Publications

You, Wei; Zang, Zengliang; Zhang, Lifeng; Li, Zhijin; Chen, Dan; Zhang, Gui (2015). Estimating ground-level PM10 concentration in northwestern China using geographically weighted regression based on satellite AOD combined with CALIPSO and MODIS fire count. REMOTE SENSING OF ENVIRONMENT, 168, 276-285.

Abstract
Satellite measurements have been widely used to estimate particulate matters (PMs) on the ground and their effects on human health. However, such estimation depends critically on an established relation between aerosol optical depth (AOD) and ground level PMs. In this study we performed an experiment at Xi'an to establish a relation between AOD and PM10. A first test of satellite AOD and PM10 data showed a low correlation between the two properties, especially in spring and summer seasons with relatively many high AOD values paired to low ground-level PM10 values, which hinted that AOD may 'contaminate' by summer agricultural burning season and dust storm intrusion leading to much elevated aerosol. MODIS fire count and CALIPSO vertical feature mask data were used to detect fire emissions and elevated aerosol to substantiate this hypothesis. A later test showed that the correlation between ADD and PM10 has strongly been improved by excluding the paired AOD-PM10 data with known biomass burning and elevated aerosol layers. This result indicates that summer agricultural burning and aloft aerosol should be considered when the AOD-PM10 association is explored. Finally, a geographically weighted regression model was developed to further improve estimation of ground-level concentration by combining AOD with meteorological parameters. The results showed that the meteorological parameters can greatly improve model performance. The overall cross-validation R-2 is 0.77 and the root mean squared prediction error (RMSE) is 16.91 mu g/m(3). These results are useful for developing satellite AOD based model to estimate ground-level PMs especially for the region with known strong biomass burning events. (C) 2015 Elsevier Inc. All rights reserved.

DOI:
10.1016/j.rse.2015.07.020

ISSN:
0034-4257