

Joubert, L, et al. 1993. Municipal training for water quality protection.
Contribution #2845. College of Resource Development, University of
Rhode Island/Rhode Island Agricultural Experiment Station.
Massachusetts  Geographic Information System.  1993.  MassGIS
Datalayer descriptions and guide to  user services.  Boston, MA: Ex-
ecutive Office of Environmental Affairs.

McCann, A., et  al. 1994. Training municipal decision-makers in the
use of Geographic Information Systems for water resource protection.
Contribution #2927. College of Resource Development, University of
Rhode  Island/Rhode Island Agricultural Experiment Station.

Narragansett  Bay Project.  1992. Comprehensive conservation and
management  plan: A summary (January).

U.S. EPA.  1983. Results of the Nationwide Urban Runoff Program
(NURP), Vols. I and II. Final report prepared by Water Planning Divi-
sion. NTIS PB84185537. Springfield, VA: National Technical Informa-
tion Service.

U.S. EPA.  1992. Compendium of watershed-scale models for TMDL
development.  E PA/841 /R-92/002. Washington, DC (June).

        The GIS Connection to Residential Yard Soil Remediation
                       Jennifer L. Deis and Robin D. Wankum, P.E.
                          Black & Veatch Special Projects Corp.
                              Overland Park, Kansas 66211
By using innovative approaches to geographic information system (GIS) applications, the U.S.
Environmental Protection Agency (EPA) and Black & Veatch Special Projects Corp. (BVSPC), as
a contractor to EPA, were able to implement a site investigation concurrently with the site's
cleanup. This effort ultimately saved the EPA approximately $1.2 million dollars. The purpose of
the investigation was to locate and prioritize the residential yards that were adversely affected by
mining activities. BVSPC used an environmental data management system (EDMS) to
consolidate x-ray fluorescence (XRF), global positioning system (GPS), and laboratory analytical
data into a unique and flexible electronic, GIS-compatible database. The application of the EDMS
with a desktop GIS allowed effective completion of the investigation of the Oronogo-Duenweg
Mining Belt Site. This paper will present the GIS approach used to expedite the investigation and
cleanup activities at the site and will identify the benefits of this process.

Site History
The Site is part of the Tri-State Mining District, which covers hundreds of miles in southwestern
Missouri, southeastern Kansas, and northeastern Oklahoma. Mining, milling, and smelting of lead
and zinc ore began in the  1850s and continued in the District until the 1970s. The activities
generated approximately 9 million tons of mine and mill wastes and smelter related materials.
These wastes, which contain high levels of lead,  cadmium, and zinc, were deposited throughout
Jasper County. Additionally, air emissions during smelting operations resulted in the
contamination of soil surrounding the smelters. Approximately 6,500 residences are now located
within the 60 square-mile Superfund Site where lead and zinc were mined.

The Missouri Department  of Health (MDOH) conducted an exposure study at the Site to evaluate
health effects on residents living in the area (1). The study concluded that the most significant
source of contamination resulting in elevated blood-lead levels was residential yard soils.
Preliminary characterization of the surface soils within a 3/4-mile radius of the largest smelter
indicated that 86 percent of those yards exceeded the 800 parts per million (ppm) target cleanup

level for lead. In addition, previous investigation activities conducted by others concluded that 50
percent of the homes within 200 feet of mine and mill waste pile locations also exceeded the
target cleanup level for lead (2). The site location and its designated areas, areas where mining
activities and/or wastes were known to occur, are shown on Figure 1.  The above findings led the
EPA to develop an overall strategy to prioritize and expedite Site cleanup. EPA contracted the
U.S. Army Corps of Engineers (USAGE) to manage the cleanup activities at the site concurrently
with the BVSPC investigation of the site.

Investigation Objectives
The ultimate goal of the investigation was to identify and locate those  properties exceeding the
established 800 ppm surface soil cleanup level for lead within the designated areas. These areas
included the radius of the main "smelter zone," other small smelter areas, and specific mine and
mill waste areas. The area to be investigated around the smelter zone was estimated by
comparing the data from previous characterization activities at the Site. The lead smelter
location and the historical smelter stack height were compared with  the prevailing southeast
wind direction to determine the impacted zone.

A secondary objective of the investigation was to prioritize the cleanup of residential yards.
Considering census information obtained from each property access agreement and the XRF
data, special attention was directed to the following two scenarios:
   •   Residences with toddlers and children under 7 years of age; and
   •   Residences with soil lead levels greater than 2,500 ppm.
The first scenario was selected based on the findings of the MDOH study. Fourteen percent of
children under the age of 7 years in the study area had elevated blood-lead concentrations
resulting primarily from residential yard soils (1). The second scenario was chosen because those
residential yard soils exceeding 2,500 ppm were considered to pose an excessive risk to residents
of all ages coming into contact with them.

Figure 1. Site and Designated Area Location Map

To prevent any further exposure to the elevated lead levels, EPA desired the cleanup activities to
be conducted at the same time as the investigation. The flexibility of the EDMS and the desktop
GIS allowed this to occur by providing the USAGE with timely information to focus their cleanup
activities on the areas of concern.

The GIS Approach
The use and ease of applying GIS were supported by the following activities:

   •   Availability of the background database and mapping resources;
   •   Using ArcView GIS™, a desktop GIS, to maintain the sample location information; and
   •   Field reconnaissance during  the investigation to confirm and catalog sample locations not
       previously listed in the background information.

BVSPC, with assistance from EPA, requested existing database coordinate files from Jasper
County officials. These files were acquired in two databases. The first database was the City of
Joplin 911 database, and the second database was a developing rural Jasper County 911
database. The databases  received by BVSPC were a modified subset of the original database,
containing address and coordinate information only. This subset served to maintain the privacy of
each individual resident.

In addition, electronic US Geological Survey Maps of Jasper County were obtained from the
USAGE during their time-critical cleanup activities at the site. Images of the areal extent of the
main smelter zone and the designated mine waste areas were imported from AutoCAD into
ArcView. These AutoCAD files were created during previous project activities at the site. The
USGS maps and the AutoCAD images provided the base map for the sampling areas  of concern.

With the use of ArcView, the combination  of the 911 databases and the base map enabled
BVSPC to identify the residential properties within the areas of concern. Those properties falling
within the areas of concern shown on the base map were selected to separate them from the
main databases that contained properties for the entire County. ArcView allowed the selected
subset of properties to be  exported to Microsoft Excel™, where the new database of geographical
data was sorted by various fields for ease of reference. This subset initially contained 8,000

         #  Property Requring Sampling
         H  Denied Access
              and Mining Areas
                       Figure 2. Residences "To Be Sampled" Map
properties requiring sampling within the Site for determination of surface soil lead levels. Figure 2
shows a portion of the properties to be sampled.
Field reconnaissance of these initial 8,000 properties was conducted as part of the access
agreement process. Access agreements were required from the owner of each property to be
sampled and included obtaining the owner's permission and taking a census of the residents living
at the property. The EDMS significantly reduced the effort for field reconnaissance by
incorporating the database information into a GIS presentable format, allowing BVSPC to easily
identify the boundaries of the investigation. During reconnaissance, approximately 2,900
properties were recorded as being commercial properties, previously remediated properties, or
properties where the owner denied access for sampling. These properties were omitted from the

initial subset of 8,000. In addition to the remaining 5,100 properties in the database that required
sampling, approximately 1,400 other residences were identified. These additional residences were
mainly identified outside the city of Joplin, in smaller rural cities and towns. The county 911
database did not include the properties of homes within "city limits" since it was established as a
rural property identification system. The rural 911 database was in the process of being developed
and was incomplete in certain areas. The additional 1,400 identified residences would also require
sampling and surveying to add them into the GIS database.

Portable GPS units, provided by EPA, were used to survey the 1,400 individual properties. EPA
provided BVSPC with the  coordinate location of the BVSPC Project Office in Joplin to ensure that
the data points obtained at each property could be within + 3 feet. Some interference was
encountered during the GPS survey from various "line of sight" obstructions, including trees and
electrical lines in the neighborhoods. GPS verification was obtained after several survey attempts
for all additional locations  requiring data. The GPS data was added to the existing geographical
database for the properties that were sampled.

GIS Data Manipulation
The EDMS for this project, as demonstrated in Figure 3, was a comprehensive data management
package that allowed BVSPC to manage large quantities of data. The data types included
analytical, hydrogeological, and geographical information related to specific sampling locations. It
served as the focal point of a "hands off" approach for input and output of environmental and GPS
data. This "hands off" approach was an essential element during the project due to the extensive
volume of data collected during the investigation. On average, three to five samples were
collected per residential yard during the investigation. The data for each sample location, including
XRF, laboratory confirmation, and GPS, were associated with the corresponding residential
property. In essence, the EDMS eliminated the potential of transcription errors and data entry
errors for 35,000 environmental data records and 6,500 coordinate and  GPS records, while
diminishing the duplication efforts of BVSPC personnel to manually review data and prioritize
contaminated residential properties.

                                 Zone Verification
           X-Ray Fluorescence (XRF)
          - 25,000 analytical data points
                                                                 Electronic Data
                                                                   data pts
Global Positioning System (GPS)
  - 6,500 positioning records
                                            Microsoft Office
                        Lead Result
                      Notification Letters
                                   Remedial Action
                                    Priority Query
File Conversions
                     Figure 3. Electronic Data Management Flow Diagram
During this investigation, the EDMS allowed the analytical data to be sorted through a querying

process. The following list includes examples of the types of queries run during the project:

•   Lead concentrations above and below the action level;

•   Lead concentrations indicating necessary remediation in ascending or descending order; and

•   Areal sorts according to street name, city, or zip code, to further pinpoint the areas of

    greatest remediation need.


Each of these queries was performed using Access and exported to Excel as a spreadsheet
file. Throughout the project, various specific preliminary queries were conducted at the request
of the EPA. Representative queries included the following:

•   Specific residence queries;
•   Continuous updates of those residences having soil lead concentrations exceeding action
•   Updates of residences with children under age 7 and high lead levels; and
•   Updates on homes with soil lead concentrations above 2,500 ppm.

XRF Data. Field environmental data was downloaded from XRF equipment into an electronic file
that was transferred from the field office via e-mail to the main office. In the main office, the XRF
file was formatted by an Excel macro to allow uploading into the EDMS. The data upload was
simplified with the help of another macro written in Access that requested information from the
database manager before searching for a specific data file to incorporate it into the EDMS. This
brief process  of transferring and uploading approximately 25,000 XRF data points allowed access
to the data within three days of sampling. All data was immediately referenced to the
corresponding geographical data to evaluate the existence of the "hot zones".

Laboratory Data. Laboratory environmental confirmation data was uploaded directly into the
EDMS using a pre-designed electronic data deliverable (EDO) package provided by the
laboratory. Prior to receipt of the EDO, the laboratory data was validated and qualified by an
outside contractor. During the upload process into the EDMS,  an analytical verification process
was accomplished that noted missing data. Approximately 10,000 confirmation data points were
entered into the EDMS through this process.

GPS Data. GPS data was downloaded directly from the field GPS units into an electronic file that
was transferred from the field office via e-mail to the main office. The GPS data underwent a post-
processing procedure to correlate latitude and longitude with each residential property. BVSPC
converted the latitude and longitude data to the State Plane-83 projection format used in ArcView.
Once these procedures were completed, the 6,500 coordinate and GPS records were
incorporated into the EDMS and the final presentation of data could begin.

GIS Presentation of Data

Presentation of the data was achieved through the availability of the created geographical
database and several additional software packages, including Surfer, Microsoft Office™, and
Surfer. Surfer, a 2-dimensional gridding and contouring package, was utilized to verify the main
"smelter zone". After approximately 15,000 analytical samples were collected, data was queried
from the EDMS and converted to a format able to be gridded and contoured in Surfer. With the
additional sampling data obtained during the investigation, a contour map of soil lead
concentrations was produced. The contour of the additional coverage area was exported from
Surfer as an AutoCAD file that could be imported into ArcView for comparison with the original
smelter zone. Upon review of the new contours, changes in the coverage of the original estimated
smelter zone were identified (Figure 4). The same identification process, as described in previous
paragraphs, was used to define the additional properties requiring sampling. The smelter zone
was expanded in several areas and reduced in  others, optimizing investigation efforts and
maximizing the protection of public health.

Microsoft Office Package.  The Microsoft Office Package gave BVSPC the flexibility of converting
the various types of data into acceptable formats. These formats were used  to prepare notification
letters to property owners, to present requested queried information to EPA, and  to query
information for use in Surfer and ArcView for final geographical information.

Property owner notification letters were prepared with the help of Excel files and Word mail-merge
capabilities. The letters informed the owners of the XRF lead sampling results for the high and low
concentrations in their soils and gave a tentative timetable in which the remediation efforts would
be conducted, if deemed necessary. These notices were  sent to property owners within one
month or less of the sampling date for their property. Without these electronic capabilities, each of
the thousands of letters would have been completed individually. The effort for individual letters
would have required much more personnel time and would not have been an expedient response
to concerned residents.

 M 336000-
       2774000 2776000 2778000 2780000 2782000 2784000 2786000 2788000 2790000 2792000
                 Figure 4. Smelter Zone Verification demonstrated in Surfer
The preliminary queries described earlier and the quick turnaround of the property owner
notification letters allowed reassurances to concerned property owners and allowed the
remediation prioritization for homes with children and those posing an excessive danger to

ArcView GIS. ArcView was the final step in the presentation of all information to be used by EPA
and the remediation contractor. It was used to create over 40 full-size maps of the remediation
area (Figure 5) for the final design. ArcView easily illustrated the individual residences that
exceeded the established  lead soil cleanup level for this Site as well as those residences with lead
concentrations below the action level. In addition, each residence identified on the map was linked
to a table containing pertinent information, including property owner information, ages of any
children living at the residence, and values of each XRF lead sample analysis (Figure 6). This
procedure allowed a visualization of the extent and location of residential yards requiring soil
remediation. Commercial locations, previously remediated properties, and those properties where

sampling access was denied were also illustrated on the investigation summary maps. The maps
allowed BVSPC personnel to confirm, track, and suggest the future progress of the soil
investigation. These maps were a great asset to USAGE, who used them to track the yards
requiring remediation and those yards that had already been remediated. EPA also used the
maps as illustrations at public meetings held in the area to keep residents informed of the

The investigation was completed effectively and efficiently due to the flexibility and adaptability of
the desktop GIS and our EDMS. As demonstrated above, the benefits of using  ArcView in
conjunction with the EDMS included the following:

=> Prioritized cleanup activities based on investigation results.
=> Maintained concurrent investigation and remediation efforts.
=> Provided a visual geographical tool for the investigation personnel.
=> Provided a visual geographical tool for the USAGE cleanup crews.

Cost requirements were minimized due to the flexibility of ArcView GIS and the EDMS resources.
The efforts for obtaining the geographical distribution of the contaminated properties  would have
required a much greater cost consideration during the implementation of the investigation and
continuation of remedial efforts had a GIS system not been implemented during the project.

      Jasper County
Yard Soil Remedial Design
          Area 7
Sampled Property
 o  No Remediation Required (< BOO ppm Pb)
 •  Remediation Required (> 800 ppm Pb)
 A  Properties already remediated
 *  Commercial Properties
 a  Denied Access
|   | Joplin City Streets
     Figure 5. ArcView Presentation of Data

file  Edil  View  Iheme  Eraphics Window  Help
                                                                                              Scale 1:| 14,207      2.JS6-Z
   /\/ County Streets/Roads
   VV/Smelter and Mining Areas

           ^Identify Result
jjflstart| Task Tracker 6 03	| jj|lnbox - Microsoft Outlook  | ^Microsoft Ward	11 '.\ArcView CIS Version 3....
                  Figure 6.  Linked Attributes Table to ArcView Property Location


1. Missouri Department of Health (MDOH), 1994, Jasper County, Missouri Superfund Site Lead
and Cadmium Exposure Study, May 1994.

2.  Dames & Moore, 1994, Residential Yard Assessment Report, Seven Designated Areas,
Jasper County Site, September 1994.

3.  U.S. Environmental Protection Agency, 1996, Record of Decision Declaration for the Oronogo-
Duenweg Mining Belt Site, Operable Units 2 and 3, Jasper County, Missouri, USEPA Region VII,
August 1996.

4.  Smith, Robert A., W. Todd Dudley, and Thomas L. Rutherford, 1995, GIS Applications in
Hazardous Waste Remediation, Black & Veatch, 1995.

5.  Black & Veatch Special Projects Corp., 1997, Oronogo-Duenweg Mining Belt Site, Remedial
Design, Statement of Work, Residential  Yard Soils, February 1997.

  Decision Support System for Multiobjective Riparian/Wetland Corridor Planning
                                Margaret A. Fast and Tina K. Rajala
                               Kansas Water Office, Topeka, Kansas
Kansas has numerous programs that affect riparian cor-
ridors and associated  wetlands. These programs in-
clude planning,  monitoring, assistance, research, and
regulatory activities. Although administration of these
programs often overlaps, integration of program objec-
tives into a holistic, multiobjective approach to resource
planning and  management has been lacking. A large
amount of resource data was routinely collected and
compiled, but no effective way had been developed to
integrate these data into the decision-making process.

The Kansas Water Office (KWO) was awarded a grant
in September 1992 from the U.S. Environmental Protec-
tion Agency (EPA) to develop a geographic information
system (CIS) decision support system (DSS) that would
enable the state to augment its ability to manage ripar-
ian/wetland corridors. The project used CIS to differen-
tiate between reaches of a stream corridor to evaluate
their environmental sensitivity. The Neosho River basin,
one of 12 major hydrologic basins in Kansas, was used
as a pilot to demonstrate the feasibility of the concept.

The KWO will use the DSS to help target  sensitive areas
in the Neosho basin for further planning activities. The
project will also  benefit  other state agencies in their
riparian/wetland corridor efforts. The implementation of
planning objectives may involve  local units of govern-
ment and, ultimately, private landowners.

Major phases of the project included:

• A needs assessment study

• A feasibility analysis

• A system design

• Construction of the DSS for the Neosho River basin

• A final evaluation of the DSS capabilities

An interagency project advisory group (IPAG), consist-
ing  of representatives  from eight agencies directly or
indirectly involved in riparian and wetland  protection
activities, was  formed  to assist in  project design and

Major steps involved in designing the DSS included:

• Selection and CIS development  of databases used
  for riparian corridor evaluation.

• Creation of riparian corridor segments.

• Development of an analysis methodology to apply to
  corridor segments.

• Evaluation of the DSS.

Databases Selected for Decision Support
System Development

Many types of data were reviewed for the DSS. Several
were  not used due to  the costs associated with geo-
graphically referencing  the data, given the current data

The databases listed in  Table 1 are available in the DSS.

During the system design phase of the project, the IPAG
identified the need to develop a pilot study area  for the
DSS.  The IPAG had difficulty understanding how a DSS
would use geographically referenced data sets (cover-
ages). Before committing to a  design for the develop-
ment  of a basinwide system, the IPAG decided  first to
develop a pilot study area, with a specific focus (appli-
cation), that could be on-line and demonstrated early.
This would allow time for further refinement of the scope
of work and identification of coverages to be developed
prior to basinwide development of the DSS. For the pilot
study application, the IPAG chose to assess the value
and vulnerability of the riparian areas  in two 11-digit
hydrologic unit code (HUC11) watersheds to allow the
user to evaluate a corridor segment and compare  be-
tween segments and to prioritize or target segments for
further planning activities.

As development of data layers progressed  for the  pi-
lot,  the IPAG quickly determined that the DSS project

Table 1.  DSS Database List
DSS Name      Data Description
Neosho River basin boundary
Riparian corridor
Stream channelization
Conservation easements
Water contamination
Riparian corridor
County boundaries
Dam structures
Water appropriations
United States Geological
Survey (USGS) stream gaging
Surface geology
11 -digit hydrologic unit
Kansas water quality action
targeting system
Land cover
Land cover statistics
Minimum desirable stream
flow monitoring gages
Nonpoint source pollution
Perennial hydrology
Populated places
Public lands
Section corners
1981 stream evaluation
Soil Conservation Service (SCS) HUC11 drainage basins; 1 :100,000-scalea
Original buffer on mainstem Neosho and Cottonwood; 147 corridor segments split
on tributary confluences
Division of Water Resources (DWR) legal description of locations
Locations of important natural resources that could be purchased by the state from
willing landowners for conservation protection
Kansas Department of Health and Environment (KDHE) contamination locations3
Final riparian corridor; 63 corridor segments developed from HUC11 boundaries
Kansas Geological Survey (KGS) cartographic database; 1 :24,000 scale3
DWR legal descriptions of locations
DWR legal descriptions of locations3
USGS latitude-longitude descriptions; CIS cover developed by USGS
KGS 1 :500,000-scale3
SCS HUC11 drainage basins; 1 :100,000-scale3
USGS 1:100,000-scale digital hydrology3
KDHE target valuable and vulnerable scores by HUC11 drainage basin
1 :100,000-scale developed from satellite imagery by the Kansas Applied Remote
Sensing Program, University of Kansas3
Summary statistics on land cover by corridor segment
Subset of USGS gaging stations
Target watersheds identified in the Kansas Water Plan
Reselected perennial streams from 1:100,000 USGS digital hydrology
Urban land cover (from landc) with 1980 and 1990 Census population data
Geographic names information system (GNIS) entries for Kansas; GIS cover
developed by USGS
State and federally owned land digitized from 1:100,000-scale USGS quad maps
USGS 1:100,000-scale digital roads3
KGS cartographic database; 1 :24,000-scale3
U.S. Fish and Wildlife stream evaluation study; Kansas Department of Wildlife and
Parks (KDWP) provided data on paper maps
T_and_e         Threatened and endangered
Temussel        Threatened and endangered
Tigrcity          City boundaries
Twp             Townships
Watrfowl         Water fowl locations

Wq_eff          Water quality: effluent
Wq_grnd        Water quality: ground water
Wqjake         Water quality: lake
Wq_strm         Water quality: stream
Stream locations of state and federal identified threatened and endangered
species; KDWP provided data on paper maps
Locations of state endangered floater mussels; KDWP provided data on paper maps

U.S. Census 1:100,000-scale TIGER line data; boundaries only, areas not named
(use with PPL)
KGS cartographic database; 1:24,000-scale3
KDWP locations and counts of annual waterfowl migration; data developed from
paper maps (Restrict public distribution of data per KDWP request.)
KDHE sampling sites; GIS cover developed by KDHE
KDHE sampling sites; GIS cover developed by KDHE
KDHE sampling sites; GIS cover developed by KDHE
KDHE sampling sites; GIS cover developed by KDHE
  Data available at the Kansas Data Access and Support Center (DASC).

parameters would  have to  be limited  to the riparian
corridor along the mainstem  of the Neosho and Cotton-
wood Rivers.  The costs associated with developing ri-
parian corridor segments for all perennial waters in the
Neosho basin was fargreaterthan the available funding.

Creation of Riparian Corridor Segments

A buffer width of one-half mile (one-quarter mile from
each stream bank) for the mainstem of the Neosho and
Cottonwood Rivers was used to produce the riparian
corridor. If more time and funding had  been available,
riparian corridors for all perennial streams in the Neosho
basin could have been developed. The development of
this second view of data, organized by the HUC11 wa-
tershed, would then  have  been useful for individual
watershed analysis because  all perennial streams in the
watershed could be analyzed.

The  intersection of the HUC11  basin boundaries seg-
mented the corridor. In several instances, small sliver
polygons were produced where the HUC11  boundary
paralleled the  river within the  1/4-mile corridor. The sliver
polygons were dissolved  into the majority  HUC11. In
other words, this project assumed that the 1/4-mile cor-
ridor buffer was more accurate and useful than the
1:100,000-scale HUC11 boundary.

Many of the HUC11 boundaries that the Soil Conserva-
tion Service (SCS) developed actually follow the course
of the Kansas  streams, ratherthan intersect them. When
this occurred along the Neosho and Cottonwood Rivers,
we found that the resulting opposing corridor segments
did not always balance with  an equivalent length. Also,
some HUC11   boundaries would first follow the  river,
then cross the river. This resulted in corridor segments
that encompass both sides of the river  for a portion of
the  segment  and follow only one side  of the river for
another portion of the segment. To address  these situ-
ations, the KWO arbitrarily added intersections to create
equivalent left and right bank corridor segments and to
create corridor segments that encompassed either one
side  of the river or both sides of the river.

Once the corridor segments were finalized and numbered,
the  corridor segment identification number (corrseg-id)
was  attached  to the other CIS  covers.  This allows the
reselection of data for a given corridor  segment, using
Boolean expressions in the DSS.

Development of an Analysis
Methodology: Land Use

The  IPAG determined  that one of the most significant
factors associated with the quality of the riparian corridor
is land cover.  Land cover was analyzed  for the riparian
corridor segments; the CIS cover lc_stats contains sum-
mary statistics for each corridor segment. The calcula-
tions discussed in the following paragraphs identify the
data found in the lc_stats cover. Due to the size of the
land cover data set in the Neosho River basin, the DSS
includes only the land cover within the riparian corridor.

One way of identifying corridor segments  in need of
protection or remedial action is to determine the ratio of
the number of acres in the corridor segment that contain
the preferred riparian land cover types (grasses, woods,
and water) to the number of acres that contain the least
preferred types of land cover (crops and urban areas).
The corridor segments can then be ranked according to
that ratio.

Other calculations are useful:

• bad_pct: percentage of the corridor segment that con-
  tains crop and  urban land cover types.

• bad_tbad: percentage of all crop  and  urban land
  cover for the entire riparian corridor that resides in
  the corridor segment.

• 'type'_pct: percentage of the corridor segment that is
  crop, grass, wood, water, and urban. Type' refers to
  each of the  five land cover types; lc_statsuses:  a
  separate value for each (e.g., crop_pct).

• 'type'_t'type': percentage of each  type of land cover
  for the entire riparian corridor that resides in the cor-
  ridor segment (e.g., crop_tcrop).

• 'type'_acres: total acreage of each type of land cover
  in the corridor segment (e.g., crop_acres).

• good_acres:  total acreage of grass, wood, and water
  in the corridor segment.

• bad_acres: total  acreage of crop and urban in the
  corridor segment.

Another significant benefit of the DSS is the ability to see
where the  land cover types are in relation  to the river.
As an example, the ability to identify corridor segments
that have crop land extending to the river on both banks
is useful because they are the segments  most vulner-
able to  bank erosion. Those segments can then be
targeted for further remedial activities planning.

Decision Support System Requirements

The DSS data sets were developed and analyzed using
ARC/INFO on a UNIX-based workstation. The final cov-
ers were then exported and transferred to a microcom-
puter for use in ARC/VIEW. Hardcopy prints are printed
to a Tektronix Phaser III color wax printer with 18 Mb of
RAM, running in Postscript mode.

The DSS data  sets total 26 Mb. ARC/VIEW version  1
requires 8 Mb of RAM to load the program. To run the DSS
efficiently, a 486DX-66 with 16 Mb of RAM  is preferred.
The DSS is slower on a 486DX-33 with 8 Mb of RAM. It
was not tested  on any other PC configuration, so a con-
figuration in between the two  may be satisfactory.

Processing GIS Data

Reselecting the perennial streams in the Neosho basin
and further identifying the mainstem of the Neosho and
Cottonwood Rivers using United States Geological Sur-
vey (USGS) 1:100,000-scale hydrography can be time
consuming. Perhaps the River Reach III covers should
replace that data in the future.

Attaching census data to the urban land cover polygons,
as was done  for the pop cover,  is not  recommended.
Use of the TIGER line files and cover would give a more
accurate  distribution of the population. Because less
than 5 percent of the riparian  corridor had urban land
cover, the KWO did not use the pop cover in its evalu-
ation. Several summary covers of TIGER and  census
data will soon be available from DASC.

Clipping the other ARC/INFO covers to  the  Neosho
basin and attaching the corrseg-id, using the  identity
command, was unremarkable.

Processing Non-GIS Digital Data

Channels and dams were in digital format  but were not
in ARC/INFO format. The files were processed using the
LeoBase conversion software from the KGS, then gen-
erated into ARC/INFO covers.  Some records were lost
in the conversion. The LeoBase program fails to convert,
or incorrectly  converts,  legal  descriptions for sections
that do not have four section corners (e.g.,  northeastern
Kansas). The Division  of Water  Resources is in the
process of attaching latitude-longitude to the  point loca-
tions. Processing these data should take only a few
hours at most.

Processing Nondigital Data

Several covers were developed on contract from paper
maps or legal descriptions. They were:  conservation
easements (con_ease), public lands (publand), stream
evaluation (streamev), threatened and endangered spe-
cies (t_and_e and temussel), and waterfowl  (watrfowl).
Most of the data for these covers were drafted  on a
1:100,000-scale USGS quad map and digitized. The
stream  evaluation data  were  developed  using  a
scanned paper map of the coded streams as a backdrop
for the 1:100,000-scale hydrography; the digital streams
were reselected and coded.

In  summary,  KWO's  GIS  personnel  needed  approxi-
mately 275 hours to develop the riparian corridor seg-
ments, process  the  land cover  data  and summary
statistics,  export the  covers,  transfer and import the
covers for ARC/VIEW, and assist  in  the development
and presentation of the DSS demo. Contract personnel
spent approximately 183 hours developing GIS covers
for the DSS. This does not include the time spent iden-
tifying the perennial and  mainstem  hydrology in the
USGS 1:100,000-scale hydrology.
Final Evaluation of the Decision
Support System

In its final evaluation of the system, the IPAG determined
the system to be useful and an excellent start at consoli-
dating a variety of data that have application for riparian
corridor/wetland  issues.  Many IPAG members  found
ways to use the DSS in their own programs. Additional
comments on the system evaluation are as follows:

• Concern  about the lack of complete wetland data.
  The land  cover data available could not identify wet-
  land areas.

• Need  for more detailed woodland data. Again, the
  resolution of the land cover data precluded detailed
  identification of woodland areas. The Kansas Biologi-
  cal Survey (KBS), the  KWO, and EPA are now pur-
  suing  options to  develop more detailed land  cover
  data, including wetlands and woodlands.

• The lack of information  on the tributaries did not allow
  full basin analysis, which  would be desirable. This
  issue  is addressed in  the "construction" discussion

• Desirability of expanding the project with elevation
  and temporal data.

• Lack of definition of the floodplain. Federal  Emer-
  gency Management Agency (FEMA) floodplain data
  are not easily  incorporated into a GIS. Other options,
  including  satellite imagery of the flood  of 1993, will
  be evaluated.

• Project development requires extensive communica-
  tion between  program  people and GIS technicians.
  This  can be a daunting task  due to  the technical
  vocabularies  involved and the many other ongoing
  activities  of the participants.

• Consideration of the requirements for transferring the
  project to other potential users. GIS applications gen-
  erally  use large  databases.  User microcomputers
  may not have the CPU, RAM,  and storage capacity
  necessary for the DSS application and often have a
  limited number of options for data transfer.

• Concern  about costs and time associated with the
  expansion of the DSS to other basins  in the  state.
  This  project was focused on one of the  12  major
  hydrologic regions in Kansas. Funding options, pro-
  ject scope, and system refinements based on the
  physical characteristics of the  other basins need to
  be pursued.

The  KWO  learned  that clearly defining a  single DSS
application  at the outset of the project is critical. The
KWO originally believed that the DSS could be devel-
oped with general descriptions of the broad range of
program applications, utilized by multiple agencies, that
could benefit from the DSS. Each participating agency

could bring its programs and needs to the IPAG for    areas for further planning activities, the  IPAG became
discussion; the resulting DSS would then  serve those    more confident in its advisory role. Upon completion of
multiple programs and needs. Instead, the  ambiguity of    the  project, the IPAG  members could readily identify
the objective confused the IPAG. Once the IPAG chose    how the DSS could be enhanced, modified, or directly
to focus on a single  application, the assessment  of    used in their own programs.
riparian corridor value and vulnerability to target priority

    Integration of GIS and Hydrologic Models for Nutrient Management Planning
                     Clyde W. Fraisse, Kenneth L. Campbell, James W. Jones,
                            William G. Boggess, and Babak Negahban
         Agricultural Engineering Department,  University of Florida, Gainesville, Florida

Recent evidence that agriculture in general, and animal
waste in particular, may be an important factor in surface
and ground-water quality degradation has increased the
interest in nutrient management research. The presence
of nitrogen and phosphorus in surface water bodies and
ground water is a significant water quality problem  in
many parts of the world. Some forms of nitrogen and
phosphorus, such as nitrate N and soluble P, are readily
available to plants.  If these forms are released into
surface waters, eutrophic conditions that severely impair
water quality may result. Advanced  eutrophication (pH
variations, oxygen  fluctuations or lack of it  in  lower
zones, organic substance  accumulation) can  cause
physical and chemical changes that may  interfere with
recreational use and aesthetic appreciation of water.  In
addition, possible taste and odor problems caused by
algae can make water less suitable  or  desirable for
water supply and human consumption  (1).

Increases in nutrient loadings to water resources have
recently been observed  in the  southeastern United
States, where well-drained sandy soils with low nutrient
retention capacity and high  water table conditions are
found in  most coastal areas. Those  increases  were
associated  statistically with nutrient sources  such as
agricultural  fertilizers and dense animal populations (2,
3). Repetitive occurrences of extensive blooms of blue-
green algae that threatened the overall health of Lake
Okeechobee, located in southern Florida, were attrib-
uted to an increase in nutrient loadings to the lake. The
South Florida Water Management District (SFWMD) re-
ported an increase of phosphorus concentrations  in the
lake water from an annual average of  0.049 milligrams
per liter in 1973 and 1974 to a peak of 0.122 milligrams
per liter in 1988 (4).

Most water  quality problems  concerning  phosphorus
result from transport with sediment in surface runoff into
receiving waters. Continuous high loadings from animal
waste on sandy soils with low retention capacity,  however,
may contribute significant quantities of labile phospho-
rus to subsurface drainage. Ground-water aquifers may
also become polluted due to recharge of high loadings
of nitrogen. Drinking water with nitrate N concentrations
higherthan 10 milligrams per liter may lead to methemo-
globinemia in infants. Ground-water monitoring of the
Middle Suwannee River area in Florida has shown high
concentrations of nitrate nitrogen near intensive agricul-
tural operations. The U.S. Geological Survey has inten-
sively monitored dairy and poultry farms and has found
high nitrate levels below these operations compared with
nearby control wells (5).

Animal waste management has always been  a part of
farming, but historically has been relatively easy due to
the buffering capacity of the land. In fact,  land applica-
tion of animal  waste  at acceptable rates can  provide
crops with an adequate level of nutrients, help reduce
soil erosion, and improve water holding capacity. As the
animal industry attempts to meet the food requirements
of a growing population, however,  it applies new tech-
nologies that reduce the number of producers, but cre-
ate larger, more  concentrated operations.  That,  in
addition to the decreasing amount of land available for
waste application, has increased the potential for water
quality degradation.

Successful planning of  an animal  waste management
system requires the ability to  simulate the impact of
waste production, storage, treatment, and  use on water
resources. It must address the  overall nutrient manage-
ment for the operation, including other nutrient sources
such as supplemental fertilizer applications. Livestock
operations are highly variable in their physical facilities,
management systems, and the soil, drainage, and cli-
matic conditions that  affect the risk of water pollution
from animal wastes (6). Linkage between geographic
information systems (GIS) and  hydrologic models offers
an excellent way to represent  spatial features of the-
fields being simulated and to improve results. In addition,
a GIS containing a relational database is an excellent way

to store, retrieve, and format the spatial and tabular data
required to run a simulation model.

This paper examines some of the  issues related to the
integration  of hydrologic/water quality models and CIS
programs.  In addition, the paper discusses  the ap-
proaches  used in  the  Lake Okeechobee Agricultural
Decision  Support System (LOADSS), which  was  re-
cently developed to evaluate the effectiveness  of differ-
ent phosphorus  control practices  (PCPs) in the  Lake
Okeechobee  basin.  The  paper  also details  a  dairy
model, designed to simulate and evaluate the impacts
of alternative waste management  policies for dairy op-
erations, that is currently under development.

Hydrologic Models and GIS

By using models, we can better understand or explain
natural phenomena and, under some conditions, make
predictions in a deterministic or probabilistic sense (7).

A hydrologic model is a mathematical representation of
the transport of water and its constituents on some part
of the land surface or subsurface  environment. Hydro-
logic models can be used as planning tools for determin-
ing   management  practices  that minimize  nutrient
loadings from an agricultural activity to water resources.
The  results obtained  depend  on an accurate  repre-
sentation of the environment through which water flows
and of the spatial distribution of rainfall characteristics.
These  models have  successfully  dealt with time, but
they are often spatially  aggregated or lumped-parame-
ter models.

Recently, hydrologists have turned their attention to GIS
for assistance in studying the movement  of water and
its constituents in the hydrologic cycle. GIS programs
are computer-based tools to capture, manipulate,  proc-
ess, and display spatial or georeferenced data.  They
contain geometry data  (coordinates and topological in-
formation) and attribute  data (i.e., information describing
the properties of geometrical objects such as points,
lines, and areas) (8). A GIS can represent  the spatial
variation of a given field  property by using  a  cell grid
structure in which the area is partitioned into regular grid
cells (raster GIS) or by  using a set of points, lines, and
polygons (vector GIS).

A close connection obviously  exists between GIS and
hydrologic models, and integrating them produces tre-
mendous benefits. Parameter determination  is currently
one of the most active  hydrology-related areas in GIS.
Parameters such as land surface slope, channel length,
land use, and soil properties of a watershed are being
extracted from both raster and vector GIS programs,
with a focus on raster-based systems. The spatial nature
of GIS  also provides an ideal structure for modeling. A
GIS can be a substantial time saver that allows differ-
ent modeling approaches to be tried, sparing manual
encoding of parameters. Further, it can provide a tool for
examining  the spatial  information from  various user-
defined perspectives (9).  It enables the user to selec-
tively analyze the data pertinent to the situation and try
alternative  approaches to analysis. GIS has been  par-
ticularly successful in addressing environmental prob-

Approaches for Integrating GIS and Models

A significant amount of work has been done to integrate
raster and vector GIS with hydrologic/water quality mod-
els. Several strategies and approaches for the integra-
tion have been tried. Initial work tended to use simpler
models such as DRASTIC  (10) and the Agricultural
Pollution Potential Index (11). In these cases, the mod-
els were implemented within the GIS themselves. These
studies attempted to  develop  CIS-based screening
methods to rank nonpoint pollution potential. The use of
more complex models requires that the GIS be used to
retrieve, and possibly format, the model data. The model
itself is implemented separately and communicates  with
GIS via data files. Goodchild (12) refers to this mode as
"loose  coupling," implying that the GIS and  modeling
software are coupled sufficiently to allow the transfer of
data and perhaps also of results, in the reverse direc-
tion. Fedra  (8) refers to this level of integration as "shal-
low coupling" (see Figure 1). Only the file formats  and
the  corresponding input and output routines, usually of
the model, must be adapted. Liao and Tim (13) describe
an application of this type, in which an  interface was
developed  to generate topographic data automatically
and simplify the data input process for the Agricultural
Nonpoint Source (AGNPS) Pollution Model (14), a water
quality model.
Shared Databases and Files



User Interface



User Interface
Figure 1.  Loose or shallow coupling through common files (8).

Higher forms of connection use a common interface and
transparent file or information sharing and transfer be-
tween the respective components (see Figure 2). The
dairy model, currently under development, is an appli-
cation of this kind. It will link the Ground-Water Loading
Effects of Agricultural Management Systems (GLEAMS)
(15) model and CIS to evaluate potential leaching and
runoff of both nitrogen and phosphorus.

LOADSS  is an extension of this type of application
because it includes an optimization module that enables
the system to select the best PCPs at the regional scale,
based on goals and constraints defined by the user.

Both applications use ARC/INFO's arc macro language
(AML), a high-level application language built into the
CIS. A subset of functions of a full-featured  CIS, such
as creation of maps (including model output) and tabular
reports,  as well as model-related analysis, are embed-
ded in the applications, giving the system great flexibility
and performance. Fedra (8) describes a deeper level of
integration  that would merge  the two  previous ap-
proaches, such that the  model becomes one of the
analytical functions of a CIS, or the CIS  becomes yet
another  option to generate and manipulate parameters,
input and state variables, and  model output, and  to
provide additional display options. In this case, software
components would share memory rather than files.

The choice between integrating a water quality model
with a raster or vector CIS depends on the importance
of spatial interactions in the process being studied and
the nature of the  model itself. Some water quality mod-
els, such as GLEAMS, are field-scale models that pro-
vide  edge-of-the-field values for surface  runoff and
erosion  as well as deep  percolation of water and  its
constituents. In this case, spatial interactions between
adjacent fields are ignored and a vector CIS can  be
used to  describe the system. Moreover, important fac-
tors in the simulation process, such as land use and
management practices, are normally field attributes and
thus, are better represented in a vector structure.

Other factors playing an important role in the  hydrologic
process, such as field slope, aspect, and specific catch-
ment area, are  hard  to estimate in  vector systems,
however. A raster-based CIS is better suited for handling
watershed models in which the routing process is impor-
tant and spatial interactions are considered.  For those,
several algorithms for estimating important terrain attrib-
utes  are often incorporated in  commercially available
raster-based CIS programs.


LOADSS was developed to help address problems cre-
ated by phosphorus runoff into Lake Okeechobee. It was
designed to allow regional planners to alter land uses
and management  practices in  the Lake  Okeechobee
              Shared Databases and Files
               Common User Interface
Figure 2.  Deep coupling in a common framework (18).

basin, then view the environmental and  economic ef-
fects resulting from the changes. The Lake Okeechobee
basin coverage  incorporates information about  land
uses, soil associations,  weather regions,  management
practices, hydrologic features, and political boundaries
for approximately 1.5 million acres of land and consists
of close to 8,000 polygons.

The SFWMD, responsible for managing Lake Okeechobee,
has initiated numerous projects to develop  effective con-
trol  practices for reducing the level  of phosphorus in
agricultural runoff as part of the Lake Okeechobee Sur-
face Water  Improvement and  Management  (SWIM)
Plan.  These projects, numbering more than 30, were
designed to develop information on the control and man-
agement of phosphorus within the lake  basin and to
determine the costs and effectiveness of selected man-
agement options. Three types of control options are
being studied:

• Nonpoint source controls, such as pasture management.

• Point source controls, such as sewage treatment.

• Basin-scale controls, such as aquifer storage and

After  completing most  of these research efforts, the
need arose for a comprehensive management tool that
could integrate the results for all three classes of PCPs.
In response to these needs, design and implementation
of a decision support system was initiated with the fol-
lowing objectives (16):

• Organize spatial  and nonspatial  knowledge about
  soils, weather, land use, hydrography of the lake ba-
  sin, and PCPs under a  GIS environment.

• Develop and implement algorithms for modeling non-
  point source, point source, and basin-scale PCPs.

• Develop and implement mechanisms for evaluating
  the performance of the entire Lake Okeechobee basin
  under different combinations of PCPs applied to the

• Design and develop a user interface that would fa-
  cilitate use of the system by noncomputer experts.

The goal in developing LOADSS was to create an infor-
mation system that would integrate available information
to help regional planners make decisions. LOADSS can
generate reports and maps concerning  regional land
attributes,  call  external  hydrologic simulation models,
and display actual water quality and quantity sampling
station  data.  LOADSS  is a  collection  of different

• The regional scale CIS-based model used to develop
  and manipulate regional plans for reducing phospho-
  rus  loading to Lake Okeechobee.

• The Interactive Dairy Model (IDM) used to develop
  field-level management plans for dairies and run the
  Field  Hydrologic  and  Nutrient  Transport  Model
  (FHANTM)  simulation  model for  nutrient transport

• An optimization module that enables the  system to
  select the best PCPs at the regional scale (currently
  under development).

