1 Introduction

Air is one of the most affected components of the environment, mainly by anthropogenic pollution (Singh et al., 2021a). Air pollution is the fourth highest risk factor for global-scale deaths (Health Effects Institute, 2020). The researcher found 5.5 million premature deaths every year, and half of that occurred in two highly populated countries, i.e., China and India, in 2013 (Brauer, 2016). Most of the world’s highly polluted cities are situated in India (22 in the top 30 cities globally), and 76.8% of the population is exposed to the exceeded limit of the pollutants (PM2.5), which shows an alarming situation and requires quick response for the management of air pollution and related health issues (Balakrishnan et al., 2019; Singh et al., 2021b). Despite this, India has only 804 manual monitoring stations under the National Ambient Monitoring Programme (NAMP) and 274 real-time monitoring stations (CAAQMS), though they require about 4000 (Day, 2021; Singh et al., 2021a). Furthermore, most of these are located disproportionately in tier-1 cities; a few are in tier-2 cities.

In northern India, especially the Indo-Gangetic region (IGR), the concentration of air pollutants (particulate matter (PM2.5 and PM10), nitrogen dioxide (NO2), sulphur dioxide (SO2), among others) exceeded the national air quality standards (Balakrishnan et al., 2019; CPCB, 2009; Dey et al., 2012; Guttikunda & Goel, 2013; Hama et al., 2020; Pant et al., 2015, 2019; Sharma et al., 2020; Singh et al., 2020, 2021c) and did not get reduced to the lower level in recent past (Chowdhury et al., 2017; Gurjar et al., 2016; Pant et al., 2019; Singh et al., 2021a). Delhi (the capital of India), which is situated in IGR and faces severe air pollution, ranks first in the pollution level of capital cities on earth (IQair, 2021). The World Health Organization (WHO) declared Delhi (with the second highest population density) as the second most polluted city, based on 5 years (2011–2015) of air pollution data (Census of India, 2011; WHO, 2016).

The concentration of air pollutants in Delhi and the National Capital Region (NCR) generally increases more in the winter season when compared with the summer season due to various reasons like biomass burning, industrial operations, vehicular pollution, calm conditions (stagnant wind), inversion, and mixing layer height among others (Chowdhury et al., 2017; Kar et al., 2010; Pant et al., 2019; Singh et al., 2021c). The annual population-weighted mean of PM2.5 in Delhi was 209 μg/m3 and relatively low in neighbouring states (i.e., Uttar Pradesh, Bihar, Punjab, and Haryana), ranges from 125·7 to 174·7 μg/m3, though very high as compared to the other countries of the globe including Nepal (83.1 μg/m3), Niger (80.1 μg/m3), Qatar (76.0 μg/m3), Nigeria (70.4 μg/m3), Egypt (67.9 μg/m3), Mauritania (66.8 μg/m3), Cameroon (64.4 μg/m3), Bangladesh (63.4 μg/m3), and Pakistan (62.6 μg/m3) among others (Health Effects Institute, HEI, 2020; Balakrishnan et al., 2019). Pant et al. (2015) also reported an average (12 h) PM2.5 concentration of 277 μg/m3 in winter and 58 μg/m3 in the summer season over Delhi. Bangar et al. (2021) and CPCB (2016) reported an annual average (24 h) of 150–178 μg/m3 over Delhi, which is much lower than the winter average of 277 μg/m3.

Vehicular emissions, municipal solid waste burning, biomass burning, suspended solid, rubber particles, seasonal crop stubble burning, air abstraction in the form of high-rise buildings, and road dust particles highly contribute to the pollutants during the winter season (Sharma & Dikshit, 2016; Singh et al., 2021c) which increases the health risk of Delhi’s residents (Ravindra et al., 2016; Singh et al., 2021b). Due to the high severity and level of the pollution, Delhi implemented several policies like the odd–even policy (vehicles with odd and even no. will run on alternate days), no waste burning, Compressed Natural Gas (CNG)-based vehicles, and use of bio-decomposer for stubble management among others. It was observed that these policies have minor effects on the air pollution level. Odd–even policy mitigated air pollution, and a reduction of 4 to 10% was observed in PM2.5 concentration in 2016 (Chowdhury et al., 2017). Nevertheless, these policies were found to be effective to some extent, and the pollution level increased drastically during the winter season due to various anthropogenic and natural reasons (Kar et al., 2010; Nair et al., 2020). Especially, the rice residue burning in Punjab, Haryana, and surrounding areas are claimed bluntly for the rise in pollution levels over Delhi during the winter season. However, there are limited and contrary shreds of evidence to confirm the clear-cut reason for the rise in air pollution in Delhi. Many researchers, including Balakrishnan et al. (2019), Pant et al. (2015), CPCB (2009), Pant et al. (2019), Singh et al. (2020), Hama et al. (2020), Dey et al. (2012), Guttikunda and Goel (2013), Sharma et al. (2020), IQAir (2021), and Singh et al. (2021c) among others claimed that the rise in winter air pollution in Delhi is attributed to the rice residue burning in surrounding states of India (mainly Punjab and Haryana), while few reported that there is no contribution of such burning (Nair et al., 2020). Furthermore, this remains a big question among policymakers, researchers, and pollution/environment managers. Thus, it is necessary to analyse more evidence for a clear-cut opinion on the rise of Delhi’s air pollution (DAP). Advance methodologies, models, ground instruments, and satellite-based measurements are available for this type of analysis (Mahato et al., 2023; Nair et al., 2020; Shang et al., 2023; Singh et al., 2021c; Tian et al., 2020; Wen et al., 2024). Alam et al. (2011) reported Hybrid Single practical Lagrangian Integrated Trajectory HYSPLIT model-based evidence for air pollution. They analyse the origin of pollutants through forward and backward trajectories. Despite using satellite-based and model-based evidence, they missed using ground-based sensor network data, which makes the study incomplete. Nair et al. (2020) reported that Delhi faces severe air pollution problems due to poor metrological conditions and local anthropogenic activities and not because of SB using the data from HYSPLIT, Navy Aerosol Analysis and Prediction System (NAAPS), and Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observation (CALIPSO). They claimed that the stubble burning (SB) aerosols dispersed at 3000 m above the ground, and this is out of the human region; however, they failed to explain the sudden increase in the level of pollution during October and November, where the temperature is not so low which supports stagnation and boundary layer height conditions. Bangar et al. (2021) reported the maximum contribution of SB in post-monsoon PM2.5 of Delhi using ground, modelled, and satellite data; however, they had not explained overall air quality in their study. Similarly, Beig et al. (2020) have reported the major contribution of SB in the post-monsoon concentration of PM2.5 in Delhi; however, this study could not explain the contributions related to other air quality parameters such as CH4 and N2O. SB in the Haryana region is less as compared to the Punjab region (Singh et al., 2021c), though both are blamed equally for the DAP, and there are limited pieces of evidences on this aspect. Simultaneously, the contribution in pollution of Delhi reported also varying from 10 to 48%. Despite these studies, the question remains the same and still has a lot of scope to answer through ground-based measurements, model databases, and satellite observations. This study has unique aspects in the form of involving all the air quality parameters from ground, sensor, and satellite-based measurements.

