Estimating particulate matter using sattellite based aerosol optical depth and meteorological parameters in Malaysia

Abstract

The insufficient number of ground-based stations for measuring Particulate Matter less than 10µm (PM10), especially in the developing countries hinders PM10 monitoring at a regional scale. The present study aims to develop empirical models for PM10 estimates from space over Malaysia using Aerosol Optical Depth (AOD550) retrieval from Moderate Resolution Imaging Spectroradiometer (MODIS), Medium Resolution Imaging Spectrometer/Advanced Along-Track Scanner Radiometer (MERIS/AATSR) synergy algorithm and meteorological data that include surface temperature, relative humidity and atmospheric stability from 2007-2011. Accuracy of meteorological parameters that have been used in the estimation of PM10 are examined. The estimated relative humidity and surface temperature using satellite data agree well with ground data where coefficient of determination (R2) = 0.78 and 0.49 and Root Mean Square Error (RMSE) = 5.14% and 2.68?C for relative humidity and surface temperature respectively. Multiple Linear Regressions (MLR) and Artificial Neural Network (ANN) techniques are utilized to develop the empirical models. The models were developed using PM10 data measured at 29 stations over Malaysia. Result of the research reveals that the ANN using MODIS AOD550 provide higher accuracy with R2 = 0.71 and RMSE = 11.61?gm-3 compared to the MLR method where R2 = 0.66 and RMSE = 12.39?gm-3 or models that use MERIS/AATSR AOD data. Stepwise regression analysis performed on the MLR method reveals that the MODIS AOD550 is the most important parameter for PM10 predictions where R2 = 0.59 and RMSE = 13.61?gm-3. However, the inclusion of the meteorological parameters in the MLR increases the accuracy of the PM10 estimations. The significance of the meteorological parameters in prediction of PM10 concentrations is in the order of (i) atmospheric stability, (ii) relative humidity and (iii) surface temperature. The estimated PM10 concentrations are validated against another 16 stations dataset of measured PM10 with the ANN model to result in higher accuracy (R2= 0.58, RMSE = 10.16?gm-3) compared to the MLR technique (R2 = 0.56, RMSE = 10.58?gm-3). The higher accuracy that has been attained in PM10 estimations from space allows (i) to map the PM10 distribution at large spatial and temporal scales and (ii) permits for future estimates of PM2.5 concentrations from space for monitoring of the Environmental Performance Index (EPI).