Although these components can run independently, they
are fully integrated in the LOADSS  package and  can
exchange information where necessary. A design sche-
matic of LOADSS is given in Figure 3.

Regional-Scale CIS-Based Model

LOADSS serves  both as a decision support system for
regional planning and as a graphic user interface for
controlling the different components. One consideration
in the  design of LOADSS was the size of the database
that was being manipulated. Because the land use da-
tabase consisted of nearly 8,000 polygons, running the
simulation  models interactively would not be  a feasible
option. Thus, the CREAMS-WT (17) runoff model was
prerun for different levels of inputs and management for
each land use, soil association, and weather region (18).

Depending on the land use and its relative importance
as a contributor of phosphorus  to the lake,  anywhere
from one (background levels of inputs to land uses like
barren land) to 25 (dairies, beef pastures) levels of
inputs were selected. Each set of inputs to a particular
land use was given a separate PCP identification code.
A CREAMS-WT  simulation was performed for each
PCP, on each soil association and weather region. This
resulted in approximately 2,600 simulation runs. Annual
average results were computed for use in  LOADSS.
CREAMS-WT provides an average annual estimate of
phosphorus  runoff from each  polygon. Phosphorus
assimilation along flow paths to  Lake Okeechobee are
estimated as an exponential decay function of distance
traveled through canals and wetlands (4).

The imports, exports, and economics of each PCP are
based on a per production unit basis. Depending on the
type of polygon, the production unit can be acres (e.g.,
pastures, forests), number of cows (dairies), or millions
of gallons of effluent (waste treatment plants and sugar
mills). Developing a regional plan  in LOADSS involves
assigning a PCP identification code to each one of the
polygons in the Lake Okeechobee basin. Accessing the
results of a regional plan involves multiplying the pro-
duction unit of each polygon by its appropriate database
import, export, or economic attribute and summing the
resulting values overall polygons in the Lake Okeechobee
basin. LOADSS runs in the ARC/INFO Version 6.1.1 CIS
software on SUN SPARC stations.

Interactive Dairy Model

Although the  LOADSS level of detail  is adequate  for
regional planning, a more detailed model was necessary
to analyze individual dairies in the Lake Okeechobee
basin, as dairies were one of the large, concentrated
sources of phosphorus runoff into the  lake.  Thus, the
IDM was developed and incorporated  into  LOADSS.
IDM utilizes FHANTM  to simulate phosphorus move-
ment  in dairy fields.  FHANTM  is  a  modification  of
DRAINMOD (19) with added functions  to handle over-
land flow routing, dynamic seepage boundary, and sol-
uble phosphorus algorithms for P input, mass balance,
and transport  (20).

Unlike in LOADSS, FHANTM is run interactively,  as IDM
requires. Furthermore,  in LOADSS, the user can only
select from a number of predefined PCPs, while  in IDM,
the user has access to more than 100 input and man-
agement variables, all of which  can take a  range  of
values. This allows for the development and evaluation
of detailed  dairy  management plans  that  otherwise
would be impossible at a regional scale. While LOADSS
only provides average annual results, IDM displays daily
time series simulation results. IDM uses the same as-
similation algorithm and can produce the same phos-
phorus budget maps and reports as LOADSS.

Optimization Module

A variety of factors must  be considered in  planning
nutrient management programs.  Production  and envi-
ronmental goals need to be balanced, and these goals
are  often incompatible. Performing this exercise on a
regional scale, comprising many fields for which a variety
of land uses and management options can be assigned,
is a tremendously time-consuming, if  not impossible,

         CIS Databases
                                                 LOADSS User Interface
Input Attribute
                                       NONPOINT SOURCE
                                       PCP'S CREAMS-WT
                                       POINT SOURCE PCP'S
                                       PROCESS ANALYSIS
                                       BASIN TREATMENTS
                                       PROCESS ANALYSIS
                                         INDIVIDUAL FIELD

                                        SIMULATION MODEL
                                        Optimization Module
                                       Models and
                                     Analysis Tools
                                          IDM and
                                      Analysis Tools
Figure 3.  LOADSS design schematic (16).
task. The optimization component of LOADSS, currently
under development, will determine the best combination
of agricultural, environmental, and regulatory practices
that  protects  and maintains  the  health  of  Lake
Okeechobee and also maintains the economic viability
of the region. The optimization process will provide an-
other method for modifying the PCPs assigned to indi-
vidual fields.  Different  optimization solution methods,
such as linear programming and integer linear program-
ming, will be available for solving the optimization prob-
lem that the user defines.

Dairy Simulation Model

The dairy model was expected to be fully functional by
the end of 1994. It is designed to be an additional tool
for answering questions related to the environmental
costs and impacts  of dairy operations. A design  sche-
matic of the dairy model is given in Figure 4. It differs
from the LOADSS/IDM  model in the following aspects:

•  It is designed to be generic so that any dairy repre-
  sented by a coverage for which relevant data, such
                                    as topography, soil characteristics, weather, and field
                                    boundaries, are available can be simulated.

                                  • The GLEAMS water quality model will be used for
                                    simulating nutrient transport of nitrogen and phosphorus.

                                  • The user will be able  to assign a larger variety of
                                    crops and crop management practices to the individual
                                    fields, including crop rotation.

                                  GLEAMS (15) is a field-scale water quality model that
                                  includes hydrology, erosion/sediment  yield, pesticide,
                                  and nutrient transport submodels. GLEAMS was devel-
                                  oped  to use the management oriented CREAMS (21)
                                  model  and  incorporate  more  descriptive  pesticide
                                  subroutines and more extensive treatment of the flow of
                                  water and chemicals in the root zone layers. The water
                                  is routed through  computational soil layers to simulate
                                  the percolation through the root zone, but the volume of
                                  percolation in each layer is saved for later routing in the
                                  pesticide component. A minimum of three and a maxi-
                                  mum  of 12 layers with variable thickness may be used.
                                  Soil parameter values are provided by soil horizon, and

      GIS Databases
                                                                     DAIRY MODEL User Interface
                                          Dairy Model
                                         Analysis Tools
           GLEAMS Simulation
              Shell Controller
Figure 4.  Dairy model design schematic.
the crop root zone may have up to five horizons. The
values for parameters, such as porosity, water retention
properties, and organic content, are automatically fitted
into the proper computational layers. Two  options are
provided in the model to estimate potential evapotran-
spiration, the Priestly-Taylor method (22) and the Pen-
man-Monteith method (23). The nutrient component of
the model simulates land application of animal wastes
by creating appropriate nitrogen and phosphorus pools
for mineralization.  It  considers ammonia  volatilization
from surface-applied animal waste by using a  relation-
ship developed by Reddy et al. (24).

The graphic interface is designed to help the user plan
a balanced nutrient management program for the dairy
being simulated.  First, total nutrient production and ac-
counting are estimated, based on information related to
the dairy management such as herd size, confinement
system, waste characterization, and handling.  Figure 5
shows the general structure of the graphic interface and
a first version of the menu used to  estimate the total
amount of nitrogen and phosphorus available for assign-
ment to the various fields. Nutrient losses during waste
storage and treatment vary widely depending on the
method of collection, storage,  and treatment. Climate
can also have a great effect on the losses. Covering all
possible methods of storage and treatment is practically
impossible, especially in an application that is designed
to be generic and applied in any part of the country. A
simplification was made to overcome this problem: the
user must provide the  percentage of original nitrogen
and phosphorus that is retained after waste storage and
treatment. The  menus designed to enter information
related to the management of fields and crops are given
in Figure 6.

For each field, a sequence of crops can be defined in
the Field Management Table,  and  for each  crop, the
sequence of practices or field operations  is defined in
the Crop Management Table. Every time a waste appli-
cation operation is defined or a field is used as pasture
for a certain period, the corresponding amount of nutri-

(Select a Dairy V^Qverall Dairy ttanaeenent <7)fPlan Manager ?}( Practice Menus vX^^nu^at^on Results X^6?"^ Manager ?)( Print Map ^(Utilities ?V Ouit) J


imp lYrE:
EOOJ^[3^^=^^=> LOO
HnsTE r.HiiRnr.TERT7nTTnn a NUTRIENTS pRnniicnnH
  • s/llll/i!.itj> ; 0.lii.r.is (Llia/nil/il.i.j) : 0.07 Hitroecn Phosphorus Lhn/nll/ijcar Lba/licor Lba/flU/ucar Lbs/ucar [onfinenent. 01 10071 13 1533 Hrjijinc M 10071 13 1H33 TU1IIL 1GU 2014B 2G 30CG linilOLING OF MI1STE PRODUCED UNDER COHFINEICHT Percentage of t.ricinal nutrients retained after STORBGE/TRERTMENT HIITKTEHTS ncr.nuHrTHfi {pEDCEiirnnE) nuailaltlc Tnr Rssifrnncnt Treatuil/rtarkuted Crd^-inc Land llpplicaUon HITROCEH 50 M il rilOGPIIORU!; 50 ' 33 17 (SHVE HNO EXIT^ (CIIHCEL) ' S# MESSAGES Icm 5 ; Drawing! dairy nap |f^| -. C: lEI Figure 5. The nutrient production table of the dairy model interface. ents will decrease from the amount available for assign- ment and the total available for future assignment will be updated. Once the total amount of nutrients is as- signed, the model can be run for the several fields in the dairy and the results evaluated in terms of nutrient load- ings to the edge of fields and ground water. Alternative plans can be designed and saved for comparison and selection of best management options. The best solu- tions in terms of reducing nutrient loadings to surface and ground water must also consider economic aspects. A producer's decision about competing waste manage- ment practices is ultimately economically motivated. Thus, the system will eventually include a tool for eco- nomic analysis of alternative management options. Summary and Conclusions The search for solutions to the many problems concern- ing nutrient management that affect water resources implies a continued demand for the development of modeling systems that can be used to analyze, in a holistic approach, the impact of alternative management policies. The development of LOADSS exemplifies how the inte- gration of hydrologic models and a CIS can be used for analyzing nutrient control practices at different scales. The addition of optimization algorithms further enhances the ability of policy- and decision-makers to analyze the impact of alternative management practices and land uses at the regional level. The first part of LOADSS (Version 2.2) that includes the CREAMS-WT regional-scale model and the IDM com- ponents is fully functional and currently available at the SFWMD. Preliminary results show that LOADSS be- haves consistently with measured data at the lake basin scale. Some of this, however, is due to offsetting errors in model behavior at the subbasin scale, particularly in subbasins that are adjacent to or very far from the lake. Currently, projects are underway to further verify and calibrate the model at the subbasin level to improve its

  • -------
    Figure 6.  Field and crop management tables of the dairy model interface.
    performance at smaller scales (16). Initial results of the
    optimization component are currently being evaluated.
    The dairy model represents a different approach in  inte-
    grating water quality models  and CIS in that it is de-
    signed to  be generic and focused mainly on the  farm
    level. It is primarily designed to help policy- and decision-
    makers analyze the effects of alternative dairy waste
    management practices on the farm level. The framework
    can easily be adapted to handle different types of animal
    wastes (such as beef cattle and poultry) and to simulate
    the impact of other crop management practices such as
    pesticide applications.
     1.  Novotny, V., and G. Chesters. 1981. Handbook of nonpoint pol-
        lution: Sources and management. New York, NY: Van  Nostrand
        Reinhold Company.
     2.  Council on Environmental Quality (CEQ). 1989. Environmental
        trends. Washington, DC: U.S. Government Printing Office.
    3.  U.S. Geological Survey (USGS). 1987. Analysis and interpreta-
       tion of water quality trends in major U.S. rivers, 1974-81. Water
       Supply Paper 2307. Washington, DC: U.S. Government Printing
    4.  SFWMD. 1989. Interim Surface Water Improvement and Man-
       agement (SWIM) plan for Lake  Okeechobee. West Palm Beach,
       FL: South Florida Water Management District.
    5.  Reck, WR.  1994. GLEAMS modeling of BMPs to reduce nitrate
       leaching in Middle Suwannee River area. In: Proceedings of Sec-
       ond Conference on Environmentally Sound Agriculture, Orlando,
       FL. American Society of Agricultural Engineers, pp. 361-367.
    6.  Fedkiw,  J. 1992. Impacts of animal wastes on water quality: A
       perspective from the  USDA.  In: Proceedings of the National
       Workshop on Livestock, Poultry, and Aquaculture Waste Manage-
       ment, Kansas City, Ml (1991).  American Society of Agricultural
       Engineers (ASAE) Pub. 03-92.  pp. 52-62.
    7.  Woolhiser, D.A., and D.L. Brakensiek. 1982. Hydrologic system
       synthesis.  In: Haan, C.T., H.P. Johnson, and D.L.  Brakensiek,
       eds. Hydrologic modeling of small watersheds. ASAE Monograph
       No. 5. St. Joseph, Ml: American  Society of Agricultural Engineers.
       pp. 3-16.

     8.  Fedra, K. 1993. GIS and environmental modeling. In: Goodchild,
        M., B. Parks, and L. Steyaert, eds. Environmental modeling with
        CIS.  New York, NY: Oxford University Press, pp. 35-50.
     9.  Tim, U.S., M. Milner, and J. Majure. 1992. Geographic information
        systems/simulation model linkage: Processes, problems, and op-
        portunities. ASAE Paper No. 92-3610. St. Joseph,  Ml: American
        Society of Agricultural Engineers.
    10.  Whittemore, D.O., J.W. Merchant, J. Whistler, C.E. McElwee, and
        J.J. Woods.  1987. Ground-water protection planning using  the
        ERDAS geographic information system: Automation of DRASTIC
        and time-related capture zones. In: Proceedings of the National
        Water Well Association  (NWWA FOCUS) Conference on Mid-
        western Ground-Water Issues, Dublin, OH.  pp. 359-374.
    11.  Petersen, G.W, J.M. Hamlett, G.M. Baumer, D.A. Miller, R.L. Day,
        and J.M. Russo. 1991. Evaluation of agricultural nonpoint pollu-
        tion  potential in Pennsylvania using  a  geographic information
        system. ER9105. University Park, PA: Environmental Resources
        Research Institute.
    12.  Goodchild, M. 1993. The state of CIS for environmental problem-
        solving. In: Goodchild, M., B.  Parks, and L. Steyaert, eds. Envi-
        ronmental  modeling with  GIS.  New York, NY: Oxford University
        Press, pp. 8-15.
    13.  Liao,  H.H., and U.S. Tim.  1992. Integration of geographic infor-
        mation system (GIS) and hydrologic/water quality  modeling: An
        interface. ASAE Paper No. 92-3612.  St. Joseph, Ml: American
        Society of Agricultural Engineers.
    14.  Young, R.A.,  C.A. Onstad, D.D. Bosch, and W.P. Anderson. 1989.
        AGNPS: A nonpoint source pollution model for evaluation of ag-
        ricultural watersheds. J. Soil and Water Conserv. 44(2):168-173.
    15.  Leonard,  R.A., WG.  Knisel, and  D.A.  Still. 1987.  GLEAMS:
        Ground-water loading effects of agricultural management sys-
        tems.  Trans, of the Amer.  Soc.  of  Agricul. Eng.  (ASAE)
    16.  Negahban, B., C. Fonyo, W Boggess, K. Campbell, J. Jones, G.
        Kiker, E.  Hamouda, E.  Flaig,  and H.  Lai. 1993. A CIS-based
        decision support system for regional environmental planning. In:
        Proceedings of the Conference on Application of Advanced Infor-
        mation Technologies:  Effective  Management  of  Natural  Re-
        sources,  Spokane,  WA.  American  Society  of  Agricultural
        Engineers (ASAE). pp. 169-178.
    17.  Heatwole, C.D., K.L. Campbell, and A.B. Bottcher. 1987. Modified
        CREAMS hydrology model for coastal plain flatwoods. Trans, of
        the Amer. Soc. Agricul. Eng. (ASAE) 30(4):1,014-1,022.
    18.  Kiker, G.A., K.L. Campbell, and J. Zhang. 1992. CREAMS-WT linked
        with GIS to simulate phosphorus loading. ASAE Paper No. 92-9016.
        St. Joseph, Ml: American Society of Agricultural Engineers.
    19.  Skaggs, R.W. 1980. DRAINMOD reference report.  U.S. Depart-
        ment of Agriculture, Soil Conservation  Service.  Fort Worth,  TX:
        South National Technical Center.
    20.  Tremwel,  T.K., and  K.L. Campbell. 1992. FHANTM, a modified
        drainmod: Sensitivity and verification results. ASAE Paper No. 92-
        2045. St. Joseph, Ml: American Society of Agricultural Engineers.
    21.  Knisel, WG. 1980. CREAMS: A field-scale model for chemicals,
        runoff, and erosion from agricultural management systems. Con-
        serv. Res. Rep. No. 26. Washington, DC: U.S. Department of
    22.  Priestly, C.H.B.,  and R.J. Taylor. 1972. On  the assessment of
        surface heat flux and evaporation using  large-scale parameters.
        Monthly Weather Rev. 100:81-92.
    23.  Jensen, M.E., R.D. Burman,  and R.G. Allen, eds. 1990. Eva-
        potranspiration and irrigation water requirements. Manuals  and
        reports on engineering practice. Amer. Soc. Civil Eng. No. 70.
    24.  Reddy, K.R., R.  Khaleel, M.R. Overcash, and P.W. Westerman.
        1979. A nonpoint source model for land areas receiving animal
        waste: II.  Ammonia volatilization.  Trans,  of the Amer. Soc. Agric.
        Eng. (ASAE) 22(6):1,398-1,405.

                Using GIS To Rank Environmentally Sensitive Land
                               In Orange County, Florida
                           Michael J. Gilbrook, HDR Engineering, Inc.1
    In 1994, the Orange County (Florida) Planning Department selected HDR Engineering to conduct a
    GIS-based environmental constraints and development suitability study to support a new, proactive
    planning initiative. The goal of the project was to conduct a McHargian overlay analysis which
    would identify environmental constraints to development, and rank all non-urban lands within
    Orange County according to their environmental value. The project required development of a
    database consisting of eleven environmental constraint factors including vegetative cover,
    wetlands, wildlife habitat preferences, habitat corridors, floodplains, aquifer  recharge potential and
    septic tank use suitability. A raster-based spatial model prepared in ARC/INFO GRID was used to
    identify and rank environmental constraints to future urban development based on the distribution
    and coincidence of the various environmental constraint factors. The HDR environmental science
    staff worked closely with the County Planning Department and the County's Environmental
    Mapping Committee (comprised of scientists, regulatory personnel, environmental activists,  land
    owners, and planners) to prepare the GIS model and weighting scenarios. The final products
    included two baseline maps (one for ecological constraints, and one for physicochemical factors),
    and several maps generated by alternative weighting strategies.
    In December 1993, Orange County (Florida) hosted a Growth Management Exposition as a
    forum for unveiling the County's proposed Development Framework. The County's Planning &
    Development Division Newsletter released concurrently with the Exposition noted that:
           "The Orange County Comprehensive Policy Plan commits the County to the
           preparation of a Development  Framework Element as a vision statement to  guide
           the pattern of future development. ...The Comprehensive Policy Plan should
           contribute to, and be an embodiment of, the common vision that represents  the

           shared values and beliefs of the community for guiding future growth and
    A "environmental suitability analysis" was among the tools cited by Orange County as
    necessary to the preparation and implementation of the Development Framework:
           "...[t]he addition of the concept of environmental suitability analysis into the
           Development Framework will strengthen the local planning process, facilitate
           efforts to obtain public ownership of environmentally sensitive lands, and
           increase the effectiveness of existing regulatory processes and procedures.  It will
           also help separate and differentiate between planning for the future and
           regulating development."
    That last point is crucial to understanding the importance of the County's environmental
    constraints analysis project. A typical regulatory approach to environmental planning addresses
    potential impacts piecemeal, as development is proposed, and attempts to minimize or mitigate
    ecological impacts. In contrast, Orange County wanted to develop a pro-active planning
    approach driven by an appreciation of the "carrying capacity" of the environment through
    composite mapping of environmental constraints to development. The environmental constraints
    mapping would then provide a long term guide for shaping future development patterns. Maps
    of environmental constraints can also be construed, inversely, as maps of development
    suitability (Twiss, 1975). Of course, a complete evaluation of development suitability would
    require combining the findings of the environmental constraints analysis with information on
    transportation systems, district plans, urban services,  and proposed capital improvements.
    In summary, the purpose of this study was to advance the County's movement towards a more
    holistic, ecosystem oriented approach to environmental planning. Specifically, HDR was tasked
    to "[d]evelop an environmental suitability index and map for Orange County that will compare
    and rank areas for environmental compatibility and development suitability according to the
    ecological characteristics of the region."
    1 HDR Engineering, Inc., 201 S. Orange Avenue, Suite 925, Orlando, FL 32801-3413; Voice (407) 872-
    7801, ext. 230; FAX (407) 872-0603; Email mgilbroo@hdrinc.com

    In his book, Design With Nature, planner Ian McHarg popularized the notion of map overlay
    analysis to evaluate land development suitability (McHarg, 1971). His method involved mapping
    landscape features which imposed some kind of limitation on urban development (such as
    wetlands, floodplains, or erosive soils) onto transparencies. For each map, the relative
    development suitability (or, conversely, environmental sensitivity) of landscape features would
    be ranked in order from high to low. Each map would be drawn at the same scale and
    registered to the same geographic area. The unsuitable areas on each map would be shaded in
    such a way that when the maps were overlaid one on another and viewed simultaneously on a
    light-table, areas which were mapped as unsuitable on all the contributing maps would appear
    very dark. Land areas which were unsuitable for development on fewer maps would appear
    proportionally lighter, providing a graduated scale from completely suitable to entirely
    McHargian overlay analysis, as it came to be known,  is a powerful planning technique.
    However, it was also inefficient and often impractical to perform by manual means. In the 25
    years since the publication of McHarg's book, his technique has  been refined though the use of
    computerized Geographic Information Systems (GIS). Examples of McHargian-type analyses
    using GIS are now quite common; the author had personally conducted two (Gilbrook, 1989a;
    Gilbrook, 1989b) prior to this study. The first step in conducting a McHargian  analysis for
    Orange County was determining what environmental  values would be incorporated into the
    evaluation process. The County Planning Department identified a tentative list of environmental
    criteria for the study. Working from that list as a starting point,  HDR identified the following
    environmental  issues which formed the basis for the environmental constraints analysis in
    Orange County:
        •  Floodplain  Protection Development in floodplains places  people and property at risk.
          Cumulative flood storage losses can exacerbate flooding elsewhere in the watershed.
          Alteration of floodplains adversely affects the hydrological, biological and
          physicochemical relationship between surface waters, wetlands, and  uplands.
        •  Protection of Wetlands Wetlands provide critical habitat to a number of native central
          Florida species. Wetlands provide flood  storage, hydrologic attenuation and erosion
          control  functions.

    Maintaining Viable Wildlife Populations Wildlife populations must maintain minimum
    numbers in order to resist extinction from genetic inbreeding depression, disease, or
    climatic catastrophes. Viable populations require a minimum amount of suitable habitat
    to support them. Without protection (and maintenance) of adequate habitat, wildlife
    populations will likely become extinct over time.
    Preserving Biodiversity The natural environment is comprised of a complex assemblage
    of interdependent and co-evolved species. The effects of urbanization (e.g., forest
    fragmentation, enhancement of edge effects, fire suppression, invasive exotic species)
    serve to simplify and eliminate this diversity. Simplified ecosystems support fewer
    species and are less resilient to environmental challenges (e.g.,  storms, disease).
    Conservation Lands and Insular Ecology Conservation lands may maintain intact flora
    and faunal assemblages representative of the ecosystems present within those
    preserves as long as they enjoy contiguity with similar (or at least compatible) adjacent
    landscapes. Once isolated from similar habitat, conservation lands become essentially
    terrestrial islands subject to the species-area effects described by the science of insular
    ecology; the preserves lose species in geometric proportion to their final, isolated size.
    Groundwater Protection Drinking water supplies in central Florida are drawn from the
    Floridan aquifer, a deep, water-bearing geological stratum. The aquifer is recharged only
    by rainfall on relatively rare areas which have the necessary geologic conditions
    conducive to movement of water from the  surface  to the deep aquifer below. However,
    these areas  are also conduits for potential contamination  of the water supply for the
    same reason that they are good recharge  sites. Similarly, the potential for contamination
    at the locations of drinking water wellheads and drainage wells (i.e., stormwater disposal
    wells to the aquifer) were also concerns for groundwater protection.
    Surface Water Protection Surface waters may be contaminated by pollutants borne by
    stormwater runoff from artificial impervious surfaces, made turbid by eroding sediment,
    or affected biologically by inappropriate  management of shoreline wetlands or littoral
    vegetation. Septic tanks sited in areas of unsuitable soils  may cause contamination of
    surface waters, presenting both an environmental  hazard and a human health threat.

    To conduct the study, each of these themes were translated into specific layers of digital data
    for analysis with GIS.
    The environmental constraints analysis tapped a number of existing GIS data sources. Orange
    County and the two Water Management Districts which have jurisdiction over the County (St.
    Johns River WMD and South Florida WMD) were the source for most of the GIS data sets.
    Other important sources included the East Central Florida Regional Planning Council (Existing
    Land Use), the Florida Game and Fresh Water Fish Commission (Strategic Habitat
    Conservation Areas and Biodiversity "Hot Spots"), and the Florida Natural Areas Inventory
    (element occurrence records for rare plants and animals).
    One of the most important data themes required was a current digital map of land cover to
    supply the information about wetlands, rare vegetative communities, and vegetative biodiversity
    needed to perform the environmental constraints analysis. A detailed digital Existing Land Use
    (ELU) map of Orange County,  last updated in 1989, was available. Before these data could be
    applied to  the environmental constraints analysis they had to be updated to reflect urban
    development which had taken  place over the previous five years. To conduct the 1994 update,
    HDR scanned and georeferenced recent aerial photography for use in performing "heads up"
    digitizing of new urban land uses not reflected in the ELU 1989  data. The final Existing Land
    Use map appears in Figure 1; an inset of the map showing some of the features in the
    Econlockhatchee River basin of east Orange County appears in Figure 2.
    Many of the GIS data sets were prepared from map sources compiled at a scale of 1:24,000
    (i.e., 1" = 2,000') or smaller. This placed constraints on how finely the data could be interpreted.
    For example, the Florida Game and Fresh Water Fish Commission's (GFC's) Strategic Habitat
    Conservation Areas were developed from Landsat satellite imagery which had a minimum pixel
    resolution  of 30 meters (about 98 feet, or 0.22 acres per pixel).  In contrast, the Existing Land
    Use 1994  (ELU 94) data developed by HDR were captured at 1:24,000 scale, but had a
    minimum mapping unit of 2 acres; polygons of homogenous ground cover smaller than 2 acres
    were not necessarily mapped.  Although these relatively low resolution input data sets limited
    their applicability for examining environmental constraints on small,  individual parcels of land,
    their scale was appropriate for the regional  analysis required by Orange County for the purpose
    of guiding  comprehensive land planning.

                                   Existing Land Use, 1994  Update
    Figure 1.  Existing Land Use 1994 Map. Dark gray areas are urbanized, other color
             represent undeveloped land cover. Inset area of Econlockhatchee River
             headwaters area appears in Figure 2 and subsequent map figures.

      Figure 2. Inset of Existing Land Use 1994 in the Econlockhatchee River Headwaters Area,
              Southeastern Orange County.
    This study employed ARC/INFO 6.1.1. GIS software manufactured by Environmental Systems
    Research Institute (ESRI). The software ran on a Sun Microsystems SPARC 10 Model 40
    workstation under the Solaris 2.2 operating system (a variant of Unix System V).
    ARC/INFO is principally a vector GIS system in which linear features (lines or polygons) appear
    as smooth lines connecting the many vertices whose coordinates define the shape of those
    features. In this respect an ARC/INFO visual display looks little different from a map prepared
    using Computer Aided Design (CAD) software. Nearly all of the  preliminary GIS manipulation
    needed to produce the  input data themes used in the study was conducted within the
    ARC/INFO vector environment. However, one drawback to the vector GIS environment is that it
    is a computationally intensive process. For ARC/INFO to combine two sets of countywide
    polygon data to create a third using a vector overlay process (e.g., UNION or INTERSECT)
    would have required considerable time. Overlays involving multiple input themes would have

    had to be processed in a series of pairs, until the final desired product was achieved. Even then,
    the GIS analyst would have had to use tabular database functions to resolve the meaning of the
    multiple layers of polygon data which had been combined.
    Although the environmental constraints analysis could have been performed this way, a better
    approach was available through use of ARC/INFO's GRID module. Combining and manipulating
    GIS data in GRID was computationally much easier (and faster) than the topologically complex
    process of resolving the output from overlays involving many arbitrarily shaped vector polygons.
    Furthermore,  the GRID process allowed for direct mathematical modeling of the input data sets
    to generate the desired composite output map;  for example, differential weighting of the input
    data was as easy as multiplying all cells in an individual data layer by a numerical constant.
    Given its clear advantages, HDR used GRID for modeling  in this study. Each of the vector
    polygon GIS input data themes was converted to a grid with a cell size of 98 feet. The 98 feet
    corresponded to the 30 meter resolution of the Landsat data which formed the basis for the
    GFC's Strategic Habitat Conservation Areas, Biodiversity "Hot Spot" areas, and the SJRWMD's
    Regionally Significant Habitat areas. A cell resolution of 98 ft resulted in a minimum mapping of
    0.22 acres, which we  deemed small enough to adequately represent details in the various input
    coverages collected at 1:24,000 scale.
    Following examination of the  available GIS data sets, consideration of the study's objectives,
    discussions with the Orange County Planning Department staff, and input from the
    Environmental Mapping Advisory Committee, HDR determined that the most appropriate end
    product would not be  a single map of environmental constraints, but two composite maps: an
    "Ecological Constraints Map" and a "Physical Constraints Map." The Ecological Constraints
    Map was derived from those factors whose attributes (mostly biological) denoted resources that
    were sensitive to any land alterations. For example, a forested area that was important for the
    maintenance  of wildlife populations was likely to be adversely affected by substantial clearing.
    The type of urban development (e.g., residential, commercial or industrial) makes no difference;
    it is the loss of forest habitat,  the reduction in forest interior, or the increase in fragmentation and
    edge effects which are important. Protection of such areas would require attention to
    development  intensities.
    The Physical  Constraints Map was generated from input data themes which denoted constraints
    based on physical factors. Protection of these resources would require attention to the type of

    land use or the development standards imposed on that development. Aquifer recharge areas
    are the best example. The presence of a high recharge area does not necessarily preclude
    urban development. Instead, a recharge area may influence the type of land use considered
    suitable (e.g., no chemical industries), the enactment of special development standards (e.g.,
    higher on-site runoff retention requirements), or both. This paper will only address the
    methodology used to prepare the Environmental Constraints Map; the process for generating
    the Physical Constraints Map was identical, differing only in the types of inputs involved.
    Several GIS layers contributed to each of the two composite maps of environmental constraints.
    In both models, we used a five-point scale or index to rank environmental sensitivity for a given
    GIS data layer. In all cases,  a "1" indicated presence of "Very Low" environmental constraints
    for urban development, a "3" a "Moderate" level of environmental sensitivity, and a "5" indicated
    the presence of "Very High"  environmental constraints. In the GIS analysis, the various
    contributing themes were combined or overlaid in such a way that each spot on the map
    represented an average of all the environmental constraint scores contributing to that map.
    Areas which had many high  environmental sensitivity scores from contributing input data
    themes received  a high total score, while areas with few or no constraints received a low
    composite score.
    Initially, all factors were given equal weight in the overlay procedure to produce two "baseline"
    maps. After the "baseline" maps were produced, weighting factors were assigned to each of the
    GIS input layers to reflect the relative importance of their contribution to the ranking of
    environmentally sensitive lands in the County. The value of the weighting analysis was two-fold.
    First, application  of weighting to a particularly important input data layer would permit it to "shine
    through" the muddle which might otherwise result from the combination of so many disparate
    input data sets. Second, the use of weighting tested the robustness of the findings  of the GIS
    analysis. That is, the environmental sensitivity of areas which ranked similarly on several maps
    despite alternative weighting schemes could be interpreted with confidence. For instance, one
    would conclude that an area which showed up as having "Very High" environmental constraints
    in every map, regardless of weightings, was clearly dominated by environmentally sensitive
    factors. The weighting  factors were developed in a workshop  meeting of the Environmental
    Mapping Advisory Committee in  concert with HDR and the Orange County Planning Department
    staff. The particulars of the weighting process are described below, following  the discussion of
    the input data themes.

    We identified nine different GIS input data layers, or themes, for use in generating the
    Ecological Constraints Map (Table 1). With two exceptions, all the input data layers were
    "derived" by combining two or more of the original GIS data sources in various ways, or by re-
    casting the original data set in a more useful form. Each input data layer is described below.
                                         TABLE 1
                 Input Data Layers For Ecological Constraints Map GIS Model
                                            Environmental Constraint Factor
    Low Moderate High Very High
    GIS Input Data Layer Very Low (1) (2) (3) (4) (5)
    Floodplain Areas
    Wetland Areas
    Ecological Integrity
    Vegetative Community
    Habitat Corridors/Biological
    Wetland Dominance
    Floodplain Dominance
    Vegetative Biodiversity
    All Other Land
    All Other Land
    Very Low
    Proximity Or
    < 20 percentile,
    Urban or Water
    < 20 percentile,
    Urban or Water
    < 20 percentile,
    Urban or Water
    RSH Areas
    > 2%,
    < 1 0%
    Wet Prairie
    5 - 7+ Focal
    FNAI S3 or
    41 -60
    41 -60
    41 -60
    100 Year
    FNAI S2 or
    > 0.5%,
    61 -80
    61 -80
    61 -80
    FNAI & GFC
    FNAI S1 or
    Area < 0.5%
    Very High
    Wetland or
    Notes: (1) Class 5 reserved for Regulatory Floodways in Floodplain Dominance theme, but not used in
    this study; (2) SJRWMD = St. Johns River Water Management District; (3) GFC = Florida Game
    & Freshwater Fish Commission; (4) FNAI = Florida Natural Areas Inventory; (5) SHCA =
    Strategic Habitat Conservation Area; (6) RSH = Regionally Significant Habitat; (7) The
    Environmental Sensitivity score for Habitat Corridors/Biological Connectivity depended on both
    habitat type and proximity to the most direct path between two preserves (see Table 2).

    Floodplain Areas. This was one of the two "non-derived" data themes. Floodplains were a
    "dichotomous" factor: areas above the 100-year floodplain boundary were ranked "1," whereas
    all lands identified as being within the 100-year floodplain (i.e., those designated "Zone A" on
    FEMA Flood Insurance Maps) were assigned a environmental constraint factor of "4." There
    were no intermediate classes. Class "5" was reserved for regulatory floodways, which would
    have received the greatest protection from alteration. Regulatory floodway boundaries were not
    available for this study, but their spot was reserved so that they could be inserted into the
    analysis and the model run again at a later time.
    Wetland Areas. This is the other non-derived theme. Wetlands were ranked in order of their
    relative difficulty to maintain or re-create.  Forested wetlands as a rule take a considerable time
    to grow to maturity, and are not easily re-created by humans; consequently, they were ranked
    "5." Non-forested wetlands (i.e., marshes) are extremely productive wetland environments, but
    are somewhat more readily replaced than forested wetlands; hence, they were assigned a value
    of "4." Since vegetative communities similar to natural wet prairie areas are commonly created
    on pastures, such areas were assigned a "3." All other (upland) areas were designated as "1."
    Scientists, planners and developers often disagree on whether wetland size is a valid  criterion
    for evaluating wetland value. While it is true that small, isolated wetlands in an urban setting
    may have little or no wildlife habitat value, this is emphatically not true of such wetlands
    immersed in a matrix of natural upland vegetation. Small, ephemeral wetlands are essential to
    the life cycles of many amphibians which  cannot survive predation  by fish in larger ponds.
    Furthermore, small wetlands are essential to the feeding (and nesting) success of wading  birds,
    particularly the endangered wood stork. And although large wetlands might seem to dominate
    surface water hydrology, the depressional storage capacity of many small wetlands may be
    considerable. Since there was no good way to rank the ecological importance of wetlands
    based on size, size did not contribute to the ranking of the Wetland Areas theme for this study.
    Ecological Integrity. This is a term which has gained in popularity to describe the process of
    protecting natural diversity, at scales ranging from populations to entire ecosystems (Minasian,
    1994). Consistent with that concept, this input data layer was comprised of several data sources
    which had been combined to represent areas of hierarchically greater or lesser importance to
    the maintenance of natural floral and faunal populations in the County. Most sensitive on this
    scale (i.e., a ranking of "5") were the reported locations  of species listed as endangered,

    threatened or species of special concern by the U.S. Fish & Wildlife Service, Florida Game and
    Fresh Water Fish Commission, or by the Florida Natural Areas Inventory. These point
    occurrences were obtained from the FNAI Element Occurrence database, and were
    represented by a 1,000 ft. radius circular area generated as a buffer around each point location.
    These circles, which are comparable to those used by Cox, et al. (1994) to highlight point
    occurrences of listed species or other significant wildlife, encompassed an area of about 72
    acres each.
    The map data used to identify lands ranked "4" for this data theme were obtained from digital
    maps of Strategic Habitat Conservation Areas (SHCAs) prepared by the Florida Game and
    Fresh Water Fish Commission (Cox, et al., 1994). Using a statewide map of vegetative cover
    derived by interpretation of Landsat satellite imagery collected during the mid- to late 1980's, the
    GFC identified polygons of vegetative cover which represented the habitats of 30 "focal
    species," most of which were listed as endangered or threatened. (Seventeen of the GFC's
    focal species occurred in Orange County, including the red-cockaded woodpecker, Florida
    scrub jay and gopher tortoise.) By mathematically modeling the minimum viable population
    sizes  needed to ensure survival of the selected focal species for 200 years,  the GFC then
    estimated the minimum necessary habitat required to maintain such populations in perpetuity.
    Following a GIS analysis of the Landsat-derived vegetative cover maps, the GFC located
    SHCAs which, if preserved, would secure the long term survival of the 30 focal species
    Using the same Landsat-derived vegetative cover maps, the GFC prepared a map of
    biodiversity "hot spots." To identify areas which  might jointly serve a number of important wildlife
    species, the GFC overlaid the individual habitat maps of their focal species and identified areas
    whose habitat could support seven or more species, five to six species, or less than five
    species. Those areas identified as supporting either 5-6 or 7+ species were combined and
    ranked "3" in the Ecological Integrity data theme.
    The St. Johns River Water Management District used the SHCA data as the basis for its
    identification of Regionally Significant Habitat (RSH). Since the method the GFC employed to
    identify SHCAs was tied to the habitat requirements of individual species, their process
    sometimes identified polygons of vegetative cover which were surrounded by other native
    vegetation not included in the SHCA. While preparing their RSH maps, the SJRWMD

    recognized the need to identify areas which could be either effectively regulated, or publicly
    acquired and managed; disjunct SHCAs embedded in other natural land cover wouldn't qualify.
    Consequently, the SJRWMD used the Cox (1994) GIS data in combination with vegetative
    cover from Orange County's 1989 Existing Land Use to "extend" the SHCA boundaries to the
    limits of immediately adjacent natural vegetative communities. The SJRWMD  RSH areas were
    assigned a ranked of "2," and all other lands not included in one of the above classes were
    ranked as  "1." A small part of the Ecological Integrity theme appears in Figure 3.
    FNAI were assigned to ranks 5, 4, and 3, respectively. All other community types were ranked
    as "1." To identify rare community types in Orange County, HDR used the 1994 Existing Land
    Use  (ELU '94) data to calculate the total acreage for each non-urban, non-agricultural cover
    type. Based  on these data, we assigned a rank of 5 to those communities comprising 0.5% or
    less  of the total natural vegetative cover of approximately 291,000 acres. Rank 4 was assigned
    to communities representing between 0.5% and 1.0% of the total  natural land  cover, while those
    communities with 1% to 2% were assigned a rank of "3," and a rank of "2" attributed to
    communities falling between 2% and 10%. Natural land cover types with greater acreage were
    all assigned  ranks of "1." Where the FNAI and Orange County ELU '94 derived ranks disagree,
    the vegetative community was assigned the higher (i.e., more sensitive) rank.  Figure 4
    illustrates a portion of the Vegetative Community Rareness theme.
    Habitat Corridors/Biological Connectivity. This GIS data layer identified areas which may be
    important to  the maintenance of biological connectivity between managed conservation areas.
    Put another way, this layer identified areas which should be protected to prevent the adverse
    effects of forest fragmentation and biological isolation in natural preserves. The idea of
    considering conservation lands as the anchor points for habitat corridors has a strong following
    in Florida (Noss, 1991; Harris and Atkins,  1991). We constructed this data theme from a
    SJRWMD  coverage of existing and proposed conservation lands (hereafter called "preserves"),
    and habitat value as derived from the 1994 Existing Land Use data. Using the ARC/INFO GRID
    functions COSTDISTANCE and CORRIDOR, we identified areas representing the most direct
    connections  between pairwise sets of proposed or existing preserves. We eliminated from
    consideration any preserve pairs for which there were either interposing urban or preserve
    areas, or which were more than 12 miles apart, and rated areas within the corridor in
    descending order according  to their proximity to the shortest distance path: 1,000 feet, 0.5
    miles, 0.75 miles and 1 mile. The corridors were overlaid with the habitat rank grid derived from

                                                                            J_       .
    Figure 3. Ecological Integrity Theme. Dark colors represent areas of highest environmental
             constraints, light areas lowest. Dark circles are the 1000-foot radius circles around
             the observed locations of endangered and threatened species.

    Figure 4. Vegetative Community Rareness Theme. Common natural communities appear in
    light colors. Darker shades of brown indicate communities of increasing rarity.
                                       TABLE 2
            Environmental Constraint Values Assigned To The Habitat Corridor
    Input Data Theme Based On Habitat Value And Corridor Centerline Proximity Scores
    2 (< 1.0 mi.)
    3 (< 0.75 mi.)
    4 (< 0.5 mi.)
    5 (< 1000 ft.)
    Habitat Value Rank

    the ELU '94 land cover data to differentiate among corridor alternatives based on preferred
    vegetative cover. The proximity/land cover matrix used to assign final environmental constraint
    rank values to grid cells for this theme appears in Table 2, and a part of the Habitat Corridors
    map appears in Figure 5.
    Wetland Dominance. The Wetland Areas GIS data layer addressed those parts of the
    landscape actually occupied by a wetland community type. However, in planning for the
    protection of natural resources in Orange County, it will  not be possible to delineate every
    individual wetland for protection. Nonetheless, large upland areas characterized by many small
    wetlands are themselves ecologically valuable. Faunal and floral biodiversity of many small
    wetlands interspersed throughout an upland matrix will likely be greater than that of a single
    large wetland of equivalent acreage. Furthermore, the life cycles of many species are
    dependent  upon smaller wetlands. Amphibians prefer small, ephemeral wetlands for breeding
    because such wetlands do not support a large population of predatory fish. Wading birds
    (notably the wood stork) benefit from the concentration of fish and amphibian prey within small,
    ephemeral  wetlands. To address these issues, we used ARC/INFO to calculate the ratio of
    wetlands to non-urban uplands within each of the 1,000 one-square mile sections of the Public
    Land Survey (PLS).  We assigned each PLS section a rank of 1 to 5 based on  its placement in a
    quintile distribution.
    Floodplain  Dominance. As with wetlands, isolated floodplains may collectively have great
    hydrological or ecological value if they are sufficiently numerous or collectively large enough. To
    identify those areas  which have a large portion of their land area within the 100-year floodplain,
    we generated this GIS data theme using exactly the same procedure as that for Wetlands
    Dominance, only using the Floodprone Areas GIS layer as the starting point.
    Vegetative  Biodiversity. All other factors being equal, an area with many different vegetative
    cover types (i.e., a high biodiversity) will typically be more ecologically valuable than an equal
    sized area with fewer different vegetative communities. To evaluate biodiversity, we could have
    the GIS simply count the number of different kinds of vegetative cover types in a PLS section,
    then use the quintile procedure outlined above. However, the mere number of different kinds of
    community  types does not tell the whole story. For example, two sections might have the same
    number of different vegetative community types, but one might have six equally sized polygons,
    while the other section has one very large polygon and five very small ones. Clearly, the section

      Figure 5. Habitat Corridors/Biological Connectivity Map. The darker colors represent areas of
               high habitat value near to the most direct path linking two existing or proposed
               preserves. Dark green solid lines border existing preserves, light green dashed lines
               surround proposed preserves.
    with more equally sized polygons is more diverse, since the other section is almost entirely
    dominated by one cover type. In order to quantify landscape diversity, we applied the Simpson
    C' Index, which is generally accepted as a measure of population diversity (Krebs, 1989). Using
    the acreage information for vegetative communities within the ELU '94 GIS data, we calculated
    Simpson C' indices for all PLS sections, then subjected those values to a quintile analysis like
    that used for Wetlands and Floodplains  Dominance, above. PLS sections were assigned the
    appropriate 1 through 5 ranks, with "5" representing the most diversity (highest Simpson's C'
    indices),  and "1" the least diverse assemblages of community types. Barren or agricultural areas
    that did not contribute to the biodiversity measurement were assigned a value of "1," thereby
    ensuring that an entire PLS section did not receive a high biodiversity score based on a small
    natural remnant on an otherwise barren landscape. A sample of the Vegetative Biodiversity Map
    appears in Figure 6.