The major objective of this paper is to identify and analyse evidence to answer this question using ground sensors, models, and satellite measurements. The winter pollution was measured using data from ground-based sensors, geophysical models (HYSPLIT, NAAPS), and space-born satellites (CALIPSO, Moderate Resolution Imaging Spectroradiometer- MODIS, Visible Infrared Imaging Radiometer Suite-VIIRS, and TROPOspheric Monitoring Instrument-TROPOMI) (Table 1) for a better conclusion on the rise of DAP.

Table 1 List of data, purpose, sources, and related spatial resolution
Full size table

2 Study Area and Methodology

2.1 Study Area and Site Selection

The study area lies in the northern part of India between latitude 27°39′ 25.969 to 32°34′ 35.008´´N and longitude 73°53′ 25.905 to 77°35′ 31.343´´E and constitutes part of Indo-Gangetic Plain (IGP) which is among the highly polluted regions (Singh et al., 2014, 2021b) in the world because of various natural, and anthropogenic activities. It includes Districts of Haryana (Haryana is a developing state of India, so the major source of air pollution is an anthropogenic activity like industries, transportation, and SB), Punjab (the main occupation is agriculture in Punjab, and the highest wheat is produced by Punjab), Delhi (Vehicular and industrial pollution) and Union Territory of Chandigarh (Fig. 1).

Fig. 1
figure 1

Geographical area of research work (Punjab, Haryana, Delhi)

Full size image

A total of 69 ground stations are working in the study area for the air pollutant measurement under the maintenance of the Central Pollution Control Board (CPCB). These ground stations are well distributed in Haryana (23), Punjab (9), and Delhi (37) (Fig. 1). The total geographical area is 9,582,589.67 hectares largely covered by agriculture and built-up/habitation. Due to the availability of water (groundwater), free electricity, and heavy subsidies on fertilizers, wheat and rice cropping dominates over the region, and a huge amount of stubble gets generated, which burnt in the open, and makes the area hazy and polluted especially during winters (Bangar et al., 2021; Singh et al., 2021c) (Fig. 2 a to c).

Fig. 2
figure 2

Clear natural colour composite from MODIS before burning starts i.e. on 8 October (a). MODIS natural colour composite with smoke during high burning episode i.e. of 28 October (b). Active fire points-based Fire Radiative Product (FRP in MJ) from VIIRS data during October and November 2020 along with scatter plot between fire count and FRP (c)

Full size image

The climate of the area is a subtropical humid type with very high temperatures, up to 45ºC in summer, and much cooled, below 0 °C during winter. The annual average rainfall is 635 mm. Soil is loamy to loamy silt. The northern part is rich in forest, while the southern and south-western part is surrounded by the desert of Rajasthan.

2.2 Data Acquisition and Processing

Data from October and November (i.e., during peak SB episodes) for 2019 and 2020 from various sources were analysed to understand the contribution of SB in DAP. The list of data used and their sources are presented in Table 1. Each data and related specifications are described in detail in subsequent subsections.

2.2.1 Ground-Based Data

CPCB Ground Data and Air Quality Index

Since our priority was to assess the SB effect on DAP through ground-based data, the hourly average air pollutants (PM2.5, PM10, NO2, CO, and SO2) and metrological parameter (temperature, wind speed, wind direction, and Relative humidity) data at 69 stations network from 1 October to 30 November for 2019 and 2020 was acquired from the from CAAQMS network available at https://app.cpcbccr.com/ccr/#/caaqm-dashboard-all/caaqm-landing/caaqm-comparison-data. Researchers reported less than 5% error in the measurements of pollutant concentration through these networks (CPCB, 2010; Singh et al., 2021b; Tyagi et al., 2016).

The hourly concentration of pollutants and metrological parameters were converted into daily average and further at 10-day average and are used for the analysis of spatio-temporal variation in air quality during the October and November months of 2019 to 2020. October and November months were taken because (a) most of the SB occurs in these 2 months (Fig. 3), (b) the temperature gets decreased, and (c) people face visibility problems due to smog during these days (Balakrishnan et al., 2019; Bangar et al., 2021; Pant et al., 2015; Singh et al., 2021c; Fig. 2).

Fig. 3
figure 3

Decadal air quality variations over the study area from 1 October to 30 November for the year 2019

Full size image

Criteria air pollutants, including PM2.5, PM10, NO2, and SO2, were taken to estimate the Air Quality Index (AQI) following Kanchan et al. (2015) and Singh et al. (2021b). The inverse distance weighting (IDW) method of spatial interpolation was used to assess the spatial variability of AQI during these months of 2019 to 2020, as suggested by Kumar et al. (2016) and Singh et al. (2021b).

Ground-Based AOD

Ground-based AOD data were measured by the CIMEL sun-photometer, which is part of Aerosol Robotics NETwork (AERONET) and is located at more than 500 ground sites around the world. AERONET provides globally distributed observations of spectral aerosol optical depth (AOD), inversion products, and precipitable water in diverse aerosol regimes. There are 2 AERONET observatories in the study area, which have the capability to measure ground-based AOD and providing data for the last several years (Table 2). However, for the current study, only Amity University locations were taken based on the recent data availability.

