Applying surface reflectance to investigate the spatial and temporal distribution of PM<inf>2.5</inf> in Northern Thailand

Abstract

The MODIS surface reflectance (SR) product (MOD09) was used to predict PM2.5 concentrations with a regression model. The predicted results were compared with the MAIAC-AOD model and the ground based measurements. The output from the MODIS product is regularly employed to predict air pollution and emissions, while the AOD model is normally used to predict PM2.5 concentrations. This study investigated PM2.5 concentrations in Northern Thailand by using SR via a linear regression model. The results showed that the highest value of SR was observed in Band-2 (0.17–0.27), followed by Band-1 and Band-4 (0.10–0.14) and Band-3 (0.07–0.10). Moreover, the correlation coefficient between SR-band-2 versus the measured PM2.5 from the master stations with PM2.5 sensor was greater than those of the other bands. The correlation coefficients between the predicted PM2.5 by the MODIS-SR and by the MAIAC-AOD models and the measured PM2.5 from the master stations varied between 0.3871–0.8588 and 0.3913–0.7802, respectively. The range of prediction efficiency by the SR model was 10.8%–27.2%, which was greater than the AOD model. It should be concluded that the distribution of spatial PM2.5 concentrations obtained from surface reflectance and MAIAC-AOD predictions was similar.