      Figure 6. Vegetative Biodiversity Theme. Dashed lines represent section lines from the Public
               Land Survey. Each one-square mile section is color coded, from light to dark, based
               on its quintile rank among the distribution of all Simpson's C' indices calculated for
               the 1000 square miles in Orange County.
    Areal Analysis Of Environmentally Sensitive Input Data. Figure 7 shows the relative amounts of
    undeveloped Orange County which were assigned to each of the five classes of environmental
    sensitivity for each of the input data layers contributing to the Ecological Constraints Maps. For
    most of the input data themes, the majority of County land was rated "Very Low," with a much
    lower percentage assigned to each of the more environmentally sensitive classes. The Wetland
    Dominance, Floodplain Dominance and Vegetative Biodiversity themes illustrated a nearly
    equitable distribution of land in each of the five constraints classes, owing to the quintile
    assignment methodology employed for those layers.

                                               Figure 7
                           Comparison of Acreage Assigned to Environmental Sensitivity Classes
                               for Each of the Ecological Constraints Input Data Themes
                                                                      Input Data Themes
                  Constraint Value
    Table 3 depicts the three alternative weighting schemes used for the Ecological Constraints
    composite maps. The first was dubbed the "Ecological Integrity" weighting option, since it
    weighted those factors most closely associated with that concept (i.e., the Ecological Integrity
    layer and the Habitat Corridor layer). The "Habitat Diversity" model weighted the Vegetative
    Biodiversity and Vegetative Community Rareness layers. Finally, the "Wetlands" model
    weighted the wetland  boundary and wetland dominance layers.
    In all cases, a maximum weight of "2" was used. This was the minimum whole-integer weight that
    can be applied, yet have a demonstrably visible effect on the outcome of the composite maps. To
    evaluate the appropriateness of the "2" factor, we conducted a sensitivity analysis on the
    Ecological Constraints map by running the Ecological Integrity model with weights of 2, 3, 4 and 5.
    We found that increases in the weight beyond 2 merely had the effect of making the output
    composite map look more like the weighted layers, and diminished the value of all other inputs.
    In contrast, the weight of 2 left most composite map effects intact, while emphasizing certain
    features from the weighted layers.

                                          TABLE 3
                   Ecological Constraints Map Alternative Weighting Factors
                                              Alternative Weighting Schemes
                  GIS Input Data Layer
    Floodplain Areas
    Wetland Areas
    Ecological Integrity
    Vegetative Community
    Habitat Corridors/Biological
    Wetland Dominance
    Floodplain Dominance
    Vegetative Biodiversity
    The areas ranked as most environmentally sensitive by the Baseline Ecological Constraints
    model (Figure 8) encompassed many locations already recognized by conservation planners as
    having high ecological value, including: Wekiwa Springs State Park, Kelly Park, Moss Park, Split
    Oak Mitigation Park, the various public lands associated with the St. Johns River, and the
    various areas proposed for acquisition within the headwaters of Reedy Creek, Shingle Creek
    and the Econlockhatchee River. The fact that the Baseline Ecological Constraints model
    identified existing and proposed conservation areas as among the most environmentally
    important lands  in Orange County validated the model's credibility.
    The three composite maps produced using weighting schemes were mostly very similar to the
    Baseline  map, while exhibiting some minor differences attributable to the weighting scenario.
    The similarity of the weighted composite maps further reassured us that the basic premise of
    the environmental constraints map was sound, since the general pattern of environmentally
    sensitive  land was relatively insensitive to changes in the input weights. However, the variations
    in the weighted maps did afford an opportunity for County planners to further evaluate those
    areas which appeared different and determine  if they required more (or, perhaps, less)

    consideration in the development of land use plans or environmental protection ordinances.
    Some brief comments about the three weighted maps follow:
    Ecological Integrity Weighted Map. The pattern on this map looked very much like that on the
    Baseline Model map. Areas along the potential habitat corridor between the Econlockhatchee
    River and the St. Johns River conservation lands ranked slightly higher than those same areas
    in the Baseline model, illustrating the effect contributed by the Habitat Corridors/Biological
    Connectivity theme to this map.
    Habitat Diversity Weighted Map. As with the Ecological Integrity Weighted Map, the pattern of
    this map appeared very similar to the Baseline. It showed much less area designated as having
    "Very High" environmental constraints, but the loss occurred mostly within the limits of existing
    and proposed public lands. This effect reflects the fact that many public lands were dominated
    by large wetlands, whereas the Habitat Diversity Weighted model was more sensitive to rare
    upland communities, or diverse mixes of uplands and wetlands.
    Wetland Importance Weighted Map. This map had much greater areas of "Very Low" and "Low"
    environmental constraints, especially in eastern Orange County. In other words, by emphasizing
    the importance of wetlands, the ecological value of upland areas was diminished. Most of the
    areas identified as "High" or "Very High" in the Baseline Map appeared similarly ranked here,
    but a greater amount of these areas followed wetland boundary lines. Furthermore, this map
    showed an increased tendency for "Moderate" ranked areas to conform to PLS section lines, an
    effect no doubt produced by the section-based wetland dominance layer. The  headwaters areas
    of Reedy Creek, Shingle Creek and Lake Sheen still ranked very high.
    Relative Acreages For Ecological Constraints Composite Maps. Figure 9 compares the percent
    Orange County acreage which fell to each of the class ranks for each model type. The overall
    pattern for all maps appears almost identical: Maximum acreage was associated with the "Low"
    category, with gradually diminishing acreage values for the higher ranks. The most visible
    difference in the amount of acreage in each  constraint class appeared in the "Very High" level,
    in which the Wetland Importance model ranked nearly 8% of the County's non-urban, non-water
    area, compared to just under 5% for the Baseline and Ecological Integrity models, and 2.4% for
    the Habitat Diversity model.

                                               Ecological  Constraints
                                                      Baseline Map
    Figure 8. Baseline Ecological Constraints Composite Map. Gray areas represent existing
             urbanization. Environmental constraints in undeveloped areas are represented by
             shades of brown, from lightest ("Very Low Constraints") to darkest ("Very High

                                           Figure 9
                    Comparison of Percent of Undeveloped Acreage Assigned to
                           Each Environmental Sensitivity Rank for the
                                  Ecological Constraints Models
                       I  20.0
                   Constraint Level
                                                             Wetland Importance
                                                         Habitat Diversity
                                                     Ecological Integrity
                                                Baseline Ecological
                                                               Composite Models
    The ARC/INFO GRID based environmental constraints model proved to be an efficient and
    effective way to prepare a composite map of environmental suitability as envisioned in Orange
    County's Development Framework. The modular nature of the GIS model provides for relatively
    easy updating  of the input data layers and production of updated constraints maps as new data
    become available.
    The maps produced by the alternative weighting models for the Ecological Constraints Map
    produced results which were very similar to that of the Baseline Map. The constancy of certain
    core areas on all weighting models bolsters the conclusion that these areas are truly
    environmentally sensitive. The areas which were affected by changes in weighting parameters
    provide an opportunity to evaluate which areas, outside of the obvious core areas, should be
    included in future public lands acquisition plans, special future land use planning, or other
    appropriate protection mechanisms.

    The author wishes to recognize the members of the Environmental Mapping Advisory
    Committee for donating their valuable time and input to the successful completion of this study:
    Jack Amon, Lester Austin, III, Jim Bradner, Vera Carter (Chair), Nancy Christman,  Michael
    Dennis, Barbara Durkin, Jeff Jones, Nancy Prine, John Richardson, Jim Sellen, P.K, Sharma,
    Bill Stimmell, Jack Stout, Renee Thomas, Jim Thomas (Vice Chair), Rick Walker, and John
    Winfree. The insightful comments and suggestions of this group frequently found their way into
    the final product.
    The staff of the St. Johns River Water Management District provided invaluable assistance in
    supplying digital data, particularly Jim Cameron,  Stuart Dary,  Mike Kleinman and Linda McGrail.
    John Stys of the Florida Game and Fresh Water Fish Commission quickly supplied us with the
    digital raster files of the Commission's gap analysis study. Thanks also to Scott Taylor of the
    Florida Natural Areas Inventory for his assistance in preparing copies of FNAI's element
    occurrence data.
    Finally, I'd like to  recognize the assistance and cooperation afforded by the Orange County
    Planning Department staff, especially that of Andre Anderson and Danielle Justice, who many
    times facilitated the progress of the project. This project was funded by Orange County Contract

    Cox, J., R. Kautz, M. MacLaughlin, and T. Gilbert. 1994. Closing the Gaps In Florida's Wildlife
           Habitat Conservation System. Office of Environmental Services, Florida Game and
           Fresh Water Fish Commission. Tallahassee, Florida.
    Gilbrook, M.J. 1989a. Marina Siting Suitability in the Coastal Estuaries of East Central Florida.
           Florida Department of Environmental Regulation, Coastal Zone Management Program,
           Contract CZM-200. East Central Florida Regional Planning Council. Winter Park,
    Gilbrook, M.J. 1989b. Wekiva River Basin Acquisition Study. Final report to the St. Johns River
           Water Management District. East Central Florida Regional Planning Council. Winter
           Park, Florida.
    Harris, L.D. and K. Atkins. 1991. Faunal movement corridors in Florida. In: Hudson, W.E. (ed.).
           Landscape Linkages and Biodiversity. Island Press. Washington, D.C.
    Krebs, C.J. 1989. Ecological Methodology. Harper & Row, Publishers, Inc. New York.
    McHarg, I. 1971. Design With Nature. Doubleday/Natural History Press. Doubleday & Company,
           Inc. Garden City, New York.
    Minasian, L. 1994. Interpreting and applying ecosystem management  principles. Environmental
           Exchange Point. Florida Department of Environmental Protection, 4(3): 17-23.
    Noss, R.F. 1991. Landscape connectivity: Different functions at different scales. In: Hudson,
           W.E. (ed.). Landscape Linkages and Biodiversity. Island Press. Washington, D.C.
    Twiss, R.H. 1975. Commentary- Nine approaches to environmental planning. In:  Burchell, R.W.
           and D. Listokin (eds.). Future Land Use: Energy, Environmental and Legal Constraints.
           Center For Urban Policy Research, Rutgers University. New Brunswick, New Jersey

                           GIS Watershed Delineation Tools
                     James A. Goodrich1, Lucille Garner1, Jill Neal1, Lee Bice2,
                              Rick Van Remortel2, Ramon Olivero3
                             1USEPA, NRMRL, WSWRD, Cincinnati,
                                2Lockheed Martin, Las Vegas, NV
                                  \ockheed Martin, RTP, NC
    The 1996 amendments to Section 1453 of the Safe Drinking Water Act require the states to
    establish and implement a Source Water Assessment Program (SWAP). Source water is the
    water taken from rivers, reservoirs, or wells for use as public drinking water. Source water
    assessment is intended to provide a strong basis for developing,  implementing, and improving a
    state=s source water protection plan. This program requires individual states to delineate
    protection areas for drinking water intakes, identify and inventory significant contaminants in the
    protection areas, and determine the susceptibility of public water  supply systems to the
    contaminants released within the protection areas. SWAP can be used to focus environmental
    public health programs developed by federal, state, and local governments, as well as efforts of
    public water utilities and citizens, into a hydrologically defined geographic area.
    The Environmental Protection Agency is assisting the states in conducting source water
    assessment by identifying potential sources of data and pointing to methods for assessing
    source waters. This presentation provides guidance to states, municipalities, and public water
    utilities for assessing source waters using geographic information system (GIS) technology. The
    GIS platforms used to organize, analyze,  and manipulate available data and generate new data
    for source water protection areas, as well as provide capabilities for presenting the data to the
    public in various forms, including maps and tables are also discussed. In addition, the National
    Risk Management Research Laboratory is developing a coordinated research approach for
    watersheds that include contaminated sediments, urban watersheds, ecological restoration,  and
    source water protection. Included in the full report, as appendices, are three case studies
    demonstrating the use of selected GIS-based software and hydrologic models to conduct
    hypothetical source water evaluations. Contamination of water supplies may be responsible for

    more human sickness than any other anthropogenic activity (Anderman and Martin, 1986). Since
    limited water resources are increasingly shared by competing consumers, there is a growing
    concern about the quality of source waters. This concern has led to the establishment of laws
    and programs designed to help protect drinking water sources. Frequent evaluation and
    identification of sources of contamination are required by federal and state rules. A successful
    SWAP reduces the cost of water treatments and disinfections required. Following enactment of
    the SDWA, a number of programs were developed for public water supply protection and
    supervision, including watershed protection and control, sanitary surveys, and WHPPs. An EPA
    document titled AStates Source Water Assessment and Protection Programs Final Guidance®
    (1997a) discusses how a SWAP can use information provided by the current water programs.
    Some of the programs include:
           - a watershed control program (WCP) under the Surface Water Treatment Rule (EPA,
           - sanitary surveys (EPA, 1995), and
           - wellhead protection programs (EPA, 1995).
    Using a GIS for any application involves following some basic steps including:
           - designing the GIS database,
           - building the GIS database,
           - using the GIS to analyze the data and show results.
    For assessing source waters, elements of the design of a GIS database include:
           - establishing the study area,
           - delineating the watershed,
           - determining data needs,
           - inventorying data sources,
           - determining coordinate system  and scale, and
           - deciding on the GIS infrastructure.
    Building the database requires collecting data to characterize the study area and inventorying
    sources of contamination. Analyzing the data entails assessing potential sources of
    contamination, delineating source water protection areas, and producing  display products.

                              CHARACTERIZE THE STUDY AREA
    After deciding on the data requirements of the GIS database, the data should be obtained and
    converted to the chosen projection and units (feet, meters). The data types include descriptions
    of physical watersheds and contamination sources and types. To understand how contamination
    from a source reaches a drinking water intake, the factors that affect its flow should be
    described. These factors include, but are not limited to terrain, soils, hydrography, land use and
    land cover, and contaminant characteristics. For example, after a precipitation event, the type(s)
    of contamination  resulting from surface runoff into a stream depends on the land use and land
    cover interactions (e.g., pesticide and fertilizer from agriculture, salts and grease from parking
    lots). The directional flow of surface runoff depends on the topography, and soil infiltration
    properties affect how much surface water reaches the groundwater. The following sections
    provide information about some of the data sets needed for assessing source waters.
    Watershed Boundaries
    Watershed or HUC boundaries are available from the USGS. The HUC boundaries are available
    at 1:2,000,000 scale and 1:250,000 scale. The USGS also provides information describing the
    hydrologic unit coding scheme. A watershed boundary data set can be created by delineating
    the  boundary on large-scale maps that have elevation contour lines; the boundary can then be
    Terrain data can  be derived from Digital Elevation Models (OEMs). OEMs are digital records of
    terrain elevations for ground positions that are horizontally spaced at regular intervals. The
    SPOT Image Corporation provides OEMs at 10-meter spacing created by digital autocorrelation
    of SPOT satellite image stereopairs which are stored in a format known as  Terrain Access Made
    Easy (TAME) (ESRI, 1992). The USGS also provides 30-meter spaced OEMs at four scales:
    7.5-minute,  15-minute, 2-arc-second, and 1-degree. The 7.5-minute (large-scale) data are
    produced in 7.5- by 7.5-minute blocks from digitized cartographic map contour overlays or from
    scanned National Aerial Photography Program (NAPP) photographs. The DEM data are stored
    as profiles in which the elevations are spaced 30 meters apart. The number of elevations
    between each profile will differ because of the variable angle between the quadrangle's true
    north and the grid north of the Universal Transverse Mercator (UTM) projection coordinate
    system.  The DEM data for 7.5-minute units correspond to the USGS 7.5-minute topographic

    quadrangle map series for all of the United States and its territories, except Alaska. The
    15-minute DEM (large-scale) data correspond to the USGS 15-minute topographic quadrangle
    map series of Alaska. The unit size changes with the latitude. The 15-minute DEM data are
    referenced horizontally to NAD27. The elevations along profiles are spaced 2 arc-seconds of
    latitude by 3 arc-seconds of longitude. The first and last data points along a profile are at the
    integer degrees of latitude.
    The U.S. Department of Agriculture (USDA) Natural Resource Conservation Service (NRCS),
    formerly the Soil Conservation Service (SCS), has three soil geographic databases of varying
    scales. The data include physical and chemical soil properties for approximately 18,000 soil
    types. Each database has three categories: soil properties (particle size, bulk density, available
    water capacity, organic matter, salinity, and soil recreation), locational properties (flooding, water
    table depth, bedrock depth, and soil subsidence), and use and management properties (sanitary
    facilities, building site development, recreational development, rangeland potential, construction
    material, crops, woodland suitability, and wildlife habitat suitability). The most detailed level of
    information is provided by the Soil Survey Geographic data (SSURGO), which is available in 7.5-
    minute topographic quadrangle units  (1:24,000) and is distributed as coverages for soil survey
    areas, usually containing over ten quadrangle units. State Soil Geographic data (STATSGO) is a
    coarser database designed for regional, multi state, river basin, state, and multi county resource
    planning, monitoring, and management. The STATSGO database is at 1:250,000 scale (1- by 2-
    degree quadrangle) and  is distributed as statewide coverages. National Soil Geographic data
    (NATSGO) is a database which is suitable for national or regional resources assessment and
    planning. With a scale of 1:5,000,000, the NATSGO database has information about the major
    land resource areas.
    Hydrography is available from several federal sources at a 1:24,000 scale and may be available
    in greater detail from state and local government agencies. The USGS digital line graphs (DLGs)
    are readily available and provide information on 5 main types of data categories: boundaries,
    public land survey, transportation (including pipelines and power lines), hydrography (streams
    and water bodies) and hypsography (elevation contours). The DLG data can be converted into
    other formats compatible with GIS software. The EPA Reach File system has a series of
    hydrologic databases that uniquely identify and interconnect stream segments (reaches) for the

    nation. RF3-Alpha is the latest and most detailed version of the reach file system, containing
    more reaches than the previous versions, RF1 and RF2. Stream segments have unique reach
    codes for determining the upstream and downstream reaches and identifying the stream name
    for each reach. River Reach data can be obtained from the STORE! User Assistance Group in
    the EPA Office of Water.
    Land Use and Land Cover
    Land use and land cover data are available from several federal sources. In many cases, the
    federal data will be either out-of-date or not detailed enough. More detailed (large-scale)  land
    use data may also be obtained from county assessor maps, which are available at various
    scales  (e.g., 1:200, 1:2,400, 1:4,800). County assessor maps may provide better detail for
    inventorying contamination sources in urban areas. The various departments of highways and
    transportation can provide maps for city streets and other local and regional road maps.
    Inventory Potential Sources of Contamination
    Potential sources of contamination, also known as sanitary defects, are conditions that may
    result in contamination of a water supply. These may be point and nonpoint source pollutants,
    connections to unsafe water supplies, raw water bypasses in treatment plants, improperly
    designed or installed plumbing fixtures, or water and sewer pipes leaking into the same ditch. All
    known  and potential sources of contamination should be included in the GIS database.
    Pollutants may be classified into  categories depending on the likelihood of their introduction into
    the water supply and the level and  significance of contamination that can result from them.  A
    contaminant inventory can include  records of operation, discharge, disposal, construction, and
    other permitted activities, as well as zoning and health records obtained from local government
    agencies. All relevant information should be gathered while focusing the search for
    contamination sources at sites of particular concern. These include, but are not limited to (EPA,
           - discharge sites: septic tanks, irrigation pipes
           - storage, treatment, or disposal sites: landfills, underground tanks, mine tailings
           - substance transporting sites: pipelines
           - activities that result in discharges: highway construction, fertilizer application
           - natural sources impacted  by anthropogenic activities

    Further information on contaminant inventory activities is provided in the EPA Guide for
    Conducting Contamination Source Inventories for Public Drinking Water Supply Protection
    Programs (EPA, 1991 a). Some of these data, such as Toxic Chemical Release Inventory (TRI)
    data can  be obtained from the EPA. Other data may need to be obtained through field surveys.
    Table 1 lists federal data sets that can be accessed for much of the above mentioned data.
                            Table 1. Federal Spatial Data Set Sources
                         U. S. Environmental Protection Agency (EPA)
    Web Page: http://www.epa.gov/enviro/html/nsdi/spatial_extent.html
    Description: The EPA Envirofacts Warehouse - Geospatial Data Clearinghouse
    Web Page: http://www.epa.gov/OWOW/watershed/landcover/lulcmap.html
    Description: This EPA Office of Water site contains land cover digital data
    Web Page: http://earth1.epa.gov/oppe/spatial.html
    Description: This EPA Office of Policy, Planning and Evaluation site contains access to GIS
    spatial data sites at the federal, state, and local levels.
    Web Page: http://www.epa.gov/OWOW/NPS/rf/rfindex.html
    Description: This EPA Office of Water site contains information on the EPA river reach files.
                                    U.S. Geological Survey
    Web Page: http://nhd.fgdc.gov/
    Description: U.S. Geological Survey site containing information on the Digital Line Graphs
    (DLG) hydrography files and the EPA Reach File Version 3.0 (RF3).
    Web Page: http://edcwww.cr.usgs.gov/doc/edchome/ndcdb/ndcdb.html
    Description: U.S. Geological Survey site containing FTP file access to Digital Elevation Models
    (DEM), Digital Line Graphs (DLGs), and Land Use and Land Cover (LULC).

    Web Page: http://mcmcweb.er.usgs.gov/
    Description: U.S. Geological Survey site for the Mid-Continent Mapping Center in Rolla,
    Missouri, containing information on Digital Raster Graphics (DRG) as well as other products.
    Web Page: http://water.usgs.gov/public/GIS/background.html
    Description: U.S. Geological Survey Water Resources site containing metadata and FTP file
    access to numerous national coverages commonly used in water resources studies.
    Web Page: http://edcwww.cr.usgs.gov/webglis/
    Description: The U.S. Geological Survey Global Land Information System (GLIS) site provides
    descriptions and prices for geospatial data available from the USGS.
                                 U.S. Fish and Wildlife Service
    Web Page: http://www.nwi.fws.gov/nwi.htmDescription:
    Description: The U.S. Fish and Wildlife Service National Wetland Inventory (NWI) site provides
    access to NWI data.
                                U.S. Department of Agriculture
    Web Page: http://www.ftw.nrcs.usda.gov/nsdi_node.html
    Description: U.S. Department of Agriculture (USDA) National Resources Conservation Service
    site containing FTP access to soils and other USDA data.
                                  U.S. Bureau of the Census
    Web Page: http://www.census.gov/ftp/pub/mp/www/rom/msrom12i.html
    Description: The U.S. Census Bureau site provides brief descriptions of the TIGER/Line files,
    1997 version. The data is available for the entire U.S. on 6 CD-ROMs for $1,500 or S250/CD-
    ROM for different sections of the country. Data is in TIGER/Line format.

    Contamination Source Risk Analysis
    After the GIS database has been built, the data can be analyzed to assess the risk associated
    with potential sources of contamination, delineate protection areas, and develop display
    Classify the contaminant data into risk groups depending on the threat of contamination they
    pose to the source water. A method for prioritizing and weighing the level of risk from various
    forms of contamination is described in an EPA document (EPA, 1991b). Similar approaches may
    be adopted for surface water sources. The tasks in this phase may reveal the need for a new
    inquiry or a more thorough data gathering effort with respect to particular sites or contaminants.
    For more information, see Managing Groundwater Contamination Sources in Wellhead
    Protection Areas: A Priority Setting Approach (EPA, 1991b). Susceptibility analysis identifies the
    location, frequency, and significance of potential contaminants in the source water protection
    area and determines the likelihood the PWS will be contaminated by these sources. Water
    quality models may be used for estimating contamination levels and determining the significance
    of selected contaminants in the protection area or in the watershed.
    Proximity Analysis and Delineation of Protection Areas
    After potential sources of contamination are identified, their proximity to the water supply intakes
    can be mapped. A set of maps at various scales can be produced  from the GIS database
    illustrating the proximity of potential pollutants to the water supply system. With data
    documenting geographic locations of actual and potential contaminants, a source water
    protection area can be delineated.  Surface water sources used for drinking water supplies may
    be protected by delineating a protection area around or upstream from the source intake. Three
    approaches for delineating a protection area for surface water systems are topographic area,
    buffer distance, and stream-flow time of travel (TOT) (EPA,1997b). For systems using
    groundwater sources, approaches  for delineating a WHP area are based on fixed-radius,
    hydrogeologic/geomorphic characteristics, and modeling, which includes analytical, semi-
    analytical, numerical flow and solute transport models (EPA, 1993). The appropriate method for
    a particular system is chosen as a balance between ease of use, level of detail needed, and
    available resources. The PWS systems using a combination of groundwater and surface water
    sources may consider conjunctive delineation of source water protection areas. Conjunctive
    delineation is the integrated delineation of the zone of groundwater contribution and the area of
    surface water contribution to a PWS.  Further information on this subject can be found in

    Delineation of Source Water Protection Areas, a Discussion for Managers; Part 1: A Conjunctive
    Approach for Groundwater and Surface Water (EPA, 1997c).
    Topographic Area
    Topographic area is defined as the watershed for the surface water feature. Watersheds are
    delineated by drawing a line along the highest elevation around the surface water feature. In this
    case, the study area itself is the source water protection area.
    Buffer Zone
    A buffer zone may be delineated for the purpose of protecting drinking water intake and is
    typically dependent on the hydrogeology, topography, and stream hydrology. A protection buffer
    for a source surface water intake is an upstream strip of vegetated land along the shore of the
    stream or lake. Buffer widths vary from 15 to 60 meters (approximately 50 to 200 ft) depending
    on topographic, land use, political, and legal factors (EPA, 1997b). Buffer zones reduce water
    quality impacts from runoff, increase wildlife habitat and improve stream-bank integrity. Systems
    with groundwater sources may use a fixed-radius protection area (buffer) around source wells
    depending on aquifer properties. The radius could be fixed arbitrarily or based on TOT (EPA,
    Time of Travel
    Water supply systems tapping rivers that are designated for commercial transportation or for
    industrial and municipal wastewater discharge may use TOT for source water intake protection.
    The time it takes a pollutant introduced into an upstream section of a river to travel to a source
    water intake is estimated using the stream-flow TOT. The contamination level of the pollutant at
    the intake can be evaluated using various water quality models. The TOT method provides tools
    for predicting impacts from spills or discharges at sections upstream of a drinking water intake,
    thereby enhancing protection strategies for emergency spills. A TOT is also used for delineating
    protection areas for groundwater-based systems by estimating contaminant transport into
    drinking water wells. Groundwater flow is significantly slower than that of surface water (e.g.,
    years versus hours or days, respectively), allowing more response time for controlling or
    remediating spills  and other plumes. The EPA (1993) provides comparisons  of TOT-based
    methods used for delineating WHP areas.

    Surface runoff and groundwater models can be used for delineating a source water protection
    area. Analytical, semi-analytical, and numerical flow and solute transport models can estimate
    the potential water quality impacts from one or more pollution sources upstream of a drinking
    water intake. With knowledge of land uses (e.g., agricultural, industrial, residential), soil
    properties, and precipitation rates in an area, potential contaminant loadings from runoff or
    infiltration can be estimated. Modeling  provides analytical tools for assessing water quality
    impacts resulting from land use changes, and may be used to identify effective water quality
    protection strategies. Some models need site-specific data which may, in turn, require field
    Stream Network Analysis: Water Quality Study
    Stream network analysis provides tools for studying  how contaminants are transported in
    streams. Distributions of contaminate concentrations along a stream can be studied using the
    physical and chemical properties of the contaminant as well as the hydraulics of the stream.
    Most GIS software packages, such as  ARC/INFO=s network analysis, have capability for
    modeling linear processes. More complex analyses can be performed by linking appropriate
    water quality models in ARC/INFO (e.g., Grayman et al.,1993).
    Generate Display Products
    Maps are graphic representations of geographic information, and, as such, provide powerful
    visual communication of ideas. The Surface Water Assessment Program requires strong public
    participation in all processes involving  development  of methods for, and implementation of,
    source water assessment. State agencies proposing or conducting a SWAP may use sets of
    maps for displaying the geographic extent of the SWAP program. For example, maps for public
    presentation can show stream segments with highlighted buffer areas and marked with potential
    pollution sites. A GIS provides the capability for generating such  maps at various scales with
    selected sets of themes.
    The GIS hardware includes the computer on which the GIS operates and the peripherals used
    for data entry, transfer, and output. A wide range of hardware types are used, from centralized
    computer servers to desktop computers used as stand-alone stations or  in networked
    configurations. The type and number of components in a system is dependent on the needs of
    the organization. Software vendors can help in recommending appropriate system

    configurations. The input and output devices (e.g., digitizers and plotters) are usually shared
    within an organization with more than one GIS user. Centralized computer servers and
    networking software can be used to enable multiple users to share GIS hardware and software.
    Hardware costs are not provided because costs are constantly changing, usually in favor of the
    buyer. Examples of GIS hardware components are listed below.
    GIS Workstation
    A GIS workstation should at a minimum include a high-speed central processing unit,  keyboard,
    mouse, disk space, high-resolution color monitor for graphics display, and a compact disk read-
    only memory (CD-ROM). An external disk drive may be used for additional disk space. The GIS
    workstation can be either an IBM-compatible personal computer (PC) using a Windows
    operating system or a high-end graphics workstation using a Unix operating system. The Unix
    systems provide a more powerful environment for GIS than PCs. Unix workstations are usually
    faster than PCs in the analysis and display of complex digital data. However, they also cost more
    ($ 15,000+ vs. < $ 10,000 for a PC). A set of workstations loaded with GIS software may use a
    common server with a large amount of disk storage space. Also, data input and output devices
    may be attached to the server so all users can share them.
    Data Transfer and Backup Devices
    A GIS should include one or more data transfer and backup devices such as a compact disk
    writer, tape drive, or disk drive. These devices allow the user to transfer GIS data to a compact
    medium that can be easily stored or physically transferred. These devices are useful for
    performing data backups or transferring data between workstations or organizations that are not
    Data Output Devices
    Output devices allow the user to print data and displays from the GIS. Printouts of GIS data are
    useful for data quality assurance and quality control (QA/QC) checks and for displaying results.
    The common GIS output devices are printers and plotters. These devices are available in a
    variety of sizes, produce output in color or black and white, and can vary widely in price. Most
    organizations will want at least a standard laser jet printer as well as a large-format color output
    device for plotting color maps for display and presentations.

    Data Input Devices
    GIS data input devices include digitizing tables, scanners, and GPS receivers. These devices
    enable a user to capture geographic information in digital form. A digitizing table is used for
    generating vector-based coordinate information directly from hard copy maps or photographs. A
    scanner is used to generate raster-based data from hard copy maps or photographs. A GPS
    receiver enables the user to capture coordinate data for features in the field. Once captured,
    GPS data must be post-processed on a workstation with specialized software to generate real-
    world coordinates.
    Three categories of information processing software are used to assess source waters when
    using GIS technology: GIS, image processing, and relational database management. Examples
    of software for each of these categories are listed below. A software package listed in one
    category may also be capable of performing functions in another category. For example, a GIS
    package such as GRASS can be used for image processing. Similarly, some of the image
    processing software packages can be used as GIS tools. The names of the software are listed
    for informational purposes only and do not indicate endorsement. The PC-based software
    packages such as GRASS and ArcView can range in cost from free or low-cost ($200-$300) to
    several thousand dollars. High-end software packages such  as Arclnfo, ERDAS Imagine, or
    Intergraph GIS will cost $10,000-$20,000.  Prices for all software packages depend on current
    market value, whether the purchaser is eligible for discounts, and what additional modules are
    purchased in addition to the baseline package.
    GIS Software
    The GIS software is used for storing, analyzing, and  displaying geographic data. The main
    components of a  GIS software are the tools for data  input and manipulation, database
    management, geographic query and analysis, and visualization and output. Several GIS
    packages are presented below for information.
    Arc/Info is a commercial software package developed by the Environmental Systems Research
    Institute (ESRI) and Henco Software, Inc. (Henco). Arc/Info provides tools for automation,
    management, display, and output of geographic and associated data. Arc/Info is a vector-based

    GIS software that runs on Unix and Windows NT workstations. Arc/Info costs between $10,000 -
    $20,000. For more information contact ESRI at http://www.esri.com.
    A rcView
    ArcView is also produced by ESRI and is a menu-driven GIS with a subset of the functionality
    provided by Arc/Info. What ArcView lacks in functionality, it makes up for in a less steep learning
    curve and an easy-to-use graphical user interface (GUI). ArcView is a vector-based GIS
    software that runs on Unix or PC workstations. ArcView costs approximately $1,000. For more
    information contact ESRI at http://www.esri.com.
    The Geographic Resources Analysis Support System (GRASS) is a public-domain, raster-based
    GIS software used for geographic data management, image processing, graphics production,
    spatial modeling, and data visualization. GRASS was written by the U.S. Army Construction
    Engineering Research Laboratories (USA-CERL) branch of the U.S. Army Corps of Engineers
    and is currently maintained at the Department of Geology at Baylor University. GRASS runs on
    Unix and PC workstations. More information on GRASS can be found at
    http://www.baylor.edu/~grass. Additional information on some of the hydrology models that have
    been integrated into the GRASS GIS is available on
    http://soils.ecn.purdue.edu/~aggrass/models/ hydrology.html.
    IDRISI is a raster-based GIS software that provides GIS, image processing, and spatial statistics
    analytical capabilities on DOS and Windows-based PCs.  IDRISI provides analytical functionality
    of GIS, remote sensing, and databases for resources management. IDRISI was developed and
    is maintained by Clark Labs, a non-profit research organization within the Graduate School of
    Geography at Clark University. A commercial/private single-user license for IDRISI costs $990.
    Licenses for non-profit, government, and academic institutions cost less. For more details see
    Intergraph GIS
    Intergraph provides Windows-based software and a range of computing services for
    engineering, design, modeling, analysis, mapping, information technology, and creative
    graphics. The GIS MGE package provides data collection and editing, data import, image

    display and analysis, advanced spatial query and analysis, and cartographic quality maps. MGE
    costs approximately $10,000-$20,000. More information on Integraph GIS is available at
     Image Processing Software
    Image processing software is used to process raster data, particularly remote sensing imagery
    data such as satellite imagery.
    The Environment for Visualizing Images (ENVI) is an image processing system which provides
    analysis and visualization of single-band, multispectral, hyperspectral, and radar remote sensing
    data. ENVI can process large spatial and spectral images, and runs on Unix; LINUX; Windows
    3.1, NT, 95; the Macintosh; and the Power Mac. For more details contact ENVI at
    ERDAS Imagine
    The ERDAS Imagine software is an image processing and raster GIS package that has a variety
    of applications ranging from simple image mapping to advanced remote sensing. Imagine
    provides tools for geometric correction, image analysis, visualization, map output,
    orthorectification, radar analysis, advanced classification tools, and graphical spatial data
    modeling. Imagine runs on Unix workstations and Windows platforms. ERDAS Image costs
    approximately $10,000-$20,000. More information on Imagine is available at
    ER Mapper
    ER Mapper provides integrated mapping software featuring image processing, map production,
    3-D presentations, and GIS integration for Windows 95/NT and Unix. The ER Mapper software
    uses a concept that separates data from the image processing steps allowing the user to apply
    and view results from a single enhancement procedure in real time. The PC version of ER
    Mapper costs $4,300; the Unix version of ER Mapper costs $18,300. See
    http://www.ermapper.com for more information.

    EASI/PACE image processing provides a variety of applications including image processing,
    geometric correction, vector utilities, and multilayer modeling. PCI implements the Generic
    Database (GDB) concept, which allows PCI programs to access image and other external data
    files without import and export. Contact PCI for more details at http://www.pci.on.ca.
    TNTmips is a map and image processing system that contains fully featured GIS, CAD, and
    spatial database management systems. TNTmips has tools that interactively integrate elements
    of on-screen image processing and photo interpretation, and provides a diverse set of tools for
    registering, rectifying and stitching imagery, which are particularly useful for low-altitude aerial
    photography and videography. More information on TNTmips is available at
    Relational Database Management Software
    Relational database management system (RDBMS) software enables large amounts of data to
    be entered, updated, related, viewed, queried and, otherwise, managed in an efficient manner.
    The data in an RDBMS is stored in a series of related tables which are designed to optimize the
    effort required for data entry, maintenance, and retrieval. RDBMS software is available for use
    on PCs, Unix workstations, networked systems, and mainframe computers. Most GIS software
    packages use an RDBMS to manage data such  as maintaining topology and providing ways to
    efficiently enter,  update, and query attribute data. Major RDBMS software includes Info, dBASE,
    MS Access, Ingres, Informix, Oracle, and Sybase.
                                SOFTWARE SUPPORT TOOLS
    There are numerous software support tools available for use in assessing source waters. These
    tools operate within specific operating and software system environments. The  information
    presented here is not an endorsement of any of these products. New products and
    improvements to existing products are continuously being introduced;  therefore, users should
    conduct their own investigation of software tools to ensure they are getting the latest information.
    The selections are considered some of the more promising and potentially useful that were
    encountered during this GIS evaluation. It should be noted that there are hundreds of available

    hydrologic models described in the scientific literature, but many of these will probably not be
    suitable for use in a source water assessment. Principal purveyors of other downloadable
    software and hydrologic models not listed here include the EPA Center for Exposure
    Assessment Modeling (http://ftp.epa.gov/epa-ceam/wwwhtml/softwdos.htm), the USGS Water
    Resources Division  (http://water.usgs.gov/software/), and the U.S. Army Corps of Engineers=
    Hydrologic Engineering Center (http://www.hec.usace.army.mil/). A selection of the software
    support systems to consider include:
           - Better Assessment Science Integrating Point and Nonpoint Sources (BASINS)
           from the EPA Office of Science and Technology (OST). Full documentation of BASINS
           Version 2.0 is available in detail at http://www.epa.gov/ostwater/BASINS/.
           - Riverine Emergency Management Model (REMM) was originally developed for
           hydrologic modeling in the upper Mississippi River in Minnesota.  REMM is public-
           domain software and is freely available by the U.S. Army Corps of Engineers office in St.
           Paul, Minnesota (e-mail: webmaster@mvp-wc.usace.army.mil).
           - Watershed Modeling System (WMS) Model was developed by the Environmental
           Modeling Research Laboratory of Brigham Young University in cooperation  with the U.S.
           Army Corps  of Engineers Waterways Experiment Station. The WMS is proprietary
           software and is available via the Engineering Computer Graphics Laboratory at Brigham
           Young University in Provo, Utah (http://www.ecgl.byu.edu). The software cost ranges
           from $500 to $2,600 depending on the desired modules.
           - Underground Storage Tank (UST)-Access Software was developed using the
           Microsoft Access 2.0 relational data base management system. All UST-Access
           installation files are stored as self-executable archive files on the Cleanup Information
           (CLU-IN) Bulletin Board System of the EPA Office of Solid Waste and Emergency
           - Spatially Referenced Regressions on Watersheds (SPARROW) Model is an
           extension from the Hydrologic Simulation Program - Fortran (HMPF) modeling
           framework. The HMPF and SPARROW models are public-domain software freely
           available through USGS (http://www.usgs.gov) for the cost of pressing a CD (about $35).

           - Hydrology Extension for the ArcView Spatial Analyst Software is a new Hydrology
           Extension for ArcView=s Spatial Analyst 1.1. ArcView Spatial Analyst 1.1 is proprietary
           software distributed by ESRI (http://www.esri.com). Price varies widely depending on the
           user=s affiliation, such as with government or industry.
           - MassGIS Watershed Tools for the ArcView Spatial Analyst Software was
           developed by the State of Massachusetts, Division of Watershed Management, GIS
           Division (MassGIS) for use with the ArcView Spatial Analyst. ArcView Spatial Analyst is
           proprietary software distributed by ESRI (internet: www.esri.com). Price varies widely
           depending on the user=s affiliation with government or industry. The MassGIS watershed
           tools are public-domain software (john.rader@state.ma.us).
                                PERSONNEL REQUIREMENTS
    To use a GIS effectively in any project, it is important to have personnel with a variety of specific
    skills. All of the software mentioned above (GIS, image processing, and RDBMS) require lengthy
    learning curves to be used effectively.
    Data Entry Technician
    Data entry includes automation or digitizing of maps, creating attribute tables, and importing
    databases. The data entry technician should have some knowledge of spatial concepts and
    experience in basic GIS use for creating thematic layers, and attribute data entry. Depending on
    the amount of data entry required, one or more technicians may be needed.
    Spatial Data Analyst
    The spatial data analyst is skilled in manipulating geographic data to retrieve pertinent, project-
    specific information such as mapping sources of contamination and their proximity to source
    waters, and delineating protection areas. This person must have a thorough understanding of
    the concepts presented in this Chapter and be experienced in using GIS and image processing
    technology. The spatial data analyst should also have some experience in working with utilities,
    hydrogeology, soils, environmental engineering, or sanitary engineering.
    Field Surveyor

    A field surveyor may be required if geographic or attribute data is not available and must be
    gathered in the field. The surveyor should be skilled in field survey management, GPS
    technology, and database development and have knowledge of sanitary or environmental
    engineering, soil science, or hydrogeology. Depending on the amount of field surveying required
    and the size of the area being surveyed, the field surveyor may require a support staff to assist
    with gathering information.
    Soil Scientist
    A soil scientist may be needed to evaluate the condition and physical properties of soils in the
    survey area. The Natural Resources Conservation Service (NRCS) formerly called the Soil
    Conservation Service may be contacted for technical assistance in this area.
    System Administrator
    A system administrator may be needed to administer the GIS and its peripherals such as
    digitizers, printers, and plotters. This is especially true for systems that require a network and
    have multiple users. A system administrator can help with hardware and software  maintenance
    and replacement, network maintenance, system backups, and other administrative duties.