Table 2 List of AERONET stations in Haryana, India, and their locations
Full size table

The CIMEL sun-photometer is a multi-channel, automatic sun-and-sky scanning radiometer that measures the direct solar irradiance and sky radiance at the surface of the earth only during daylight hours (sun above the horizon). The measurements are provided after the application of the cloud screening and quality control procedures. The AERONET program is a coalition of ground-based remote sensing aerosol networks established by NASA and PHOtométrie pour le Traitement Opérationnel de Normalisation Satellitaire (PHOTONS), which were further expanded by other collaborators from institutes, universities, national agencies, individual scientists, and partners. AERONET AOD products (version 3) include three levels of data viz. Level 1.0 (unscreened), level 1.5 (cloud-screened and quality controlled), and level 2.0 (quality-assured) (Tao et al., 2015).

AERONET-observed aerosol data of level 2.0 at Amity University, Gurugram (28.317 N, 76.916 E) were used for the analysis. Four characteristics of aerosols data were retrieved from the AERONET site, including AOD (Aerosols Optical Depth) at 500 nm, AAE (Absorption Angstrom Exponent) and EAE (Extinction Angstrom Exponent), volume size distribution (VSD), and SSA (single scattering albedo) for October and November 2020. The analyses of all the characteristics were performed following the method suggested by Nair et al. (2020). The data for the same period of 2019 was absent and thus not considered for further analysis; however, the data for April 2019 was analysed.

2.2.2 Model-Based Data

HYSPLIT Back Trajectories

A trajectory is the path of a single hypothetical point that is carried passively with the mean wind. Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) model provides both forward and backward trajectories to assess the fate of air mass from or to a particular location. HYSPLIT encompasses a series of digital codes that read various meteorological data files to compute trajectories, air concentrations, and particle dispersion. Meteorological data such as wind direction, speed, temperature, and moisture are provided on a spaced grid at multiple levels in the atmosphere. The meteorological data are updated at each integration step. Dispersion is introduced by calculating the trajectory for many points. However, each trajectory is perturbed by the random atmospheric turbulence along its path. Air concentrations are computed by adding together the mass of the computational particles and dividing by the volume of their horizontal and vertical distribution.

HYSPLIT backward trajectory of 20 October to 8 November 2020 (during peak SB) was derived for the understanding of air mass flow and its origin. It is expected that this particular analysis can provide evidence for the contribution of SB on DAP. For that, the trajectory must follow the path of high SB regions.

NAAPS Data

Navy Aerosol Analysis and Prediction System (NAAPS) is a 3D model system developed by the Naval Research Laboratory (NRL) for the analysis of horizontal dispersion and produce deterministic 6-day forecasts of a combined anthropogenic SO2, Sulphate, and biogenic fine, smoke, sea salt, and dust aerosol on 25 vertical levels at 1/3◦ every 6 h (Srivastava et al., 2012). NAAPS is used operationally by the US Navy to predict visibility-reducing conditions. A close analysis of this data may provide a means to understand the contribution of SB. In this study, concentration and horizontal distribution trends of smoke and dust during October and November 2020 were analysed (https://www.nrlmry.navy.mil/aerosol_web/loop_html/globaer_india_loop.html).

2.2.3 Satellite-Based Data

Satellite-Based AOD

The Moderate Resolution Imaging Spectroradiometer (MODIS) Aerosol products monitor the ambient aerosol and some properties of the aerosol over the cloud-free, snow-free, and ice-free land and ocean surfaces. Data from two different algorithms are provided over land, known as Deep Blue and Dark Target Land. The primary data product is the aerosol optical depth (AOD), also known as aerosol optical thickness (AOT), at a wavelength of 550 nm. In addition, each algorithm provides selected additional information about the aerosol, such as single scattering albedo, spectral AOD, or descriptions of relative aerosol size, as well as quality assurance information.

There are two MODIS Aerosol product files from the Terra and Aqua platforms. MODIS (Terra) level 2 is available from year 2000 onwards, and MODIS (Aqua) level 2 is available from year 2002 onwards. Here, we have used MODIS level 2 3 × 3 km Terra and Aqua data together for our study because of finer resolution. The aerosol retrieval makes use of seven wavelength bands (Table 3) and a number of other bands to help with cloud and other screening procedures. The wavelength ranges included in Table 3 are estimates of the central wavelength in each band (obtained by integration of the channel-averaged response functions). An average data of 10 days during October and November of 2019 and 2020 were used to assess the spatial variations of AOD over the study area.

Table 3 Characteristics of MODIS channels used in the aerosol retrieval
Full size table

Fire Radiative Power

Visible Infrared Imagine Radiometer Suite (VIIRS-375 m) fire count and Fire Radiative Power (FRP) data was acquired from the Suomi-National Polar-Orbiting Partnership (Suomi-NPP) satellite, which is distributed through a Fire Information for Resource Management System (FIRMS) system (https://firms.modaps.eosdis.nasa.gov/download/create.php). VIIRS has a swath of 3040 km with high resolution 375 m data and is used in fire management, which allows the detection of the number of fire points, even of small intensity. Suomi-NPP satellite passes through 1:30 pm and 1:30 am local time and has a fire band of 4.05 µm wavelength (NASA Earthdata, 2020). Due to its high resolution, VIIRS datasets were used by researchers in air pollution studies (Nair et al., 2020; Singh et al., 2021c; Vadrevu & Lasko, 2018; Zhang et al., 2017). FRP data for October and November 2019 and 2020 was analysed to assess the periodic phenomenon of increase in the SB, which is reported to increase the air pollution in the Delhi region with a combined effect of metrological conditions, industrial operations, and other anthropogenic activities like transportation (Nair et al., 2020; Singh et al., 2021c). It was expected that the high FRP values would be observed in nearby regions of Delhi during the SB period. This may provide an idea about the contribution of SB in DAP.

Spatial data of FRP from MODIS with a spatial resolution of 1 × 1 km were also used. It was converted into a cumulative sum of October and November 2019 and 2020 to identify the regions with high SB or high FRP. This was done to get information related to the regional contribution of SB, particularly from Punjab and Haryana in DAP.

