Abstract
Clarifying the spatio-temporal evolution of PM2.5 concentration law and its driving mechanism is crucial for the prevention and control of air pollution in urban agglomerations, also helping promote their high-quality development. Based on remote sensing and statistics of urban agglomerations in China's Beijing-Tianjin-Hebei (BTH), Yangtze River Delta (YRD), and Pearl River Delta (PRD) from 2005 to 2020, the paper analyses the evolution characteristics of the pollution concentration pattern and identifies the influencing factors through spatial analysis method combining the geodetector and geographically weighted regression (GWR) model. As the results show, during the study period: (1) Temporal Trends: annual PM2.5 concentrations exhibited significant declines, with BTH decreasing from 1004.71 μg/m3 (2006) to 528 μg/m3 (2020), YRD from 1434.81 μg/m3 (2008) to 621 μg/m3, and PRD from 405.02 μg/m3 (2007) to 292 μg/m3. The ranking remained YRD > BTH > PRD throughout the study period. (2) Spatial Heterogeneity: Spatial clustering (Moran's I: 0.286-0.729, p < 0.05) dominated all regions. BTH showed a "high-south" pattern (e.g., Xingtai: 78.3 μg/m3 vs. Qinhuangdao: 34.2 μg/m3), YRD displayed "high-northwest" characteristics (Hefei: 68.5 μg/m3 vs. Ningbo: 42.1 μg/m3), while PRD exhibited a west-east gradient (Foshan: 49.8 μg/m3 vs. Shenzhen: 25.6 μg/m3). (3) The evolution of PM2.5 concentration in three urban agglomerations is generally positive autocorrelative aggregative distribution, and aggregation types include "high-high", "low-low" and "high-low". (4) The measurement of geographical detector indicates the differentiation of PM2.5 concentration is affected by both natural geography and socio-economic factors, and the former ones have stronger driving forces. (5) The measurement of GWR model indicates temperature, precipitation, vegetation coverage, urban expansion, industrial structure, and energy efficiency are main influencing factors of PM2.5 concentration pattern, and the degree of influence of these factors is different.