Estimations of PM 2.5 concentrations based on the geographically weighted regression from Himawari-8 AOD

Abstract

Polar-orbit satellites have the limitation of observed frequency, so the method of PM 2.5 concentration estimation based on MODIS AOD is difficult to capture the variation in trend and distribution characteristics of PM 2.5 in a day. Himawari-8 geostationary satellite with the advantage of high temporal resolution can obtain the hourly aerosol optical depth(AOD).This paper selected AOD from 2km Himawari-8 Standard Data to estimate the PM 2.5 concentration based on Geographically Weighted Regression(GWR)model, and used the ground-level PM 2.5 concentrations as training samples of the model. In addition to AOD, the model also took meteorological data such as height of the planetary boundary layer (HPBL) and relative humidity (RH) as parameters to participate in model training to improve the reliability of the results. At the end of the paper, it utilized the model to estimate PM 2.5 concentration over eastern China in July 2018 and executed the validation with ground-level PM 2.5 concentrations. The result indicated there is a basic consistency in the change trend between them (R=0.74, RMSE=10.07μg/m 3 ).