Multi-timescale analysis of air pollution spreaders in Chinese cities based on a transfer entropy network

Abstract

Cross-regional air pollutant spillovers aggravate air pollution in China. To mitigate air pollution, identifying and monitoring air pollution spreaders (APS) is a vital strategy that helps locate the source of air pollution and guides the Joint Prevention and Control of Air Pollution. In this paper, we define an APS as a city with a high spillover impact (CHSI) of air pollution and propose a transfer entropy network to investigate the APS from a multi-timescale analysis perspective. Taking the time series of PM2.5 concentration of 358 Chinese cities from 1 January 2015 to 31 December 2020 as the sample, they are decomposed into short, medium, and long timescales, corresponding to an average period of 12, 111, and 530 days, respectively. Then, we use transfer entropy networks to analyze APS’s spatial distribution and temporal variation patterns on each timescale. The results demonstrate that air pollution spillover widely exists in Chinese cities, and the short-term air pollution spillover dominates all spillovers. The CHSIs form large agglomeration areas in Central and East China on short and medium timescales, while the results of the undecomposed data show a more discrete distribution. In addition, the cities’ air pollution spillover impact is usually high in winter and spring and low in summer. Moreover, the spillover impacts of half of the cities have a lead-lag relationship between short and medium timescales. All results suggest that combining short-term controls and longer-term strategies helps China mitigate air pollution and develop sustainably.