    Anderman, W.H. and G.Martin. 1986. Effect of public sewers on watershed contamination,
    Journal of Environmental Health 4(2):81-84.
    EPA. 1997a. States source water assessment and protection programs final guidance. EPA
    816-R-97-009. U.S. Environmental Protection Agency, Office of Water, Washington, DC.
    EPA. 1997b. State methods for delineating source water protection areas for surface water
    supplied sources of drinking water. EPA 816-R-97-008. U.S. Environmental Protection Agency,
    Office of Water, Washington, DC.
    EPA. 1997c. Delineation of Source Water Protection Areas, a Discussion for Managers; Part 1:
    A Conjunctive Approach for Ground Water and Surface Water, U.S. Environmental Protection
    Agency, Office of Water, Washington, DC. (Expected August 1997)
    EPA. 1995. EPA/State Joint Guidance on Sanitary Surveys. U.S. Environmental Protection
    Agency, Office of Water, Washington, DC.
    EPA. 1994. The quality of our nation=s waters. EPA 841-S-95-004. U.S. Environmental
    Protection Agency, Office of Water, Washington, DC.
    EPA. 1993. Guidelines for delineation of wellhead protection areas. EPA 4405-93-001. U.S.
    Environmental Protection Agency, Office of Water, Office of Groundwater Protection,
    Washington, DC.
    EPA. 1991 a. Guide for Conducting Contamination Source Inventories for Public Drinking Water
    Supply Protection Programs. EPA 570/9-91-033. U.S. Environmental Protection Agency, Office
    of Water, Washington, D.C.
    EPA. 1991b. Managing Groundwater Contamination Sources in Wellhead Protection Areas: A
    Priority Setting Approach. EPA 570/9-91-023. U.S. Environmental Protection Agency, Office of
    Groundwater and Drinking Water.

    EPA. 1990. Guidance manual for compliance with the filtration and disinfection requirements for
    public water systems using surface water sources, U.S. Environmental Protection Agency, Office
    of Drinking Water, Washington, D.C.
    ESRI, Inc. 1992. Understanding GIS: The ARC/INFO Way. Environmental Systems Research
    Institute, Redlands, CA.
    Grayman, W.M., S.R.  Kshirsagar, and R.M. Males. 1993. A Geographic Information System for
    the Ohio River Basin,  Risk Reduction Engineering Laboratory, Office of Research and
    Development, US Environmental Protection Agency, Cincinnati, Ohio.

                                   A GIS for the Ohio River Basin
                                           Walter M. Grayman
                         W.M. Grayman Consulting Engineers, Cincinnati, Ohio
                                          Sudhir R. Kshirsagar
                               Global Quality Corporation, Cincinnati, Ohio
                                            Richard M. Males
                             RMM Technical Services, Inc., Cincinnati, Ohio
                                           James A. Goodrich
              Risk Reduction Engineering Laboratory,  Office of Research and Development,
                         U.S. Environmental Protection Agency, Cincinnati, Ohio
                                             Jason P.  Heath
                    Ohio River Valley Water Sanitation Commission, Cincinnati, Ohio
    Much of the information used in the management of
    water quality in a river basin has a geographic or spatial
    component  associated with  it. As  a  result, spatially
    based computer models and database systems can be
    part  of an  effective water  quality  management and
    evaluation process.  The Ohio River Valley Water Sani-
    tation Commission (ORSANCO) is an interstate water
    pollution control agency serving the  Ohio River and its
    eight member states. The U.S. Environmental Protec-
    tion Agency (EPA) entered into a cooperative agreement
    with ORSANCO to develop and apply spatially based
    computer models and database systems in the Ohio
    River basin.
    Three computer-based technologies have  been ap-
    plied and integrated: geographic information systems
    (GIS), water quality/hydraulic modeling, and database
    GIS serves as a mechanism for storing,  using, and
    displaying spatial data. The  ARC/INFO GIS,  EPAs
    agencywide standard, was  used in the study, which
    assembled databases of land and stream information for
    the Ohio River basin. GIS represented streams in hydro-
    logic catalog units along the Ohio River mainstem using
    EPAs new, detailed  RF3-level Reach File System. The
    full Ohio  River basin was represented using the less
    detailed  RF1-level reach file.  Modeling provides a way
    to examine the impacts of human-induced and natural
    events within the basin and to explore alternative strate-
    gies for mitigating these events.
    Hydraulic information from the U.S. Army Corps of En-
    gineers' FLOWSED model enabled EPAs WASP4 water
    quality model  to be embedded in a menu-driven spill
    management system to facilitate modeling of the Ohio
    River  mainstem under emergency spill conditions. A
    steady-state water quality modeling component was
    also developed under the ARC/INFO GIS to trace the
    movement and degradation of pollutants through any
    reaches in the RF1 representation of the full Ohio River
    Database management technology relates to the stor-
    age, analysis,  and display of data. A detailed database
    of information on dischargers to the Ohio River mainstem
    was assembled under the PARADOX database manage-
    ment system using EPAs permit compliance system as
    the primary data source. Though these three technolo-
    gies have been widely used in the field of water quality
    management,  integration  of these tools into a holistic
    mechanism provided the primary challenge of this study.
    EPAs Risk Reduction Engineering Laboratory in Cincin-
    nati, Ohio, developed this project summary to announce
    key findings of the research project, which is fully docu-
    mented in a separate report of the same title.

    During the past 25 years, computers have been actively
    used in water quality management, demonstrating their
    potential to assist in a wide range of analysis and display
    tasks. Technologies such  as geographic information
    systems (CIS), database management systems (DBMS),
    and mathematical modeling have  been applied in the
    water quality management field and have proven to be
    effective tools. For computers to achieve their full poten-
    tial, however,  they must become integrated  into the
    normal programmatic efforts of agencies and organiza-
    tions in the planning, regulation, and operational areas
    of water quality management.
    Recognizing this  need for routine use of computer-
    based tools,  the  Ohio River  Valley Water Sanitation
    Commission (ORSANCO) and the Risk Reduction En-
    gineering Laboratory (RREL) of the U.S. Environmental
    Protection Agency (EPA) commenced  a study in 1990.
    The goals of the study included the adaptation, devel-
    opment, and application  of modeling and spatial data-
    base  management (DBM)  tools  that  could  assist
    ORSANCO in its prescribed water quality management
    objectives. These goals were consistent with EPAs on-
    going programs involving the use  of CIS and modeling
    technology. The study's goals also coincided with EPAs
    Drinking Water Research Division's work over the past
    decade, which applied similar technology to study the
    vulnerability of water supplies on the Ohio and Missis-
    sippi Rivers to upstream discharges.
    Methodology Overview
    To address the goals of this project, three basic tech-
    nologies have been applied and integrated: CIS, water
    quality/hydraulic modeling, and DBM.  CIS serves as a
    mechanism for storing,  using, and displaying spatial
    data. Modeling provides a way to  examine the impacts
    of human-induced and natural events within the basin
    and to explore alternative strategies for mitigating these
    events. DBM technology relates to the storage, analysis,
    and display of data. Though these three technologies
    have  been widely used in the field  of water quality
    management,  integration of these tools into a holistic
    mechanism provided the primary challenge of this study.
    CIS Technology
    The guiding principle in developing the CIS capability
    was to maximize the use of existing CIS technology and
    spatial databases. The  study used  ARC/INFO CIS,
    EPAs  agencywide  standard.  Remote  access  of
    ARC/INFO on a VAX minicomputer facilitated the initial
    work. Subsequently, both PC ARC/INFO and a worksta-
    tion-based system were obtained.
    EPA has developed an extensive spatial database re-
    lated to water quality and demographic parameters. This
    served as the primary source  of spatial data for the
    study. Following is a summary  of spatial data used  in
    this study:
    • State and county boundaries.
    • City locations and characteristics.
    • Water supply locations and characteristics.
    • Locations and characteristics of dischargers to
      water bodies.
    • Toxic loadings to air, water, and land.
    • Dam locations and characteristics.
    • Stream reaches and characteristics.
    The  primary organizing concept for the  water-related
    information was EPAs Reach File System (1). This sys-
    tem  provides a common  mechanism within EPA and
    other agencies for identifying surface water segments,
    relating water resources data, and traversing the nation's
    surface water in hydrologic order within a computer envi-
    ronment. A hierarchical hydrologic code uniquely identi-
    fies each reach. Information available  on each reach
    includes topological identification of adjacent reaches,
    characteristic information  such as length and stream
    name, and stream flow and velocity estimates. The origi-
    nal reach file (designated as RF1) was developed in the
    early 1980s and included approximately 70,000 reaches
    nationwide. The most  recent version  (RF3)  includes
    over 3,000,000 reaches nationwide.
    As part of this project, an RF1-level database was es-
    tablished for the entire Ohio River basin. The RF3 reach
    file was implemented for the Ohio River mainstem and
    lower portions of tributaries. River miles along the Ohio
    River were digitized and established as an ARC/INFO
    coverage to provide a linkage between the reach file and
    river mile indexing used by ORSANCO and other agen-
    cies along the  river. Figure 1 shows the RF1 reach file
    representation of the Ohio River basin along with state
    The study incorporated several EPA sources of informa-
    tion on dischargers to water bodies. The industrial facil-
    ity  discharger  (IFD)  file  contains locational  and
    characteristic data  for National  Pollutant Discharge
    Elimination System (NPDES) permitted discharges. De-
    tailed permit limits and  monitoring  information was ac-
    cessed from the permit compliance system (PCS). The
    toxic release inventory (TRI) system includes annual
    loading of selected  chemicals to water, land, air, and
    sewer for selected  industries based on quantity dis-
    charged. All  water data are referenced to the NPDES
    permit number, which is spatially located by reach and
    river mile, and  by latitude and longitude.

                             0 meter,
    Figure 1. RF1 reaches in the Ohio River basin.
    Spill Modeling
    An important role that  ORSANCO fills on the Ohio River
    relates to the monitoring and prediction of the fate  of
    pollutant spills. Typically, ORSANCO serves as the over-
    all communications link between states  during  such
    emergency conditions. ORSANCO coordinates and par-
    ticipates in monitoring and serves  as the information
    center  in gathering  data and issuing predictions about
    the movement of spills in the river. In the past, a series
    of time-of-travel  nomographs,  based  on  National
    Weather Service flow forecasts, Corps of Engineers
    flow-velocity relationships,  and  previous  experience,
    were used to predict the movement of spills. This project
    combined a hydraulic  model with a water quality model
    to serve as a more  robust method for making such
    The U.S. Army Corps of Engineers' FLOWSED model
    was selected as the means of predicting daily flow quan-
    tities and water levels  along the mainstem  and portions
    of major tributaries  near their confluence with the Ohio
    River (2). The Ohio River Division of the Corps of Engi-
    neers applies FLOWSED daily as part of its  reservoir
    operations  program.  The  Corps can  generate 5-day
    forecasts of stage and flow for 400 mainstem and tribu-
    tary segments, and ORSANCO can access the results
    via phone lines.
    EPAs WASP4 water quality model was selected for use
    in the project (3). WASP4  is a  dynamic compartment
    model that can be  used to analyze a  variety of water
    quality problems in a diverse set of water bodies. Because
    the primary use of the  model in this project is quick
    response under emergency situations,  only the toxic
    chemical portion of the model with first order decay is
    being used. The FLOWSED and WASP4 models have
    been combined into a user-friendly spatial decision sup-
    port system framework  described later in this project
    Discharger Database  Management System
    EPA's  PCS  and  historical records maintained  by
    ORSANCO furnish a rich source of data on dis-
    charge  information for the Ohio River. To  organize
    these data and make them available for analysis, a
    database was  developed  using the PARADOX DBM
    The database was established using a relational struc-
    ture with a series of related  tables (two-dimensional flat
    files). Individual tables contain  information on facilities,
    outfalls, permit  limits, monitoring data, and codes used
    in the other tables. The NPDES permit number is used
    as the primary key in each data table. A mechanism for
    downloading  and reformatting data  from the national
    PCS database  has  been developed along with custom
    forms for viewing and editing data, and custom reports
    for preparing hard copy summaries. Latitude and longi-
    tude values for each facility can provide  the locational
    mechanism for use of this data in conjunction with CIS.
    Integration of GIS/Modeling/Database
    A major objective of this study was the  integration of
    CIS, modeling, and DBMS  technologies  into a holistic
    tool for use by ORSANCO.  Several integration mecha-
    nisms were implemented as summarized  below.
    Steady-State Spill Tracing
    The NETWORK component of the ARC/INFO CIS pro-
    vides a steady-state, transportation-oriented routing  ca-
    pability.  This capability  was used  in an arc macro
    language (AML) program to construct a routing proce-
    dure for determining downstream concentrations  and
    travel times. The pollutant may be treated as a conser-
    vative element or represented by a first order expo-
    nential  decay  function. This  capability  has been
    implemented for use with  the  RF1 reach file repre-
    sentation of the full Ohio  River basin. The  user may
    select from six flow regimens: average flow,  low flow,
    and four multiples of average  flow ranging from one-
    tenth  to  10 times  average flow.  This system gives
    ORSANCO the ability to estimate the arrival time of a
    spill from any RF1 tributary to the Ohio River mainstem.

    Sp/7/ Management System
    A PC-based spatial decision support system (SDSS)
    was built as a spill management system to be a quick
    response tool for analyzing and displaying the results of
    pollutant spills  into  the Ohio River. The schematic in
    Figure 2 illustrates the components in this computerized
    spill management system. The system is implemented
    in the C language using a commercial menuing system
                             and  a series of graphic display routines developed at
                             EPA. Custom, written routines have been used to read
                             the  output from  the  U.S. Army Corps of Engineers'
                             FLOWSED  model, to generate input files for  EPAs
                             WASP4 model, to create output reports and output plots,
                             and  to provide an animated  representation of the con-
                             centration profiles moving down the river. Figure  3 pre-
                             sents an  example of a  graphic output the system
                             generated. Additionally, the system generates a file in
             5-DAY RIVER FLOW
             DAILY BY CORPS OF
             ENGINEERS USING
              FLOWSED MODEL
     via phone line
                  ARC/INFO CIS
                                            SPATIAL DECISION SUPPORT SYSTEM
                                               FOR TOXIC SPILL MODELING
    Menu driven user interface linked to
    EPA WASP 4 water quality model.
    Tabular /graphical screen and hard
    copy outputs, spill animation and
    GIS DBF format output file
                    Arc View Spatial data
                    base display system
                                                          Segment & Time
    Figure 2. Schematic representation of spill modeling system process.
                                                                                            141 ;47 1991
    Figure 3. Graphic output from the basinwide network spill model.

    DBF format that may be read by ARC/VIEW (the com-
    panion software to ARC/INFO for user-friendly viewing
    of spatial data).
    Hardware Platform
    Within the study, the initial hardware platform was a
    combination of local PCs (in Cincinnati) and a remote
    access terminal to  a VAX computer located at EPA's
    National Computer  Center in Research Triangle  Park,
    North Carolina. The final platform, and the one on which
    the completed system was installed, comprised a UNIX-
    based Data General workstation and a PC workstation.
    The full hardware configuration is shown schematically
    in Figure 4.
    The application of  computer-based display, analysis,
    and modeling  tools  in conjunction with CIS technology
    proved to be an effective strategy for water quality man-
    agement. This study used an existing CIS package and
    DBMS in  conjunction with  existing water  quality and
    hydraulic models. The study focused  primarily on as-
    sembling available spatial and relational databases and
    integrating the  systems to provide a usable, effective tool.
    1. Horn, R.C., and W.M.  Grayman. 1993. Water-quality modeling with
      EPA reach file system. J.  Water Res. Planning and  Mgmt.
    2. Johnson, B.H.  1982. Development of a numerical modeling capa-
      bility for the computation of unsteady flow on the Ohio River and
      its major tributaries. Vicksburg, MS:  U.S. Army Engineer WES.
    3. Ambrose, R.B., T.A. Wool, J.P. Connolly, and R.W  Schanz. 1988.
      WASP4, a hydrodynamic and water quality model: Model  theory,
      users' manual, and programmers' guide.  Athens, GA:  EPA Envi-
      ronmental Research Laboratory. NTIS PB88185095.
    I \
    8 MM
    1/4" TAPE
    Figure 4.  Hardware configuration.

         Data Quality Issues Affecting GIS Use for Environmental Problem-Solving
                                               Carol B. Griffin
                                Henrys Fork Foundation, Island Park, Idaho
    "Abandon hope, all ye who enter here." Dante's quote
    might well be the advice that experienced geographic
    information system (GIS) users give to nonusers about
    to confront data quality issues  associated  with GIS
    use. Indeed, after reading this paper, some  decision-
    makers might abandon attempts  to use a GIS because
    of the error associated with it. Others may want to spend
    an inordinate amount of time and money trying to elimi-
    nate all error associated with GIS use. Neither option is
    Data quality is important because it affects how reliable
    CIS-generated information is in  the  decision-making
    process. Too often, the availability of inexpensive digital
    data overshadows data quality concerns;  people fre-
    quently use digital data because they are available, not
    because they have the necessary accuracy.
    A GIS can help decision-makers use spatial informa-
    tion more fully than manual methods allow, but some-
    times data quality issues cause concern about using
    CIS-generated  outputs.  Making  environmental  deci-
    sions without adequate consideration  to data quality
    may lead to an erroneous decision, erode public confi-
    dence, or cause an agency to incur liability. This paper
    attempts to encourage  decision-makers  to become
    more aware of data quality issues, including the sources
    and magnitude of error.
    GIS  error research  has necessarily progressed in a
    linear fashion, beginning with  identifying and classifying
    sources of error. This paper discusses both inherent
    (source) error and the error that GIS operations  intro-
    duce  (operational error)  during data  input, storage,
    analysis/manipulation,  and output (1).  Strategies  for
    coping with error and research into error reduction tech-
    niques have only recently received attention. Unfortu-
    nately, the answers to error management  questions
    such as, "How will the error affect decision-making?" are
    not clear.  The  end of this paper covers several  error
    management suggestions and anticipated software im-
    provements designed to reduce errors,  however.
    Data Quality Concepts and Their
    Data quality is a major issue for CIS-generated maps,
    much more so than it is for paper maps. In part, this is
    because a GIS can perform operations on spatial data
    that would be nearly impossible without a GIS because
    of scale, complexity, and generalization issues (2). Car-
    tographers adjust for these problems when they manu-
    ally manipulate and instantly combine  paper maps by
    adhering  to  long-standing cartographic principles, but
    GIS personnel may not be fully trained in these princi-
    ples. A GIS enables an analyst, whether trained in car-
    tographic principles or not,  to combine or manipulate
    data in appropriate  or in inappropriate, illogical,  and
    erroneous ways. Lack of training coupled with the speed
    of spatial data manipulation can have serious conse-
    quences for an agency whose personnel produce and
    use CIS-generated maps.
    Limited scientific understanding, limited ability to meas-
    ure data,  sampling error, inherent variability, and inade-
    quacy of mathematical representations all contribute to
    uncertainties associated with spatial data. Uncertainty
    about spatial data consists of two parts: ignorance and
    variability. Ignorance means that variables have a "true"
    value, but it is unknown to us, whereas variability means
    one value cannot represent the variables.
    Data quality defies a simple definition. For this paper,
    data quality can roughly mean  how "good" the data are
    for a given purpose. People usually think of data quality
    in terms of error, but the term is broader and encom-
    passes the six components outlined  in the next section.
    Error can mean the difference between the observed
    values and the "true" value. The "true" value of a variable
    is usually unknown and unknowable, but for this paper's
    purposes, "true" could  be the known value or the value
    one would obtain from field measurements (the discus-
    sion of data collection tries to dispel the notion that there
    is one "true" value for many variables, such as soil type
    in a given area orwatertemperature in a lake). Imperfect

    equipment or  observers  and  environmental  effects
    cause spatial error. According to Thapa and Bossier (3),
    errors fall into three categories:
    • Gross errors and blunders (people or equipment).
    • Systematic errors (which introduce bias).
    • Random errors (due to imperfect instruments and
    In addition, another view divides spatial error into two
    different components: accuracy and precision. Accuracy
    means  how close a  value is to the "true"  value or  a
    known standard (absence of bias). Precision can have
    two definitions:  it can be a measure of dispersion  (stand-
    ard deviation) of observations about a mean, or it can
    refer to the number of decimal digits used to represent
    a value (4). In  the first definition of precision, a meas-
    urement of 6 feet  plus or minus 1 foot is more precise
    than one of 6 feet plus or minus 3 feet. In the second
    definition, a value of 6.1794 feet is more precise than
    one of 6.1 feet. Figure 1 provides a graphic explanation
    of the difference between error, accuracy, and precision.
      True Value
    4  1    Mean  2  3
    Figure 1.  Relationship between error, accuracy, and precision.
    Data are not accurate or inaccurate. Instead, data accu-
    racy exists on  a continuum, ranging from low to high
    accuracy. Although people strive for accurate (error-
    free) data, obtaining  100-percent accurate data  is im-
    practical. The list below provides some of the reasons
    why total accuracy is  not obtainable  (5):
    • Objects to  be measured are often vaguely defined.
    • Some phenomena  are variable in  nature.
    • Classification schemes are imprecise.
    • Measurements are  inherently imprecise.
    • Gross errors of a nonstatistical nature can occur during
    • Attributes encoded on an ordinal  scale  (high, me-
      dium, low) are approximate.
    • Data represent a past state of reality.
    Users of geographic data should strive for data that are
    only as accurate as they need. A variety of factors, of
    course, can determine need:
    • Intended use of the data
    • Budget constraints
    • Time constraints
    • Data storage considerations
    • Potential liability
    The main  barrier to highly accurate data is lack of funds.
    Male (6) suggests that  rather than abandoning a CIS
    project because funds are not sufficient to achieve the
    desired accuracy,  an agency should collect data at the
    desired accuracy  from  smaller  areas, such as  areas
    being developed or redeveloped. Overtime, data collec-
    tion at the desired accuracy can expand to include areas
    that lacked data due to budgetary constraints. Smith and
    Honeycutt (7) outline the use of a value of information
    approach  in  determining the need for more  data (or
    more accurate data) based  on the expected costs and
    benefits associated with data collection.  If the  benefits
    of increased data accuracy are  greater than  the ex-
    pected costs, additional funds should be  allocated to
    obtain more accurate data.
    The intended use of data affects the type of data, as well
    as the data quality needed. Beard (8) divides CIS appli-
    cations into six types (see Table 1). The specific type of
    data quality one needs (e.g., positional accuracy, attrib-
    ute accuracy) also varies with the intended application.
    Analysts with inventory applications such as agricultural
    production are less  concerned about positional  accu-
    racy than with an accurate  assessment of anticipated
    crop yields (attribute  accuracy).  Decision-makers must
    Tabl e 1.  Types of GIS Applications (8)
    Application                   Example
    Siting        Finding optimal location (fire station, waste site)
    Logistic       Movement or distribution through space
                (emergency response, military movement)
    Routing       Optimal movement through a known network (mail,
                school bus)
    Navigation    Way finding; may or may not involve a known
                network (ground, sea, air)
    Inventory     Count and location of objects for a given time
                (census, tax rolls)
    Monitoring/    Examining processes over space and time
    Analysis      (ecological, zoological, geological,  epidemiological

    decide which data quality component is the most impor-
    tant for their use because optimizing all six components
    can be very expensive  (9). An obvious conflict arises
    when local and state governments must meet multiple
    application needs simultaneously and thus feel forced to
    try to optimize several data quality components.
    The nature of the decision may also help decision-makers
    determine the data quality they need. Beard (8) lists sev-
    eral of these factors (see Table 2). A political, high-risk
    decision requires higher quality data than a nonpolitical,
    low-risk decision because more public attention focuses
    on the former decision.
    Table 2.  Factors That May Affec  t the Data Quality Needed for
            Decision-Making (8)
    Lower Data Quality
    Possibly Needed
    Higher Data Quality
    Possibly Needed
    Minimal risk
    Local implication
    High risk
    Global implication
    Components of Data Quality
    The  National Committee for Digital Cartographic Data
    Standards (9) identifies six components of digital carto-
    graphic data quality. This section discusses each  of
    these components:
    • Lineage
    • Positional accuracy
    • Attribute  accuracy
    • Logical consistency
    • Completeness
    • Temporal accuracy
    Most components of data quality apply to both source
    and operational error.
    Because uses and users of data change, those at the
    national level have noted a recent push to include docu-
    mentation when disseminating spatial data. Data  line-
    age, also known as metadata or a data dictionary, is data
    about data. Metadata consists of information about the
    source data such as:
    • Date of collection
    • Short definition
    • Data type, field length, and format
    • Control points used
    • Collection method, field notes, and maps
    • Data processing steps
    • Assessment of the reliability of source data
    • Data quality reports
    Access to this information can help CIS personnel de-
    termine if the data are appropriate for their use, thereby
    minimizing risks associated with using the wrong  data
    or using  data  inappropriately. According to Chrisman
    (10), the only ethical and probably best legal strategy for
    those who produce spatial data is to reveal more infor-
    mation about the data  (metadata)  so that users can
    make informed decisions. Eagan and Ventura's article
    (11) contains a sample of a generic environmental  data
    lineage  report.  The  U.S. Environmental  Protection
    Agency's  (EPA's) new locational data  policy requires
    contractors to estimate data accuracy and provide infor-
    mation about the lineage of the data (12).
    Positional A ecu racy
    Anyone who has used a  map has probably come across
    features that are not located where  the map says  they
    should be located and has experienced low positional
    accuracy. (Undoubtedly, they have  also detected  fea-
    tures that were not on the map, but that is a different
    issue.) Positional accuracy,  frequently referred to as
    horizontal error, is how close  a location on a map  is to
    its "true" ground position. Features may be located inac-
    curately on maps for many reasons, including (13):
    • Poor field work.
    • Distortion of the  original paper  map (temperature,
    • Poor conversion from raster to  vector or vector to
      raster data.
    • Data layers are collected at different times.
    • Natural variability in data (tides, vegetation, soil).
    • Human-induced  changes   (altering  reservoir water
    • Movement of features (due to scale of the map and
      printing constraints) so they can be easily discerned
      by the  map reader.
    • Combining maps with different scales.
    • Combining maps with different projection and coordi-
      nate systems.
    • Different national horizontal datum  in source materials.
    • Different minimum mapping units.
    Positional accuracy has  two components: bias and pre-
    cision. Bias reflects the  average positional error of the

    sample points and indicates a systematic discrepancy
    (e.g., all locations are 7 feet east of where they should
    be). Estimating precision entails calculating the stand-
    ard deviation of the dispersion of the positional errors.
    Usually, root mean square error (RMSE) is reported as
    the measure of positional accuracy, but it does not dis-
    tinguish bias from precision (14). RMSE  is  frequently
    monitored during digitizing to minimize the introduction
    of additional positional error into the CIS.
    To determine positional accuracy, one must compare the
    location of spatial data with an independent source of
    higher accuracy. Federal agencies that collect data and
    produce maps adhere to National  Map Accuracy Stand-
    ards (NMAS)  for positional accuracy. Maps such  as
    United States Geological Survey (USGS) topographic
    maps that conform to NMAS carry an explicit statement
    on them. Other groups also  have developed  standards
    for large-scale mapping (15).
    NMAS for positional accuracy require that not more than
    10 percent of well-defined points can be in error by more
    than one-thirtieth  of an inch for maps at a scale of
    1:20,000 or larger. For smaller scale maps,  not more
    than 10 percent of well-defined points can be in error by
    more than one-fiftieth of an inch (16). Thus, less than 10
    percent  of  the  well-defined locations on   a  USGS
    1:24,000 map can stand more than  40 feet  from  their
    "true" location; the other 90  percent of the well-defined
    points must stand less than 40 feet from their "true"
    location. Table 3 shows the  acceptable positional accu-
    racy for commonly used maps. Note that as scale de-
    creases from 1:1,200 to 1:100,000, positional accuracy
    Several  important issues relate to NMAS. First, not all
    maps adhere to NMAS, which means their positional
    accuracy may be lowerthan  NMAS or may be unknown.
    Second, NMAS do not indicate the location of points in
    error. Third, 10 percent of the well-defined points can
    have a positional error greater than the standards allow,
    but neither the location nor the  magnitude of these
    errors are known. Fourth, NMAS apply to well-defined
    points; therefore,  areas that are  not well defined  may
    Tabl e3.  NMAS Horizontal (Positional) Accuracy
    1 :2,400
    1 :4,800
    1 :24,000
    1 Inch = x Feet
    Accuracy +/- Feet
    have even lower positional accuracy. The implication of
    these errors in location is that users should use caution
    in  making decisions that require high positional accu-
    racy. Positional accuracy issues are particularly trouble-
    some for CIS operations on small-scale maps or when
    combining large-scale maps (1:1,200) with small-scale
    maps (1:100,000).
    Recently, global  positioning systems (GPS), which the
    U.S. military developed, have  helped  to obtain more
    accurate feature locations. GPS is not without error,
    however. The  list below  notes  some  of the possible
    sources of error associated with GPS use, some of
    which can be controlled while others cannot (17):
    • Errors in orbital information.
    • Errors in the satellite clocks.
    • Errors in the receiver clocks.
    • Ionospheric  or tropospheric refraction.
    • Deliberate degrading of the satellite signal.
    • Obstructions that block the signal.
    • Reflection of the GPS signal off buildings, water, or
    • Human error.
    The importance of positional accuracy depends on the
    intended use  of the  data.  In an urban area, a posi-
    tional error of 1  foot  on a  tax map may be unaccept-
    able because  1 foot may be worth millions of dollars.
    In  a  rural  area, however, tax  boundaries mapped
    within 10 feet of their surveyed location may be accu-
    rate enough (6). Somers  (18)  reports that  positional
    accuracy of 10 to 20 feet may be sufficient for envi-
    ronmental  analysis.  She says  the cost of increasing
    accuracy to 5 feet could increase the cost of data  col-
    lection by a factor of  10. The decision-maker must de-
    termine the needed positional accuracy.
    A ttribute A ccuracy
    Attribute accuracy refers to how well the description of
    a characteristic of spatial data matches what actually
    exists on the ground. For some spatial data, the location
    does not change overtime, but the value of the attribute
    does  (e.g.,  the  location of a  census  tract does  not
    change,  but the population  within a  census tract
    changes). Attribute accuracy is  reported differently for
    continuous data (i.e., elevation, which has  an infinite
    number of values) or discrete data (i.e., gender, which
    has a finite number of values).
    NMAS exist for elevation  contour lines on topographic
    maps.  NMAS for vertical accuracy state that not more
    than 10 percent of the points tested shall be  in error by
    more than one-half of the contour interval (16). A well-
    defined point on a USGS topographic map with a 10-foot

    contour interval could vary by 10 feet because the actual
    elevation could be 5 feet higher or lower than the map
    indicates. The implications of these errors are similar to
    the ones for positional accuracy.  In addition, errors in
    elevation are important because small changes in ele-
    vation may significantly affect some CIS analysis opera-
    tions  such  as the  determination  of  aspect,  slope,
    viewshed, and watershed boundaries.
    NMAS do not exist for discrete variables such as land
    use derived from satellite imagery. Instead, a classifica-
    tion matrix reports attribute accuracy. Field checking or
    checking a portion of the classified image against a map
    of higher accuracy determines the accuracy of the land
    use classification. The result of the comparison is a table
    from which to calculate overall, producer's, and user's
    accuracy. Table 4 is an example of a classification accu-
    racy matrix.
    Table 4.  Example of a Classification Accuracy Matrix (19)
                Reference Data ("GrouriH-uth")
                       Number of Cells
    Classified Data
    (Satell ite Imacp)
    Number of Cells
             Overall Accuracy (sum of the main diagonal)
                             = 63%
    Producers Accuracy
    (column total)
    Forest = ^ = 93%
    Users Accuracy
    (row total)
    Forest = -
    = 49%
    Water   =   ^   =   50%
    Urban   =   ^   =   50%
    Water   =   ^
    Urban   =   —
    Overall accuracy is the percentage of correctly classified
    cells calculated as the sum of the main diagonal (19).
    Producer's accuracy is the total number of correct pixels
    in a category divided by the total number of pixels of that
    category as derived from the  reference data (column
    total). It corresponds to how well the person classifying
    the image (the "producer") can  correctly classify or map
    an area on the earth. In this example, the producer most
    accurately classified forested land (93 percent).  User's
    accuracy describes the probability that a sample from
    the classified area actually represents that category on
    the ground. The  map  "user" is concerned  about  the
    map's reliability. In this example, the most accurately
    classified land use from the user's perspective is urban
    (91 percent).
    The significance of overall, producer's, and user's accu-
    racy depends on  the intended use of the data. As an
    example, Chrisman (20)  says that the error in distin-
    guishing wetland from pasture may not matter to some-
    one estimating open space, but the difference is critical
    if the person is estimating the amount of wildlife habitat
    available. Story and Congalton (19) provide an example
    of how  to  interpret a classification matrix. A  forester
    looks at the classification matrix and sees that forest
    classification is 93 percent accurate (producer's ac-
    curacy); therefore, the analyst did not identify only 7
    percent of the forest on the ground. Once the  forester
    field checks the supposed forested area, she finds that
    only 49  percent (28 cells) of the sites mapped as forest
    are actually forest; the rest are water (14 cells) or urban
    (15 cells) areas.
    A report of overall, producer's, and user's  accuracy can
    help decision-makers determine the appropriateness of
    the classified image for their use by identifying potential
    errors in classification. This can help  direct field work,
    which can improve the classification  of the image and
    perhaps subsequent images. Because CIS analysis fre-
    quently  uses land use, decision-makers need to know
    that significant variability  can result when several  ana-
    lysts classify the same image. Bell and Pucherelli (21)
    found that consistency in classification can improve by
    having one person classify the entire image. McGwire (22)
    even found  significant differences between analysts in
    unsupervised classification of Landsat imagery. Com-
    puters  primarily  perform  unsupervised  classification,
    which implies that different analysts would classify the
    same image in the same way.
    Logical Consistency
    Logical  consistency focuses on flaws in the logical  rela-
    tionships among data elements. For example, a vector
    CIS should  label  all polygons with only one label per
    polygon, and  all  polygons should  be closed.  Logical
    inconsistency can also occur by collecting  data layers at
    different times or from different scale maps with different
    positional accuracies. For example, the edge of a  lake
    on  the  hydrology data layer should  coincide with the
    edge of land in the land use data layer. If data on the
    lake were collected  during a wet year  rather than a dry
    year, the lake's volume would be higher than  normal,
    affecting its location on the map. If land use data for the
    same area were collected during a dry year, the bound-
    ary of the lake on  the two  layers would not be the same.
    Logical  inconsistencies usually do not appear until the
    two maps are overlaid and the boundaries do not coin-
    cide (see Figure 2). The user must determine the  "cor-
    rect" location of the feature that appears misaligned on
    one or more data layers. The inconsistency between the

        Land Use
    Figure 2.  Logical inconsistency in lake and forest location.
    location of the two layers resolves through a process
    called  conflation. All  maps are  adjusted  so that the
    feature on each data layer lines up with the same feature
    on the base map.
    Completeness focuses on the adequacy of data collec-
    tion procedures. Robinson  and  Frank (5) discuss two
    kinds of uncertainty associated with collecting spatial
    data that can lead to error. One type of uncertainty is the
    inability to measure or predict an inherently exact char-
    acteristic or event with certainty. Examples of this are
    blunders in data collection or measurement error, nei-
    ther of which can be accurately predicted. The other kind
    of uncertainty is associated with concepts that are inher-
    ently  ambiguous. Crisp data sets,  such as  property
    boundaries, have little ambiguity; the only issue related
    to error  is the positional  accuracy in measuring the
    boundary. Because  land use data are not crisp data
    sets, the challenge is to accurately represent an inher-
    ently inexact concept.
    Although we know spatial data are variable, our classi-
    fication  systems generally  ignore the second type  of
    uncertainty. Analysts map data as though all variables
    had  exact boundaries and all polygons consisted  of
    homogeneous data. Burrough (4) reports that spatial
    variation of natural phenomena is "not just a local  noise
    function or inaccuracy that can be removed by collecting
    more data or by increasing the precision of measure-
    ment, but is often a fundamental aspect of nature that
    occurs at all scales. . .  ."
    Mapping spatial  data is a function of how humans ag-
    gregate and disaggregate data either in space, catego-
    ries, quantities,  or time;  spatial  data seldom exist  in
    nature the way maps depict them (23). Data and rela-
    tionships between data are sensitive to the scale and
    the zoning system in which the data are reported (24,
    25). The modifiable  area unit  problem occurs because
    an analyst can recombine a given set of units  or zones
    into the same total number of units producing very dif-
    ferent results (see Figure 3).
    The scale problem occurs because an analyst can com-
    bine a set of small units into a smaller number of larger
    units, which can change the inferences that can be
                   Figures.  Modifiable area unit. (Number of units is  constant;
                           location of units changes.)
                   made from the data.  In Figure 4, the area containing the
                   highest values changes from the southwest corner in the
                   first picture to the northern half in the second picture.
                   For example, water quality data are scale-dependent
                   because they vary based on the size and location of the
                   collection area (e.g., adjacent to a point source dis-
                   charge, a stream segment, the entire river, or the lake
                   the river discharges  into).
                   Kennedy (25) reports on a similar problem known as the
                   small number problem. This problem occurs when cal-
                   culations use a percentage,  ratio, or rate for a geo-
                   graphic  area for which the  population  of interest
                   (denominator) is sparse or the numerator is a rare event
                   (1 case of cancer  per 1 million people).  CIS-generated
                   maps may highlight a statistically insignificant change in
                   rare events. Small, random fluctuations in the numerator
                   may  cause large fluctuations in the resulting  percent-
                   age,  ratio, or rate.  If  policy-makers use these  maps,
                   priorities for public health policy may change because
                   of the erroneous belief that an area is experiencing more
                   unwanted rare events.
                   Data can be collected using a tag- or count-based sys-
                   tem, which affects their usefulness.  The tag approach
                   categorizes items  based  on the dominant or average
                   attribute and  is  ideal for planners who  want only one
                   value for each area. For example, each polygon in a
                   county soil survey is tagged with one soil type. Accord-
                   ing to soil taxonomy rules, however, only about 35 per-
                   cent of a delimited area on a soil survey must match its
                   classification, and  up to 10 percent may be a  radically
                   different soil (26).  Although the  text in the soil survey
                   sets limits on data accuracy by listing major impurities
                   found with each soil type, the CIS seldom carries that
                   information because analysts only digitize soil  bounda-
                   ries and  label data  with  the dominant attribute. This
                   Figure 4.  Scale problem (number of units changes).

    leads to the depiction of apparently homogeneous soil
    units although the text specifies that the  data are not
    homogeneous (27).
    Some soil or land cover phenomena, even though pre-
    sent in small quantities and thus not mapped, may have
    great  significance for hydrologic models, which makes
    the tag approach to data collection troublesome. Data
    collected using the count system allow the analyst to
    tabulate the frequency of occurrence or areal extent of
    a particular phenomenon. Environmental modelers pre-
    fer count data but are usually forced to use tag data,
    which can  introduce  error into their models (26). The
    new digital soils databases, STATSGO and SSURGO,
    are collected and depicted using a count format, which
    will help experienced analysts use the data more fully.
    Figure 5 shows the difference between tag and count
    methods of data collection.
    Data are seldom complete because analysts use clas-
    sification rules to indicate how homogeneous  an area
    must  be before  it is classified  a particular way (e.g.,
    more than 50 percent, more than 75 percent).  Another
    decision an analyst must make is where to draw the
    boundary between two different areas; it is  seldom clear
    where a forest leaves off and a rural development be-
    gins. Analysts must also decide how or if to show inclu-
    sions  (e.g., a forested area in  the middle of agricultural
    land uses).
    Temporal A ecu racy
    Collecting  data at different times introduces error be-
    cause the variable may have changed since data collec-
    tion. The  effect  of time,  reported as  the date of the
    source material, depends on  the intended use of the
    data.  Some natural resource  data have daily, weekly,
    seasonal,  or annual cycles that are important to con-
    sider.  For example, obtaining  land use data from re-
    motely sensed imagery in November for North Dakota
    produces a very different land use map than data ana-
    lysts obtain during the July growing season.
    In addition, demographic and land  use  information
    changes quickly in a  rapidly urbanizing area. Data  col-
    lected at several times can  produce logical  inconsis-
    tency between data layers, forcing the analyst to adjust
    the location of features to coincide with the base map.
    Another problem with collecting data at different times
                 Soil A
       Soil B
    Soil B 55%'
    Soil C 45%
    Soil A 30%
    Soil B 25%
    Soil C 20%
    Soil D 25%
              Tag                       Count
    Figure 5. Tag and count methods of data collection.
    is that data may be collected using different standards,
    which may not be apparent to the user (4).
    Source Errors in a GIS
    Source (or inherent)  error derives from errors  in data
    collection. The amount of error present in collected data
    is a function of the assumptions, methods, and proce-
    dures used to create the source map (28).  Primary data
    refers to  data  collected  from  field sampling or remote
    sensing. Causes of the errors associated with this data
    are (3, 4, 8, 14,29):
    • Environmental conditions (e.g., temperature, humidity).
    • Sampling system (e.g., incomplete or  biased data
    • Time constraints.
    • Map projection.
    • Map construction techniques.
    • Map design  specifications.
    • Symbolization of data.
    • Natural variability.
    • Imprecision due to vagueness (e.g., classifying a forest).
    • Measurement error from unreliable,  inaccurate, or
      biased observers.
    • Measurement error from unreliable,  inaccurate, or
      biased equipment.
    • Lab errors  (e.g., reproducibility between lab proce-
      dures and between labs).
    The process of converting primary data to secondary
    data (usually a map) introduces additional error. Many
    of the data layers that a GIS analyst acquires are sec-
    ondary data. Some of the errors associated with map-
    making are (3):
    • Error in plotting control points.
    • Compilation  error.
    • Error introduced in drawing.
    • Error due to map generalization.
    • Error in map reproduction.
    • Error in color registration.
    • Deformation of the material (temperature, humidity).
    • Error introduced due to using  a uniform scale.
    • Uncertainty  in the definition of a feature (boundary
      between two land uses).
    • Error due to feature exaggeration.
    • Error in digitization or scanning.