CALIPSO Data

Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO) satellite launched in collaboration between the French Agency National Centre for Space Studies (CNES) and the American space agency NASA on April 28, 2006, and provided vertical profiles of cloud and aerosols (Winker et al., 2007). CALIPSO launched at an altitude of 705 km with an inclination of 98° with an equatorial crossing time of about 13:30 LST (ascending node) (Nair et al., 2020) and a 16-day repeated cycle at a particular location (Shaik et al., 2019). To understand the vertical uplift of aerosol particles or to identify the smoke aerosol position over the study regions, CALIPSO level 2.0 version 4.20 aerosol subtype and Vertical Feature Mask (VFM) and version 4.10 are used to retrieve the total attenuated backscatter at 532 nm with a spatial resolution of 5 km (horizontal) × 0.30 km (vertical) (Omar et al., 2009). Two days of daytime data were retrieved based on availability for the Haryana and Punjab regions on October 29, 2019, and for the Delhi region for November 11, 2019. The CALIPSO data for the year 2020 was not available and hence was not utilised.

TROPOMI Data

The TROPOspheric Monitoring Instrument (TROPOMI) is the instrument onboard the Copernicus Sentinel-5 Precursor (S5P) satellite. The S5P is the first of the atmospheric composition Sentinels, launched on 13 October 2017, with the expectation of a mission of the next seven years. It provides various parameters like nitrogen dioxide (NO2), sulphur dioxide (SO2), methane (CH4), carbon monoxide (CO), and ozone (O3). It uses forecasted meteorological data like wind speed and wind directions from the European Centre for Medium-Rang Weather Forecasts (ECMWF) in the real-time parameters prediction. The data for the years 2019 and 2020 were analysed for October and November. An average of 10 days was estimated for each parameter, including CH4, CO, SO2, and NO2, as these are important pollutants generated from SB (Singh et al., 2021c). The wind forecast provided as ancillary data was used for the assessment of wind and its effect on pollution dispersion.

3 Results

3.1 Ground-Based Evidences of Pollution Contribution

3.1.1 Variability in Pollutants, Meteorology, and AQI

Ground-based analysis for the regional contribution of air pollution in Delhi was done based on five criteria air pollutants (viz. PM2.5, PM10, NO2, CO, and SO2) and Air Quality Index (AQI). Four meteorological parameters (Temperature, wind speed, wind direction, and Relative humidity) were also analysed to understand the effect of meteorological conditions on pollutant persistence and dispersion conditions. Both pollution and meteorology were taken for October and November 2019 and 2020 at 69 ground stations from the CPCB. For this analysis, 24-h average data for October and November were taken because the literature shows the highest concentration of pollutants these days due to burning and metrological conditions (Nair et al., 2020; Zhang et al., 2017; Vadrevu & Lasko, 2018, Singh et al., 2021b). Nair et al. (2020) identified the highest concentration of pollutants in the last week of October and a higher concentration of PM throughout November 2017. The same pattern was observed in this research during 2019 and 2020 spatially. Similar findings have been reported by Singh et al. (2021c) also for the Kurukshetra district of Haryana.

Table 4 shows the concentration of various pollutants. High concentrations of pollutants were observed during 2020. The highest concentrations of the pollutants were observed in the first 10 days of November 2019 and 2020 which followed the SB season. Temperature starts decreasing from October first week. Average temperature ranges from 19 to 24.77 °C and 16.64 to 27.04 °C, respectively, in 2019 and 2020 (Table 5). Wind speed was less (below 1.4 m/s) in the first week of November, and pollutant concentration was high during the same time, indicating that the meteorological conditions put the pollution at the worst level (Table 4 and 5). Furthermore, the analysis shows a low dispersive capacity of the wind due to low velocity, resulting in a stagnant condition over the study area. This resulted in a long time persistence of the pollutant over this region both from local and regional sources. The same has been reported in other studies for some other places in India where metrological parameters have an impact on PM concentration and dispersion of the other pollutants in the atmosphere (Dhaka et al., 2020; Nair et al., 2020).

Table 4 Comparison of pollutant concentration during October and November for 2019 and 2020 over the study area
Full size table
Table 5 Comparison of metrological parameters during October and November 2019 and 2020
Full size table

Spatial variation in average AQI over Haryana, Punjab, Chandigarh, and Delhi can be depicted from Map in Fig. 3 (for 2019) and Fig. 4 (for 2020). Significant variation in AQI over Delhi is clearly visible in both years, with high values during the last 10 days of October and the first 10 days of November (i.e., during the peak burning time). Delhi has poor AQI as compared to Haryana and Punjab throughout the study period. This shows that the main reason for air pollution in Delhi could be local emissions from various sources like vehicular emission, industrial activity, coal-based fuel ignition, and Diwali festival, along with the regional climate. However, at the end of the SB season, the AQI was reduced in both years, which indicates that SB also influences the pollution level in Delhi (Table 6). Less wind speed and relative humidity during 2020 resulted in a high AQI in November as compared to November 2019.

Fig. 4
figure 4

Decadal air quality variations over the study area from 1 October to 30 November for the year 2020

Full size image
Table 6 Average AQI and percentage change from 2019 to 2020 over Delhi
Full size table

It is also to be noted here that the local contribution of pollution may be high over the Delhi region; however, the local sources of pollution remain the same for these 2 months, or minor variations are expected. A periodic increase in the pollutant concentrations and AQI during the last week of October and the first week of November (Singh et al., 2021a, 2021b, 2021c), which matches with the SB episode, shows the contribution of SB. Nevertheless, the meteorological conditions (Table 5) make the conditions worse during these days over Delhi (Table 6).

Table 6 shows the percentage changes in AQI from 2019 to 2020. October and November months are divided into six categories, and out of these six, only one shows high AQI during 2019 (i.e., on 11–20 November). Maximum AQI was observed during 1–10 November, i.e., 355 and 425 in 2019 and 2020, respectively, following the reported pattern of earlier studies such as those by Nair et al. (2020) and Singh et al., (2021a, 2021b, 2021c).

3.1.2 AOD Characteristics from Aerosol Robotic Network (AERONET)

AOD is the most important parameter to identify the burden of aerosols in the atmosphere. Thus, version 3 AOD500 data from AERONET were analysed for October and November 2020. The analysis for the year 2019 was also done due to the non-availability of the data for April 2020, considering the peak burning duration observed every year in the region due to Wheat stubble burning. Values of AOD were ranging from 0.2 to 3.00 on a daily scale (Fig. 5). High values were observed during the last 10 days of October and in the first and second 10 days of November 2020 (Fig. 5).

Fig. 5
figure 5

Variation in AOD500 during October and November 2020 from the ground station (Amity University)

Full size image

More than 75% of observation days have AOD values greater than 0.5. AOD value range recorded in the last 10 days of October was 0.4 to 1.5, with a mean of 0.93. The average value of AOD during 1–10 November and 11 to 20 November was found to be 1.49 and 0.94 respectively. Greater than 2 values of AOD were found on 4, 8, and 9 November.

