Spatial Temporal Analysis of Air Quality and Aerosol Optical Depth of India using Machine Learning

Rising PM2.5 levels are a major concern for public health and thus monitoring its concentration effectively is important. This study aims to analyze Aerosol Optical Depth time series area-averaged data from MODIS Terra satellite and pollutant and meteorological data from ground stations in 435 cities of India from the year 2014 to 2023 using various Machine Learning and Deep Learning techniques. In total, 9 different models are compared in this paper. The data was divided into North and South to improve spatial accuracy and comparison. Stacking Ensemble outperformed all the models with an R2 score of 0.737 and RMSE value of 29.56 for North India data and R2 score of 0.763 and RMSE value of 8.69 for South region. The best performing individual model was Random Forest with an R2 score in the range of 0.72(North) and 0.74(South) and RMSE equivalent to 30.3 for North and 9.1 for South. This study aims to analyze PM2.5 by combining AOD data from MODIS Terra with ground measurements and using ML and DL techniques for North and South India separately. The results are good compared to recent studies in other regions, thus offering a thorough approach to predicting PM2.5. By contributing to improved air quality monitoring, this study aims to contribute to the Sustainable Development Goals (SDGs) related to health, sustainable cities, and climate action.