    Converting paper maps to digital data for entry into a
    CIS (tertiary data) introduces still more error (the errors
    generated from converting  paper maps into a digital
    format are discussed in the section on input error), in
    part because the  purpose for which the data was col-
    lected differs from the intended use of the data.
    Many types of error are associated with data collection:
    • Data for the entire area may be incomplete.
    • Data may be collected  and mapped  at inappropriate
    • Data may  not be relevant for the intended application.
    • Data may  not be accessible because  use is restricted.
    • Resolution of the data may not be sufficient.
    • Density of observations may not be  sufficient.
    The following discussion explains these types of errors.
    Data for the Entire Area May Be Incomplete
    An incomplete data record may be due to mechanical
    problems that interrupt recording devices, cloud cover
    or other types of  interference, or financial constraints.
    Possible solutions to this problem include collecting ad-
    ditional data for the incomplete area, using  information
    from a similar area, generalizing  existing  large-scale
    maps to match  the less detailed data needed, or  con-
    verting existing small-scale maps to large-scale maps to
    obtain data  at the desired scale. Collecting additional
    data may not be a feasible solution because of time or
    money constraints. Extrapolating data from the surro-
    gate area to the desired  area can cause problems be-
    cause  the  areas  are  not  identical   and  the  scale,
    accuracy, or resolution  of the surrogate area data
    may be inappropriate for the intended use. The sec-
    tion on analysis/manipulation of data within a  CIS
    covers the  effect of generalization  on data quality
    as well as the effect of converting small-scale maps
    to large-scale maps.
    Data May Be Collected and Mapped at a Scale
    That Is Inappropriate for the Application
    A variety  of guidelines suggest the appropriate  map
    scale to use  for various applications (see Table 5). Also,
    Table 5. Relationship Between Map Scale and Map Use (6)
          Map Scale               Map Use
          1:600 or larger
          1:720to 1:1,200
          1:2,400 to 1:4,800
          1:6,000 and smaller
    Engineering design
    Engineering planning
    General planning
    Regional planning
    some maps and digital databases suggest the type of
    application  for which  they  are  appropriate (e.g.,  the
    STATSGO digital soil database is suitable for state and
    regional planning, whereas SSURGO is suitable for lo-
    cal level planning). Tosta (30) cites an example of com-
    bining wetland data with parcel boundaries to determine
    ownership of the land  containing a wetland. If wetland
    mapping was done to plus or minus 100 feet positional
    accuracy and parcels are 40 feet wide, then the scale of
    the  wetland map is inappropriate for determining  if a
    wetland is located on a specific parcel.
    Identifying the optimal scale of  the necessary data is
    crucial because at some point, the cost of collection and
    storage exceeds  the  benefits of  increasing  the map
    scale.  Lewis  Carroll (1893) summed  up the quest for
    data mapped at an ever larger scale and the problems
    associated with large-scale maps:
        "What do you consider the largest map that would
        be really useful?"
        "About six inches to the mile."
        "Only six inches!" exclaimed Mein Herr. "We very
        soon got to six yards to the mile. Then we tried a
        hundred  yards to the mile.  And then came  the
        grandest idea of all! We actually made a map of the
        country, on the scale of a mile to the mile!"
        "Have you used it much?" I enquired.
        "It  has never been  spread out, yet," said Mein Herr.
        "The farmers objected: they said it would cover the
        whole country, and shut out the sunlight! So now we
        use the country itself,  as its own map, and I assure
        you it does nearly as well."
    Data Collected May Not Be Relevant for the
    Intended Application
    Frequently,  using surrogate data is quicker or cheaper
    than collecting  needed data (e.g.,  Landsat imagery
    rather than  data field collection used to determine land
    use) (4). The accuracy and classification scheme used
    in collecting the data depends on the intended use of
    the  data, which may  not coincide with the  analyst's
    purpose. For instance, soil maps were developed to aid
    farmers in determining what crops they should plant and
    for estimating crop yield. Soil maps, however, see wide
    use for very different  purposes (e.g., hydrologic and
    other environmental models). In addition, STORETdata,
    collected at points, are typically extrapolated to repre-
    sent water quality in an entire stream stretch.
    Data May Not Be Accessible Because
    Use Is Restricted
    An example of restricted data is Census data on individ-
    ual households. An agency may not want to release data
    that reveal the location of endangered species. Another

    example is that people may not even want the informa-
    tion mapped. For example, some cavers do not want to
    reveal the location of caves to the U.S. Forest Service,
    which is charged  under the  federal Cave Resources
    Protection Act with protecting caves, because they think
    the best way to protect  the caves  is to not map them
    (31). The National Park Service is putting the location of
    petroglyphs in the Petroglyph National Monument into a
    CIS. Making their location known to the public, however,
    is  troublesome because this may,  in fact, encourage
    their vandalism (32). Other problems in obtaining data
    include  difficulty in acquisition even if access is  not
    restricted, expensive collection or  input,  or unsuitable
    format (4, 14).
    Resolution of the A vailable Data
    May Not Be Sufficient
    Spatial resolution is the  minimum distance needed  be-
    tween two objects for the equipment to record the  ob-
    jects as two entities; that is, resolution is the smallest
    unit a map represents. To obtain an approximation of a
    map's resolution, divide the denominator of the map
    scale by 2,000 to get resolution in meters; for instance,
    a 1:24,000-scale map has a resolution of approximately
    12 meters (33).
    Resolution relates to accuracy  in  that different map
    scales conform to different accuracy standards. Two air
    photos shot from the same camera at the same distance
    above the ground have the same scale. If one photo has
    finer grain film,  however, smaller details are evident on
    it, and this photo produces a map with higher resolution
    (34). According to Csillag (33), analysts cannot simulta-
    neously optimize attribute accuracy and spatial resolu-
    tion. As spatial resolution increases, attribute complexity
    increases (35). Also, the finer the spatial resolution, the
    greater  the probability that random error significantly
    affects a data value.
    Resolution of the data is not necessarily  the same as
    the size of a raster cell  in a database. Statistical sam-
    pling theory suggests using a  raster cell size that is half
    the length (one-fourth of the area) of the smallest feature
    an analyst wishes to record. Raster data have a fixed
    spatial resolution that depends on the  size of the  cell
    employed, but a CIS analyst can divide  or aggregate
    cells to  achieve a different cell size.  Frequently, an
    analyst transforms data collected at one level of resolu-
    tion to a higher level  of resolution than existed in  the
    original  source  material. According to Everett  and Si-
    monett (23), "Geographic analysis,  however, can be no
    better than that of the smallest  bit of data which  the
    system is capable  of detecting." Vector data are limited
    by the resolution of input/output devices,  limits  on data
    storage,  and the accuracy of the digitized location for
    individual points (36). The spatial  resolution of the data-
    base and the  processes that operate on it should be
    reduced to a level consistent with the data's accuracy.
    The spatial resolution needed depends on the intended
    use of the data, cost, and data storage considerations.
    As resolution increases, so does the cost of collection
    and storage.  Resolution  sufficient to detect an object
    means that an analyst can reveal the presence of some-
    thing. Identification, the ability to identify the object or
    feature,  requires three times the  spatial resolution of
    detection. Analysis, a  finer level  of identification, re-
    quires 10 to 100 times the resolution that identification
    needs (23). Increasing resolution increases the  amount
    of data for storage, with storage requirements increas-
    ing  by the square of the resolution of  the data. For
    example, if the resolution of the data needs to  change
    from 10-meter to 1-meter pixels, file size increases by 102
    or 100 times (14).
    Density of Observations May Be Insufficient
    The density of observations serves as a general indica-
    tor of data reliability (4). Users need to know if sampling
    was done at the optimum density to resolve the  pattern.
    Burrough determined that boulder clay in The  Nether-
    lands could be resolved by sampling at 20-meter inter-
    vals or less, whereas coversand showed little variation
    in sampling from 20- to 200-meter intervals.
    Some strategies for reducing data collection errors are to:
    • Adhere to professional standards
    • Allocate enough time and money
    • Use a  rigorous sampling design
    • Standardize data collection procedures
    • Document data collection procedures
    • Calibrate data collection instruments
    • Use more accurate  instruments
    • Perform blunder checks to detect gross errors
    Documenting data collection procedures and distribut-
    ing them along with data allows potential users to deter-
    mine if the data are suitable for their purposes. By not
    documenting  procedures, errors in the source material
    are essentially "lost" by inputting the data to a CIS, and
    the errors become largely undetectable in subsequent
    CIS procedures. The result is that agencies that make
    decisions based on the CIS-generated map assume the
    source data are accurate, only to discover later  that the
    map contains substantial errors in part due to errors in
    the source material.
    Operational Errors in a GIS
    Data input, storage, analysis/manipulation,  and output
    can introduce operational errors.  Digital maps, unlike

    paper maps,  can accumulate  new  operational errors
    through CIS operations (8). Even if the input data were
    totally error-free, which the last section demonstrated is
    not the case, CIS operations can produce positional and
    attribute errors. The CIS operation itself determines to
    a large extent the types of errors that result.
    Input Errors
    The  process of inputting  spatial and attribute data can
    introduce error. The  major sources  of input error are
    manual entry  of attribute  features and scanning or dig-
    itizing spatial  features. Manual entry errors include in-
    complete entry of attribute  data, entering the wrong
    attribute data, or entering the right attribute data at the
    wrong location. Digitizing errors originate from equip-
    ment, personnel, or the source  material (see Table 6).
    Digitizing errors, such as under- and overshoot of lines
    and  polygons that are not closed, can  introduce  error
    (see  Figure 6). CIS software  can "snap" lines together
    that  really do  not connect. Depending on  the tolerance
    Table 6.  Types of Digitizing  Errors (4, 14, 37)
    Personnel Errors
       Changes in the origin
       Incorrect registration of the map on the digitizing table
       Creation of over- and undershoots
       Creation of polygons that are not  closed
       Incomplete spatial data when data are not entered
       Duplication of spatial data when lines are digitized twice
       Line-following error (inability to trace map lines perfectly with the
       Line-sampling error (selection of points used to represent the map)
       Physiological error (involuntary muscle spasms)
    Equipment Errors
       Digitizing table (center  has higher  positional accuracy than the
       Resolution of the digitizer
       Differential accuracy depending on cursor orientation
    Errors in Source Material
       Distortion because source maps have not been scale-corrected
       Distortion due to changes in temperature and humidity
       Necessity of digitizing sharp boundary lines when they are gradual
       Width of map  boundaries (0.4 mm) digitized with a 0.02-mm accu-
       racy digitizer
          Undershoot       Overshoot         Polygon Not Closed
    Figure 6.  Common digitizing errors.
    selected, this can result in the movement of both lines,
    which can decrease the accuracy of the resultant map.
    Despite the long  list of personnel errors associated with
    digitizing, a good operator probably contributes the least
    error in the  entire digitizing process  (38). Giovachino
    discusses methods that can help determine equipment
    accuracy, including checking the  repeatability, stability,
    and effect of cursor rotation.  Digitizing accuracy var-
    ies based on the width, complexity, and density of the
    feature being digitized but typically varies from 0.01
    to 0.003 (3).
    One problem with digitized data is that the data  can
    imply a false sense of  precision.  Boundaries on paper
    maps are frequently 0.4 mm wide but are digitized with
    0.02-mm accuracy. The result is that the lines are stored
    with 0.02-mm accuracy, implying a level of precision that
    far exceeds the original data.
    Minimizing digitizing errors is  important  because  the
    errors can affect subsequent CIS analysis.  Campbell
    and Mortenson (39)  provide a list of  procedures they
    used to  reduce  errors  associated  with digitizing and
    • Use log  sheets to  ensure consistency and  account-
      ability, and to provide documentation.
    • Check for completeness in  digitizing  all  lines and
    • Check for complete and  accurate polygon labeling.
    • Set an acceptable RMSE term for digitizing (usually
    • Always  overshoot  rather than  undershoot  when
    • Overlay  a plot of the digitized  data with the  source
      map to check lines and polygons. If light passes be-
      tween the digitized  line  segment and  source map,
      redigitize it.
    • Check digitized work immediately  to provide feed-
      back to the digitizer operator and to  help identify and
      correct systematic errors.
    • Limit digitizing  to less than 4 hours  a day.
    • Involve people in doing CIS-related jobs other than
      digitizing to decrease turnover and increase the level
      of experience.
    Storage Errors
    Data storage in a CIS  usually involves two main types
    of errors.  First,  many  CIS systems  have  insufficient
    numerical precision,  which can introduce error  due to
    rounding. Integers are  stored as 16 or 32 bits, which
    have four significant figures. Real numbers are stored
    as floating point numbers either in single precision  (32 bit,

    7 significant figures) or double precision (64 bit, 15 or 16
    significant figures).  If the data in  a CIS  range from
    fractions of a meter to full  UTM coordinates, typical
    32-bit CIS systems cannot store all the numbers. Using
    double precision (64 bits) reduces this problem but in-
    creases storage requirements.
    Second,  CIS  processing and storage  usually ignore
    significant digits (data precision). As a result, the preci-
    sion of CIS processing frequently exceeds the accuracy
    of the data (40). When a CIS converts a temperature
    recorded and entered as 70 degrees Fahrenheit (near-
    est degree) to centigrade, the CIS stores the tempera-
    ture as 21.111 degrees rather than 21 degrees, which
    the significant figures in the original temperature meas-
    urement would dictate.  Using the accuracy of the data,
    not the precision  of floating  point  arithmetic, partially
    resolves  this but requires the user to make a special
    effort  because  the  CIS does not  automatically  track
    significant figures.
    Analysis/Manipulation Errors
    CIS analysis/manipulation functions, designed to trans-
    form or combine data sets, also can introduce errors.
    These errors  originate from  the measurement scale
    used  or during data conversion (vector to raster and
    rasterto vector), map overlay, generalization, converting
    small-scale to large-scale maps, slope, viewshed, and
    other  analysis functions. One of the  biggest problems
    associated with CIS use is that data in digital  form are
    subject to different  uses than data  in paper form be-
    cause the user has access to multiple data layers.
    Measurement Scale
    Four measurement scales can depict spatial data: nomi-
    nal, ordinal, interval, or ratio scales. A name or letter
    describes nominal data (e.g.,  land use type, hydrologic
    soil group C). Performing mathematical operations such
    as addition and subtraction on nominal data is meaning-
    less. Ordinal or ranked data have an order to them such
    as low, medium, and high. Interval data have  a known
    distance between the intervals such as 0, 1 to 5, 6 to 9,
    more  than 9. Ratio  data  are similar to interval data
    except ratio data have a meaningful zero (e.g., tempera-
    ture on the Kelvin scale).
    Often  during  CIS  operations, analysts convert interval
    or ratio data  into nominal data (e.g., low slope is 0 to
    3 percent, medium slope is 4 to 10 percent), resulting in
    a loss of information. Analysts should preserve the origi-
    nal slope values in the CIS in case the user later wants
    to modify the  classification  scheme.  Robinson  and
    Frank (5) describe the tradeoff between information con-
    tent and the meaning that can be derived from it, which
    partly helps explain why interval data are frequently
    converted to nominal data. The authors identify a con-
    tinuum progressing from nominal data at one end that is
    highly subjective, has low information content, and high
    meaning (low slope means something to the average
    user) to ratio data that has low subjectivity, high infor-
    mation content, and low meaning (a slope of 7 percent
    may not mean much to the average user).
    Data Conversion
    Errors can occur in converting a vector map to a raster
    map or a raster map to a vector  map.  For instance,
    remotely sensed data are collected using a raster-based
    system. Using a vector CIS, however, requires conver-
    sion from raster to  vector data. The size of the error
    depends on the conversion algorithm, complexity of fea-
    tures, and grid cell size and orientation (13).
    A line on a vector map converted to  a raster map has
    lower accuracy in  the raster representation because
    vector data structures store data  more accurately than
    raster ones. When polygons in a vector CIS are con-
    verted  to a  raster  CIS, the coding rule usually used
    assigns the value that covers the largest area within the
    cell of a categorical map to the entire cell (see Figure 7).
    For example, when placing a grid over a vector map with
    an urban land polygon adjacent to an agricultural poly-
    gon, the  cell placement can include part of both poly-
    gons. If the  resultant cell comprises  51 percent urban
    and 49 percent agricultural land, the cell is assigned 100
    percent urban.  Converting  a  numerical map between
    raster and vector systems requires spatial interpolation
    procedures.  CIS software packages use different inter-
    polation methods that can  produce a different  output
    even when using the same input data.
                Vector              Raster
    Figure 7.  Polygon conversion from vector to raster data.
    Map  Overlay
    Map overlay, used extensively in planning and natural
    resource management, is the combining of two or more
    data layers to create new information.  In a vector CIS,
    slivers or spurious polygons can result from overlaying
    two data layers to produce a  new map (slivers cannot
    be formed in a raster-based CIS). When combining the
    data layers, lines do not coincide, resulting in the crea-
    tion of a new polygon or sliver that did not exist on either
    layer (see Figure 8). Unfortunately, as accuracy in digit-
    izing  increases, so does the number of slivers (41).
    Positional error in  the boundaries can occur because of

    Figure 8. Sliver example.
    mistakes in measuring or converting the data to digital
    form, incremental expansion or recession of a real world
    boundary over time, or the fact that certain boundaries
    are  difficult  to  determine and  thus are generalized
    differently (42).
    The  number of map layers, accuracy of each map layer,
    and the coincidence of errors at the same position from
    several map layers all  determine the accuracy of the
    map overlay procedures (43). Using probability theory,
    Newcomer and  Szajgin determined that the highest ac-
    curacy to expect from  a map overlay  is equal to the
    accuracy of the least accurate map layer. The  lowest
    accuracy in  map overlay occurs when errors in each
    map occur at unique  points.
    In the quest for more accurate  results,  CIS modelers
    have increased the  complexity  of  their models and
    therefore have  increased  the number of data  layers
    needed.  Guptill (44) states,  "Conventional  wisdom
    would say that as you add more data to the solution of
    a problem, the likelihood of getting an accurate solution
    increases. However,  if each additional data layer de-
    grades the quality of the combined data set,  and hence
    the accuracy of the solution, then additional data sets
    may be counterproductive."
    Monmonier  (45) provides  an extensive discussion of
    geometric and content generalization procedures used
    in map-making.  Table 7 lists common types of generali-
    zation. Generalizing data on a map  helps to focus the
    user's attention  on  one or two types  of information and
    to filter out irrelevant details. Generalizing is performed
    by reducing the  scale of the data; a 1:24,000-scale map
    can be generalized to a 1:100,000-scale map so that all
    data layers  have the same scale. With generalizing,
    areas on  a large-scale map become point  or line
    features on  a small-scale map (35). Obtaining some
    measurements  from small-scale  maps, however, re-
    quires caution.  For  example, a map may depict  a
    40-foot wide road as  a single line one-fiftieth of an inch
    wide. On a  1:100,000  map, one-fiftieth  of an inch
    translates into  a 160-foot wide  road—four times the
    actual width  of the  road.
    Several studies have pointed to  errors that can result
    from generalization.  Wehde  (46) compared soil maps
    generated from 0.017-acre grid cells and 11 progres-
    sively increasing grid cell sizes. He  found that as grid
    Table?.   Generalization Operations (45)
                   Geometric Generalization
    Generalizing a Line
    Generalizing a Point
       Graphic association
       Area conversion
    Generalizing an Area
       Point conversion
       Line conversion
                    Content Generalization
    cell size increased, map accuracy decreased. More re-
    cently, Stoms (47) found that generalizing a habitat map
    from  1 to 25, 100, and 500  hectares decreased the
    number of habitat types and the number of species
    Transforming Small-Scale Maps to
    Large-Scale Maps
    Converting small-scale maps (1:250,000) to large-scale
    maps (1:24,000) is advisable only if the analyst fully
    appreciates the effect  of this procedure on map quality.
    Data mapped at a small scale are subject to different
    accuracy standards than data mapped at a large scale.
    Connin  (48)  reports,  "Problems with accuracy arise
    when positions are reported to decimal parts of a foot or
    meter, but the method of data capture may cause the
    positional error to be  as much as hundreds  of feet or
    meters."  Yet when converting the data from small- to
    large-scale, the data appear to have the accuracy of the
    large-scale map. Theoretically, data should not be trans-
    formed and used at a scale larger than the scale of the
    document from which the data are derived (3).

    Slope andMewshed
    CIS software packages use a variety of algorithms to
    calculate slope and viewsheds and  can produce very
    different results. Algorithms are an unambiguous set of
    rules or a finite sequence of operations used to carry out
    a procedure. Smith,  Prisley, and  Weih (49) used six
    different CIS algorithms to determine slope on  5,905
    acres of land in order to calculate the amount of land
    deemed unsuitable for timber harvesting. They  found
    that unsuitable land  varied from 175 to 1,735 acres,
    indicating that different algorithms produce very different
    results. Felleman and Griffin (50) found  that CIS pack-
    ages with different algorithms generate  alternate view-
    sheds (the area that can be seen from a point).
    Output Errors
    A variety of errors are associated with data output:
    • Output devices create error.
    • Paper shrinks and  swells.
    • Line implies  certainty that  may not  exist because
      boundaries are gradual.
    • A cell or polygon implies homogeneity.
    • Scale can be modified to imply higher accuracy than
      exists in the source data.
    • Precision can be modified to  imply higher precision
      than exists in the source data.
    • Depiction  of symbols and colors may  not  follow
    An important problem associated  with  CIS-generated
    maps is that users make informal assessments  about
    data quality, partially based on how they perceive the
    quality of the  output. A hand-drawn map connotes a
    lower level of accuracy than a five-color, CIS-produced
    map complete with scale  and  agency logo. Another
    problem with output is that distinguishing highly  accu-
    rate data from  less accurate data is impossible on a
    CIS-generated map. Users want the  output from  a CIS
    to  look like  maps they usually  see,  perpetuating the
    notion that lines mark exact boundaries and that poly-
    gons or cells are homogeneous. Maps that federal map-
    ping  agencies  produce  frequently follow NMAS,  but
    CIS-generated maps seldom adhere to  published map
    accuracy standards. An agency  could require that CIS
    map products meet NMAS, which  would establish and
    maintain data standards from data  collection to output.
    A pen stroke of one-fiftieth of an inch on an output device
    translates to an error of 40 feet on the ground for a
    1:24,000-scale map (6). Small changes  in paper maps
    due to changes in temperature and humidity can  repre-
    sent several feet on  the ground. As previously noted,
    analysts can modify the scale of CIS maps to whatever
    they desire. The basic rule of informational integrity is
    that  the  implied precision of data output should  not
    exceed the precision (spatial, temporal, or mathemati-
    cal) of the least precise input variable (26).
    CIS-generated maps probably do not differ significantly
    from paper maps in their implication that lines and poly-
    gons on the map represent certainty and homogeneity.
    CIS-generated maps, however, may not depict standard
    symbols, sizes, shapes, colors, and orientation. For ex-
    ample, paper geological maps use dashed lines to show
    inferred, rather than actual, field collected data, but geo-
    logical maps in a CIS may not follow the same conven-
    tion (27). Cartographers conventionally use blue lines to
    indicate water, but a CIS map-maker can show water as
    red rather than blue.
    Even more troublesome  are the  color schemes that
    some analysts use in depicting model output. Analysts
    often give little thought to assigning the colors to model
    results depicted as ordinal rankings. For example, areas
    of high erosion might be blue, medium erosion might be
    red,  and  low erosion might be green. This selection of
    colors ignores the intuitive meaning that people assign
    to colors. It has been suggested that the color ordering
    used in stop lights might provide a better option. In that
    case, areas of high erosion would be red, medium ero-
    sion would be yellow, and low erosion would be green.
    Error Reduction Techniques
    Although CIS users and researchers develop error re-
    duction strategies, ultimately users must rely on CIS
    software  developers to  implement new error reduction
    techniques in CIS packages. Error reduction techniques
    range from simple software  warnings to prohibiting a
    user from performing selected CIS  procedures. Dutton
    (51)  predicts  that future CIS programs  will automate
    data manipulation (i.e.,  size, format,  and placement
    of feature labels  on  maps)  in  keeping with  standard
    cartographic principles.  Dutton (51) and Beard (8) also
    predict that future CIS packages will enforce metadata-
    based constraints such as operations that are illegal or
    illogical (e.g., determining the average value of nominal
    data such as  land use), or are inadvisable (e.g., over-
    laying maps with widely different scales).
    Another change Dutton anticipates is that software ven-
    dors will  include information  in manuals that explains
    how executing a specific command may affect the data-
    base. Graphic techniques to  depict error are being  de-
    veloped for nonexpert users  while experts tend to use
    spatial statistics. Felleman (52) and Berry (53) present
    an interesting graphic portrayal of an error map that may
    indicate the future of error maps. Additional  research
    must determine what effect errors will have on decision-

    Error Management
    Ultimately, the decision-maker must determine what to
    do with the information in this paper. A decision-maker
    has a variety of possible courses of action, ranging from
    prudent steps that attempt to minimize  error and  the
    effect it has on decisions, to other less useful options.
    Possible actions are to:
    • Abandon use of a CIS.
    • Ignore the error associated with CIS use.
    • Attempt to collect "error-free" data.
    • Determine  if the data are accurate enough for  the
      intended purpose.
    • Develop and use data quality procedures.
    • Obtain and  use an  error report with CIS-generated
    • Ask that CIS-generated maps show potential errors.
    • Continually educate users about the appropriate use
      of spatial data.
    First, the decision-maker could abandon any attempt to
    use a CIS because of the errors associated with its use.
    At times, this may be the appropriate strategy, but this
    approach ignores the potential benefits associated with
    CIS  use.
    Second, the decision-maker could ignore the error associ-
    ated with CIS use and continue to  use  the CIS for deci-
    sion-making. This type of "head in the sand" approach is
    not advisable because of the potential liability associated
    with  making decisions  based on inaccurate data.
    Third, the decision-maker could engage in an expensive
    and time-consuming effort to collect highly  accurate er-
    ror in hopes that error becomes a nonissue. Depending
    on the intended  use of the data, the cost  of collecting
    more accurate data may exceed the benefit.
    Fourth, the decision-maker could  assess  whether  the
    information available  is accurate enough for the  in-
    tended purpose. If data quality is too low, the decision-
    maker may opt to collect new data  at the desired quality.
    If collecting additional data is not possible, the decision-
    maker can explore what types of decisions  are possible
    given the attainable data quality.  For  instance, Hunter
    and Goodchild (54) found that the  data they were  using
    were suitable  only for initial screening rather than  for
    regulatory and land-purchasing decisions.
    Fifth, procedures to ensure high quality  data could be
    developed and used in the data collection, input, and
    manipulation stages of building a CIS database.
    Sixth, the decision-maker could require a quantitative or
    at least a qualitative report on the  sources, magnitude,
    and  effects of errors.  The absence of an  error report
    does not mean the map is error-free (36). Dutton (51)
    predicts that in the near future users of geographic data
    will demand error reports,  confidence limits, and sensi-
    tivity analyses with CIS-generated output.
    Seventh, the decision-maker could ask for CIS-generated
    maps that adequately portray the error in the final map. For
    example, areas where the uncertainty is high could appear
    in red on maps. Another option is to place a buffer around
    lines to indicate the relative  positional accuracy of a line or
    to show transition zones. Finally, an analyst can present
    the output in ways other than a  dichotomous yes or no;
    instead, the analyst may use yes, maybe, or no depictions
    or even more gradations.
    Finally,  Beard (8)  introduced  the concept of  directing
    efforts  toward educating  users about use error.  She
    defines use error as the misinterpretation of  maps or
    misapplication of maps to  tasks for which they are not
    appropriate. "We can't assume that CIS  will  automat-
    ically  be less susceptible to  misuse than traditional
    maps,  and it may, in fact, exacerbate the problem by
    expanding  access to  mapped information." Beard ar-
    gues that money directed to reducing source and opera-
    tional error, while important, may not matter if use error
    is large.
    CIS is  a powerful tool for analyzing spatial data. Every-
    one who uses CIS-generated output, however, must be
    aware  of  source errors and  operational errors  intro-
    duced during data input, storage, analysis/manipulation,
    and output. Increased awareness  of the sources and
    magnitude of error can help decision-makers determine
    if data are appropriate for their use. Decision-makers
    cannot leave data quality concerns to CIS analysts be-
    cause  efforts to improve  data quality are not without
    cost, and the decision-makers typically control funding.
    Decision-makers must not  get caught up in the glamour
    of the  spatial analyses and  outputs that a  CIS can
    produce. These attributes  may lead decision-makers to
    ignore  issues associated with uncertainty, error, accu-
    racy, and precision. Inexpensive digital data can make
    analysts and decision-makers ignore data quality. If sub-
    sequent management decisions are made based on
    poor quality data, the resultant  decisions  may turn out
    wrong. This would give decision-makers a jaded view of
    the usefulness of CIS. An adequate  understanding of
    data quality issues can  help decision-makers ask the
    right questions of analysts and avoid  making decisions
    that are inappropriate given the  data quality.
     1. Walsh, S.J., D.R. Lightfoot,  and  D.R. Butler. 1987. Recognition
       and  assessment of error in geographic information systems. Pho-
       togrammetric Eng. and Remote Sensing 53(10):1,423-1,430.

     2. Openshaw, S. 1989. Learning to live with errors in spatial data-
        bases.  In: Goodchild, M., and S. Gopal,  eds. The accuracy of
        spatial databases. Bristol, PA: Taylor and  Francis, pp. 263-276.
     3. Thapa, K., and J. Bossier. 1992. Accuracy of spatial data used
        in geographic information systems.  Photogrammetric Eng.  and
        Remote Sensing  58(6):835-841.
     4. Burrough, PA. 1986. Principles of geographical information sys-
        tems for land resources assessment. New York, NY: Oxford Uni-
        versity Press.
     5. Robinson, V.B., and A.U. Frank. 1985. About different  kinds of
        uncertainty in collections of spatial data. In: Proceedings of Digital
        Representations of Spatial Knowledge, Washington,  DC, Auto-
        Carto 7. Falls Church, VA: ASPRS and ACSM. pp. 440-449.
     6. Nale, O.K. 1992. Do traditional map standards conflict with a  CIS
        landbase? CIS World 5(7):50-53.
     7. Smith, G.R., and  D.M. Honeycutt. 1987. Geographic data uncer-
        tainty, decision-making, and the value of information. In: Proceed-
        ings of CIS'87, San Francisco, CA (October 26-30). Falls Church,
        VA: ASPRS and ACSM. pp. 300-312.
     8. Beard,  K. 1989.  Dimensions of use and value  of geographic
        information. In: Onsrud, H., H.W  Calkins, and N.J. Obermeyer,
        eds. In: Proceedings of use and value of geographic information:
        Initiative four specialist meeting, Tenants Harbor, ME  (August).
        Vol. 89-7. Santa Barbara, CA: NCGIA.
     9. National  Committee for  Digital  Cartographic Data Standards.
        1988. The proposed standards for digital cartographic data. Amer.
        Cartographer 15(1):11-140.
    10. Chrisman, N.  1994. A vision of digital libraries for geographic
        information, or how I stopped trying to find the on-ramps to the
        information superhighway. Geo. Info. Sys. 4(4):21-24.
    11. Eagan, P.O., and S.J. Ventura. 1993.  Enhancing value of envi-
        ronmental data: Data lineage reporting. J. Environ. Eng. 119(1):5-
    12. U.S.  EPA. 1990. Information resources management policy man-
        ual. Washington,  DC.
    13. Lunetta, R.S., R.C. Congalton, L.K.  Fenstermaker, J.R. Jenson,
        K.C. McGwire, and L.R. Tinney. 1991. Remote sensing and geo-
        graphic information system data integration:  Error sources  and
        research  issues.  Photogrammetric  Eng.  and  Remote  Sensing
    14. Aronoff, S. 1989.  Geographic information systems: A manage-
        ment perspective. Ottawa, Ontario: WDL Publications.
    15. ASPRS. 1990. ASPRS accuracy standards for large-scale maps.
        Photogrammetric Eng. and Remote  Sensing 56(7):1,068-1,070.
    16. Thompson, M.M.  1987. Maps for America: Cartographic products
        of the U.S. Geological Survey and others, 3rd ed. Washington,
        DC: U.S. Geological Survey.
    17. Ferguson, J. 1993. What is it we  obtain from GPS? Professional
        Surveyor 13(4):3-4.
    18. Somers, R. 1994. Is a CIS department the answer?  Geo. Info.
        Sys.  4(6):29-32.
    19. Story, M.,  and R.G. Congalton. 1986.  Accuracy  assessment: A
        user's perspective. Photogrammetric Eng. and Remote  Sensing
    20. Chrisman, N.R. 1987. The accuracy of map overlays: A reassess-
        ment. Landscape and Urban Planning  14:427-439.
    21. Bell, A.D., and M.J. Pucherelli. 1992. Audubon  Lake island ero-
        sion study using aerial photography and geographic information
        systems.  Report No. R-92-12. Denver, CO:  U.S. Department of
        the Interior, Bureau of Reclamation.
    22. McGwire, K.C. 1992. Analyst variability  in labeling  of unsuper-
        vised classifications. Photogrammetric Eng. and  Remote Sensing
    23. Everett, J.,  and D.S. Simonett. 1976. Principles, concepts, and
        philosophical problems in remote sensing. In: Lintz,  J., and D.S.
        Simonett, eds. Remote sensing of the environment. Reading, MA:
    24. Fortheringham, A.S. 1989. Scale-independent spatial analysis. In:
        Goodchild, M., and S. Gopal, eds. The accuracy of spatial data-
        bases. Bristol, PA: Taylor and Francis, pp. 221-228.
    25. Kennedy,  S. 1989. The small number problem and the accuracy
        of spatial  databases. In: Goodchild, M.,  and S.  Gopal, eds. The
        accuracy  of spatial databases. Bristol, PA: Taylor and Francis, pp.
    26. Brown, D.A., and PJ. Gersmehl. 1987. Maintaining relational ac-
        curacy of  geocoded data in environmental modeling.  In: Proceed-
        ings of CIS '87, San Francisco, CA (October 26-30). Falls Church,
        VA: ASPRS and ACSM. pp. 266-275.
    27. Fisher, PF.  1987. The nature of soil data in GIS—error or  uncer-
        tainty. In:  Proceedings of the International Geographic Informa-
        tion Systems (IGIS) Symposium, Arlington, VA (November 15-18).
        Vol. Ill, pp. 307-318. Arlington, VA:  Association  of American Ge-
    28. Walsh,  S.J. 1989. User considerations in  landscape charac-
        terization. In: Goodchild, M., and S. Gopal, eds. The accuracy of
        spatial databases. Bristol, PA: Taylor and Francis, pp. 35-43.
    29. Stoms, D. 1987. Reasoning with uncertainty in intelligent geo-
        graphic information systems. In: Proceedings  of GIS '87, San
        Francisco, CA (October 26-30). Falls Church, VA: ASPRS and
        ACSM. pp.  693-700.
    30. Tosta. 1993. The data wars:  Part II. Geo. Info. Sys.  3(2):22-26.
    31. Anonymous. 1994. Between a rock and a hard place. Backpacker
    32. Fletcher, M., and D. Sanchez. 1994. Etched in stone: Recovering
        Native American rock art. GPS World 5(10):20-29.
    33. Csillag, F. 1991. Resolution revisited. In: 1991 ACSM-ASPRS
        Annual Convention, Baltimore, MD (March 25-28). Auto-Carto 10,
        Volume 6. Bethesda, MD: ACSM and ASPRS. pp. 15-28.
    34. Star, J.L.,  and  J.E. Estes. 1990. Geographic information systems:
        An introduction. Englewood Cliffs, NJ: Prentice  Hall.
    35. Goodchild, M.  1993. The state of GIS for environmental problem-
        solving. In: Goodchild, M.R.,  B.O. Parks, and L.T Steyaert, eds.
        Environmental modeling with GIS. New York, NY: Oxford Univer-
        sity Press, pp. 8-15.
    36. Veregin, H. 1987. Error modeling in geographic information sys-
        tems: A review. Unpublished manuscript.
    37. Keefer, B.J., J.L. Smith, and T.G. Gregoire.  1991. Modeling and
        evaluating the effects of stream mode digitizing errors on map
        variables. Photogrammetric Eng. and Remote Sensing 57(7):957-
    38. Giovachino, D. 1993. How to determine the accuracy of your
        graphic digitizer. Geo. Info. Sys. 3(3):50-53.

    39. Campbell, W.G., and D.C. Mortenson. 1989. Ensuring the quality
        of geographic information system data: A practical application of
        quality  control.  Photogrammetric  Eng.  and  Remote  Sensing
    40. Goodchild, M., and S.  Gopal. 1989. Preface. In: Goodchild, M.,
        and S. Gopal, eds. The  accuracy of spatial databases. Bristol,
        PA: Taylor and Francis.
    41 . Goodchild, M. 1 993. Data models and data quality: Problems and
        prospects. In: Goodchild, M.F., B.O. Parks, and L.T Steyaert, eds.
        Environmental modeling with GIS. New York, NY: Oxford Univer-
        sity Press, pp. 94-103.
    42. Lester, M., and N. Chrisman. 1991.  Not all slivers are skinny: A
        comparison of two methods for detecting positional error in cate-
        gorical maps. In: Proceedings of GIS/LIS '91, Atlanta, GA (Octo-
        ber 28-November 1). Bethesda, MD: ASPRS. pp. 648-658.
    43. Newcomer,  J.A., and J. Szajgin. 1984. Accumulation of thematic
        map errors  in  digital overlay analysis.  Amer. Cartographer
    44. Guptill, S.C. 1989. Inclusion of accuracy data in a feature-based,
        object-oriented data model. In: Goodchild, M., and S. Gopal, eds.
        The accuracy of spatial databases. Bristol, PA: Taylor and  Fran-
        cis. pp. 91-98.
    45. Monmonier, M. 1991. How to  lie with maps. Chicago, IL: Univer-
        sity of Chicago Press.
    46. Wehde, M. 1982. Grid cell size in relation to errors in maps and
        inventories produced by  computerized map processing. Photo-
        grammetric Eng. and Remote Sensing 48(8): 1, 289-1 ,298.
    47. Stoms, D.M. 1992. Effects of habitat map generalization in bio-
        diversity assessment. Photogrammetric Eng. and Remote Sens-
        ing 58(11):1,587-1,591.
    48. Connin, R.  1994. The trouble  with data. Professional Surveyor
    49. Smith,  J.L., S.P. Prisley, and R.C. Weih. 1991. Considering the
        effect of spatial data variability on the outcomes of forest man-
        agement decisions. In: Proceedings of GIS/LIS '91, Atlanta, GA
        (October 28-November 1). Bethesda, MD: ASPRS. pp. 286-292.
    50. Felleman, J.,  and C.B.  Griffin. 1990.  The  role of  error  in CIS-
        based viewshed determination: A problem analysis. Report No.
        IEPP-90-2.  Syracuse, NY: Institute of Environmental Policy and
        Planning, State University of New York, College of Environmental
        Science and Forestry.
    51. Dutton, G. 1991. Improving spatial analysis in GIS environments.
        In: 1991 ACSM-ASPRS  Annual  Convention,  Baltimore, MD
        (March 25-28). Auto-Carto 10,  Volume 6. Bethesda, MD: ACSM
        and ASPRS. pp. 168-185.
    52. Felleman, J. 1992. Letter to the editor. Geo. Info. Sys. 2(8):12-13.
    53. Berry, J.K. 1994.  The this, that, there rule.  GIS World 7(7):22.
    54. Hunter, G.J., and M.F. Goodchild. 1993. Managing  uncertainty in
        spatial  databases: Putting theory  into practice. In: Proceedings
        of the  annual meeting of the  Urban and  Regional Information
        Systems Association, Atlanta, GA (July 25-29). Washington, DC:
        Urban and Regional Information Systems Association, pp. 1-15.

                 Expedition of Water-Surface-Profile Computations Using GIS
                       Ralph J. Haefner, K. Scott Jackson, and James M. Sherwood
                   Water Resources Division, U.S. Geological Survey, Columbus, Ohio
    Water-surface profiles computed by use of a step-back-
    water model such as Water Surface PROfile (WSPRO)
    are frequently used in insurance studies, highway de-
    sign,  and development planning  to  delineate flood
    boundaries. The WSPRO model requires input of hori-
    zontal and vertical  coordinate data that define cross-
    sectional  river-channel  geometry.  Cross-sectional  and
    other hydraulic data are manually coded into the WSPRO
    model, a  labor-intensive procedure. For each cross sec-
    tion, output from the model assists in approximating the
    flood boundaries and high-water elevations of floods with
    specific recurrence intervals  (for example,  100-year or
    500-year). The flood-boundary locations along a series of
    cross sections are connected to delineate the flood-prone
    areas for the selected recurrence intervals.
    To expedite  the data collection and  coding  tasks re-
    quired for modeling, the geographic information system
    (GIS), ARC/INFO, was used to manipulate and process
    digital data  supplied in  AutoCAD drawing  interchange
    file (DXF) format. The  DXF files, which were derived
    from aerial photographs, included 2-foot elevation data
    along topographic contours with  +0.5-foot resolution
    and the  outlines of  stream  channels.  Cross-section
    lines, located according to standard step-backwater cri-
    teria, were digitized across the valleys. A three-dimen-
    sional surface was generated from the 2-foot contours
    by use of the GIS software,  and the digitized section
    lines were overlain on this surface. GIS calculated the
    intersections of contour lines  and cross-section lines,
    which provided most of the required cross-sectional ge-
    ometry data for input to  the WSPRO model.
    Most of the data collection and coding processes were
    automated,  significantly reducing  labor costs and hu-
    man error. In addition, maps  at various scales can  be
    easily produced  as needed after digitizing the flood-
    prone areas  from the WSPRO model into GIS.
    Introduction and Problem Statement
    Losses due to flood  damage generally cost the Ameri-
    can public hundreds of millions of dollars  annually. In
    1968, the National Flood Insurance Act established the
    National Flood Insurance Program (NFIP) to  help re-
    duce the cost to the  public and provide a framework to
    help reduce future losses. The Federal Emergency Man-
    agement Agency (FEMA) administers the NFIP. As listed
    in Mrazik and Kinberg (1), the major objectives of the
    NFIP are to:
    • Make nationwide flood insurance available to all com-
      munities subject to periodic flooding.
    • Guide future development, where  practical, away
      from flood-prone areas.
    • Encourage state and local governments to make ap-
      propriate land  use adjustments to restrict develop-
      ment of land that is subject to flood damage.
    • Establish a cooperative program involving the federal
      government and the private insurance  industry.
    • Encourage lending institutions, as a matter of national
      policy, to assist in furthering program objectives.
    • Authorize the continuing studies of flood hazards.
    Studies of flood-prone areas typically  involve using
    step-backwater computer algorithms (digital models) to
    estimate river water-surface profile elevations and flood-
    inundation patterns  along the topography  of the river
    and its overbanks. FEMA recognizes the  U.S. Geologi-
    cal  Survey's (USGS's) step-backwater  model,  Water
    Surface PROfile (WSPRO),  as a suitable computer
    model for  use  in flood insurance studies (2, 3). Basic
    data input for step-backwater models includes:
    • Estimates of flood  discharge and initial  water-surface
    • Stream cross-sectional geometry.
    • Roughness coefficients for cross sections.