Based on important parameters viz. Absorption Angstrom Exponent (AAE), Extinction Angstrom Exponent (EAE), volume size distribution (VSD), and Single Scattering Albedo (SSA), various types of aerosols are characterised by their origin. In a certain combination, these characteristic parameters may provide important information about the SB contribution. Some researchers classified the aerosols on the basis of the AAE vs. EAE relation as dust (0.01 < EAE < 0.4 and 1.0 < AAE < 3.0), SB (0.8 < EAE < 1.7 and 1.1 < AAE < 2.3), urban/industrial (0.8 < EAE < 1.6 and 0.6 < AAE < 1.3) in the Indo Gangetic region (Nair et al., 2020). Some other classify the aerosol types as prominent dust type 1 (0.05 < EAE < 0.8 and 1.3 < AAE < 2.2), type 2 (0.45 < EAE < 0.76 and 0.7 < AAE < 1.3) with prominent black carbon and vehicle coated moderate dust, and type 3 (0.85 < EAE and 0.7 < AAE < 1.4) include organic carbon or black carbon due to incomplete combustion of stubble and fossil fuels (Bibi et al., 2016, Dey et al., 2012; Mishra et al., 2014). In common, the aerosols from SB show medium/high EAE and low/medium AAE values (Mishra et al., 2014). Similarly, VSD and SSA also provide evidence of pollution increase during the SB episode.

Thus, the AAE, EAE, VSD, and SSA were obtained from ground-based AERONET data over Haryana/Delhi region (Amity University station data) to characterise the aerosols. Scatter plot for AAE (440–870 nm) vs. EAE (440–870 nm) for the Delhi region during April 2019 (period of wheat SB) and October and November 2020 (Rice SB) shows the contribution of SB when characterised based on Nair et al. (2020) and others classification, as in Fig. 6a and b respectively. Figure 6a represents the range of the aerosols during April 2019 as 0.3 < EAE < 1.2 and 1.3 < AAE < 2.5, which indicates the contribution of dust and wheat SB. In April 2019, predominant dust particles were observed, and on the 28 and 29 April, black carbon particles were observed, which may be due to incomplete combustion of fossil fuel. However, it could not be suggested that the contribution is coming from SB alone; rather, a combined contribution from local and regional sources is suggested. The main reason for the incomplete explanation may be meteorological mixing and favourable dispersion conditions, which is high during April month in general. Figure 6b represents the range of the aerosols as 1.18 < EAE < 1.4 and 1 < AAE < 1.28 and 1.07 < EAE < 1.6 and 0.9 < AAE < 1.5, respectively, during October and November 2020. This distribution indicates the presence of pollutant particles from SB. Approximately 90% of particles observed belonged to type 3 categories, which signifies SB contribution. Furthermore, the poor metrological conditions make the situation more critical. Though it is difficult to identify the different types of aerosols under metrological mixing conditions (Bibi et al. 2016), mostly calm condition exists during our study period of October and November, which provides the opportunity to identify various types (i.e., dust, SB, urban/industrial) of aerosols (Fig. 6b).

Fig. 6
figure 6

Scatter plot between AAE vs EAE (440–870 nm) for 04–30 April 2019 (a) and 20 October–30 November 2020 (b). The signs of type-3 aerosol were present in April, but most of the values showed less magnitude for EAE, while all the values supported the contribution during October and November

Full size image

For further decision on the identification of the aerosol type and its origin, VSD and SSA were analysed through a line graph (Fig. 7 a and b and 8a and b). Some researchers reported the categories of aerosols such as polluted dust (PD), polluted continental (PC), black carbon (BC), organic carbon (OC), and non-absorbing aerosols based on these criteria parameters (Dey et al., 2012; Mishra et al., 2014; Nair et al., 2020). Figure 7 (a and b) shows the plotted graph of VSD (dVr/dlnr) for the last 10 days of October (Fig. 7a) and the first 10 days of November month (Fig. 7b), as these days showed high fire events (Fig. 2). The difference in the radius of the particle during October and November 2020 was visible. Aerosol particles were classified into three classes, viz., fine, course, and mixed particle, following Xu et al. (2015). The outcome showed that the coarse particles were dominant in October month (on 20, 21, 24, 25, and 31 October 2020) (Fig. 7a), and fine particles were dominant in November month (on 2, 6, 8, and 9 October 2020) (Fig. 7b). The mixed combination of fine and coarse particles is observed on 22 and 29 October and 1 and 7 November.

Fig. 7
figure 7

Plot showing volume size distribution graph for Oct. (a) and Nov. (b)

Full size image

The VSD range for the coarse particle concentration was 0.09 to 0.18 µm3/ µm2 at 4 µm during October, and the fine particle concentration range was 0.02 to 0.31 µm3/ µm2. It is marked that type 3 aerosols (SB generated) showed the highest fine concentration of the range 0.12–0.31 μm3/μm2 at 0.15–0.23 μ during 4–9 November 2020 which was peak fire time with the highest fire events (Fig. 2c). Similarly, type 2 aerosol (mixture of dust and absorbing aerosols) indicated a maximum coarse mode concentration of 0.18 μm3/μm2 over the Delhi region during October. This justifies the dominance of mixed types of aerosols (type 2) that majorly fall under the classification of polluted continental (PC).

This also justified that when fine particles were dominant in the last week of October and in the first week of November (Fig. 7b), the SB particle (BC + OC) was found in Delhi (Fig. 6b). Since there are limited internal sources of SB in the Delhi region and the dominance of SB is visible in the Punjab and Haryana region (Fig. 6 a and b), the presence of SB particles in the aerosols at Delhi station indicates the significant contribution of SB in DAP. Concluding such a remark from the above observations does not justify that there is no contribution of local emission source from the anthropogenic activity that occurred in Delhi. Nevertheless, the internal sources of pollution (like transport, industrial operations, construction activities, and biofuel burning) remain constant or may vary minutely during October and November, and there are limited chances of a sudden increase in these internal pollution sources, the presence of SB particles in the aerosol data confirms the SB contribution.

