Spatial self-aggregation effects and national division of city-level PM2.5 concentrations in China based on spatio-temporal clustering

With growing haze episodes in China, comprehensive air quality management has been frequently proposed and implemented during major events or heavy pollution episodes. However, except for such heavily polluted regions as the Beijing-Tianjin-Hebei region, regional integration of air quality management in other parts of China has rarely been discussed, due to limited research on the spatio-temporal aggregation of PM2.5 concentrations. To fill this gap, we employed a repeated-bisection method, which supports high dimensional datasets and bootstrap clustering, for spatio-temporal clustering of city-level PM2.5 concentrations in China using time-series PM2.5 data and the test of geographical detector proved the reliability of the clustering. Since no weighted geographical information was employed during the clustering process, this research suggested that PM2.5 concentrations in China were of strong spatial self-aggregation effects, which proved the necessity for regional integration of air quality management. Based on the spatio-temporal clustering of PM2.5 concentrations, we further proposed six divisions of PM2.5 concentrations across China, within which PM2.5 concentrations display similar variation patterns and specific emission-reduction measures can be implemented accordingly. The division output of PM2.5 concentrations was highly consistent with the recent “2017 air pollution prevention and management plan for the Beijing-Tianjin-Hebei region and its surrounding areas” plan, indicating the reliability and practical significance of the national division of PM2.5 concentrations based on spatio-temporal clustering. The findings and methodology from this research provide useful reference for improving regional air quality management by better understanding spatio-temporal aggregation of PM2.5 concentrations.