    • Contracted opening geometry if bridges  or culverts
      are located along the  study reach.
    Obtaining meaningful model results typically requires
    numerous stream cross  sections referenced to a com-
    mon elevation datum along a stream reach. The data-
    collection efforts to obtain  these  cross-sectional data
    require  costly, labor-intensive  fieldwork.  Study  efforts
    along lengthy stream reaches may, however, involve the
    generation of a contour map using aerial photogram-
    metric mapping techniques. Processing the spatial data
    may still require extensive labor to extract the  cross-
    sectional data needed for the WSPRO model.
    The  development  of geographic  information systems
    (CIS) technology has greatly enhanced analyses  of spa-
    tial data such as topography. In an effort to improve the
    quality of mapping  and delineation of flood-prone areas
    in Summit County, Ohio, the USGS developed a method
    of using a CIS as a pre-  and postprocessor of the input
    and  output data for the WSPRO model. This paper
    describes the steps  the USGS used to develop this
    interface and discusses some difficulties encountered
    during the process.
    Several steps were taken that resulted in the delineation
    of a  flood-prone area in Summit County, Ohio.  These
    steps are shown in  a flow chart (see Figure 1) and
    described below.
    Data were obtained for this study via aerial photography
    during April 1990.  These data include mappable fea-
    tures at the given scale  including  topography at 2-foot
    contour intervals, stream boundaries, roads, and build-
    ings. The data are estimated to be vertically accurate to
    +0.5 feet. The data were put into AutoCAD and were
    prepared for delivery to the USGS on 3.5-inch  floppy
    disks in  AutoCAD drawing interchange file (DXF) ASCII
    format. ARC/INFO was used to convert the DXF file into
    two separate data layers containing only the topography
    and traces of stream banks within  the study area.
    A three-dimensional  surface was generated from  the
    topographic data using the ARC/INFO software package
    Triangulated  Irregular  Network  (TIN).  Cross-section
    lines were digitized over the topography data layer. The
    cross sections were placed  according to standard step-
    backwater criteria (4) and were generally:
    • Perpendicular to stream flow
    • At major breaks  in streambed profiles
    • At minimum and maximum cross-sectional areas
    • At major changes in stream  conveyance
    • Spaced about one cross-section width  apart
                     Put Data in DXF Format
              Process DXF Data With CIS (ARC/INFO)
           Establish Relation of Attribute Data to Spatial Data
               Generate Three-Dimensional Surface
                   Overlay Cross Sections and
                     Calculate Intersections
             Verify Elevations With Topographic Controls
               Generate Input Files for Model With
                     Cross-Section Data
            Use WSPRO Model To Compute Water-Surface
                 Elevations for Each Cross Section
           Plot Model Output on Topography Data Layer and
                Connect Endpoints of Cross Sections
    Figure 1.  Flowchart of data conversion and processing for use
             in the Water Surface PROfile.
    The cross-section lines were then overlaid on the three-
    dimensional surface of topography, and CIS calculated
    the intersections of the contour lines and cross sections.
    The locations and elevations of these intersections were
    output as an ASCII file and slightly modified for input into
    the WSPRO model.
    These CIS data were  used along with the aforemen-
    tioned required data as  input to the WSPRO model.
    Input  for the model included estimates of the 100-year
    flood  discharge (5),  stream cross-sectional  geometry
    (supplied  by this work),  and estimates of roughness
    coefficients for cross sections. The WSPRO model was then
    run, providing output in  the form of water-surface eleva-
    tions at specific distances along section lines correspond-
    ing to the simulated elevation of a 100-year flood.
    Points corresponding to the flood elevations along the
    cross-section lines were plotted on the topography data

    layer and were connected manually (to delineate flood
    boundaries) by interpolating the elevations with respect
    to adjacent contours. A polygon of the flood "surface"
    was generated and drawn on a map (see Figure 2).
    The supplied topographic data were of sufficient quality
    and resolution to substitute  for field-surveyed  eleva-
    tions; however, field surveys to verify the  elevations
    along the  cross sections would augment this quality
    control process (see Figure 1). Typically, a crew of two
    individuals may take up to 4 days to survey and reduce
    the field  data for the study area chosen for  this study.
    Because aerial photography is commonly substituted for
    land surveying, the most significant effort and source of
    error may come from manually extracting elevations and
    distances along cross sections for input into the WSPRO
                                                                 100 200  300 400 500 Meters
    1,500   2,000  2,500 Feet
                                                                              100-Year Flood-Prone
                                                                              Topographic Contours
                                                                              (Contour Labels Omitted
                                                                              for Clarity; Contour
                                                                              Interval 2 Feet)
                                                                              Cross-Section Lines
                                                                              Stream-Bank Trace
                                                                              Intersection of 100-Year
                                                                              Flood-Prone Area and
                                                                              Cross-Section Line
    Figure 2.  Watershed showing delineation of 100-year flood-prone area.

    model. Initial development of the method to use CIS for
    this analysis took approximately 1 week to refine; how-
    ever, future analyses would probably only take one per-
    son 1 day to perform. This represents a significant cost
    savings. Additionally, reducing the amount of human-in-
    duced error can substantially improve the reliability and
    accuracy of the  computer-generated flood-prone area
    Because topography, stream traces, and other features
    are supplied in the DXF file, these data  can easily be
    brought into CIS. Maps can be made that show these
    features  in relation to the predicted flood-prone  area.
    Maps showing a variety of features can be produced at
    any scale, with accuracy limited only by the accuracy of
    the source  scale. Additionally,  CIS  can  calculate the
    intersection  of map features that  may  lie  within the
    flood-prone  area, such as buildings that may contain
    hazardous materials. CIS can also overlay land use
    data layers within the flood-prone area to define areas
    that should not be developed or that have already been
    overdeveloped in accordance with the aforementioned
    NFIP objectives.
    FEMA now requests that future flood-study mapping be
    completed using CIS format, a  common goal that both
    the USGS and FEMA are working toward.  These data
    are important to  land planners, flood-plain regulators,
    and  insurance companies that  rely on accurate esti-
    mates of flood-prone areas.  By increasing the accessi-
    bility  of the data by  using  CIS, we can substantially
    improve our ability to analyze spatial data efficiently.
    Problems Encountered
    Problems using data supplied in DXF format in conjunc-
    tion with CIS resulted primarily from the fact that the
    DXF data were prepared for the purpose of making a
    topographic map, not a CIS data layer. The contour lines
    were segmented; that is, where ends of segments met,
    they were not physically connected to form a topologi-
    cally viable  data layer. The data  layer needed to be
    edited because CIS requires topology for spatial-data
    processing. Additionally, in areas where the topographic
    gradient was particularly steep, contour lines were omit-
    ted. In both  cases,  an attempt was made to allow CIS
    to establish  a physical connection  of contour lines, but
    subsequent manual interpolation  was also  required.
    This may have introduced error into the data set. If future
    work requires the use of DXF data, the request for data
    should specifically state that  all topographic contours be
    AutoCAD stores data differently from CIS, so a relation-
    ship needed to be established  between the  data file
    containing elevations and the data file associated with
    the lines that make up the topography data layer. Sev-
    eral lines from the DXF file did not have any data asso-
    ciated  with  them, thus  necessitating  the  addition of
    contour elevation  data by context with the adjacent con-
    tours that did have data. This step may also have intro-
    duced  errors, but quality-control measures to verify the
    topographic contours and contour elevations could help
    to minimize these errors.
    Output from the WSPRO model is in the form of a series
    of points along cross sections that were connected by
    manual interpolation. This step also may introduce some
    error,  but the same process must be performed when
    not using CIS.
    This report documents an example  of how CIS can be
    used  to facilitate step-backwater modeling  of flood-
    prone areas. The results of the study show that signifi-
    cant savings may be expected in the form of reduced
    labor  requirements. Furthermore, FEMA now requires
    the use of CIS to conduct flood-study mapping, thus
    providing a means to conduct additional spatial analy-
    ses more efficiently. As aerial  photography and CIS
    technology improve, although additional sources of error
    may arise, the overall accuracy, reliability, and  repro-
    ducibility of the model input and results should also
     1. Mrazik, B.R., and H.A. Kinberg. 1991. National flood insurance
       program: Twenty years of progress toward decreasing nationwide
       flood losses. Water Supply Paper 2375. U.S. Geological Survey.
     2. Shearman, J.O.,  W.H. Kirby, V.R. Schneider, and H.N. Flippo.
       1986.  Bridge  waterways analysis model. Research Report
       FHWA/RD-86/108. Federal Highway Administration.
     3. Shearman, J.O. 1990. Users manual for WSPRO, a computer
       model for water-surface profile computations. FHWA-IP-89-027.
       Federal Highway  Administration.
     4. Davidian, J. 1984. Computation of water-surface profiles in open
       channels: Techniques of water-resources  investigations of the
       United States Geological Survey. In: Applications of Hydraulics,
       Vol. 3.
     5. Koltun, G.F., and  J.W Roberts. 1990. Techniques for estimating
       flood-peak discharges of rural unregulated streams in Ohio. In-
       vestigations Report 89-4126. U.S. Geological Survey, Water Re-

     Reporting on the Development of an Environmental GIS Application
             Wetlands Restoration in the Central Valley of California
                                       David T. Hansen1
    Appropriate documentation of information used in addressing environmental problems is a
    common issue. This paper focuses on information supporting geographic information system
    (GIS) applications on wetlands in the Central Valley of California. It identifies the role GIS
    played in the preparation of a report to Congress on wetlands and water supply in the Central
    Valley. It also describes the role that GIS is playing in the dissemination of information on
    wetlands and potential wetland habitat development to communities and groups in the Central
    Valley. Information or data documentation are major elements supporting this data and these
    GIS applications. For GIS data, this information is commonly referred to as metadata in
    conformance with the "Content Standards for Digital Geospatial Metadata (FGDC, 1994 and
     GIS is an integrative technology. It typically has involved specialists from a variety of disciplines
    who associate and integrate different data sets into a spatially referenced system. The
    development of GIS data typically follows a series of steps:
       1.  Based on a conceptual model of the environmental issue, geographic features or data of
           interest are identified.
       2.  These features are mapped or the data is spatially referenced in a map coordinate
       3.  The mapped features or data are digitally captured and processed in a software system
           to create a GIS data theme containing a graphical representation of the feature and
           associated information as attributes.
       4.  This digital data are reviewed and assessed as to fitness for the application.
       5.  The digital data is then ready for display, query, and analysis with other digital  data for
           that geographic area.
    1 U.S. Bureau of Reclamation, Mid Pacific Region, Dept. of the Interior, 2800 Cottage Way, Sacramento,
    CA 95825-1898; (916) 978-5268; (916) 978-5290 FAX; dhansen@mp.usbr.gov

       6.  The resulting data output from GIS analysis then is again reviewed and assessed
           against the requirements for the application.
    Increased speed and capacity of computer systems and the development of graphical user
    interfaces (GUI) have brought GIS and geospatial analysis to the computer desktop.
    Technologies such as global positioning systems (GPS), remote sensing, and scanning
    technologies have combined some of these steps and shortened the time required for digital
    data development. Many of these steps which were essentially GIS back office operations can
    now be performed by managers and increasingly the public. This has enabled managers and
    the public to access and use data in a map or geographic format to address environmental
    In the case of the Central Valley Joint Venture, desktop GIS has permitted the direct
    involvement of managers and partners in data development and GIS analysis. Desktop GIS
    applications developed as part of this program permit analysis of the data at the local level with
    community groups. Many GIS applications are robust enough to permit the loading and use of
    locally developed data in modeling environmental systems. Providing information supporting this
    data and the GIS application are critical in the dissemination of the data and application down to
    the local level. This case study will follow these steps identifying the information needed by the
    Central Valley Joint Venture partners for evaluating the data for application and use. This
    information is metadata and terms used will follow the elements of the "Content Standards for
    Digital Geospatial Metadata" (FGDC,  1994 revised 1998). However, this represents only a
    subset of the elements in these standards. The relationship between FGDC metadata elements
    and GIS data development  are further described in a draft guides for documenting the
    development of a GIS data  theme (Hansen, 1998) and reporting metadata for data
    management, data catalogs, and data transfer (Hansen, 1998).
    Conceptual Model - Identification of Data Requirements
    The Central Valley Joint Venture was established as part of the North American Waterfowl
    Management Plan signed by Canada and the United States in 1986. In 1990, the Central Valley
    Joint Venture issued an  implementation plan (CDFG, 1990) for wetland habitat restoration and
    enhancement. Wetland and adjacent upland habitat are important wintering areas for waterfowl
    in the Pacific Flyway. By the mid 1980's, waterfowl populations were approaching 30 percent of
    long term averages.  Much of this decline  is associated with the loss of wetland habitat since the

    turn of the century. The Central Valley is a semi arid area. Successful wetland habitat
    restoration in the valley requires a dependable and adequate water supply. Water supply for
    wetlands must be balanced against environmental requirements as well as water requirements
    use for agriculture and municipalities. The Central Valley is a major agricultural producing area
    for California and the nation. Surface water flowing primarily from snow melt in the Sierra
    Nevada mountains provides approximately 70 percent of the water used for agriculture and
    municipalities in the State. The plan attempts to balance water requirements for wetlands,
    agriculture and municipal use without impairing the supply for other aquatic and terrestrial
    In 1995, the Central Valley Joint Venture partners prepared a statement of work for a report to
    Congress based on the implementation plan (USFWS, 1995). Main topics for this report are the
    identification of methods for improving the reliability of water supply for existing private wetlands
    and identification of water requirements for an additional 120,000 acres to be restored to
    wetland habitat in the Central Valley. As part of the statement of work, the cooperator in the
    report was to incorporate any appropriate data into a desktop GIS.
     The statement of work identified a variety of information needed for the report. Much of this was
    geographic in nature such as the location of existing wetlands, lands under wetland or
    conservation easements, and lands suitable for wetland habitat development. Information was
    also identified that was not strictly geographic in nature such as the identification of constraints
    affecting the protection or restoration of wetlands or the reliability of water supply for wetlands.
    The statement of work recognized that not all information required to address issues in the
    report were suitable or could be developed in time for analysis as GIS data themes. It
    recognized that other tools would be needed to address some issues for the report to  Congress.
    As part of the statement of work, metadata on the collected data was to be provided by the
    contractor. The statement of work, itself, provided some of the initial information called for in the
    "Content Standards for Digital Geospatial Metadata" (FGDC, 1994  and 1998). The statement  of
    work and the implementation plan prepared in 1990 form the basis  of the conceptual model for
    identifying data and information requirements. Data themes were identified and the purpose for
    collecting this information. The time period of data content, target source scale  (1:24,000), and
    GIS data format and system were identified.  Information not explicitly identified in the statement
    of work included actual data sources, map coordinate system, and  method of coordinate control.

    The statement of work was a working document which was revisited as information was
    developed to adjust for changes in the availability of data for the report.
    Mapping, Digital Capture, and Database Development
    Mapping, digital capture, and database development are distinct process steps. Increasing with
    new technologies, these steps occur concurrently. Such a theme for the partners in the Joint
    Venture was the identification of existing wetland habitat on the valley floor. The program
    participated in  a joint project with other organizations which identified wetlands, riparian habitat,
    and other land use (DU,  1997). This GIS theme achieved a minimum resolution for wetlands at
    about 0.8 hectares (2 acres) using remote sensing techniques. Many of the other GIS data
    themes were already mapped independently of the program requirements. The cooperator
    following the statement of work digitally captured the features represented on these maps,
    merged separate sources for a particular theme together, and constructed databases of
    attributes for the digital features.
    Focus areas or areas for analysis is another GIS theme that required definition, mapping, and
    digital capture  for the Joint Venture partners. The floor of the Central Valley covers
    approximately  4 million hectares (10 million acres). This was too large an area for data
    development within the time constraints of the program. One of the initial tasks of Joint Venture
    was to narrow  the focus of data collection efforts to smaller areas within the Valley floor. The
    wetland habitat GIS theme as it was being developed and the personal knowledge of the Joint
    Venture Partners assisted  in identifying focus areas for intensive data collection efforts. The
    resulting focus areas represent approximately 0.8 million hectares (2 million acres) of the valley
    floor. These areas include virtually all  the areas of existing  Public and private wetlands.
    For the partners in the Joint Venture Program, digital capture and database development
    represented the black box  phase of GIS data  development. Key information from this stage for
    the Joint Venture partners  included the sources of data for the GIS themes, definition of criteria
    used in digitally capturing the features, database definitions, and criteria or rules for classifying
    the attributes of those features. Since the area of interest or focus areas for Joint Venture cover
    such a broad area, multiple sources of data were required for each GIS theme. Often, these
    sources represented  different time periods for mapping. Different sources also raised issues of
    consistency in  mapping criteria and in attributes identified for the mapped features.

    Evaluation of Digital Data for Application and Use
    Using the desktop GIS, information collected and digitally captured as a GIS data theme could
    be reviewed directly by the Joint Venture partners. This evaluation occurred repeatedly during
    data development. This was helpful to the Joint Venture partners as well as the GIS data
    developer. Managers as well as staff could  directly evaluate the data against the issues
    identified in the statement of work and their own knowledge of the area. GIS analysis could be
    interactively performed to evaluate the various data themes for addressing issues required in
    the report. This review identified GIS themes that were not useful in addressing the issues. The
    Joint Venture could then focus attention on  other methods for developing information to address
    those issues. Surrogates to represent information that could not be directly represented in GIS
    could be addressed and evaluated. The Joint Venture partners could visually review:
        •   Extent of coverage  of a  particular GIS theme for the areas of interest,
        •   Data gaps between GIS themes for  the same area,
        •   Attributes carried by the GIS themes and the definitions for those attributes, and
        •   Attributes relationship to the issues identified for the report.
    Information that could not be directly displayed in GIS were:
        •   Sources used to construct each GIS theme,
        •   Time periods represented by the GIS theme,
        •   Consistency of a GIS theme for all areas of the Valley,  and
        •   Criteria used in classifying the attributes of the GIS themes.
    This metadata was not available at this stage to the Joint Venture partners.
    This information represents a subset of information identified in the "Content Standards".
    Although available to the data developer, this information was  not in a form easily provided to
    the data users. While the data provider had some experience in spatial data capture, the
    provider had little experience in working directly with data users and in recognizing information
    that they might need. The data  developer was deferring metadata compilation until the end of
    the data development. The "Content Standards" had been recently adopted by FGDC and the
    data provider had little experience in  addressing and applying the standards. This hindered the
    evaluation of the GIS data  by the Joint Venture partners.

    GIS Display, Query, and Analysis and Evaluation of Analysis
    For the Joint Venture program, GIS analysis and evaluation occurred concurrently. With the GIS
    desktop application, the partners in the Joint Venture program were involved directly in applying
    GIS to address some of the issues for the report. This included display, query and reporting of
    the following information for the report:
       •   Location and extent of public managed wetlands,
       •   Private lands under easements for wetland habitat and conservation,
       •   Private lands managed for waterfowl or duck clubs, and
       •   Major water supply agencies for those lands.
    The Joint Venture partners ran a variety of different scenarios using the desktop GIS to identify
    lands suitable for wetland habitat restoration. These scenarios were based on criteria defined
    and run by the Joint Venture partners at their meetings. These scenarios were primarily based
    on the following GIS data themes:
       •   Soil characteristics suitable for wetland habitat development,
       •   Land use,
       •   Existing Publicly managed wetlands,
       •   Lands with easements for wetland habitat or wildlife conservation, and
       •   Private wetlands.
    A variety of other GIS data themes were available for display with these themes for review with
    the results of the scenarios.
    The Joint Venture partners could evaluate the results of the scenarios for issues required in the
    report. Criteria for the scenarios could be evaluated and adjusted to meet specific needs for
    different areas of the Central Valley. At the time of these meetings, the desktop GIS was not
    exploited to  the full in documenting the various scenarios that the participants posed.
    Documentation of the final scenarios do form the basis for the some of information reported to

    Application Development - Public Outreach
    Several of these GIS data themes and other data developed in concert with the Joint Venture
    program led to the development of a stand alone application. This application contains several
    GIS data themes and software for assessing the suitability of areas for wetland habitat
    restoration (Ducks Unlimited, 1998). The application is basically a modeling tool for ranking and
    weighting various GIS themes as to their suitability for developing and maintaining wetland
    habitat. The user selects the themes that they want to use in the analysis, weights or assigns
    values contained in the attribute table, and ranks the theme on basis of its importance for
    wetland habitat. The application converts the themes into a raster data structure and combines
    the raster data sets into new data set or surface whose cell values represent suitability of that
    cell for wetland habitat. The user selects the cell size for processing. This application has been
    issued on a CD-ROM containing GIS data sets, metadata, and some sample scenarios. The
    data on the CD requires desktop GIS software, but the application is open in the sense that
    other spatial data can be loaded and used in the analysis.
    This  tool was developed not only to assist offices of the individual partners in the Joint Venture
    program but also as an outreach tool to local community groups and for education. This
    application provides the opportunity to address issues at the local level. Locally developed  as
    well as other data can be used in the application. Metadata describes the GIS  data themes and
    a user guide has been prepared for using the application. While metadata is included with the
    GIS data themes, it is expected that this information will not address all questions posed by the
    users of the data. Describing the GIS application following the "Content Standards of Digital
    Geospatial Metadata" is beyond the scope of these standards. As is true for most guides, the
    guide for the application focuses on the mechanics of operating the application. The guide
    probably does not address all questions on  how the various themes could be ranked or
    The local user is responsible for evaluating the results of running the analysis.  It is up to the
    user to decide on the cell size for appropriate for the selected GIS themes. The user evaluates
    the resulting output surface against the cell  size used, themes selected, and the values
    assigned to the themes. The guide can not answer all questions on what spatial analysis is, how
    the model  process actually runs, or what uncertainty can be assigned to the results of analysis.
    To some users, it will appear to be another black box. Access to additional information can be

    provided by citations to other documents and providing contact information for individuals and
    organizations involved with the data or the application.
    This has been a review of the information needed at different stages in the development of GIS
    data and applications for the Central Valley Joint Venture Program of California. Partners in
    Joint Venture were able to evaluate the data, to run spatial analysis, and to review results from
    multiple scenarios. Issues were identified that could not be adequately addressed as GIS data
    themes. Resources could then be focused on other methods to develop this information. Some
    of these GIS themes along with other GIS themes supported the development of an
    independent GIS application. With this application, community groups and the public can
    develop their own scenarios for identifying wetlands at the desk top level.
    The development of relatively easy to  use desktop GIS software provides the opportunity for
    final data users to be directly involved  in data evaluation and spatial data analysis. New and
    improved technologies have compressed the steps and time required for spatial data
    development and the incorporation of that data into GIS applications. Environmental GIS
    applications are increasingly available for use by local communities and the public. Many are
    robust enough to permit the loading of locally developed data. This represents current trends as
    described in "The Future of Spatial Data and Society" (NRC, 1997). Under this trend, spatial
    data proliferate rapidly and the tools to use this data are widely available. Under this scenario,
    key issues are education or training in using spatial data and access to information to evaluate
    and apply the data. Lack of adequate information can to lead to increased uncertainty in the use
    of GIS data and increased litigation.
    Joint Venture partners were hindered in their evaluation and application of the GIS data themes
    because some information was not available at their meetings. This information was not
    available because the data developer  lacked experience in working concurrently with users in
    developing and applying data. As these applications are reaching down to community groups
    and the local  level, the GIS users are further removed from data  producers and application
    developers. To effectively run these applications, GIS data users can be expected to need more
    information than can be easily contained in our existing metadata descriptions or in a user

    Documentation during this development is critical for the effective application and use of the
    data or the application. The "Content Standards for Digital Geospatial Metadata" (FGDC, 1994,
    revised 1998) provides a lexicon of commonly accepted terms for describing GIS data. It
    identifies what metadata should accompany the digital data when it has been completed and
    transferred. It does not address how metadata is to be developed or presented during data
    development. These standards are a comprehensive list of elements of metadata for geospatial
    data but they are not exhaustive. These standards do not address the development of GIS
    applications for use at the local level.
    We as GIS data or application developers need to be directly involved in the application  and
    development of these standards. The involvement of data developers and application
    developers is needed for guides on the development and use of GIS data and applications. The
    Environmental Protection Agency has been an active participant in the development of many
    standards related to environmental quality (Johnson, 1996).  For environmental data, there are a
    variety of standards and guides often specific to a particular  discipline supporting the data
    collection and modeling efforts. The "Content Standards for National Biological Information
    Infrastructure Metadata" (USGS-BRD,  1995) has been issued for describing biological data. The
    process of developing of GIS environmental applications is similar to the development of a
    ground-water model to a site specific application (ASTM D5979) and the steps followed in
    environmental site characterization (ASTM D5730).  For water quality and monitoring, there are
    a host of commonly adopted standards. This paper is offered to continue discussion and to
    encourage involvement in the development of guides for describing environmental data and
    applications using GIS.

    ASTM, 1996, "D 5979 - Guide for Conceptualization and Characterization of Ground-Water
    Flow Systems" ASTM, 100 Barr Harbor Drive, West Conshohocken PA 19428-2959.
    ASTM, 1996, "D 5730 - Guide for Site Characterization for Environmental Purposes with
    Emphasis on Soil, Rock, the Vadose Zone and Ground Water" ASTM, 100 Barr Harbor Drive,
    West Conshohocken PA 19428-2959.
    ASTM, 1996, "D 5714 - Specification for Content of Digital Geospatial Metadata" ASTM, 100
    Barr Harbor Drive, West Conshohocken PA 19428-2959.
    California Department of Fish and Game, California Waterfowl Association, Ducks Unlimited,
    February, 1990,  "Central Valley Habitat Joint Venture Implementation Plan", Sacramento
    California (CDFG, 1990).
    Ducks Unlimited, January, 1997, "California Wetland and Riparian Geographic Information
    System Project', Prepared for California Department of Fish and Game, California Wildlife
    Conservation Board, U.S. Bureau of Reclamation; Sacramento, California (DU,  1997).
    Ducks Unlimited, November 1998, "Central Valley Project Improvement Act GIS Model Version
    1.0, User's Guide", Sacramento, California (DU, 1998).
    Federal Geographic Data Committee, June 1994 (revised 1998,), "Content Standards for Digital
    Geospatial Metadata", Washington, D.C. (FGDC 1994 and 1998).
    Hansen, David T., 1998, "Guide for Documenting the Development of Geospatial Data for Site
    Investigations", Draft for review by ASTM Technical Sub Committees of D18, 100 Barr Harbor
    Drive, West Conshohocken PA 19428-2959 (Hansen,  1998)
    Hansen, David T., 1998, "Guide on Reporting Geospatial Metadata for On-line Data
    Management, Data Catalogs, and for Data Transfers", Draft for review by ASTM Technical Sub
    Committees of D18, 100 Barr Harbor Drive, West Conshohocken PA 19428-2959 (Hansen,

    Johnson, A. I., 1996, "The Accelerated Development of Standards for Environmental Data
    Collection", Sampling and Environmental Media, ASTM STP 1282, James Howard Morgan Ed.,
    ASTM, 100 Barr Harbor Drive, West Conshohocken, PA 19428-2959 (Johnson, 1998).
    Mapping Sciences Committee, National Research Council, 1997, "The Future of Spatial Data
    and Society: Summary of a Workshop", National Academy Press, Washington, D.C. (NRC,
    U.S.  Fish and Wildlife Service, 1995, "Scope of Investigations and Plan of Work under
    Cooperative Agreement for Wetland Water Supply Investigations", Sacramento California.
    (USFWS, 1995).
    U.S.  Geological Survey, Biological Resources Division, 1995, "Content Standard for National
    Biological Information Infrastructure Metadata", formerly National Biological Service,
    Department of Interior, Washington, D.C. (USBS-BRD, 1995).

        Using a Geographic Information Systems Application to Implement Risk
                           Based Decisions in Corrective Action
            Lesley Hay Wilson, P.E., Andrew P. Romanek, and David R. Maidment, Ph.D., P.E.,
                                 The University of Texas at Austin
                             James R. Rocco, Sage Risk Solutions LLC
    The implementation of site-wide corrective action using risk-based decision making at large and
    complex industrial, energy or defense facilities presents a number of challenges. To address these
    challenges, a spatial environmental risk assessment methodology has been developed by
    connecting Geographic Information Systems (GIS), relational databases and spreadsheets. The
    methodology is based on a description of the facility and a site conceptual model. A case study site
    with multiple potential sources, transport mechanisms and receptors has been used to evaluate the
    methodology. The case study facility is an approximately 300-acre crude oil refinery and petroleum
    products terminal that has operated since the early 1900's. This paper will discuss the development
    of two key elements of the spatial environmental risk assessment, the Digital Facility Description and
    the Spatial Site Conceptual Model using the GIS application.
    Risk-based decision making provides a mechanism to determine the necessary and cost-efficient
    strategies for protection of human health and the environment. It is an iterative process that begins
    with a planning phase that incorporates risk management decisions with a site conceptual model.
    The process then proceeds to an evaluation phase where data collection, and fate and transport
    analysis, provide the basis for evaluation of the exposure pathways identified in the site conceptual
    model. Finally, the process ends with a decision phase in which the plan is compared to the results of
    the analysis to determine whether further evaluation or remedial action is warranted or if "no further
    action" is appropriate.  Risk-based decision making requires the interrelationship of three basic
    processes: risk assessment, risk management and risk communication.
    Risk Assessment
    Risk assessment is a process that quantifies the potential for adverse effects to human health and
    the environment caused by an exposure to a chemical of concern released to the environment.
    Where there are no current or potential exposures to a chemical of concern, or where the
    concentration of a chemical of concern is not harmful to human health or the environment then, the

    risk assessment will conclude that there is no unacceptable risk. Risk assessment is accomplished
    by collecting information to construct exposure hypotheses for a chemical of concern, or a group of
    chemicals of concern, and evaluating those hypotheses to determine the potential for adverse effects
    from human or ecological exposures to chemicals of concern in the environment. This process is
    based on the National Academy of Sciences Risk Assessment paradigm (MAS, 1983).
    Effective risk assessment is based on several important activities. First, a comprehensive site
    conceptual model is needed to provide the working hypothesis for a site. The site  conceptual model
    is the understanding of the potential exposure pathways based on the chemical characteristics and
    the physical setting of the site. In order for the site conceptual model to be comprehensive, it needs to
    identify all of the potential exposure pathways and be updated, as new information becomes
    available. It provides the mechanism for determining the necessity and scope of data collection, and
    a template for evaluating exposure pathway completeness. Second, effective data collection is
    needed to evaluate the exposure pathways identified in the site conceptual model. Data collection
    can be both qualitative (e.g., location of source areas, historical release information) or quantitative
    (e.g., concentration of chemicals of concern in environmental media, hydrogeological characteristics).
    It requires spatially defined and relationally organized property characteristics such as physical
    features, and information related to chemical and media characteristics such  as analytical results  and
    hydrogeological information. Data collection needs to be focused on developing and updating the site
    conceptual model, evaluating exposure pathways, determining appropriate initial response actions
    and comparing site conditions to the corrective action goals. Third, the value of the information that is
    being collected must be considered. Only the quantity and quality of data necessary to provide a
    sound basis for the decisions to be made should be collected. Collecting data for the sake of more
    data is not necessary or cost-effective. Data collection must consider the use of the data to be
    collected and the potential for that additional data to change the decisions that will be made. Fourth,
    the fate and transport of chemicals of concern in the environment needs to be considered when
    evaluating exposure pathways. However, the results of fate and transport analysis must be confirmed
    through the collection of empirical data. Finally, "no further action" is not always the appropriate result
    of a risk assessment. Interim remedial action, remedial action as well as further evaluation are
    alternatives to be considered.

    Risk Management
    Risk management decisions are necessary to support site-specific determinations that are protective
    of human health and the environment and to provide a means for similar decisions to be made for
    similar circumstances. Many risk management decisions rely on the application of scientific
    methodologies such as the determination of the chemicals of concern to be considered in the risk
    assessment; the appropriate toxicity factors for the chemicals of concern; and the appropriate data
    quality and quantity. Other risk management decisions rely on non-scientific factors for definition such
    as the determination of a process for stakeholder involvement; an approach for ground water
    resource and use; and a consistent set of criteria for the objective comparison of alternatives. Clearly,
    a risk-based decision cannot be made without having defined the appropriate risk management
    Risk Communication
    Risk-based decision making requires the identification and involvement of the individuals,
    organizations and other entities that are directly affected by the corrective action process,  commonly
    referred to as the stakeholders. Since risk management decisions are necessary to support
    determinations that are protective of human health and the environment and require the
    consideration of a combination of scientific, social, political, personal and economic factors, it is
    critical that the risk management decisions are acceptable to most if not all of the stakeholders. In
    addition, the application of the risk management decisions within the risk assessment process must
    be clearly understood and accepted by all of the stakeholders. Therefore, early, effective and regular
    risk communication is critical to the successful implementation of risk-based decision making.
    Risk communication, however, has been the most overlooked component of risk-based decision
    making and all too often undertaken after the evaluation has been completed and the decisions have
    been made rather than as part of the process. When applying risk-based decision making there will
    typically be a number of alternatives for solving or addressing the environmental condition of a
    property. Each alternative will have characteristics that define its benefits and its risks to the
    stakeholders. There will also be stakeholders with differing interests, objectives, and levels of
    knowledge. However, the perception of these benefits and risks may vary among the stakeholders
    and it cannot be assumed that all stakeholders will have the knowledge or background to  effectively
    participate in the process. Therefore, effective risk communication must make information concerning
    a corrective action easily accessible to the stakeholders, provide a mechanism for the stakeholders to
    visualize the results and be interactive to allow for participation in the process by  the stakeholders.

    Spatial Environmental Risk Assessment
    In recent years, the use of risk assessment and risk-based decision making in environmental
    management has gained increasing attention (Rocco and Hay Wilson, 1998,  Washburn and
    Edelmann, 1998). However, there are a number of significant challenges to the practical application
    of risk-based decision making at large, complex industrial, energy or defense facilities. For these
    types of facilities, the implementation of a risk-based approach has been difficult and a practical
    methodology has not been demonstrated (Hay Wilson etal., 1998).
    The most significant challenges arise from the multiple potential sources, multiple chemicals of
    concern and multiple potential receptors. In these large facilities, there can literally be hundreds of
    potential exposure pathways to analyze. In practice, at complex facilities, sources are evaluated
    individually or in small groups. As an example, the corrective action program under the Resource
    Conservation  and Recovery Act (RCRA) encourages this piecemeal evaluation through its focus on
    individual solid waste management units (SWMU). In general, the implementation of program-
    specific (e.g.,  air, water, waste) regulations by the Environmental Protection Agency (EPA) has  also
    perpetuated this non-holistic approach. The same is true for the regulation of  Department of Defense
    and Department of Energy facilities. Individual areas of a facility are studied using a straightforward
    process to analyze exposure pathways for each source-receptor pair. However, there is typically no
    attempt to understand the interaction of all of the sources and pathways on facility-wide risks or the
    affects of these multiple sources and pathways on the environmental management decisions. It has
    also been the case in the past that the goals for corrective action projects were based on very low to
    non-detect concentrations of chemicals of concern, so the need to understand all of the potential
    exposures was not as great. In addition, often many individuals are involved, over a number of years,
    at a significant cost, in the calculations of the risks for each of the different areas for a facility. Many of
    these facilities are also regulated under different regulatory programs, with different regulatory
    agencies and no one investigator examines all of the  results (Hay Wilson, 1998).
    The availability of cost-effective computing power and information systems applications can provide
    the foundation for the development of computer-based systems to manage information and perform
    calculations for large numbers of pathways. In particular, the information processing capabilities of
    Geographic Information Systems (GIS), relational databases, spreadsheets and other computer
    code-based models can provide a mechanism to construct a methodology to implement risk
    assessment at large, complex facilities (Hay Wilson, 1998). In the spatial environmental risk
    assessment methodology the ESRI software ArcView® is being used for the GIS functions (ESRI,

    1998). The Microsoft Office products Excel® and Access® are being used for the spreadsheet and
    relational database components, respectively (Microsoft, 1997).
    Building the Digital Facility Description
    In order to conduct the analyses required for making risk-based decisions and to provide a common
    point of reference for all of the stakeholders, an understanding of the features of the facility and
    surrounding area must be developed. This understanding can best be represented in digital files so
    that analyses may be conducted, data managed and calculations performed efficiently. These digital
    files make up the digital facility description, that is, the description of natural and man-made
    features and information about the physical, geological, hydrological and chemical characteristics of
    the facility and the region. The digital facility description is the foundation upon which the site
    conceptual model is developed and all of the exposure analyses conducted.
    The digital facility description consists of two major components; a spatial database and a tabular
    database. The spatial database contains the GIS shape files and coverages of geographic features
    related to regional information (e.g., surface water flow, geology) and facility information (e.g.,
    locations of current and former process areas and storage tanks). It must include the coverages
    necessary to conduct the exposure pathway analyses and should include coverages that define the
    physical setting  of the facility and provide contextual information (Romanek, 1999). The regional
    information can  be obtained from various websites, including the United States Geological Survey
    (USGS) the Environmental Protection Agency (EPA), and includes topography, land use, and digital
    elevation data coverages. The facility information can be obtained from digital aerial
    orthophotographs or can be developed based on CAD files that have been converted to GIS files.
    The tabular database contains the facility feature descriptions and chemical and physical data (e.g.,
    soil property measurements, chemical of concern concentration data) linked to each other through
    common fields,  thus forming a relational database. The relational database provides an easily
    manageable format for storing, adding, and retrieving different types of information (Romanek, 1999).
    The information included in the digital facility description for the case study is a comprehensive
    compilation of available information for the facility from  environmental, geo-technical and other
    investigations or activities that have been conducted at the facility over the past ten to fifteen years
    and the available regional spatial data. The regional information for the case study digital facility
    description includes data from regional data sources, primarily state agency web sites and EPA web
    sites. The facility information for the case study digital facility description includes facility data

    sources, an environmental measurements database developed by the facility and the facility
    geographic coverages developed from a 1997 aerial survey. In addition, facility information has been
    collected from site investigation reports,  historical maps, engineering drawings and older aerial
    The case study facility is an approximately 300-acre crude oil refinery and petroleum products
    terminal that has operated since the early 1900's. For this case study site, an aerial survey was
    conducted and the facility coverages of operating units, tanks, waste areas, surface waters, etc.,
    were digitized from the resulting orthophotographs. Figure 1 includes six facility scale coverages of
    the physical site features. When displayed together these depict a typical site plan. The types of
    analyses that are expected to be conducted will dictate the resolution needed for the mapping. As an
    example, for the case study site it was important to understand the potential for surface water runoff
    to the adjacent creeks and the river, so a digital terrain model (DTM) was developed from the
    orthophotographs to provide a detailed description of the land surface.
    Developing the Spatial Site Conceptual Model
    The site conceptual model is a critical  component of the risk assessment process and the focal
    point of risk-based decision making. It provides the working hypothesis of all of the potential exposure
    pathways associated with the chemicals of concern identified at the many potential sources, their
    movement in the environment and their  relationship to potential receptors. The site conceptual model
    is a synthesis of spatial and observational data. The challenge in assessing environmental risk at a
    large, complex facility lies in capturing the complexity of multiple sources and receptors.  Typically,
    simplified site conceptual models are used to represent the relationships between the sources and
    receptors at facilities. However, using a simplified site conceptual model can potentially lead to an
    inaccurate understanding about the effects on receptors and expected results from implementing a
    remedial action alternative (Koerner etal.,  1998). This is particularly true when assessing the
    environmental risk to the many potential receptors existing on and off of a facility resulting from
    multiple sources and receptors, as is the case at most large industrial facilities. It is also often the
    case that the site conceptual model is developed at the beginning of the risk assessment project as a
    static display of the pathways thought to be of importance to the investigation at that time. The typical
    representation is a series of flow charts (ASTM,  1995, ASTM 1998, BP, 1997). For a facility with
    multiple sources of many chemicals of concern and various potential receptors, the presentation of
    flow charts is not very informative.

         /V/ Boundary' Line
              ' Fences and Walls
             / Railroads
               Storage Tanks
               Surface Hydrology
                Figure 1. Compilation of Six Facility Coverages to Display a Site Plan
    The public has received the process of risk assessment with skepticism, largely because it is not
    generally perceived to be an open and transparent calculation process. It has been viewed by many
    as a "black-box" approach. However, through the use of spatial and tabular databases, a spatial site
    conceptual model can be developed to describe the working hypothesis of the potential exposure
    pathways for a facility and provide a mechanism to communicate what is known and what is not
    known about the site and the exposure scenarios among the engineers, decision-makers and other
    stakeholders. The spatial site conceptual model can, therefore, be a means for different stakeholders
    to identify the exposure scenarios for which they have the greatest concerns and quickly identify the
    scenarios that have already been analyzed. To accomplish this the spatial site conceptual model is

    linked to the environmental modeling process and to the calculations of risk or corrective action
    goals. The Risk Assessment Data Model shown in Figure 2 depicts an exposure pathway evaluation
    that is tailored to the digital and spatial processes (Hay Wilson, 1998).
                                    Corrective Action Goal Calculation
                              Figure 2. Risk Assessment Data Model
    The spatial site conceptual model is developed by identifying each complete or potentially complete
    exposure pathway for a site. The exposure pathways are segmented into the functional elements of
    sources, cross-media transfer elements, geographic transport and receptors. Segmenting the
    pathways in this way allows for the connection of the elements to their computational counter parts.
    The descriptions of the sources are linked to the tabular database that describes the release history
    for each of the sources, or the sources can be linked to the spreadsheet that is used to calculate the
    representative concentration of the chemical of concern at a particular source area given the
    analytical results from environmental media sampling. Transfer and transport segments are linked to
    the spreadsheet calculations describing the environmental processes.
    The spatial site conceptual model is an evergreen description of the current understanding of the site.
    Developing it using PC-based software makes the application accessible to all of the members of the
    project team (e.g., state regulators, responsible party project managers, consultants) and provides a
    mechanism to update the entire exposure pathway evaluation as new data is added. This saves time
    during modeling runs and reduces errors in data transfers. The fact that the exposure pathway
    evaluation results are  stored in a tabular database means that the outcomes can be catalogued and

    tracked. This provides the stakeholders with the documentation they seek to identify the pathways
    analyzed and the outcomes. This process makes the risk assessment more reliable and repeatable.
    Itemizing the exposure pathways in a database and segmenting the pathways facilitates the tracking
    of many multiple sources using conservative analytical models (Koerner, 1998). The sources that
    affect an individual receptor area can be identified and remedial action scenarios can be evaluated
    through an iterative calculation procedure. Cataloguing all of the potential impacts of the many
    multiple potential sources at a facility will provide the data and understanding of the site to support a
    site-wide analysis of exposures and ultimately of risks.
    The spatial site conceptual model includes a spatial representation of the elements for multiple
    sources, pathways and receptors. The components of the spatial site conceptual model (e.g.,
    sources and receptors) are represented as individual themes or data layers in vector representation.
    The user identifies the potential sources, transport mechanisms and potential receptors within the
    GIS application. Themes for potential sources are constructed based on the digital facility description
    data, the historical  information and the environmental measurements. The sources are defined as
    point coverages. These coverages may include individual  points for releases (e.g., the location of an
    emissions stack) or they may be the center points for source areas (e.g., the center of a defined area
    of soils containing chemicals of concern). Receptor locations are identified based on the spatial data
    and the current and potential future activities on the facility and surrounding properties. The receptor
    locations are defined by points, represented by points of exposure (e.g., the location of a drinking
    water well) or areas, represented by potentially affected areas (e.g., the residential neighborhood
    near a facility). The areas for which receptor identification is needed can be identified in the GIS
    application based on land use, census data or digital ecological habitat data. The transport
    mechanisms that link the sources to the receptors are grouped in themes based on environmental
    media. Lines define the lateral transport mechanisms such as groundwater flow. All of the elements
    are drawn in the GIS application. Scripts are used to assign distances, coordinates and areas to the
    spatial objects. Figure 3 includes one source area, six potentially affected areas, six points of
    exposure, and four geographic transport mechanisms.
    The modeling of the environmental process is implemented using a tiered approach. For simple
    evaluations the single point to point relationships are analyzed using vector operations and  one-
    dimensional algorithms. Simple fate and transport algorithms have  been assembled in a group of
    spreadsheets. In higher level evaluations the modeling is accomplished as raster or grid functions.

        Figure 3. Example Spatial Representation of the Site Conceptual Model Components
    Grid-based groundwater and surface water models have been developed using the methodologies
    presented by Maidment, 1996 and Romanek, 1999.
     Risk-based decision making is an iterative process that provides a mechanism to determine the
    necessary and cost-efficient strategy for protection of human health and the environment. It requires
    the interrelationship and interoperation of risk assessment, risk management and risk
    communication. Risk Assessment is accomplished  by collecting information to construct exposure
    hypotheses for a chemical of concern or group of chemicals of concern and evaluating those
    hypotheses to determine the current and future potential for adverse effects to human or ecological
    exposures from chemicals of concern in the environment. The incorporation of risk assessment as an

    integral component of an overall decision making process provides a mechanism to determine the
    necessary and cost-efficient strategy for protection of human health and the environment. In addition,
    improved methods for engineers and scientists conducting risk assessments and greater acceptance
    of risk assessments by the stakeholders will support the general environmental regulatory agency
    and community goals of environmental protection and sustainable economic development.
    The contribution of this research is a new spatial risk assessment methodology using GIS to
    practically implement a holistic environmental risk assessment that accounts for the multiple
    exposure pathways at large, complex facilities and support risk-based environmental management
    decisions. This methodology provides an information processing system that more clearly ties the
    data and information for  a study area to the risk-based decisions that are made for an environmental
    management project. In  addition, the application of a spatial site conceptual model methodology
    helps automate the selection of exposure pathways to be considered, provides a mechanism to
    document the pathway completeness evaluation and provides connections to the transport
    calculation components  and to the site conceptual model tabular database. In this manner,  the
    process of evaluation and calculation can be more clearly understood and presented.
    BP Amoco provided funding to support this research. Dr. Robert Gilbert, Dr. Randall Charbeneau
    and Ms. Susan Sharp provided insights and support.