It is reported that the SSA below 0.9 for the entire wavelength represents the contribution of SB or the presence of urban and industrial pollution (Mishra et al., 2014). The SSA was found to be below 0.9 for most of the days of the last decade of October, indicating the presence of SB aerosol (Fig. 8a). Similar observations were obtained for 1, 3, and 10 November. However, the rest of the days of the first 10 days of November show higher SSA values, which indicate the comparative dominance of scattering particles like dust aerosols or large Sea-salt/sulphate mixed with absorbing aerosols and SB particles in the atmosphere (Fig. 8b). Simultaneously, the decreasing trend of SSA for the entire wavelength region (except 675) confirms the presence of SB/urban/Industrial pollution for both October and November (Nair et al., 2020).

Fig. 8
figure 8

Plot showing SSA vs. wavelength for October last decade (a) and November first decade (b)

Full size image

3.2 Evidences from Model Data

3.2.1 Evidences from HYSPLIT Data Analysis

HYSPLIT back trajectories (72 h) for peak SB days (20 October to 8 November 2020) showed the origin of air mass from India (Gorakhpur, Saharanpur, Punjab, Uttar Pradesh, and Madhya Pradesh), Pakistan (RCW Rohari, Bahawalpur, Khara, Multan, Larkana, and Kohat), Iran (Bandaan), and Turkmenistan (Mary) (Fig. 9). The SB contribution from the Punjab and Haryana region was visible on 5 November and 8 November 2020 (Fig. 10), which was the peak time of SB. Furthermore, the wind speed (2 to 10 m/s) and wind direction (north-west to south-east) also favour the fact of the contribution of SB particles in DAP and their origin from the Punjab, Haryana, and surrounding regions. The regional contribution from outside of Punjab and Haryana is also expected, as observed in the HYSPLIT trajectory (Nair et al., 2020), which follows the path (Punjab and Haryana) from where the smoke travelled towards Delhi (Fig. 10).

Fig. 9
figure 9

HYSPLIT backward trajectories (72 h) from 20 Oct. to 7 Nov. 2020

Full size image
Fig. 10
figure 10

HYSPLIT four days backward trajectories for the peak active fire and peak pollution days (5–8 Nov. 2020)

Full size image

Furthermore, the concentration of SB was high in the Punjab and Haryana regions, as observed from average FRP values (Fig. 10) confirmed this conclusion. However, on other days, the air mass origin looks variable and confirms the regional contribution as well, from desert areas of Pakistan, Turkmenistan, and Iran, among others. Nevertheless, in most of the cases, the origin was from agriculture-dominant areas (Multan, Fazilpur, Shaharanpur, Gorakhpur, and RWC Rohari, among others), again confirming the contribution of regional SB in DAP (Fig. 9).

3.2.2 Evidences from NAAPS Data Analysis

Dust and smoke concentrations retrieved from the NAAPS model during October and November 2020 were analysed. Dust and smoke continuously increase from 5 October to 5 November and then start decreasing (Fig. 11 a to l). The lowest smoke concentration was 02 to 04 µg/m3 on 5 October 2020, when there was limited SB (Fig. 11 b). The highest smoke concentration was observed when the SB was on peak, i.e., on 5 November 2020, which ranges from 64 to 128 µg/m3 along with low dust load (Fig. 11 h and g respectively). Since the SB started peaking up from the last 10 days of October to the first 10 days of November in the Haryana and Punjab regions, high smoke concentration is attributed to the SB contribution. This can also be seen on 15 November when there is a high smoke concentration (16–64 µg/m3) over the Punjab and Haryana regions with no/limited dust loads (Fig. 11 i and j).

Fig. 11
figure 11figure 11figure 11

NAAPS-derived dust (a, c, e, g, I, and k) and smoke (b, d, f, h, j, and l) concentration from October 5, 2020 (12 h), to Nov. 25, 2020, on 10-day interval

Full size image

Here, it is important to note that the smoke concentration was relatively high during the SB period for all dates analysed in the current study. This indicates that the smoke originated either from internal pollution sources or from regional SB, including the Punjab and Haryana regions. Since the SB is more intense in the Punjab and Haryana regions, and there is a limited chance of an increase in internal pollution sources (Saxena et al., 2021; Singh et al., 2021a), it may be concluded that the major contribution is coming from SB in Punjab and Haryana regions. Furthermore, the contribution of SB burning from nearby areas like Uttar Pradesh, Madhya Pradesh, and Pakistan cannot be ignored.

3.3 Satellite Data-Based Evidences

3.3.1 MODIS AOD

Two years (i.e., 2019 and 2020) of analyses of AOD showed high values during the peak season (11 October to 20 November) of winter SB (Figs. 12 and 13). The AOD was relatively low (< 0.6) during the first and last 10 days of October and November, respectively, in both years. Here, it is important to note that the high AOD values persist only during the SB period (11 October to 10 November). Furthermore, there are no other sources of pollution that suddenly increase either in number or intensity of emissions. Nevertheless, the meteorological conditions also remain calm with low wind speed flowing towards north-west to south-east direction, which denies the theory of long-range continental transport of the pollutants during this particular period. Thus, a question arises as to where the aerosol came from, as visualised by increased values of AOD. We consider that this increase in AOD values resulted from SB from Punjab, Haryana, and nearby regions of other Indian states and countries like Pakistan based on our logic of the absence of any significant reason except SB during the short study period.

Fig. 12
figure 12

AOD MODIS variation in 2019

Full size image
Fig. 13
figure 13

MODIS AOD pattern for 2020

Full size image

3.3.2 MODIS Fire Radiative Power Analysis and Evidences

The cumulative sum of fire radiative power (FRP) values for October and November 2019 and 2020 showed a high concentration of SB and generated power over the Punjab and Haryana region (Fig. 14 a and b). It can be depicted that the cumulative sum of FRP is mainly distributed over the Punjab and Haryana region and less at the regional level, i.e., over Pakistan. Furthermore, the FRP is high over the Punjab and covers almost the entire state, while this is distributed over the northern part of the Haryana state. From this, mainly two things may be concluded: (1) The SB from Punjab and Haryana contributes in the DAP, and (2) SB in Punjab contributes more in DAP as compared to the SB in Haryana.

