Objective To repair the deficiencies on missing historic data and real-time data of the PM2.5 with temporal-spatial limitation in the research of chronic population health impact by PM2.5 through building, testing, and evaluating a model to estimate ground-level PM2.5 of Shanghai using satellite remote sensing data.
Methods We retrieved the moderate-resolution imaging spectroradiometer (MODIS) L1B1KM data to derive Edry (dry extinction coefficient) via aerosol optical depth (AOD) by V5.2 algorithm and adjusted by height and relative humidity. We used PM2.5 concentration as dependent variable, and Edry, average temperature, average air pressure, as well as average wind speed as independent variables to build a multivariate linear regression model, followed by validation with real data and comparison with the model built by reported method.
Results The average value of AOD in spring was 0.57, 0.48 in summer, 0.41 in autumn, and 0.72 in winter. There was an obvious seasonal tendency that the AOD average values were higher in spring and winter, and the spatial distribution showed higher values in urban area than in suburban area. We built four multivariate linear regression models for four seasons:The Edry was positively correlated with the PM2.5 concentrations in four seasons; Both average air pressure and average wind speed were negatively correlated with PM2.5 concentrations in four seasons; Average temperature in winter and spring was positively correlated with PM2.5 concentrations, but that in summer and autumn was negatively correlated. We compared our models with the model built by reported method without weather factors, and found the latter's fitting precision and prediction precision were both around 50%, while our models reached 80.74%-90.83% and 70.58%-77.79% respectively, better than the latter one.
Conclusion The models built in this article based on satellite remote sensing data have higher fitting precision and prediction precision than the model built by reported method and could be used for estimating the ground-level PM2.5 concentration of Shanghai. Our results provide a solution to the research of PM2.5 monitoring and related chronic population health impact tackling the lack of historical PM2.5 concentration data and the restricted temporal-spatial data due to limited monitoring spots and poor coverage.
DownLoad: