A robust calibration approach for PM10 prediction from MODIS aerosol optical depth
- ISSN: 1680-7316
- DOI: 10.5194/acp-13-3517-2013
Investigating the human health effects of atmospheric particulate matter\n(PM) using satellite data are gaining more attention due to their wide\nspatial coverage and temporal advantages. Such epidemiological studies\nare, however, susceptible to bias errors and resulted in poor predictive\noutput in some locations. Current methods calibrate aerosol optical\ndepth (AOD) retrieved from MODIS to further predict PM. The recent\nsatellite-based AOD calibration uses a mixed effects model to predict\nlocation-specific PM on a daily basis. The shortcomings of this daily\nAOD calibration are for areas of high probability of persistent cloud\ncover throughout the year such as in the humid tropical region along the\nequatorial belt. Contaminated pixels due to clouds causes radiometric\nerrors in the MODIS AOD, thus causes poor predictive power on air\nquality. In contrary, a periodic assessment is more practical and robust\nespecially in minimizing these cloud-related contaminations. In this\npaper, a simple yet robust calibration approach based on monthly AOD\nperiod is presented. We adopted the statistical fitting method with the\nadjustment technique to improve the predictive power of MODIS AOD. The\nadjustment was made based on the long-term observation (2001-2006) of\nPM10-AOD residual error characteristic. Furthermore, we also\nincorporated the ground PM measurement into the model as a weighting to\nreduce the bias of the MODIS-derived AOD value. Results indicated that\nthis robust approach with monthly AOD calibration reported an improved\naverage accuracy of PM10 retrieval from MODIS data by 50% compared to\nwidely used calibration methods based on linear regression models, in\naddition to enabling further spatial patterns of periodic PM exposure to\nbe undertaken.