Measurement and Modelling of Particulate Pollution over Kashmir Himalaya, India

Abstract

Ground and satellite measurements of particulate pollution play an important role in determining the particulate pollutant-Aerosol Optical Depth (AOD) relationship. The daily observed PM10 and PM2.5 concentration varied from 11–757 μg/m3 and 8–630 μg/m3 with the mean concentrations of 137 ± 119 μg/m3 and 86 ± 90 μg/m3, respectively. The long-term mean annual PM10 and PM2.5 levels are several times higher than the WHO permissible limits. The 1377 satellite-derived AOD observations from the Moderate Resolution Imaging Spectrometer, ground-based particulate matter (PM) and meteorological observations from 2013–2017 were analysed to develop two-variate linear model (TVM) (AOD versus PM10 or PM2.5) and multi-variate regression models (MVMs) (AOD + meteorological parameters versus PM10 or PM2.5) for estimation of the ground level PM10 and PM2.5 in the Kashmir Himalaya, India. The model evaluation showed that the PM predication estimates are significant at 99% confidence level for all the models. The TVM predicts daily PM10 concentration better than PM2.5 explaining 82% and 74% variance in the observed data, respectively. By adding meteorological data to the regression analysis, there is an improvement of 5% and 11% in R2 for PM10 and PM2.5 estimates which inter alia reduced the RMSE by 11.8% and 20.47%, respectively. Estimation of the particulate pollution, utilising satellite-based AOD, observed PM and meteorology, would encourage satellite-based air quality monitoring in the data-scarce Himalaya. However, it is suggested that more studies are required to improve the operational prediction of PM pollution by incorporating satellite observations of other pollutants, and processes in the model using advanced approaches.