Assessments of premature deaths caused by PM2.5 exposure have important scientific significance and provide valuable information for future human health–oriented air pollution prevention. PM2.5 concentration data are particularly vital and may cause great uncertainty in premature death assessments. This study constructed an index of deviation frequency to compare differences in premature deaths assessed by five sets of extensively used PM2.5 concentration remote sensing datasets. Then, a preferred combination project of the PM2.5 dataset was proposed by selecting relatively high-accuracy PM2.5 concentration datasets in areas with significant differences. Based on this project, an index of uncertainty was constructed to quantify the effects of using different PM2.5 datasets on premature death assessments. The results showed that there were significant differences in PM2.5 attributable to premature deaths assessed by different datasets from 2000 to 2016, and the differences were most obvious in 2004. Spatially, differences were most significant in Jilin, Fujian, Liaoning, Hebei, Shanxi, Hubei, Sichuan, and Yunnan. The differences were caused by PM2.5 concentration; therefore, in order to reduce uncertainty in subsequent premature death assessments because of using different PM2.5 concentration data, the CGS3 dataset was recommended for Jilin, Sichuan, Yunnan, and Fujian, and the CHAP dataset was recommended for Liaoning, Hebei, Shanxi, and Hubei, and for other regions, CGS3, CHAP, or PHD datasets were more applicable. The CHAP dataset was the best selection for premature death assessments in the whole area. Based on the preferred combination project of the PM2.5 dataset, uncertainty in annual premature death assessments could be reduced by 31 and 159% in the whole and local area, respectively. The research results will provide a scientific basis for a reasonable selection of PM2.5 concentration remote sensing datasets in air pollution premature death assessments in China.
Full Text Finder