Exploring the association between long-term MODIS aerosol and air pollutants data across the Northern Great Plains through machine learning analysis

Aerosol optical depth (AOD) and Ångström exponent (AE) are the major environmental indicators to perceive air quality and the impact of aerosol on climate change and health as well as the global atmospheric conditions. In the present study, an average of AOD and AE data from Tera and Aqua satellites of MODIS sensors has been investigated over 7 years i.e., from 2016 to 2022, at four locations over Northern Great Plains. Both temporal and seasonal variations over the study periods have been investigated to understand the behavior of AOD and AE. Over the years, the highest AOD and AE were observed in winter season, varying from 0.75 to 1.17 and 1.30 to 1.63, respectively. During pre-monsoon season, increasing trend of AOD varying from 0.65 to 0.95 was observed from upper (New Delhi) to lower (Kolkata) Gangetic plain, however, during monsoon and post-monsoon a reverse trend varying from 0.85 to 0.65 has been observed. Seasonal and temporal aerosol characteristics have also been analyzed and it has been assessed that biomass burning was found to be the major contributor, followed by desert dust at all the locations except in Lucknow, where the second largest contributor was dust instead of desert dust. During season-wise analysis, biomass burning was also found to be as the major contributor at all the places in all the seasons except New Delhi and Lucknow, where dust was the major contributor during pre-monsoon. A boosting regression algorithm was done using machine learning to explore the relative influence of different atmospheric parameters and pollutants with PM2.5. Water vapor was assessed to have the maximum relative influence i.e., 51.66 % followed by CO (21.81 %). This study aims to help policy makers and decision makers better understand the correlation between different atmospheric components and pollutants and the contribution of different types of aerosols.