Abstract: Using MODIS L1B021KM data to obtain daily AOD data of Chengdu area, China, in 2018 firstly, then combined with PM2.5 ground monitoring data and meteorological data, a geographically weighted regression (GWR) model is constructed to obtain the monthly PM2.5 concentration of Chengdu. The results show that: (1) Compared with the multiple linear regression model, GWR model has higher credibility in the inversion of PM2.5 concentration in Chengdu area in 2018, which is specifically reflected in that R2, ERMS and EMA are 0.884, 7.8704 µg·m−3 and 6.1566 µg·m−3, respectively for GWR model in the inversion of PM2.5 concentration, which are better than 0.808, 9.7098 µg·m−3 and 7.6081 µg·m−3 of multiple linear regression. (2) On a monthly scale, PM2.5 concentration in Chengdu shows a characteristic of first decreasing and then increasing. The highest concentration reaches 67.38 µg·m−3 in February, and the lowest reaches 28.31 µg·m−3 in July. The seasonal variation of PM2.5 concentration is characterized by increasing in summer, autumn, spring and winter. (3) The spatial distribution of PM2.5 concentration in Chengdu generally presents a characteristics of “high in the middle and low on both sides”. The western region is a low-value PM2.5 concentration area, the central region is a high-value area, and Jianyang City and Jintang County in the east are the second-highest PM2.5 concentration areas.

Key words: remote sensing, geographically weighted regression, moderate-resolution imaging spectroradiometer, multiple linear regression, PM2.5