    American Society for Testing and Materials (ASTM). 1995. Standard Guide for Developing
    Conceptual Site Models for Contaminated Sites. E 1689-95.
    American Society for Testing and Materials (ASTM). 1998. Standard Provisional Guide for Risk-
    Based Corrective Action. PS104-98.
    BP Exploration & Oil Inc. (BP). 1997. "Risk-Based Decision Process Guidance Manual." BP
    Exploration & Oil Inc. Cleveland Ohio.
    Environmental Systems Research Institute (ESRI). 1998. ArcView GIS, Version 3.1, Redlands, CA.
    Hay Wilson, L. 1998. "A Spatial Risk Assessment Methodology for Environmental Risk-Based
    Decision-Making at Large, Complex Facilities," Dissertation Proposal, The University of Texas at
    Austin, Department of Civil Engineering, Austin, Texas.
    Hay Wilson, L. L. N. Koerner, A.  P. Romanek, J. R. Rocco, R. B. Gilbert. 1998. "Critical Success
    Factors for Implementing Risk-Based Decision Making at a Large Refinery Site," Contaminated Sites
    Remediation Conference: Challenges Posed by Urban & Industrial Sites Proceedings, March 23-25,
    1999, Fremantle Western Australia.
    Koerner, L. N., L. Hay Wilson, A. P. Romanek, J. R. Rocco, S. L. Sharp, R. B. Gilbert. 1998.
    "Maximizing the Value of Information in Risk-Based Decision-Making: Challenges and Solutions,"
    American Nuclear Society Conference, April  3-5,1998, Pasco WA.
    Koerner, L. N. 1998. "Development of a Site  Conceptual Model Using a Relational Database,"
    Masters Thesis. University of Texas at Austin, Department of Civil Engineering, Austin, Texas.
    Maidment, D. R. 1996. "Environmental Modeling within GIS," in GIS and Environmental Modeling:
    Progress and Research Issues. Goodchild, M. F. etal., eds.,  GIS World, Inc.  Fort Collins,  Colorado, p
    Microsoft Corporation 1997. Microsoft Office  Professional Edition 97, Seattle, WA.

    National Academy of Sciences (MAS). 1983. Risk Assessment in the Federal Government.
    Managing the Process. National Academy Press, Washington, D.C.
    Rocco, J. R. and L. Hay Wilson. 1998. "The Evolution of Risk-Based Corrective Action," Published in
    proceedings: Geo-Congress 98, Risk-Based Corrective Action and Brownfields Restoration, ASCE,
    Boston, Massachusetts, p44(11).
    Romanek, A. P. 1999. "Building the Foundation for Environmental Risk Assessment at the Marcus
    Hook Refinery using Geographic Information Systems," Masters Thesis. University of Texas at
    Austin, Department of Civil Engineering, Austin, Texas.
    Washburn, S. T. and K.G. Edelmann. 1998. "Development of Risk-Based Remediation Strategies,"
    Published in proceedings: Geo-Congress 98, Risk-Based Corrective Action and Brownfields
    Restoration, ASCE, Boston, Massachusetts, p30 (14).

        Characterizing the Hydrogeology of Acid Mine Discharges from the Kempton Mine
                            Complex, West Virginia and Maryland
                                      Benjamin R. Hayes
                       Department of Geological and Environmental Sciences
                                    Susquehanna University
                              Edgar W. Meiser, Meiserand Earl, Inc.
             Constance Lyons, Maryland Department of the Environment, Bureau of Mines
    The Kempton Mine Complex consists of an area of interconnected, abandoned underground
    coal mine workings in the Upper Freeport Coal seam in southwestern Maryland and
    northeastern West Virginia (Figure  1). The mine complex was operated by the Davis Coal and
    Coke Company from 1885 to  1950. The workings encompass an area of 31.3 km2. Groundwater
    discharges from  the abandoned mine contribute on the order of six million gallons per day of
    acid mine drainage (AMD) to the North Branch of the Potomac River and the North Fork of the
    Blackwater River.
    The Maryland Bureau of Mines is undertaking a comprehensive investigation of the Kempton
    Mine Complex to develop remedial measures to reduce AMD by (1) decreasing the quantity of
    recharge to the deep mine and/or (2) improving the quality of the discharge from  the mine pool.
    Before specific remedial measures are identified and evaluated, a GIS was developed to refine
    our understanding of the mine structure and hydrology. The visualization and computational
    capabilities of GIS enhance our ability to conceptualize the geometry and hydrology of the deep
    mine complex.
    Sources and Types of Data. Data incorporated into this study are diverse and of  varying quality.
    The watershed characteristics and  mine information included published geologic maps and
    boring data from the West Virginia Geologic Survey, Maryland Geologic Survey,  U. S.
    Geological Survey, and U.S. Bureau of Mines, as well as records from the coal company
    including mine inspection  reports, coal production reports, and water management reports.
    These data were synthesized, converted into digital form, and archived into Microsoft Access™

    databases. Selected figures and important pages from old reports were scanned and archived
    as raster images in the database. Geologic maps, structural contours, hypsography, hydrology,
    and transportation features were digitized from the Blackwater, Davis, Lead Mine, and Mozark
    Mountain USGS 7.5 minute quandrangles. Detailed maps of underground mine workings at a
    scale of 1 inch equals 100 feet that were originally surveyed by mine engineers with the Davis
    Coal and Coke Company were digitized in AutoCAD and incorporated into ArcView GIS
    coverages of the mine region. Boring locations and coal depths were incorporated into the
    coverages. Details of each coverage were documented in metadata files  in HTML format that
    can be read and written by any web browser.
    Historical aerial photographs from the 1950s through 1980s were scanned and combined in the
    GIS with recent (March, 1999) low-altitude orthophotographs of the mine  region. The recent
    orthophotos were georeferenced and rectified to control points surveyed on the ground using
    precision GPS survey techniques.
    Hydrologic and water quality data collected at field monitoring stations were organized in
    Microsoft Excel™ spreadsheets (flat files). These spreadsheets were used to analyze time
    series of mine  discharges, stream flow hydrographs, and water quality samples.
    Data Analysis. The GIS proved valuable at integrating the historical boring data and mine maps
    with the ongoing mine discharge and water quality data. Three-dimensional stratigraphic cross-
    sections and isopach maps help identify hydrogeologic features such as stratigraphic pinch
    outs, changes  in dip, and mine pool barriers. Large sets of spatial data can be integrated and
    analyzed to generate digital terrain models, correlate stratigraphic units, and compute the
    locations of outcrops of selected strata at the ground surface.  Algorithms  are currently being
    developed to generate new maps of flooded mine areas, mine pool shorelines, and directions of
    groundwater flow.
    In order to characterize  the hydrogeology of the mine complex, one must first understand the
    details of the structure of the  mine. The structural features include the location, orientations, and
    dimensions of the Upper Freeport coal and the underground workings within it.
    Specific questions about the mine structure include:

        1.  regional geologic structure of the coal seam;
        2.  precise locations of underground mine workings;
        3.  degree of coal extraction;
        4.  locations of coal outcrops; and
        5.  proximity to overlying coal seams that have been mined.
    Regional geologic structure of the coal seam. The Upper Freeport coal seam is found in a
    broad, north-east plunging  syncline in Upper Potomac coal basin. The fold axis trends about 5
    degrees and plunges 2 to 3 degrees to the northeast. The dip of the syncline limbs varies from 2
    to over 16 degrees, with minor variations.  Over most of the Kempton Mine, the Upper Freeport
    coal seam is 4 to 4.5 feet thick and pinches out on the eastern limb of the syncline (Figure 2).
    No  major faults are mapped on the regional geologic maps, but numerous faults and offsets are
    described in the mine reports.
    Figure 2. Kempton Mine and elevation contours of the base of the Upper Freeport coal seam.
    A detail structure contour map of the base of the Upper Freeport coal was prepared by the West
    Virginia Geological Survey  from over 230  borings compiled from multiple sources. The boring
    logs provide top and bottom Upper Freeport coal elevations which were  incorporated into the
    GIS database. We used the GIS to generate trend surface and three-dimensional boring
    diagrams from the boring database.
    Precise locations of underground mine workings. The ability to generate maps showing the
    precise locations of underground mine workings in relation to present-day features is one of the
    most valuable contributions of the GIS in this study.  The original (1885-1950) mine maps at a
    scale of 1  inch to 100 feet were carefully digitized by the current property owner, Western
    Pocahontas Properties in AutoCAD format. The coverages were georeferenced in West Virginia
    State Planer Coordinates and verified using a network of 12 monuments surveyed using
    precision GPS technology.  The CAD coverages were imported into ArcView GIS and draped on
    top of digital orthophotographs and (DRG) maps of the region (Figure 3).
    Degree of coal extraction. Headings and pillars are clearly visible on the old mine maps. The
    digital maps of the old mine workings were extensively cleaned and edited to generate closed
    polygon coverages of the subsurface mines (Figure  4).

    Secondary mining, i.e. removal of coal pillars, occurred over most of the mine area. The
    remaining sections of the mine apparently remain intact as a labyrinth of tunnels and pillars. A
    noteworthy exception to this is the unmined coal barrier that trends east-west, located 3,000 feet
    north of Coketon (Figure 4). This barrier has a width of approximately 400 feet and begins near
    Thomas and extends 5700 feet to the west. The western tip of this barrier lies at elevation 2870
    feet MSL according to the coal structure elevation contours.
    Locations of coal outcrops. The outcrop of the Upper Freeport coal has been surface-mined
    along its entire length. Because of the surface disturbance of the coal at its outcrop, the actual
    location of the coal crop line is difficult to define and locate in the field. By projecting the
    structural contour surface  of the coal seam onto the ground surface topography, the GIS
    enables us to generate maps showing actual position of coal crop lines. The crop lines,
    corresponding to the edge of the coal underclay, or pavement, are needed to calculate
    watershed areas contributing to the deep mine.
    Proximity to overlying coal seams that have been mined. The Bakerstown coal, which lies
    approximately 180 to 200  feet above the Freeport coal, was extensively deep-mined and
    surface-mined. Because the Bakerstown coal is relatively close to the surface, and because it
    has been surface-mined in many areas  above the Kempton Complex, it is very effective in
    capturing recharge from precipitation. Once the water has "sumped" into the Bakerstown  coal
    seam, it inevitably drains into the underlying deep mines.  We suspect however, that in some
    areas the opposite situation is true, where flow from the Kempton complex may leak upward into
    and discharge from the Bakerstown coal mines.
    Once the deep mine maps were refined accurately using GIS,  our ability to visualize the
    relationships between the  Kempton Mine and overlying aquifers and areas of deep mining in the
    Bakerstown coal was improved greatly.  The visualization  capability of GIS enables us to
    develop the fundamental concepts of the deep mine's role in the regional groundwater
    Questions about the mine hydrology include:
       1.  the extent of mine  flooding and elevations of mine pools;
       2.  interconnection between deep mine to surface water runoff;

       3.  leakage from streams and wetlands into the deep mines;
       4.  quantities of recharge and discharge; and
       5.  hydraulic gradients and groundwater flow directions in strata overlying the mine.
    Extent of mine flooding and elevations of mine pools. Flooding in the abandoned mine workings
    is divided into two distinct and separate mine pools. The northern mine pool has a surveyed
    elevation of 2656 feet MSL and discharges from an 8-foot diameter air shaft and 12-inch
    borehole into Laurel Run about one mile north of Kempton, Maryland. These discharges are
    located approximately 4.5 miles upstream of the North Branch Potomac River. The Kempton
    mine pool elevation is constant and controlled by the elevation of the air shaft.
    The southern mine pool has an approximated elevation of 2870 feet MSL and discharges from
    several deep mine entries immediately south of the village  of Coketon, West Virginia into the
    North Fork of the Blackwater River. The elevation of this pool must be controlled by the intact
    coal barrier west of Thomas, West Virginia described earlier. This barrier is relatively
    impermeable, considering that in the vicinity of Rose  Hill Cemetery northwest of Thomas, it
    narrows to 200 feet in width and impounds hydraulic  heads exceeding 200 feet.
    Interconnection between deep mine to surface water runoff. Mine discharge records show
    significant increases in response to large rainfall events, suggesting direct inflow of surface
    water into the mine. We used the GIS to delineate areas of shallow cover above the mines
    (Figure 5). These areas will be examined in future field efforts to identify any sources of direct
    Leakage from streams and wetlands into the deep mines. Leakage from wetlands  along the
    Potomac River has been observed in abandoned mine shafts at the town of Kempton,
    Maryland. Streamflow losses are predictable in streams draining the east slope of  Backbone
    Mountain, as they cross the outcrop of Upper Freeport coal where the Kempton  deep mines are
    closest to the surface. We used the GIS to overlay stream  coverages with coal outcrop maps in
    order to delineate contributing watersheds and to identify locations of potential leakage. At these
    locations, predicted streamflows can be compared with measured streamflows to determine
    whether significant leakage is occurring.
    Quantities of recharge and discharge. In October and November, 1998, eleven weirs were
    installed to gage discharges from the Kempton Mine Complex. Two weirs were installed on

    discharges from the Kempton pool to Laurel Run; one on the air shaft and the other on the
    borehole. Nine weirs were installed on discharges from the Coketon pool to North Fork
    Blackwater River.
    We are using the GIS to analyze time series of discharge records from the weirs and
    precipitation  records from nearby gaging stations. We are developing ArcView AVENUE™
    scripts to plot hydrographs for each weir and to compute the total volume of discharge for any
    specified period of time. The total discharge volumes are then compared with the respective
    watershed areas contributing to the deep mine, delineated by the GIS, to compute the mine
    recharge rates (gpm/acre). The computed mine recharge rates can be compared to measured
    precipitation, as well to as average regional recharge, to determine where recharge to the
    Kempton Mine is anomalously high.
    Based on preliminary weir measurements,  recharge rates in the Coketon mine pool watershed
    appear to be greater than recharge to the Kempton mine pool watershed. The Kempton mine
    pool discharges up to 6 millions gallons per day while the Coketon discharges a maximum of
    approximately 4.4 million gallons per day - 75 percent of the northern mine pool. However, the
    Coketon pool has a recharge area of only 2,900 acres - 63 percent of the northern mine area of
    4,600 acres.
    Hydraulic gradients and groundwater flow directions. In late summer/early fall 1999,
    piezometers will be installed in aquifers overlying  the Upper Freeport coal. Comparison of
    hydraulic heads in these aquifers to mine pool levels will dictate vertical  directions of
    groundwater flow and hydraulic gradients. The head data will be incorporated into the GIS and
    gradients computed and cross-sections generated showing groundwater flow directions.
    As field investigations continue, the GIS is  being used to locate new test borings and monitoring
    wells, delineate areas for subsidence monitoring,  integrate historical and current mine discharge
    water quality data, and characterize groundwater recharge rates and flow directions. By
    analyzing the spatial variability of the hydrogeologic data, we plan to use the GIS to discover
    relationships among various parameters. Temporal variations in certain  parameters can also be
    analyzed to discern patterns of change not apparent with static, two-dimensional maps. For

    example, areas of mine pool fluctuations can be combined with coverages of coal structure to
    delineate areas of AMD generation.
    This project is being accomplished and funded through the cooperation and team effort of
    numerous state, federal, and private entities: Maryland Department of the Environment,
    Maryland Department of Natural Resources, University of Maryland, West Virginia Department
    of Environmental Protection, Susquehanna University, U.S. Environmental  Protection Agency,
    Region 3, U.S. Department of Energy, U.S. Office of Surface Mining, Anker Coal Company,
    Buffalo Coal Company, Mettiki Coal Corporation, Western Pocahontas Properties, Meiser and
    Earl, Inc.

       Mine Discharges
       Wine Pttimetei
    V Mine Pillars
    V Undennad Mine W Drirlngs
       Davis, WVQuad
       Leadmine.Wv Quad
    Figure 1. Study location map. Inset shows GIS coverage of the mine pillars and haulways draped over USGS quadrangle maps.

    Figure 2. Kempton Mine and elevation contours of the base of the Upper Freeport coal seam.

                                                      Scale 1:14,000
                                         500   1000   1500  2000   2500   3000   3500 Feet
    Figure 3. GIS coverages of northern section of Kempton Mine complex  showing detailed mine working maps draped on top of color
       NAPP aerial photograph.

                                                          • ; SSSSSSSSSS SSSSSSSSS
       Mine Discharges
    ^f\ Mine Perimeter
       Mine Pillar;
       Undefined Mine W ork ings
       Davis, W V Quad
       Leadmine. W V Quad
                             Figure 4. Southern portion of the Kempton Mine complex showing location of intact mine barrier.

    Figure 5. Overburden thickness map of Kempton Mine Complex.

           Use of GIS Tools for Conducting Community On-site Septic
                                 Management Planning
              David Healy, Chief, GIS Services & Bruce F. Douglas, Senior Geoscientist
                                  Stone Environmental, Inc.
                                     58 East State Street
                                  Montpelier,  Vermont, USA
    This paper presents a summary of the authors' experiences in using GIS tools in support of
    Community Septic Management. This experience examines a number of Massachusetts and
    Vermont case studies in which the use of GIS was a primary tool in the planning and
    implementation of community septic management. These are the towns of Duxbury and Tisbury,
    Massachusetts, and the towns of Jericho, Vermont. Based on these case studies lessons can
    be drawn on how to further evolve the use of the technology in  aiding community environmental
    management. This effort is based on the State of Massachusetts' nation-leading efforts in
    addressing the issue of management of on-site septic systems. The state's efforts are designed
    to address current failures, the protection of environmentally sensitive areas, and to assist
    communities develop tools and capabilities for insuring that problems are addressed and
    systems remain functional at all times. The Vermont community experiences are driven by a
    desire at the local level to avoid environmental pollution and costly centralized sewer systems.
    The paper will describe all aspects of acquiring and using state and local GIS data to carry out
    the plans in four communities. Issues surrounding availability, quality, and data formats will be
    described. Each of the case study communities approached the problem slightly differently
    based on the available data. Each community used GIS to different levels for decision making
    and management.  From each of these experiences, a summary of the successes, pitfalls and
    challenges will be described.
    Decentralized wastewater management is a growing trend in the United States. Faced with
    soaring costs for large centralized treatment systems, communities are increasingly turning to
    the smarter management philosophy associated with the decentralized approach. Traditionally,

    one expensive solution has been available to communities that have outgrown or outlived their
    on-site septic systems — a sewer system and an expensive treatment facility. With federal
    funds becoming increasingly scarce, most small communities can not afford this type of
    conventional centralized approach.
    By definition, decentralized wastewater management employs all available treatment and
    disposal technologies. The appropriate technologies, in a measure that meets current needs
    and takes into consideration future growth, are matched with the treatment and disposal
    requirements that have been identified. The  end result is a unique municipal wastewater
    management solution that includes a program of preventive maintenance designed to identify
    weaknesses or potential failures before they become a problem.
    This approach also simplifies future maintenance and planning. Community-wide decentralized
    wastewater management offers an opportunity to track the condition of individual systems, the
    relationship of those systems to other community infrastructure, like drinking water sources, and
    to the environment.
    The Role of Geographic Information Systems
    Geographic Information Systems (GIS) are a powerful tool for identifying and examining
    problem areas to display information for public understanding. GIS played a significant role in
    supporting a number of elements in each of  the following case histories. The overlay capability
    of GIS was used to show environmentally sensitive areas, soil suitability for on-site systems,
    and areas of the community using subsurface disposal for prioritizing replacement systems. GIS
    was used for mapping soil types to determine suitability for siting systems, mapping community
    infrastructure and existing systems, tracking plans for future development, mapping
    environmentally sensitive areas, and tracking maintenance, repair and upgrade information for
    an entire community. Maps can be readily generated from the GIS to make decision making
    more timely and public education more effective. The maps effectively communicate complex
    information at-a-glance in a graphical format that is easily understood.
    Database Applications
    Databases are considered essential to enable the efficient management of on-site systems at
    the local government level. Databases can be used to track permits, inspections, pump-outs
    and failures of on-site systems, maintain on-site system records, and prepare mailings for

    necessary system maintenance. The database should be capable of storing, retrieving and
    reporting all data pertinent to all of the on-site systems. The foundation of the database should
    consist of the following key data areas:
           1.     System related information including components, septic tank size, application
                 and permit numbers, maintenance (septic tank pump-out dates, inspection dates
                 and inspection results), plans, and images.
           2.     Parcel map and lot designation, related information including soil test data and
                 structures, number of bedrooms and design flow.
           3.     People related information. Names and addresses of people used for lookup in
                 various sections of the database.
           4.     Easy linkage to GIS. Any database characteristic can be displayed graphically
                 through GIS.
    Jericho, Vermont
    The Town of Jericho is a suburban and rural community located in the Lake Champlain Valley
    area of Northwestern Vermont. The town includes 92 square kilometers (35 square miles) of
    land. The 5,000 residents rely on individual and cluster type on-site wastewater treatment and
    disposal systems (septic systems).
    Town officials recognized the potential for treatment and disposal problems to develop as the
    town grows and existing septic systems age. The Jericho Board of Selectmen hired consultants
    to determine the most appropriate means of addressing short-term and long-term wastewater
    disposal needs for the town. The process involves assessing existing systems, characterizing
    environmental conditions, and analyzing the community's future options.
    The study focused on 3 areas:
           Jericho Center — a traditional 19th century New England village.
           Jericho Village — a traditional village with some commercial properties.
           Route 15 district — a recently developed commercial zone.
    Data Availability: Data for Jericho came from two primary sources the town planning office and
    the Vermont Center for Geographic Information.  The primary databases used for the analysis

           SSURGO Digital Soils
           Surface Waters
           Well Locations
           Source Protection Areas
           Swimming Areas
    Quality: This data was primarily digitized from 1:5000 Orthophotos and 1:24000 USGS Quad
    Sheets. FGDC type metadata did not exist for town developed data. The later issue poses long
    term problems for understanding the source parameters used. For this community mixing the
    two scales of data did not pose a major problem. For certain the FEMA flood data is highly
    inaccurate, but was not essential for this study.
    Data Formats: Local data was  in  Maplnfo format in Vermont State Plane Feet. State provided
    data was provided in Arclnfo format. Local data was converted to Arclnfo format and put into
    Vermont State Plane Meters, NAD 83.
    The study team  conducted a thorough assessment of existing systems and conditions. They
    identified environmentally sensitive areas and parcels with special conditions or limitations and
    conducted analyses and developed maps.
    The GIS analysis revealed that soils in the study areas are generally suitable for on-site
    systems on terraces in the major stream valleys where the densest residential and commercial
    development is typically located.  In individual neighborhoods, a 2 to 30 percent rate of on-site
    system failure in a 10-year period was evaluated. The failures were predominantly due to age,
    design and construction of systems. Most of these failures have been  effectively resolved by on-
    site replacement of the septic tank and/or soil absorption system with conventional
    technologies. Due to the low density of development, centralized collection systems do not
    appear to be cost-effective. Due to existing use of the river nearest the most densely developed
    areas in town for swimming, direct discharge of treated wastewater was not an option  either.

    CIS Integration: The town has received the consolidated set of reprojected data and various
    ArcView projects developed for this project on CD-ROM. It has acquired ArcView and is using is
    for various planning and zoning/regulatory functions.
    The consultants identified specific options for community wastewater management based on the
    town's current needs and plans for future growth. They concluded that existing septic systems
    and a locally developed septic system management program would be the most cost-effective
    option. To implement this as a sustainable option, Jericho has established a local Wastewater
    Planning Committee with three objectives:
          1.     Develop a decentralized wastewater management plan based on a town wide
                 assessment of need.
          2.     Develop a community homeowner and student education program to increase
                 the local understanding of on-site systems.
          3.     Demonstrate the effectiveness of septic tank risers and septic tank effluent filters
                 in facilitating the inspection and maintenance of on-site systems.
    Jericho's forward-thinking approach to decentralized wastewater management is saving the
    town money by using existing wastewater infrastructure, protecting homeowners' investments in
    their current systems, and avoiding the high cost of developing a centralized sewer and
    wastewater treatment facility.
    Duxbury, Massachusetts
    The coastal community of Duxbury, is located 35 miles south of Boston. Duxbury has a
    population of approximately 14,000 and a land area of approximately 61 square kilometers (24
    square miles). Greater than 95 percent of the residents rely on individual on-site sewage
    systems and the community is committed to using on-site systems as a long-term solution to
    wastewater management. The town relies on an extensive sand and gravel aquifer for
    community drinking water supplies. In addition to aesthetic and recreational value of the
    freshwater and saltwater resources of the community,  there are numerous cranberry bogs in
    commercial cultivation and a significant potential for shellfish  harvesting in the coastal
    Over the past three decades, Duxbury has been making substantial efforts to protect their
    groundwater and surface water quality with a permitting program for on-site wastewater disposal

    systems. In 1996, the town dedicated two shared soil absorption systems designed to address
    severe problem areas along Duxbury Bay. These shared systems (each with less than 37,800
    litres per day design flows) handled wastewater from three parcels along the Bluefish River and
    18 parcels in the  Snug Harbor area by conveying the wastewater to sites located inland that are
    suitable for soil absorption systems. The establishment of these systems has enabled the
    opening of shellfish harvesting areas due to a decrease in bacteria in the coastal embayments.
    The town has recently voted to design and build  another 37,800 litres per day cluster on-site
    system to serve approximately 30 households in a residential area along Kingston Bay to
    improve water quality in a historic shellfish harvesting area. To continue their on-going efforts  in
    this area, the town has recently completed a Community Septic Management Plan (CSMP) to
    provide a clear process for decentralized wastewater management.
    Duxbury's CSMP consists of the following components:
           1.  Comprehensive inventory of existing systems.
           2.  Parcel information permit records and septic tank pump-out records stored in a
              database to track permitting and maintenance.
           3.  Development of a local GIS system and training of Town personnel to identify:
                 a.  Drinking water source protection areas
                 b.  Freshwater wetlands, ponds, vernal pools, rivers and streams
                 c.  Saltwater wetlands, coastal resource zones
                 d.  Buffer zones around these areas
                 e.  Parcel maps to determine environmental sensitivity of particular parcels.
           4.  Public Education and Information development including a  brochure explaining the
              existing situation and the community septic management plan.
           5.  Betterment Loan/Upgrade Program.
           6.  Ranking environmental sensitivity by category to determine priority of parcels for loan
              program to assist homeowners in upgrading failed systems.
           7.  Voluntary Maintenance program that recommends setting routine system
              maintenance in conjunction with the following requirements in local ordinance:
              detailed reporting of septic tank condition, liquid and solid levels at septic tank

    Data Availability: Data for Duxbury came from two primary sources the University of
    Massachusetts,  Boston which had completed an Open Space Plan for the town and from
    MassGIS office.  The primary databases used for the analysis include:
           Digital Soils
           Land Cover
           Surface Waters
           Well Locations
           Source Protection Areas
           Swimming Areas
           Title 5 Setbacks
    Quality: This data came from highly mixed sources. MassGIS is primarily 1:24000 scale data
    digitized from USGS Quad Sheets and 1:20000. FGDC type metadata did not exist for UMass
    Data Formats: MassGIS data was in Massachusetts State Plane Meters Feet North American
    Datum (NAD) 1983. On the other hand, Umass' GIS data was in an unknown and
    undocumented format. It turned out after much trial and error that it was in Massachusetts State
    Plane Feet, NAD 1927.
    Specific GIS Processing: The ranking system devise for the betterment loan program was
    based on a assignment of weighted values to six factors. In this community the intersecting of
    coverages with the weighted values was competed to derive a composite weighting. A "grid-
    based" math algebra operation could have just as easily been used.
    GIS Integration:  Duxbury has received a complete copy of all the data and ArcView projects
    developed for the project. Two days of training were provided to staff members in the Planning
    and Health Departments. They are currently using GIS in many diverse planning and
    management functions.

                               Figure 1: Priority Ranking Results
    Tisbury, Massachusetts
    The Town of Tisbury is a coastal resort community located on the island of Martha's Vineyard
    located 8 km (5 miles) off the southwestern coast of Massachusetts, USA. The Town covers an
    area of 54 square kilometers (21 square miles). The population of Tisbury is approximately
    3,000 year round residents and 10,000 seasonal residents. Approximately three-quarters of
    Tisbury's residents rely on groundwater from municipal wells that tap a glacial sand and gravel
    deposit underlying the western half of the island. The remaining residents utilize individual wells,
    generally tapping the same aquifer. Currently all properties in town are served by on-site
    sewage disposal systems. The Tisbury Board of Health estimates that there are currently
    approximately 2,500 individual on-site systems.

    The intent of Tisbury's wastewater management program is to provide an institutional and
    regulatory framework enabling the long-term viability of the on-site wastewater treatment and
    disposal facilities in the Town. The program includes a Community Wastewater Management
    Plan (CWMP), Watershed Management Strategy, Public Outreach and Education, Institutional
    and Regulatory Requirements, and a Program Implementation Strategy. The CWMP is in the
    final planning stage and has not been adopted by the community.
    The town is taking a very pro-active approach to on-site system  management. The local Board
    of Health and Wastewater Planning Committee is developing a management program with
    required inspections for every system and requiring pump-outs of septic tanks based on
    expected rates of solids accumulation for the specific tank size and system usage. GIS has
    been used to delineate environmentally sensitive areas and parcels; a ground water flow model
    has been used to determine water table contribution to surface waters; and a database is being
    developed to track the management program and notify residents of compliance requirements.
    This approach enables the residents to easily see the relationship between the areas that
    utilized on-site systems and environmentally sensitive areas such as aquifers, streams, coastal
    ponds and wetlands using a risk-based management strategy.
    Environmentally sensitive areas in Tisbury will be used to identify high priority areas are for the
    management of on-site systems. Initial wastewater management districts have been defined to
    address the downtown Vineyard Haven area, and the low elevation (less than 3 meters (10 feet)
    above mean sea level) areas with the potential for systems to have the least vertical separation
    to groundwater.
    Periodic maintenance is critical to ensuring the long-term success of even the most basic septic
    system. An inspection and maintenance program has been designed to assess current
    infrastructure conditions, ensure proper use and maintenance of on-site systems, and reduce
    future failures. A relational computer database, the Septic  Information Management System
    (SIMS), will be established to maintain an up-to-date inventory of all onsite systems in town and
    to track the permitting of new systems, upgrades of existing systems, and inspection and
    maintenance program. The database will also be useful to ensure that all owners are
    adequately addressing the unique needs of their system. One of the primary goals of the
    management plan is to provide for better septage (solids and liquid pumped out of septic tanks)
    management by creating a predictable and manageable production of septage. This plan

    pertains to those parcels outside of the area serviced by the centralized wastewater collection
    and treatment system. While the central service area will have management requirements, they
    will likely differ from the  rest of town and will be addressed in a separate management program.
    On-site systems are designed to treat domestic wastewater before reaching the groundwater
    and down-gradient surface waters. The residual components of this treatment process, such as
    nitrates, have an impact on the environment. However the degree of impact is relative to the
    sensitivity of the groundwater beneath the on-site systems and the sensitivity of the surface
    waters where the groundwater discharges. The watershed management strategy is designed to
    protect the environmental resources of Tisbury at an appropriate level to the sensitivity of the
    different areas in town.
    A key element of the watershed management strategy is to use a risk assessment/risk
    management approach. During the risk assessment process, the town will develop rankings
    regarding the value and vulnerability of local receiving environments to impacts from on-site
    systems, and to define the areas in town which contribute flow to the receiving environments. A
    steering committee of stakeholders will be established to address the needs of the community in
    this process. The rankings will be used to determine appropriate levels of treatment and develop
    a risk management program, in order to protect public health and the environment. For
    example, specific areas may be delineated where nitrogen removal is required for upgrades and
    new on-site systems to reduce nitrate loading to a particularly sensitive and valuable receiving
    The second program in  the watershed management strategy is the development of a long-term
    groundwater monitoring program. A network of surface and groundwater sampling stations will
    be established up to monitor trends in water quality.
    Environmental professionals, municipal departments, and community environmental groups will
    conduct the risk assessment/risk management process. The water quality monitoring program
    will be run in conjunction with the Martha's Vineyard Commission, the University of
    Massachusetts Extension program, the Town of Tisbury, and local professionals. Funding for
    the watershed management program  is will be provided through a combination of  local and
    federal sources.

    Data Availability: Data for Tisbury came from two primary sources the MassGIS office and a
    previous consultant who had worked for the town. The primary databases used for the analysis
           Building Footprints
           Title 5 Buffered Footprints
           FEMA Flood Data
           Land Cover
           Surface Waters
           Well Locations
           Source Protection Areas
           Possible Discharge Areas
           Color Infrared Orthophoto
    Quality: This data came from two sources. MassGIS is primarily 1:24000 scale data digitized
    from USGS Quad Sheets and 1:20000. The consultant's data came primarily from the town's
    CAD-based tax maps for which were assembled and rubber sheeted to fit with the MassGIS
    data. FGDC type metadata did not exist for the consultant data. The FEMA data was the least
    accurate of the data used. The focus of the study was for the village of Vineyard Haven. Given
    the coastal nature of the data, the town boundaries varied widely depending on the original
    Data Formats: Most Data was acquired in State Plane Meters, Mainland Zone, NAD 1983.The
    one exception was the digital CIR Orthophoto which came from the Massachusetts Coastal
    Zone Office in State Plane Meters, Island Zone, NAD 1983. Since we derived contours for the
    study area from this data, it had to be reprojected to State Plane Meters, Mainland Zone. These
    kinds of issues working with available data always presents problems in all studies to date.
    CIS Integration: The town has recently embarked on an effort to develop a town-wide GIS
    program. This has begun with an evaluation of  existing data and assessment of the many
    department needs.


           Geographic Information and Tools for Informed Decisions:
                    The Lake Superior Decision Support Project
                       George E. Host, Lucinda B. Johnson, Carl Richards
                              Natural Resources Research  Institute
                                         Pat Collins
                              MN Department of Natural Resources
    Recent trends in land use, such as increased population movement from urban to rural areas,
    conversion of forest land, and increased development have emerged as key issues affecting
    natural resource management in the Lake States. As units of government ranging from local
    townships to the federal governments of the US and Canada plan for the future, the need for
    data and tools for sound decision-making has become critical. Nonetheless, at the scale of the
    Lake Superior Basin, we lack synoptic information on even the most fundamental data layers
    required for sound planning. Among these are comprehensive coverages of land use/land
    cover, transportation infrastructure, hydrography, demography, and even the bathymetry and
    shorelines of Lake Superior itself.
    In addition to the lack of spatial data, smaller units of government often embark on land use
    planning exercises with few tools at their disposal. While computer simulation models, draft
    ordinances, and decision support tools are receiving wider use in planning, these tools are often
    out of reach of local governments who lack equipment and expertise required for their use.
    Finally, in natural resource management, the general public is often faced with information that
    has been slanted in favor of the perspectives of industrial or environmental advocacy groups.
    There is a critical need for a source of neutral data to allow the public to develop informed
    opinions on current issues.
    To this end, we have begun a project to help resolve some the issues of data  accessibility and
    interpretation with respect to land use planning in the Lake Superior Basin. We have three key
           1) To develop synoptic databases across the Lake Superior Basin, and to make these
             data available through the Internet and other data distribution formats.

           2)  To concurrently develop decision support tools to assist local units of government in
              land use planning activities. These tools will consist of CD-based resources,
              including spatial data, prototype planning documents, and flowcharts to guide users
              through the planning process.
           3)  To develop and deploy information kiosks in Visitor's Centers and other publicly
              accessible locations around the Lake Superior Basin.
    The following paper charts our experiences and progress to date, and provides some insights
    into issues related to integrating information and decision support tools into the planning
    This project was motivated by several key factors:
       1.  Forestry and forest products industries dominate the basin's economy.
       2.  Demand for tourism and leisure activities is replacing mining and other previous
           contributors to the region's economy, with an associated increase in development to
           leverage this demand. The Lake Superior Basin's wealth of natural features is a major
           factor driving these changes.
       3.  The Basin is a unique harbor of biodiversity, on a world-wide basis (TNC). A
           considerable portion of the Basin's biodiversity resources are located  along coastal
           shores and wetlands.
       4.  The Basin's natural and biodiversity resources are increasingly in conflict with both
           forestry and development. And increasingly, the three are in conflict at a given place
           simultaneously. People want to visit and live, by and large, where natural resources are
           rich and biodiversity is highest.
    It stands to reason that the Basin's future is directly dependent on the resolution of these land
    and resource use conflicts. But the resolution of these conflicts will  not occur as a result of a
    small number of large government agencies or private land holders developing and

    implementing policies that address land use issues. The real impacts of development result
    from the cumulative effects of a large number of small land use decisions, spread across time
    and space. When integrated across time and space, these land use decisions have major
    impacts on the Basin's resources.
    The unique role of the Lake Superior Decision Support project is to develop and foster the use
    of an infrastructure, based on GIS and computer models, upon which the Basin's incremental
    land use decisions can be made more effectively. Our target users are local governments,
    resource management agencies, commercial  interests,  citizen volunteers, advocacy groups,
    aboriginal  groups, and educational organizations. There are two fundamental missions of the
    project, both intended to sustain the Basin's resource in a truly ecosystem-based approach (i.e.,
    sustaining humans as a part of the natural world). The first mission is to provide practical,
    useable tools that can be used by people involved in day-to-day land use decisions. The second
    mission is to provide a context and a demand for these tools by providing educational and
    interpretive information for their use.
    The direct beneficiaries of this project are those organizations  and individuals that will directly
    use the GIS applications and databases developed by this project. But there will also be many
    indirect benefits that are difficult, if not impossible, to quantify.  For example:
       1.  Local governments will gain tools to more effectively create and implement zoning
           ordinances that mold day-to-day land-use decisions.
       2.  Educational and interpretive organizations will gain resources that will help them inform
           the public about these resources and energize their target audiences to encourage their
           use in the planning process.
       3.  Regional agencies (including state/provincial and federal agencies) will, over time, gain
           access to a consistent GIS database for assessing regional and basin scale issues. In
           this process, data gaps are identified.
       4. All  parties to local and regional land use decisions and policies will gain from a collective
           knowledge base and shared and unbiased information  base.

       5.  Ultimately the Basin and its people benefit from more effective land use decisions that
           help to sustain and steward the Basin's resources.
    Data development
    Data development began by prioritizing a list of key data layers critical for land use planning and
    capable of being mapped synoptically across the Lake Superior Basin. The geographic scope of
    the data was the Lake Superior
    watershed boundary plus a 50 km buffer
    (Figure 1). The buffer allows us to
    access impacts to the basin that may be
    due to activities outside the watershed
    boundary. The watershed boundary was
    constructed by merging a number of
    independently-derived boundaries
    developed at state or provincial scales -
    it was informative that neither the
    watershed boundary nor the shoreline of
    Lake Superior previously existed in a
    continuous, fine-resolution GIS coverage.
    Figure 1. Lake Superior basin boundary with 50
    km buffer.
    The base scale of the data aggregated across the whole basin was set at 1:250,000; a
    compromise between the desired accuracy of the data and the ability to easily store, manipulate
    and visualize the data. We then conducted a broad survey of agencies involved in development
    of spatial data for natural resource management to identify the source, scope, and resolution of
    key data layers.
    A critical step in the identification and acquisition of data was the development of Memoranda of
    Understanding and Cooperative Agreements with collaborative groups, such as management
    agencies and industry. A key issue is that many agencies operate on a cost-recovery basis, in
    which GIS products are sold to recover costs of development. In many cases, licensing
    agreements require that data be made available only in image format (e.g. GIF or TIF), rather
    than as spatially-referenced GIS databases. In addition,  providing appropriate
    acknowledgments, respecting publication rights, and developing well-defined dissemination

          criteria was essential for this effort. To date, we have assembled approximately 30 databases
          across all or part of the Basin, as shown in Table 1.
                                               Table 1.
    Major spatial data themes, geographic extent and key attributes for data compiled for this project.
           Data theme
    Bathymetric Model
    Census data
    Civil divisions, minor
    Ecological classification
    Ecological classification
    Ecological classification
    Forest inventory (FIA)
    Forest inventory
    Habitat megasites
    Habitat projects
    Habitat sites
    Land Cover: Forest
    Land Cover: General
    Land Cover: Original
    Land ownership
    Pollutants-point source
    Pollutants-point source
    Public Land Survey
    Satellite imagery
    Lake Superior
    MN, Wl, Ml,
    VIN, Wl, Ml
    VIN, Wl, Ml, ONT
    .ake Superior
    VIN, Wl, Ml
    N MN, N Wl
    VIN, Wl, Ml
    VII, MN, Wl, ONT
    Lake Superior
    Lake Superior
    VIN, Wl, Ml, ONT
    VIN, Wl, Ml, S.
    Great Lakes
    Lake Superior
    Lake Superior
    VIN, Wl, Ml, ONT
    Lake Superior
    Interpolated depth values
    Population, demography, housing
    Interpolated data, monthly temp,
    Digital Elevation Model
    ECS, Province to Subsection
    Land Type Associations
    Hierarchy: ecozone, region and
    Forest inventory data by sample
    point, 1990s
    GLO witness trees, section corner
    Surficial geology, landforms
    Wilderness areas, ecoregions,
    parks etc.
    Habitat projects, location,
    Habitat sites, location, description
    Lakes, streams, rivers
    24 classes of forest vegetation
    Anderson II, Land cover/use
    40 classes derived  from GLO
    survey notes
    Public/private ownership
     ndustrial point source locations
    Municipal point source locations
    Township, range and section
    Mosaic of Landsat MSS, mid
         1 km
                    5 km
                    1 km
                    1 km
                  40 acre

                                               Table 1.
    Major spatial data themes, geographic extent and key attributes for data compiled for this project.
    Lake Superior
    Satellite imagery Basin
    Lake Superior
    Satellite imagery Basin
    Soils ONT
    Soils STATSGO
    Watersheds, Major
    Watersheds, Major
    Watersheds, Minor
    MN, Wl
    MN, Wl, Ml, ONT
    .ake Superior
    MN, Wl
    AVHRR NDVI Composite
    Night image of city lights
    Drainage, surface material,
    deposition, etc
    Many attributes: texture, drainage,
    Classification of roads, railways,
    Major Watershed boundaries
    Major Watershed boundaries
    Minor watershed boundaries
    1 km
    1 km
          Documenting the source, accuracy, resolution, and other key attributes of spatial data is critical
          for its effective use. We are using the Minnesota Geographic Metadata Guidelines, developed
          by the Minnesota Land Management Information System, a state agency that serves as a
          clearinghouse for spatial information. The Geographic Metadata Guidelines are a simplified set
          of the Federal Geographic Data Committee's Content Standards for Digital Spatial Metadata.
          They consist of 99 elements divided into seven categories describing the following data
                1. What kind of a database is this?
                2. What is its quality?
                3. How is it organized?
                4. Where is the database located?
                5. What features does this database describe, and how are they characterized?
                6. Is this database distributed? If so, how?
                7. Who documented this data?
         We are implementing the metadata with a software tool called DataLogr, developed by the State
         of Michigan. The metadata for each data layer will be available on-line on the LSGIS web site.

