Satellite-Based Estimation of PM_2.5 using Meteorological and Statistical Analysis over Delhi, India

The study aims to estimate ground-level PM_2.5 concentrations over Delhi, India, using satellite-derived aerosol optical depth (AOD) and meteorological parameters. Ground-based PM_2.5 measurements were obtained from the Central Pollution Control Board/Delhi Pollution Control Committee (CPCB/DPCC) every minute, at multiple monitoring stations, including Anand Vihar, RK Puram, IIT Delhi, and others. The data was then converted to daily mean values which were used for further analysis. The annual mean PM_2.5 concentrations ranged from 90 to 105 µg/m³. Meteorological variables, including air temperature, wind speed, atmospheric pressure, and relative humidity, were acquired from the corresponding CPCB/DPCC monitoring stations, while planetary boundary layer height (PBLH) was obtained from the ERA5 reanalysis dataset. Aerosol optical depth (AOD) at 550 nm was derived from the combined Terra–Aqua MAIAC (MCD19A2 Collection 6.1) product available through NASA's LAADS DAAC. Following quality assurance filtering, valid AOD observations were temporally matched with the corresponding same-day ground-based PM_2.5 measurements. The resulting dataset was quality-controlled, temporally aligned, and used to develop statistical models for PM_2.5 estimation. A Multiple Linear Regression (MLR) model was employed to establish the linear relationship between PM_2.5, AOD, and meteorological variables, while a Generalized Additive Model (GAM) was implemented to capture potential nonlinear relationships among these predictors. Model performance was evaluated using cross-validation and standard statistical metrics (R², RMSE, and MAE), and the predictive performance of the MLR and GAM models was compared to assess the benefits of nonlinear modelling. The validated model was subsequently used to generate the spatial and temporal distribution of PM_2.5 across Delhi. A detailed analysis will be presented.