Potential application of DMSP/OLS nighttime light data for estimating ground-level PM2.5 concentrations

摘要

This study attempted to establish the relationship between ground-level PM2.5 concentrations and satellite-retrieved aerosol optical depth (AOD) using the Vegetation Adjusted NTL Urban Index (VANUI), over the New England region for the year 2013. A geographically weighted regression (GWR) model was used to predict ground-level PM2.5 on a daily basis. The study demonstrates that DMSP/OLS NTL data has the potential to be used as an alternative source for estimating PM2.5 exposure. Results from the 10-fold cross validation showed that the accuracy of daily NTL-GWR model were relatively higher than the model without NTL data. Moreover, the NTL-GWR model can illustrate more details of the predicted surface due to its continuous spatial patterns. Overall, it can be concluded that the DMSP/OLS nighttime light data is promising in delineating PM2.5, thus providing a supplemental data source to in situ monitoring and computational modeling.