Fig. 14
figure 14

Cumulative sum of FRP obtained from MODIS FRP for Oct. Nov. 2019 (a) and 2020 (b). Most of the FRP is concentrated over Punjab and Haryana, which depicts the contribution of stubble burning to Delhi’s air pollution. Slightly higher burning was observed during 2020

Full size image

3.3.3 Evidences from CALIPSO Data Analysis

Based on the availability of CALIPSO data over the study area, the data of 29 October 2019 and 09 November 2019 was used based on availability (Fig. 15 a and b). Dust, polluted dust, and elevated smoke type of aerosols were dominant in Punjab and Haryana. The presence of elevated smoke and polluted dust over the Punjab and Haryana region on 29 October 2019 and in the Delhi region on 09 November 2019, along with a large number of fire events, indicates the SB in this region (Fig. 15 a and b). However, with the limited coverage of this data, it cannot be concluded that all the pollution increases in Delhi are because of SB alone. There are contributions from other internal sources as well, and they may be detailed from source apportionment studies.

Fig. 15
figure 15

Calipso derived aerosol type: a Punjab and Haryana (29 October 2019), b Delhi (11 November 2019)

Full size image

3.3.4 Evidences from TROPOMI Data

Wind Pattern

Wind direction and speed are the important parameters that decide the fate of pollutants generated from SB in the Punjab and Haryana regions. Thus, the wind speed and direction from TROPOMY data were analysed for the peak SB days i.e. between 25 October 2020 and 07 November 2020 (Fig. 16 a to l). It was seen that the wind direction remains mostly towards the south-east from north-west i.e. towards Delhi. Thus it may be concluded that the SB-generated pollution finds its way towards Delhi through the characteristic patterns of the wind direction. Furthermore, the wind speed was low to medium, showing the stagnation of the pollutants over the region of SB for a longer time, as observed from the CLIPSO observation on 29 October 2020 (Fig. 15 a). The same was observed from ground-based measurements as well.

Fig. 16
figure 16figure 16

Wind speed and wind direction during 25 October to 07 November 2020 (a to l)

Full size image

Pattern of major air pollutants

Sentinel 5P data analysis showed a clear indication of an increase in pollutant (NO2, SO2, CO, and CH4) concentration during the SB period (Fig. 17 a to d). NO2 was found to be low during the first decade of October in the Punjab and Haryana region while high over Delhi. This starts increasing from the second decade and became highest in the first decade of November, i.e., during the peak SB period (Fig. 17 a). However, the NO2 was consistently high over Delhi (ranging from 44.32 to 78.69 µg/m3), which indicates combined contributions from SB and internal sources like industrial and vehicular emissions (Fig. 17 a to d). Furthermore, the synchronous pattern of changes in the NO2 concentration as followed by fire incidences and FRP, indicates (Fig. 2c) the significant contribution from SB. SO2 also showed a similar pattern; however, the variability was very high (Fig. 17 b). A similar pattern in the variability of CO and CH4 was seen (Fig. 17 c and d) and can be attributed to SB. Point sources of pollution, like industries, showed emission flow towards the south-east direction (Fig. 17 a).

Fig. 17
figure 17figure 17

a–d Satellite-based air pollutants showing variation over Haryana, Punjab, and Delhi during the SB period

Full size image

4 Discussion

The study focuses on the establishment of a relationship of SB with DAP using ground data, model data, and satellite data/products. The assessment of these datasets provided the answer to an important question, “Does SB in Punjab and Haryana really affect DAP?” It has been a hot topic of environmental management in the Delhi region for a long time. Various researchers tried to answer this question from time to time, resulting in two distinct groups with contradictory beliefs. One group believes that the SB in Punjab and Haryana significantly contribute (10 to 48%) to DAP (Pathi & Chhabra, 2020; Kedia et al., 2020; Beig et al., 2020; Laskar et al., 2020; Srinivasan & Abirami, 2020; Usmani et al., 2020; Bangar et al., 2021; Saxena et al., 2021) and another group does not believe this theory (Dhall et al., 2020; Goswani, 2020; Khanna, 2020; Liu et al., 2018; Nair et al., 2020; Petersen, 2019; Rasma et al., 2019).

Since the policymakers and environmental managers did not get a clear-cut conclusion on this issue, they found difficulties in combating air pollution in the Delhi region, which leads to serious health hazards for the residents of Delhi and nearby regions (Singh et al., 2021b). Thus, it was important to get the answer to this important question for better environmental management. The availability of data from ground-based measurements provided us the opportunity to understand pollution scenarios in the Delhi region.

Ground-based pollutants concentration on a 10-day average-basis showed high concentration during the peak fire events (October–November) in both 2019 and 2020 with little higher concentration during 2020.

The AQI values were found to be high during the peak SB period (Table 6). Year 2020 showed a relatively high AQI as compared to 2019 during the same period. In this, the RH and WS played a key role. However, the RH ranged from 46.65 to 48.90% for 2020. The high RH values support positively to the air pollutants, which further increases the AQI values (Table 6), and our findings deny this in the Delhi region and support the SB contribution hypothesis. This observation showed equal importance of meteorological conditions, especially RH and WS, in worsening the pollution conditions over the Delhi region rather than increasing. However, the origin of high levels of pollution, as marked during the peak SB period, still remains the same, which indicates the contribution of Punjab and Haryana-based SB in DAP (Fig. 14). This was also supported by the rise in FRP (Fig. 14a and b). We also consider this fact as there are minimum chances of an increase in internal sources of pollution like transportation, construction, and biomass burning inside Delhi during such a small time-frame. Furthermore, the presence of organic pollutants and smoke conditions (i.e., type-3 aerosols) during the peak SB period over Delhi ground station (AERONET station at Amity University) confirms our reasoning (Figs. 7 and 8). Nevertheless, the limited contributions of SB from other states like MP, UP, and Uttarakhand could not be ignored. This has also been supported by other researchers, including Pathi and Chhabra (2020), Srinivasan and Abirami (2020), Saxena et al. (2021), and Singh et al. (2021a).

Greater than 2 value of AOD were found at 4, 8, and 9 November over the region. At the same time, AQI was also high in the study area, as shown in Figs. 3 and 4. Furthermore, the fire counts and FRP were also found to be highest during these days (Fig. 2 and Fig. 14), which indicates the contribution of SB in the pollution increase over Delhi and surrounding region. The WS was 1.34 m/s, which is quite low, and the direction was dominated towards SE in the first week of November and thus has lower dispersive capacity in the atmosphere, worsening the situation of pollution in the Delhi region. Actually, farmers practice SB twice a year, i.e., during summer and at the onset of winter. The generated pollutants disperse quickly due to warm breezes during summer, while in winter (October or November), low wind speed and plummeting temperatures spread the smoke at a slow rate, resulting in air mass stagnation and high AQI (Fig. 11 a–l, Rasma et al., 2019).

