Estimating PM2.5 concentration from satellite derived aerosol optical depth and meteorological variables using a combination model

Özet

In India, PM2.5 concentration is monitored at few locations by various agencies. The measurements are sporadicover temporal and spatial scales and include a lot of missing data. Satellite-based observations can complementthe existing network of these measurements. In this study, Moderate Resolution Imaging Spectroradiometer(MODIS) derived Aerosol Optical Depth (AOD) at five cities in Maharashtra, India was used in estimating groundPM2.5 concentration during January 2016–May 2017. Meteorological parameters were also included in themodel to enhance the accuracy. A combination model that combines multiple linear regression (MLR) and residualsof MLR was developed to obtain the estimates. The applicability of the approach was assessed for the twotypes of time series; one with less frequency of missing data and the other with a high frequency of missing data.The spatial analysis suggested high AOD at Mumbai. It was observed that the inclusion of meteorology in theregression equation improved the performance of the MLR model. The combination model outperformed MLRdue to the consideration of residuals of the MLR model.