Land Use Land Cover Dynamics and Their Effect on Air Quality Across North India Using Geo‐Spatial Datasets

ABSTRACT Environmental health monitoring, its proliferation, and risk mitigation remain a priority for environmentalists, engineers, and researchers. Land use land cover (LULC) remains a major player impacting all the environmental variables. The present study aims to assess and quantify the impact of LULC and its change on air quality (AQ) and its role in causing air pollution over northern India using geospatial and remote sensing datasets. Landsat 7, 8, and 9 imagery was used for constructing different LULC maps for the years 2001, 2011, 2021, and 2018, 2020, and 2022. MODIS (Aqua and Terra) was used to acquire aerosol optical depth (AOD) (1 km) datasets for years 2001, 2011, and 2021, and TROPOMI (Tropospheric Monitoring Instrument) (Sentinel 5‐P) was used for gaseous pollutants, namely, CO, SO 2 , NO 2 , and O 3 . MLA and classification and regression tree (CART) were used for preparing region‐wise LULC maps for Delhi, Chandigarh, Haryana, Himachal Pradesh, Punjab, Uttarakhand, and Jammu and Kashmir. LULC maps were further checked for their accuracy. Accuracy assessment and kappa coefficients showed that most maps had an accuracy of more than 75%. While the AQ datasets were validated with ground‐monitored data of the Central Pollution Control Board (CPCB). The results revealed a moderate to good correlation between satellite and ground data, and most datasets had the coefficient of correlation as 0.5 < r < 0.8 and above values. Different image interpretation platforms, like ArcGIS, Erdas Imagine, Google Earth Engine (GEE), and so forth, were used for the spatial analysis of the datasets. Spatial analysis of all the images revealed a pollution distribution as follows: Built‐up > Water‐Bodies > Vegetation > Bare/Barren‐Land > Snow‐Cover. ΔLULC (LULC change) had a profound effect on ΔAQ and generally followed the same trend, that is, ΔBuilt‐up > ΔWater‐Bodies > ΔVegetation > ΔBare/Barren‐Land > ΔSnow‐Cover. Overall, the regions of Delhi, Haryana, and Chandigarh were noted to be more polluted than the Himalayan regions of Jammu and Kashmir, Himachal Pradesh, and Uttarakhand. Finally, a geographically weighted regression (GWR) was performed in Delhi between ΔAOD and ΔLULC (ΔBuilt‐up, ΔWater‐Bodies, ΔVegetation, ΔBare/Barren‐Land). The region was divided into 190 grids of 3 km × 3 km size, and a regression analysis was done for each square. The GWR showed the highest determination coefficient ( R 2 ) of 0.782 when applied between 2001 and 2021, using all the variables in the model with the RMSE values reaching as low as 8.08 (mean residual sum of squares [RSS] = 65.29). Overall, the GWR model showed decent results for Delhi.