    Delivering Information: Web sites, Kiosks, and Exhibits
    The LSGIS web site can be found at www.nrri.umn.edu/lsgis. It was developed and maintained
    with Microsoft Frontpage® and allows for the interactive guerying of maps using ESRI's Internet
    Map Server® (IMS). IMS provides a number of tools for retrieving information from maps, and
    includes the ability to provide progressively more information as a user zooms in to finer scales
    of resolution.
    In parallel with the Internet-based information, we are developing a touchscreen computer
    display that provides a similar suite of information. These touchscreen computers will be
    deployed in kiosks at several visitors' centers around the  Lake States. We are also currently
    working with the Great Lakes Aquarium at  the Lake Superior Center in Duluth, MN to integrate
    information collected in this project into a public "town meeting" forum that allows participants to
    assess data and make decisions on a series of problems facing the Lake Superior watershed.
    The touch screens use Site Explorer®, a map-oriented information utility developed by
    cooperator Mike Koutnik.
    Decision Support Projects
    A key part of this project is the implementation of the data and tools to address land use
    planning issues. We have two ongoing projects in this area, as described below:
    Hydrologic Modeling of the Miller Creek  Watershed
    Miller Creek is a relatively high quality trout stream that runs through the cities of Duluth and
    Hermantown, MN. The stream originates in wetlands and undeveloped shrublands and
    woodlands northwest of Duluth, flows through a heavily commercialized area around Miller Hill
    Mall, through residential areas as it drops rapidly towards Lake Superior, and ends in an
    industrial area near its  mouth. The upper part of the watershed is fairly flat, and the stream has
    a moderate gradient. Near the mall it is channelized for several hundred meters and starts to
    receive stormwater discharge from storm drains.  Further downstream, in the more heavily
    residential areas, multiple stormwater channels enter Miller Creek. The stream develops a high
    gradient near its mouth. The stream is routed underneath the city in stormwater conduits and
    ditches until it enters the St. Louis River  estuary of Lake Superior
    The Miller Creek watershed contains several new strip malls, and further commercial
    development is under construction. This  has generated a wide public concern over issues over

    environmental quality, particularly with respect to degradation of trout habitat resulting from
    increased stream temperatures, sediment inputs, and degraded water quality. To inform the
    decision-making process, we have assembled detailed land use maps of the Miller Creek
    watershed, and employed the EPA's PC-SWMM, a well-documented hydrologic model, to
    predict flow and pollution loads to the creek.
    Our first goal was to model, as accurately as possible, the volume and temporal distribution of
    flow in Miller Creek, based on overland, ground water, storm sewer conduit,  and natural channel
    flow using PC-SWMM. The second objective was to estimate pollution loads and concentrations
    in runoff based on land use, road densities, local street cleaning programs, and locally
    calculated and published land use-specific pollution loading rates. The modeling of these
    parameters will allow planners and others to determine the likely effects of proposed changes in
    the basin and evaluate options for reducing pollution loads to Miller Creek.
    The model predicts the hydrograph associated with rain events fairly well (Figure 2). Since a
    large part of the pollution concentration model is based on flow rates, this accurate flow model
    was needed before proceeding to the pollutant modeling stage. Hydrologic modeling was
    complicated by the large amount of wetlands  in the headwaters of the Miller  Creek basin, but
    results to date show significant differences in  water yield related to land  use. For example, the
    effects of the Miller Hill Mall commercialized area on flow in Miller Creek is shown in Figure 3:
    the impervious surfaces around the mall result in a large increase in flow volume from a
    relatively small area, in spite of flow mitigation from detention ponds. If the flow volume  needed
    to be reduced, the model could be used to determine the size of pond required to reduce the
    volume to the desired level. The pollution model is nearly complete as well, and should  provide
    similar comparative data and scenario analysis capabilities.
    70 -,
    60 -
    50 -
    40 -
    30 -
    20 -
    10 -
     0 -
                         Miller Creek - SWMM Prediction
                         Miller Creek - Gauge Data
                 Aug15  16  17  18  19  20  21   22  23  24   25   26   27   28
                                           !->„<.,-, I* r\r\o\
          Figure 2. Predicted and observed discharge from middle reach of Miller Creek.

              . Miller Hill Mall: Outlet to Miller Creek (29.2 acres)
               Subcatchment # 21; Wooded (44.6 acres)
                                  10:00    11:00    12P.M.     1:00
                                        Time (Aug. 16, 1998)
    Figure 3. Comparison of hydrographs above and below mall development area on Miller Creek.
    Land use planning resources for northern Wisconsin
    Our second prototype effort involves compiling spatial data and other resources to support local
    units of government in citizen-based land use planning exercises. The objective of this exercise
    is to compile data and tools relevant to local-scale land use planning onto a CD-ROM for use by
    local units of government. Within the watershed, these are primarily towns and townships. The
    CD will contain the following elements:
              basic information describing the processes involved in planning
              spatial data - land use, transportation, rivers and lakes, natural features, political
              boundaries, and other data layers relevant to local-scale planning.
              planning tools - example surveys and ordinances language, zoning policies,
              development/preservation strategies, and other instruments that a local government
              could tailor to its specific needs
              landscape graphics - air photo or line drawing examples of different scenarios
              illustrating housing density and patterns (clustered or dispersed), riparian buffers,
              and other land use strategies. This would provide information on what a future
              landscape might look like under particular management strategies
              Example landscape plans developed by cooperating governments

    The CD will be written in standard HTML code to allow access with available web browsers.
    Publicly available map viewing tools such as Arc Explorer® will also be provide to provide basic
    querying of map products. The CD will be designed to stand alone, but will also contain links to
    web sites to access a broader range of information.
    Education and citizen involvement are central to understanding the data and tools available on
    the CD (Figure 4). To assist in issue identification,
    example citizen surveys will be available from the Land-
    use Issues menu.  Data will be available at both
    township and county levels, the latter allowing  for
    contextual analysis of issues. Tools for assessing the
    suitability of planning elements will also be available.
    Example landscape plans will be available, along with
    information on implementing and administrating
    landscape plans once they are developed.
    This CD-based resource will be developed and tested
    in cooperation with a number of partners from local and
    state governments. The anticipated completion and
    delivery date of the CD is June 2000. The information
    on the CD will also be available through the LSGIS
    web site.
    Figure 4. Schematic of LSDSS
    planning support CD architecture.
    The Lake Superior Decision Support project is an integrated effort to provide data and planning
    tools to citizens and local units of government to assist in land use planning efforts. This work
    should provide a means of reducing the cumulative impacts to the Lake Superior Basin through
    informed decision-making at the local level.
    This project was funded by the US Environmental Protection Agency through the Minnesota
    Department of Natural Resources. We acknowledge the encouragement and support of the
    Lake Superior Ecosystem Cooperative, where the key ideas of the project originated. Mark
    White, Gerry Sjerven, Jesse Schomberg and Amy Trauger have been integral parts of this

    effort. A special acknowledgement goes to cooperator Mike Koutnik of Environmental Systems
    Research Institute for his support and contributions to data and kiosk development. Sandy
    Shultz of Ashland-Bayfield-Douglas-lron Land Conservation Department and David Lonsdale of
    the Great Lakes Aquarium have provided valuable support and advice to the project.

      A GIS-based Approach to Predicting Wetland Drainage and Wildlife
      Habitat Loss in the Prairie Pothole Region of South-central Canada
                                       David Howerter
                          Institute for Wetland and Waterfowl Research
                                   Ducks Unlimited Canada
                                       P.O. Box1160
                             Stonewall, MB, Canada ROC 2ZO and
                                         Lee Moats
                                   Ducks Unlimited Canada
                                    16064th Avenue Regina
                               Saskatchewan, Canada S4P 3W7
    The prairie pothole region covers approximately 870,000 km2 of the north-central United States
    and south-central Canada (Batt 1996). The area extends from the tall grass prairies of
    northwestern Iowa west to the short grass prairie of southern Alberta and north to the boreal
    transition aspen (Populus tremuloides) parklands of central Saskatchewan (Figure 1).
    Approximately 13,000 years ago, retreating glacial ice revealed a rolling terrain comprised of
    end, ground and lateral moraines; glacial drift and till; and eskers (Winter 1989, Batt 1996).
    Remnant ice remained within the till after the main body of the glaciers retreated. As this ice
    melted, kettles were created. Subsequently, these kettles filled with water and became the
    potholes of modern vernacular. (Pielou 1991, Batt 1996). The majority of potholes are not
    connected by a natur; Figure 1.  Prairie Pothole Ecoregion
    Prairie wetlands serve a variety of hydrological/ecological functions including storage and
    control of surface water (Winter 1989, Miller and Nudds 1996, Murkin 1998), recharge of
    groundwater supplies (Winter 1989, LaBaugh et al. 1998, Murkin 1998), filters for sediments
    and agricultural chemicals (Grue et al. 1989, Neely and Baker 1989, Gleason and Euliss 1998,
    Goldsborough and Crumpton 1998,  Murkin 1998), sinks for excess nutrients (Crumpton and
    Goldsborough 1998,  Murkin 1998), and habitat for a wide array of invertebrate, fish and wildlife
    species (e.g., Batt et al. 1989, Fritzell 1989,  Peterka 1989). This area is particularly important to
    waterfowl where > 50% of the continental duck populations are annually produced (Batt et al.

                             Figure 1. Prairie Pothole Ecoregion
    European settlement of the PPR began around 1878. Farming was the primary motivation for
    settling the area and agriculture remains a dominant economic force in the region (Leitch 1989,
    Leitch and Fridgen 1998). Accompanying this expansion of agriculture was a dramatic
    transformation of the landscape. Samson and Knopf (1994) estimate that as little as 23% of the
    pre-settlement prairie remains in Canada. Conversion in the parklands (Figure 1) has been even
    more extensive (Turner et al. 1987). Similarly, > 70% of the wetlands in the PPR have been
    drained or severely degraded (Turner et al. 1987, Dahl 1990, Batt 1996).
    These changes to the landscape have had consequences for a variety of ecosystem functions
    throughout the region. For example, Miller and Nudds (1996) attributed increased flooding within
    the Mississippi River valley over the past decades to reduced natural upland vegetation and
    wetland drainage Other hydrologic functions such as groundwater recharge and the ability to
    filter agrochemicals are similarly reduced by drainage. The focus of this paper, however, is the
    effect these changes have had on the PPR's ability to support wildlife populations-specifically
    upland-nesting duck species.

    In 1986, in response to declining duck populations attributed to land use changes throughout
    North America and exacerbated by prolonged drought on the prairie nesting grounds, the
    governments of the United States and Canada signed into law the North American Waterfowl
    Management Plan (NAWMP; joined in 1994 by Mexico). The goal of the Plan was to return
    waterfowl populations to the levels of the mid-1970's—a period with abundant waterfowl
    populations. The Plan was structured as a number of joint ventures targeting either species of
    concern or regions of special importance to waterfowl populations. Two of these joint ventures
    (Prairie Habitat - Canada; Prairie Pothole - U.S.) focus on restoring nesting habitats within the
    During the planning stages for the  Prairie Habitat Joint Venture (PHJV), one of the underlying
    tenets was that policy changes would ensure that existing habitats remained intact. Ultimately,
    this has not occurred. Draining wetlands and conversion  of lands for
    agriculture continues unabated—particularly in Canada. For instance, between 1971 and 1996
    natural land  has declined by 2.4 million hectares within the prairie pothole region of Canada
    (Figure 2) while there has been a corresponding increase in cultivation (Figure 3). Likewise,
    wetlands continue to be lost.  For example, in a single region near Wadena, Saskatchewan, the
    number of wetland hectares declined from 3,019 to 563 between 1949 and 1998. Between 1980
    and 1998, 41% of the remaining ha were drained (Ducks Unlimited, unpublished data). In the
    U.S. portion  of the PPR,  nearly 15,000 wetland ha continue to be degraded annually, despite
    provisions that have linked wetland protection to agricultural subsidies (Tiner 1984). If, as
    currently proposed, conservation provisions are de-coupled from agricultural subsidies, the
    incentive to retain wetlands will be  further weakened (J. K. Ringelman, Ducks Unlimited, Inc.,
    Bismarck, ND).
    Exacerbating the problem of continued loss of existing upland and wetland habitats, recent
    assessments of PHJV habitat programs indicate that restored grasslands may be less
    productive than existing native/naturalized cover for nesting ducks (Institute for Wetland and
    Waterfowl Research, unpublished data). Therefore, a paradigm shift was in order for habitat
    managers faced with a constantly eroding habitat base coupled with restoration efforts that may
    be less effective than envisioned. Instead of primarily converting cropland back to grassland
    vegetation, it has become increasingly clear that preserving existing habitats may be a more
    effective strategy. However, limited conservation resources preclude the possibility of securing
    all existing cover,  and conversion rates vary spatially. Therefore, to prioritize which  habitat

    Figure 2. Decline of natural land in the Canadian prairie
    pothole region 1971-1996. (Source: Statistics Canada 1996)
        Figure 3. Change in cropped areas 1971-1996 (Source:
        Statistics Canada Agriculture Census 1996)

    parcels to secure, habitat managers need to identify the areas with the highest potential for
    waterfowl production that are also at high risk of conversion to agriculture. Our objective, then,
    is to develop a spatially-explicit model to project where land conversion is most likely. This
    model ideally will be will be hierarchical allowing identification of "high risk" areas at both the
    regional and local scales. For this paper, we discuss our plan for the regional model in some
    detail and touch only briefly on additional considerations for the local model.
    Hypothesized relationships
    For conservation dollar investment decisions to be made wisely, resource managers require a
    spatially-explicit tool to project which areas are most likely to be converted to cropland in the
    near future.  To  allow quantification of the relationships that will allow optimization of resource
    expenditures, risk of conversion to agriculture can be thought of as an expected rate of habitat
    loss. In essence, the decision about where to expend resources on securement of existing
    habitats within the foreseeable planning horizon takes the form:
    8P = f(P, Cs, R), where:
           8P = change in duck production/habitat expenditure
           P =  current duck production (1 [Amount of suitable upland cover, wetland
           Cs = site-specific cost of conservation, and
            R = risk = (E[loss of habitat]).
    Our hypothesized relationships are depicted in Figure 4. In Figure 4a, we've speculated that Cs
    for a piece of land is likely correlated with its inherent fertility. Therefore, the least expensive
    land most likely has low productivity for agriculture and ducks (e.g., Manitoba's Interlake).
    Alternatively, the most costly (productive) land likely has been already largely converted to
    agricultural and/or residential purposes. As a result, the areas that are currently most productive
    for duck production likely occur on land with marginal suitability for farming and at intermediate
    land costs.
    Following similar reasoning, we have hypothesized that R, too, is highest for lands with
    intermediate costs (Figure 4b); poor quality land (low cost) likely isn't worth converting to
    cropland for economic reasons, and high cost land likely already has been converted.

          Figure 4. Hypothesized relationships between: (a) cost of securing a given
          piece of habitat and potential duck production, (b) the risk of habitat being
          converted to agriculture and cost of securing the habitat, (c) potential duck
          production and risk of conversion, and (d) all three  components.
    Because we have hypothesized that both P and R are unimodal in relation to Cs, P and R are
    positively associated when plotted against each other (Figure 4c). Combining these
    relationships into a single model results in a relationship with one realization that may resemble
    Figure 4d. Figure 4d depicts several interesting points.  First, the best land for duck production is
    also the land at highest risk of conversion. Second, happily, this land likely is of only
    intermediate cost.
    Model Development
    Several techniques have been proposed for predicting land use change. The simplest of which
    is to use historical trends to project forward in time making the assumption that past trends will
    continue into the future. Empirical models based on posited driving forces may, however,
    provide better predictions (Robinson et al. 1994) and allow for exploration of causative factors.

    To develop our model we plan to use relatively recent (e.g., since 1971) conversion data
    (source: Statistics Canada - Census of Agriculture) and a number of candidate explanatory
    variables (Table 1). An information theoretic approach (e.g. Akaike's Information Criterion;
    Burnham and Anderson  [1998]) will be used with regression analysis to select the most
    parsimonious set of explanatory variables.
    Table 1. List of candidate variables useful for predicting land conversion to agriculture in
    the prairie pothole region.
    Candidate variable
    Landowner demographics
    Soil type
    Crop profitability
    Proximity to an existing
    water conveyance
    Hypothesized relationship
    Areas with relatively steep topography
    are difficult to cultivate and/or drain
    Conversion most likely when land
    changes ownership
    Related to cropping practices
    Input costs and crop prices influence an
    individual's decisions on how best to
    use their land
    Proximity to water conveyance makes
    drainage less costly
    Data Source
    Canada Surveys
    and Mapping
    Statistics Canada
    Canada Soil
    Information System
    Provincial water
    Topography and proximity to an existing water conveyance both relate to the cost of conversion,
    while soil type and crop profitability both relate to the benefit received by the manager for
    converting. Crop profitability will require a separate "sub-model" with spatially-explicit crop
    prices and input and transportation costs. Alternatively, land prices may serve as a suitable and
    less data-intensive proxy for crop profitability for model development.
    The model we have described will be developed using (recent) historical data. We realize,
    however, that extrapolating forward based on past relationships is inherently dangerous.
    Therefore, a series of monitoring stations will be established throughout the areas targeted as
    important for nesting ducks. Through an adaptive process uncertainty will be reduced in time.

    Using empirical Bayesian techniques, these stations will allow us to iteratively update model
    parameters. This process will yield immediate results with increasing confidence overtime.
    Because much of the data used to build the regional risk model will be at a fairly coarse scale
    (50-200 km2), additional information likely will be needed at the finest geographic scale to
    accurately predict the risk of an existing patch of habitat being converted to cropland. Factors
    such as landowner demographics and topography will likely still be important predictors, but
    added factors such as size of the habitat patch and surrounding land use will also likely be
    significant. Also, while true crop profitability might be predictable through entirely empirical
    information,  it almost certainly will be much more difficult to predict the vagaries of individual
    landowner decisions. To attempt to explain how decisions are made, new data will need to be
    gathered to determine how producers determine profitability when clearing, breaking or draining
    Integrating the Model: A Decision Support System
    The model that we  have described in this paper represents one component of a decision
    support system designed to optimize expenditures of conservation dollars. Simultaneously,
    spatially-explicit models to predict duck production and the cost of securing important habitat
    areas also will be developed (Figure 5). The combined output from these 3 components should
    allow habitat managers to make informed decisions about how best to prioritize expenditures.

              Decision Support System
                                                      T            V.-".'-^.
                 Figure 5. Conceptual model of Decision Support System
    We thank K. Guyn for helpful comments on a previous draft of this paper. J. Holland and B.
    Kazmerick provided technical assistance and assisted with preparation of the figures. K.
    LePoudre provided unpublished data.

    Batt, B. D. J. 1996. Prairie ecology—prairie wetlands. Pages 77-88 in F. B. Samson and F. L.
          Knopf, editors. Prairie conservation: preserving North America's most endangered
          ecosystem. Island Press, Washington, D.C. 339pp.
         _, M. G. Anderson, C. D. Anderson, and F. D. Caswell. 1989. The use of prairie potholes
           by North American ducks. Pages 204-227 in A. van der Valk, editor. Northern prairie
           wetlands. Iowa State University Press, Ames. 400pp.
    Burnham, K. P. and . R. Anderson. 1998. Model selection and inference: a practical information-
           theoretic approach. Springer-Verlag, New York. 353pp.
    Crumpton, W. G. and L. G. Goldsborough. 1998. Nitrogen transformation and fate in prairie
           wetlands. Great Plains Research 8:57-72.
    Dahl, Thomas E. 1990. Wetlands losses in the United States 1780's to 1980's.U.S. Department
           of the Interior, Fish and Wildlife Service, Washington,D.C. Jamestown,  ND: Northern
           Prairie Wildlife Research Center Home Page.
           (Version 16JUL97).
    Fritzell, E. K. 1989. Mammals in prairie wetlands. Pages 268-301 in A. van der Valk, editor.
           Northern prairie wetlands. Iowa State University Press, Ames. 400pp.
    Gleason, R. A., and N.  H. Euliss, Jr. 1998. Sedimentation and prairie wetlands. Great Plains
           Research 8:97-112.
    Grue, C. E., M. W. Tome, T. A. Messmer, D. B. Henry, G. A. Swanson, and L.  R. DeWeese.
           1989. Agricultural chemicals and prairie pothole wetlands: meeting the  needs of the
           resource and the farmer—U.S. perspective. Transactions of the North American Wildlife
           and Natural resources Conference 54:43-58

    Goldsborough, L. G., and W. G. Crumpton. 1998. Distribution and environmental fate of
           pesticides in prairie wetlands. Great Plains Research 8:73-95.
    LaBaugh, J. W., T.  C. Winter, and D. O. Rosenbery. 1998. Hydrologic functions of prairie
           wetlands. Great Plains Research 8:17-37.
    Leitch, J. A. 1989. Politicoeconomic overview of the prairie potholes. Pages 3-14 in A. van der
           Valk, editor. Northern prairie wetlands. Iowa State University Press, Ames. 400pp.
         _, and P. Fridgen. 1998. Functions and values of prairie wetlands: economic realities.
           Great Plains Research 8:157-168.
    Miller, M. W., and T. D. Nudds. 1996. Prairie landscape change and flooding in the Mississippi
           River valley. Conservation Biology 10:847-853.
    Murkin, H. R. 1998. Freshwater functions and values of prairie wetlands. Great Plains Research
    Neely, R. K. and J. L. Baker. 1989. Nitrogen and phosphorus dynamics and the fate of
           agricultural runoff. Pages 92-131 /nA. van der Valk, editor. Northern prairie wetlands.
           Iowa State University Press, Ames. 400pp.
    Peterka, J. J. 1989. Fishes in northern prairie wetlands. Pages 302-315 in A. van der Valk,
           editor. Northern prairie wetlands. Iowa State University Press, Ames. 400pp.
    Pielou, E. C. 1991. After the ice age: the return of life to glaciated North America. University of
           Chicago Press, Chicago. 366p.
    Robinson, J., S. Brush,, I. Douglas, I.E. Graedel, D. Graetz, W. Hodge, D. Liverman, J. Melillo,
           R. Moss, A. Naumov, G. Njiru, J. Penner, P. Rogers, V. Ruttan, and J. Sturdevant. 1994.
           Land-use and land-cover projections: report from working group C.  Pages 73-92 in W. B.
           Meyer and B.  L. Turner, II, editors. Changes in land use and land cover: a global
           perspective. Cambridge University Press, Cambridge, U.K. 537pp.

    Samson, F. and F. Knopf. 1994. Prairie conservation in North America. Bioscience 44:418-421.
    Tiner, R. W., Jr. 1984. Wetlands of the United States: current status and recent trends. U.S.
           Dept. of the Interior, Fish and Wildlife service, Washington, D.C. 59pp
    Turner, B. C., G.  S. Hochbaum, F. D. Caswell, and D. J. Nieman. 1987. Agricultural impacts on
           wetland habitats on the Canadian Prairies, 1981-1985. Transactions of the North
           American Wildlife and Natural Resources Conference 52:206-215.
    Winter, T. C. 1989. Hydrologic studies of wetlands in the northern prairie. Pages  16-54 in A. van
           der Valk, editor.  Northern prairie wetlands. Iowa State University Press, Ames. 400pp.

      Onsite Wastewater Management Program in Hamilton County, Ohio
                               - An Integrated Approach
               to Improving Water Quality and Preventing Disease
                        Tim Ingram, Terry Hull, Travis Goodman, and Staff
                         Division of Water Quality, Hamilton County, Ohio
    Postwar economic and population growth of the 1950s launched Hamilton County, Ohio,
    families into suburbia. Ohio's most southwestern county and home of Cincinnati expanded
    typically. Construction of suburban residential housing outstripped the development of traditional
    urban infrastructure. But extended aeration technology made it possible to put a miniature
    wastewater treatment plant in the yard of every new homeowner. Home aeration units were
    marketed as virtually maintenance-free devices capable of producing effluent with the quality of
    drinking water. These fallacies, in concert with the natural limitations of the sites being
    developed and a community willing to believe that which was too good to be true, brought public
    health consequences.
    Since most of Hamilton County escaped the advance of the last (Wisconsinan) glacier, its steep
    shale ridges remain covered with  thin top soils and slowly-permeable silty-clay subsoils. Only
    small areas in the valleys of the three principal streams are underlain by the more-permeable
    glacial deposits. Seasonal rain and snow melt runoff cut the ridges with  innumerable erosion
    channels that join to form intermittent streams. Conditions that limited the use of leaching
    devices appeared to be optimal for use of aeration  units with surface discharges. Supported by
    exaggerated claims of operational efficiency, public officials were quick to accept the technology
    as a way to support new tax-generating development without the cost of sewer construction.
    Developers enthusiastically supported a public policy that reduced their capital costs and raised
    profits. Homeowners believed that they had an effective, low cost, state-of-the-art sewage
    treatment device. With discharges running to any downhill drainage way, out of sight/out of mind
    became the prevailing attitude, and it took a long time to change.
    By the mid-1980's, sewage contamination of the  West Fork of the Mill Creek elicited strong
    public reaction. When a hepatitis A outbreak occurred in 1989, the public clamor over lack of
    public sewers, ineffective sewer system operation,  and poorly functioning home aeration units
    intensified. Public alarm further increased when a child playing in a stream polluted by aeration

    systems caught a rare protozoan infection and was hospitalized for days.
    By the early 1990's, the total number of units was unknown, but estimates ranged from 20,000
    to 40,000 units.  Public perception regarding the public health impact was markedly changed. An
    irate homeowner persisted until the Hamilton County General Health District's Board of Health
    officially declared her neighborhood a public health nuisance and requested the State to order
    construction of a public sewer. Public concern peaked when the Board of Health authorized use
    of aeration units in a proposed subdivision where the effluent would drain directly into a
    downstream homeowner's recreation  lake. The resulting legal action brought Ohio
    Environmental Protection Agency and Ohio Department of Health sanctions against the Board
    of Health. A political response occurred in 1993 as Board members were replaced and a new
    Health Commissioner (pubic health officer) was employed. The new Board determined that
    water quality improvements and waterborne disease prevention were dependent on the
    effective use of household sewage disposal systems. Thus, the Board made a commitment to:
          •   Update its household sewage regulation, which had been in force since 1959;
          •   Reconsider its previous approval actions regarding the usage of aeration sewage
              systems in residential subdivisions;
          •   Inventory and evaluate existing home sewage systems;
          •   Build community partnerships and educate homeowners about system operation and
              maintenance requirements;
          •   Expedite complaint investigations; and
          •   Strengthen enforcement when sewage system repairs are necessary.
    This paper discusses the establishment of an operation permit program which was developed to
    achieve several  of these commitments. The purpose of this paper is to describe the components
    of the operation  permit program and the steps that have been taken to improve water quality in
    Hamilton County streams.
    Building the Operation Permit Program
    Policy Development
    In Ohio,  local health districts have sewage system permitting authority for one-, two-, and three-
    family dwellings. These districts enforce either the minimum state code or more stringent
    regulations adopted by the local board of health. Under the minimum code, a newly constructed

    system is automatically permitted for operation upon final construction approval. This operation
    permit remains in force until it is revoked by board action. Such permits are rarely revoked.
    Some health districts issue operation permits for existing sewage systems. In 1994, the
    Hamilton County General Health District (HCGHD) initiated a comprehensive operation permit
    program. Under this program operating permits are issued for both new and existing household
    sewage disposal systems.
    On December 13,  1993, the Board of Health adopted a new household sewage disposal code
    and created a new division to enforce it. Previously, the sewage permitting functions were a part
    of the Health District's  plumbing division. To implement the Board's commitments, the Division
    of Water Quality and Waste Management was established. The main function of this division
    was to implement  Regulation 529, the Household Sewage Disposal Code.
    To guide the new program, the Hamilton County Board of Health established the following
          •  No permit fee will be billed until an inspection of the household sewage disposal
              system(HSDS) is completed and the owner is provided with a written inspection
          •  All household sewage disposal systems (HSDS) with  electrical components (aeration
              units, etc.) are subject to an annual inspection and, if the system is found to be
              operating properly, a one-year operation permit will be issued.
          •  All non-mechanical or non-electrical household sewage disposal systems (HSDS)
              will be inspected once every three years and,  if found to be operating, a three-year
              operation permit will be issued (effective on Sept.  11,  1995).
          •  All newly-constructed dwellings with newly-installed household sewage disposal
              systems (HSDS), permitted and approved by the HCGHD, will be exempted from the
              operation permit program for a period of two years following construction approval.
    Critical to the success  of the operation permit program was the establishment of inspection
    criteria for determining proper operation. Considering the varying ages of the systems in use
    and the  different standards under which they were manufactured and installed, a balanced
    approach was needed that would achieve water quality objectives while maintaining public and

    political support. The Health Commissioner and staff were charged with the development of the
    inspection criteria.
    It was decided that system performance would be evaluated by observation rather than effluent
    sampling. This decision took into account the fact that many HSDS had no separate access for
    sampling, as well as the unlikelihood that most older systems could meet effluent quality
    standards established by the Board of Health. Also, considered was the cost to sample as many
    as 20,000 sewage systems and that many  individual systems were connected to common
    collector lines. Collector lines function as private sewers that transport effluent from several
    homes. The ten observational criteria selected to establish improper performance for
    mechanical HSDS, thus resulting in the designation "disapproved" are:
           1)     Motor missing,
           2)     Motor inoperable (cold),
           3)     Motor not drawing air or insufficiently drawing air,
           4)     Broken lid(s), i.e., piece missing or broken to the extent that it allows entrance of
                  surface water, or lid cannot be lifted without collapse; decayed metal grating,
           5)     Flooded filter,
           6)     Visual evidence of septic sewage, i.e., black, odorous,
           7)     Visual evidence of electric service problem,
           8)     Components are not functioning in accordance with design standard,
           9)     Discharge creates a public health nuisance, and
           10)     An access riser has not been brought to grade over each compartment requiring
    Fee Structure
    Policies concerning program fees have changed as program activities evolved and knowledge
    was gained regarding operating costs and the willingness of HSDS owners to pay. At the outset,
    the Board of Health established a single operation  permit fee of $40.00. As program
    implementation progressed, two problems became apparent. First, there was no way to recover
    the cost of the additional inspections required to  achieve and confirm system repairs when the
    initial inspection resulted in disapproval. Second, no permit was issued to HSDS owners who
    failed to pay their permit fee. Consequently, those who were delinquent soon forgot about the
    need for an operation permit as well as their indebtedness. The Health  District then entered a

    new era-marked by the need for dunning letters and small claims court appearances-in which it
    acted somewhat like a small utility company.
    In July of 1995, the fee structure was changed based on a report issued to the Board of Health
    entitled, Aeration Sewage Disposal Systems in Hamilton County. This report evaluated the
    progress made to date. Based on community input, the Board implemented a variable fee
    structure. The annual fee covering the initial inspection and permit was reduced to $30.00. A
    reinspection fee of $30.00 was established for second and subsequent re-inspections. As an
    inducement for HSDS owners to provide regular maintenance, the second reinspection fee was
    set at $15.00 for owners holding a maintenance contract with a registered and bonded service
    Delinquent permit fees rose as high as 62% but annually averaged around 13% for the 1994-
    1995 time period. This amounted to $32,320 of uncollected permit fees at the end of 1995. In
    1996 a collection agency was retained by the Health District at a cost of  $4.95 per outstanding
    account. Also, a $10 late fee was assessed against each delinquent account. Delinquent fees
    dropped to an average of 9% in subsequent years. Many homeowners were more threatened by
    a bad credit rating than criminal prosecution. Nonetheless, the percentage of unpaid accounts
    was still too high. In 1998, the HCGHD worked with local Ohio legislators, and legislation was
    passed  allowing for unpaid operation permit fees to be certified by the Health Commissioner to
    the County Auditor for placement as a lien on the property. With the passage of this legislation,
    the "playing field" has been leveled so that all HSDS owners have a financial stake in the
    operation permit program.
    Staff Development
    The development and implementation of policies regarding Health District staff and staff activity
    has been crucial to  successfully carrying out the operation permit program. These  policies have
    been established largely by management and  include  issues such as, personnel qualifications,
    inspection protocol, and training.
    Field work was done initially by registered sanitarians and sanitarians-in-training. During the first
    year staff turnover was unacceptably high.  Newly-graduated environmentalist quickly gained
    valuable field experience, but were just as quickly tired by the large volume of inspections and
    the repetitive, sometimes confrontational nature of the work. Experienced sanitarians soon

    found the work not challenging. Management found the solution rested in the employment of
    technicians as inspectors. With specialized training and sanitarian oversight, these staff have
    maintained their enthusiasm and perform their inspections in a highly competent and efficient
    manner. The water quality technicians have worked well with HSDS owners and the repair
    contractors, too.
    The effectiveness of the inspection staff can be attributed to the development of operating rules
    which emphasize thorough training, clear identity, and consistency in performance of duties.
    Each inspector receives detailed training regarding the operation of each brand of aeration unit
    that he/she will inspect in the field. Contact with repair contractors is encouraged, and where
    possible, attendance at manufacturer's training school is supported. Inspector training in
    communication and public relations is also provided, while membership in professional
    environmental associations is encouraged.
    Inspectors adhere to a dress code wearing distinctively colored shirts and jackets. Clearly
    visible photo identification badges are worn. Each inspection begins with a stop at the front door
    to explain the purpose of the visit to the citizen, followed by a brand-specific standardized
    inspection of the  sewage system, and, if the owner is present informing him of the inspection
    The Inventory - What Was Found
    Aeration sewage systems are a high maintenance type of household sewage disposal system.
    These systems contain electrical components and filters which require maintenance by the
    homeowners. Homeowners were unaware, for the most part, of these maintenance needs.
    The media reported that 40,000 aeration sewage systems discharged untreated waste water
    into homeowners' backyards. However, no one knew for sure the exact number nor their
    operating condition.  Installation permit records were not complete. While citizens worried about
    disease and pollution, an Ohio EPA connection ban and class action lawsuit put the  Hamilton
    County Board of  Health in a reactive mode. There was a rush to inventory and evaluate the
    operating status of all existing aeration sewage systems first. The Board of Health directed the
    Health Commissioner to address the following questions:

           1)     How many aeration sewage systems exist in Hamilton County?
           2)     What are the manufactured types?
           3)     Where are the aeration sewage systems located?
           4)     What is the operational status of these systems?
           5)     How much water pollution and how many health nuisances have been created
                 from these aeration sewage systems?
    From January 1994 through July 1995, sanitarians and water quality technicians blanketed the
    County inspecting home aeration systems. The staff visited over 10,000 properties - many on
    several occasions -to find aeration systems, locate collector lines, consult with homeowners, or
    reinspect systems for compliance. Inspection sheets were filled out and information entered into
    a custom-designed database for ease of tracking.
    Within eighteen months 9,145 home aeration sewage systems were located and inspected. Six
    manufactured-types of  home aeration sewage systems were found. (See Table 1). The high
    percentage of the Cavitette brand in use was a significant concern because there was no active
    manufacturer or replacement parts available. The Cavitette brand was manufactured locally in
    the late 1950's through the 1960's. The design of this system was based on less stringent
    standards than the National Sanitation Foundation  Standard 40.
                                           Table  1
                    Percent of Total Home Aeration Units by Manufacturer
                     Manufacturer                    Percent
                        Cavitette                                    22.4
                         Coate                                      10.6
                           Jet                                       36.6
                         Multiflo                                      2.4
                        Norweco                                     0.1
                         Oldham                                     27.9
    Of the 9,145 aeration systems inspected, 34 percent or 3,077 were disapproved during this time
    period. Table 2 summarizes the number of that had been repaired by July 1995 and the average
    number of reinspections to obtain compliance.

                                          Table 2
                        Disapproved Systems with Completed Repairs
    Number of Number of Mean reinspects
    systems reinspects to compliance
    Many of the home aeration sewage systems connected to a common yet private sewer line
    known as a collector line. These collector lines served as few as two homes or as many as 50
    homes. The effluent from the collector lines discharged into storm water sewers, ephemeral
    streams, and onto the ground surface. None of the collector lines had National Pollutant
    Discharge Elimination Systems (NPDES) permits. It had been observed and reported that the
    discharges from these collector lines had deteriorated the water quality in the county's streams
    and waterways. Prior to 1994, there was no documented wastewater monitoring of these
    collector lines. However, in the summer of 1994 as a part of the operation permit program,
    Project CLEAN (Collector Line Evaluation and Assessment of Needs), was implemented.
    Wastewater samples were taken from 197 collector lines. The samples were analyzed for
    biochemical oxygen demand (BOD), suspended solids (SS),  and fecal coliform bacteria. The
    data results from the initial round of sampling are shown in Table 3.

                 Table 3
    Project CLEAN Effluent Sample Data
    Township/ Number of Fecal
    Municipality samples under 5,000
    Anderson 7 2
    Colerain 47 13
    Crosby 1 0
    Delhi 6 2
    Green 104 34
    Harrison 4 2
    Madeira 13 1
    Miami 7 0
    Springfield 4 0
    Sycamore 1 1
    Symmes 2 0
    Whitewater 1 0
    Total 197 55
    Water Quality Standards
    Fecal Coliform (colonies/100 ml)
    Suspended Solids (mg/l)
    Biochemical Oxygen Demand (mg/l)
    Ammonia Nitrogen (NH3) (mg/l)
    Dissolved Oxygen (mg/l)
    5,000 to 50,000
    over 50,000 TNTC
    Ohio EPA
    1 (summer),
    3 (winter)
    No. meeting

    When the Board of Health adopted a new sewage code in late 1993, they also codified
    discharge standards for suspended solids, biochemical oxygen demand, and fecal coliform
    bacteria, which are:
       •         BOD: 20 mg / L
       •         Suspended Solids: 40 mg / L
       •         Fecal Coliform Bacteria: < 5,000 colony forming units /100 ml
    Table 3 reveals that 31 or 16% of the collector lines met the Board of Health discharge
    standards. Nearly three-fourths (72%) failed to meet the standard for fecal coliform bacteria.
    This was not surprising since disinfection devices had not been installed on many of the
    individual aeration sewage systems. The initial inventory and evaluation of aeration sewage
    systems was completed.
    Non-mechanical Sewage Systems
    A contention by County citizens early in the program was that the Health District ignored the
    pollution created by non-mechanical household sewage disposal systems, like leach lines and
    dry wells. Once the initial inventory and evaluation of aeration (or mechanical) sewage systems
    was completed, the Health District staff began the inventory and evaluation  of non-mechanical
    sewage systems in 1996. Table 4 shows the number and the type of non-mechanical systems
    inspected in 1996 and 1997, and the number of total units that failed the initial inspection.
                                           Table 4
                              Non-mechanical Sewage Systems
    Dry we 1 1
    Leach Lines
    Subsurface Sandfilter
    Dry we 1 1
    Leach Lines
    Subsurface Sandfilter
           Total initial failure 59 (4.2%)
    Total initial failure 117 (13.3%)

    The inventory and evaluation of non-mechanical sewage systems is a work-in-progress. By May
    17, 1998, a total of 2605 non-mechanical sewage systems had been assessed.
    The following inspection criteria are used to approve or disapprove a non-mechanical household
    sewage disposal system.
           Approve if:
                 a)     System components can be located and
                 b)     The system uses in-soil dissipation and there is no surface seepage of
                        gray, malodorous effluent (minor seasonal wetness is acceptable).
                 c)     The system uses surface discharge, the discharging effluent is clear, and
                        not malodorous, does not pond on the inspected property or on adjacent
                        property, and the discharge pipe is accessible with sufficient freeboard to
                        allow collection of an effluent sample.
           Disapprove if:
                 a)     Gray malodorous sewage is seeping to the ground surface creating a
                 b)     Gray, malodorous sewage is discharging from the sewage system to an
                        adjacent property or roadside drainage way.
                 c)     Any similar condition is occurring which is creating public health nuisance.
    Upon the initial inspection, the percentage of non-mechanical systems disapproved overall
    averaged 9%.
    The inventory and evaluation of these sewage systems continues. The District did not meet its
    goal  of inspecting and inventorying all non-mechanical sewage systems by December 31, 1998.
    On March 8, 1999, the Board of Health changed the non-mechanical permit from a three-year
    permit to a five-year permit.
    Building Community Partnerships
    All the media attention over the lack of governmental oversight and widespread pollution did not
    convince the vast majority of citizens with  home sewage  systems that the operation permit
    program was necessary. Media and citizenry criticism for the  Health District's historic lack of
    oversight was replaced by concern about  Health District persistence in requiring sewage system
    repairs and payment of permit fees. During the first eighteen months of the program, the Health

    District staff handled 20 to 40 telephone calls per week. About 265 citizens felt strongly enough
    to write letters protesting the program. The theme of the letters ranged from why inspect septic
    systems to a governmental invasion of their property rights and privacy.
    The Health District initiated a community education program in order to gain public acceptance
    of the operation permit program, to improve political support, and to teach homeowners about
    HSDS operation and maintenance. The education program utilized a variety of strategies.
    Brochures were developed and mailed to homeowners and handed out at neighborhood
    festivals and community halls. Numerous press releases and editorials in the local newspapers
    about the benefits of the operation permit program to the community were published. Health
    District staff conducted presentations at neighborhood gatherings or backyard barbecues about
    the importance of HSDS.
    The Metropolitan Sewer District (MSD) and the Health District forged a partnership and
    collaborated at several public meetings. The two agencies worked in concert to extend public
    sewers into those watersheds where sewage nuisances were prevalent and HSDS upgrades
    were not feasible.
    A public sewer assessment credit was established by the Board of Hamilton County
    Commissioners. This credit was the brainstorm of MSD and Health District officials as well as a
    group of western Hamilton County business leaders and elected officials. The credit program
    stated that all single family dwellings with HSDS in existence on or prior to September 20, 1995,
    were eligible for $5,000.00 credit  towards their public sewer assessment. The public sewer
    assessment credit helped convince many homeowners to sewer their neighborhoods.
    Using GIS
    In 1996, the Health District began using a geographical information system (GIS) known as
    CAGIS (Cincinnati Area Geographic Information System). CAGIS technology allows the Health
    District to place all home sewage systems, stream quality data, and communicable disease data
    on computer generated maps. CAGIS technology allows layers of information to be overlaid on
    top of each other in order to carry out analyses. For instance, the public sewer system layer can
    be overlaid with the County home sewage system layer. This allows the user to quickly observe
    the proximity of public sewers to home sewage systems. Other layers of data, such as stream
    data and communicable disease information, are overlaid to look for patterns or clusters of
    disease or pollution associated with home sewage systems. The CAGIS technology is a
    powerful public health surveillance tool for targeting resources.

    Results - Has the Operation Permit Program Made a Difference?
    Data comparison between two time periods, 1994 -1995 and 1996-1997, will be used to reveal
    program successes and failures. Table 5 compares pass / fail inspections of home aeration
    systems for the two time periods.
                                          Table 5
                   Pass / Fail Inspections of Home Aeration Sewage Systems
    Number inspected
    Percent failed
    first inspection
    33.1% (4,962 no.)
    Number inspected
    Percent failed
    first inspection
    6% (1,061 no.)
    As of January 1, 1998, the total number of aeration systems located by Health District staff had
    been 9,515. The staff continues to find additional aeration systems in remote areas of the
    When the data between the two time periods is compared, clearly there is a large reduction in
    number of systems failing the first inspection. Homeowners are assuring their aeration systems
    are maintained. The number of homeowners with private maintenance contracts increased from
    1,623 in 1995 to 2,274 in 1996. However, in 1997 the number of homeowners with private
    maintenance contracts decreased slightly to 2220.
    Additional sampling of collector lines to determine program effectiveness was carried out. A
    12% randomly selected subset (23 no.) of the original Project CLEAN sampling locations were
    sampled and analyzed for BOD, SS, and fecal coliform bacteria. The collector line discharges
    were sampled once each year over the course of a three-year period. (See Graphs, pages 14-

    Graph 1. Biochemical Oxygen Demand
    Sampling Period
         Graph 2. Suspended Solids
    Sampling Period

                          Graph 3. Fecal Coliform Levels
    Sampling Period
    Graphs 1, 2, and 3 reveal the reduction of BOD, suspende