Again, the similar pattern of increase in pollutant concentration over Delhi (Table 7), Haryana, and Punjab (Table 8) during the SB episodes in both years supports the final conclusion of this study, i.e., a significant contribution of SB in DAP. The PM2.5 and PM10 are the dominant pollutants; those are generated by SB and worsen the environmental conditions in Delhi. It can be understood that the concentration of PM2.5 is found to be much higher (247–301 μg/m3) during the SB episode than the annual average (150 μg/m3) of Delhi. If the episodic average of PM2.5 is compared with the peak 10 days average (for both 2019 and 2020), a total of 37% increase was observed, which may be considered proxy of approximate contribution from SB in DAP. This range (37%) is exactly falling within the range (1–58%) reported earlier is various studies, such as by Chowdhury et al. (2017) (28%), Guttikunda and Gurjar (2012) (37%), Rasma et al. (2019) (16–33%), Beig et al. (2020) (1–58%), and MoEFCC (15–40%) among others using source apportionment studies. The contribution of SB in the PM2.5 concentration of Haryana ranged from 14 to 52% (calculated based on the reported average of 81 µg/m3 for the year 2019 by Singh et al., 2021c); for Punjab, it ranged from 27 to 47%. This shows that the SB in Punjab and surrounding areas also affects the people of Haryana badly. However, the source apportionment studies need to be done for a better understanding, especially in National Capital Regions (NCRs).

Table 7 Pollutant concentration (µg/m3 except for CO, which is in mg/m3) over Delhi during the SB episode
Full size table
Table 8 Pollutant concentration (µg/m3 except for CO, which is in mg/m3) over Haryana and Punjab during the SB episode
Full size table

Similar increase in FRP (from satellite-based data) may be seen (Fig. 14), which advocated the contribution of SB in DAP. The contribution of SB in DAP (raging between 10 and 48%) was reported by MoEFCC and some other researchers also. The concentration of all the pollutants was the highest during the first 10 days of November. Since the internal sources of pollution were fixed, and the meteorological conditions have minor variations (Table 5), the drastic increase in the concentration of pollutants during the peak SB episode may be attributed to the SB in Punjab, Haryana, and other neighbour states and countries. Our reasoning for SB contribution in DAP further gets support from satellite-based fire count data (having a close correlation, i.e., r2 = 0.98 with FRP, as shown in Fig. 2), which was mostly concentrated over the Punjab and Haryana region and TROPOMI data for NO2 showing flow towards Delhi (Fig. 2, Fig. 14a and b, and 17a respectively). Satellite-based wind direction patterns also support the flow of air pollutants from the north-west region to the southern part of the Indo-Gangetic plain, i.e., towards Delhi.

5 Conclusions

Based on data from ground-based systems, geophysical models, and satellite systems following have been concluded.

  • An increase in the pollutant concentration during the peak SB period (20 October to 10 November) was observed from ground-based measurements. This synchronous behaviour of pollutant concentration with the SB pattern showed the contribution of SB in DAP. At the same time, the concentration of pollutants in Delhi remains high due to internal sources like transportation, industrial operations, and construction, which may contribute equally to winter air pollution. However, these internal sources must be constant during this small window of analysis, which showed the significant contribution of the SB from Punjab and Haryana, along with other neighbouring states and countries like Pakistan and Iran in DAP.

  • Ground-based AQI showed high AQI values during the last decade of October and the first decade of November, which follows the SB pattern. To ensure the phenomenon correctly, the extended study for the previous year, i.e., 2019, was done, which again confirms the same pattern with less burning.

  • Ground sensors AERONET-based AOD data and its characteristics like single scattering albedo, EAE, AAE, etc. showed the presence of organic content or type-3 aerosols, which indicates the contribution of SB in DAP. AOD from MODIS satellite also showed high concentration during the SB episode. This again confirms the contribution of SB in DAP.

  • Similarly, model data from NAAPS showed high smoke over Punjab, Haryana, and Delhi during the peak SB time. This elevated smoke also confirms the contribution of SB in DAP.

  • Three-Dimensional (3D) data from CALIPSO showed elevated smoke (on 29 October 2019) over Punjab and Haryana with a height of 3 to 4 km from the earth’s surface during the peak SB time. The same was observed in Delhi on November 09 2019. Elevated smoke, along with polluted continental/smoke during the peak SB time over Punjab, Haryana, and Delhi regions, is not possible without SB, which confirms the contribution of SB to DAP. Interestingly, the thick smoke was also observed over the MP region, which showed the combined effect of SB burning from nearby states and countries on DAP and blaming only Punjab and Haryana, which may not be correct. However, most of the SB occurs in Punjab and Haryana regions, indicating a relatively major contribution of these two states in the DAP.

  • MODIS-based AOD, fire count, and FRP showed a synchronous pattern of increase during the study period with the highest concentration at the time of peak SB time, i.e., last 10 days of October and the initial 10 days of November.

  • The prominent wind direction (from TROPOMI) during the peak SB time was towards Delhi from the north-west direction, confirming the contribution of SB in DAP. Furthermore, the wind speed which was low to medium provides stagnant or slow dispersion condition of the pollutants from Punjab and Haryana.

  • Sentinel 5P-based pollutant concentration also indicates the contribution of SB in DAP. The Decadal mean for SO2, NO2, CO, and CH4 showed high concentration during the peak SB time, i.e., during 20 October 31 October and 01 November to 10 November 2020. Simultaneously, the emissions from point sources (major factories) indicated the direction of pollutants towards Delhi.

  • The overall conclusion is that there is a significant impact of SB on DAP as the data from ground, model, and satellite remote sensing indicates traces of burnt biomass and related parameters.

  • Satellite-based observations were found efficient and have provided a great opportunity to analysis the contribution of SB burning in DAP. The current study gives an answer to a major question that has been a hot topic for several years. It is expected that the continuous dispute regarding the contribution of SB in DAP may be solved using the same analysis approach. This study may support the policymakers and environmental managers involved in DAP management.