The Spatiotemporal Evolution Patterns and Spatial Differentiation Mechanisms of PM2.5 Concentrations in East China Based on Multi-Source Fused Data

To reveal the spatiotemporal evolution of PM2.5 concentrations in East China from 2000 to 2023 and quantify the multiscale driving effects of natural and socioeconomic factors, we integrated satellite-derived and ground-based monitoring data to construct a multi-source dataset. XGBoost-SHAP screened key factors; Theil–Sen trend, Hurst index, and Moran’s I characterized spatiotemporal patterns; and the geographic detector and MGWR quantified driving mechanisms. PM2.5 declined significantly (Sen slope: −1.29 μg m−3 a −1 to −0.03 μg m−3 a −1), with accelerated decrease after 2013. The Hurst index indicatedsustainable improvement in the Yangtze River Delta core but reversal risk in southwestern Zhejiang and northern Fujian. Spatially, the north–south gradient intensified: high concentrations persisted in industrial regions (northern Jiangsu, northern Anhui, western Shandong), whereas low concentrations remained in ecological zones (southwestern Zhejiang, northern Fujian). Natural factors dominated spatial variation; temperature (q = 0.804) and precipitation (q = 0.724) showed the strongest explanatory power. MGWR further revealed the stronger negative effects of temperature and precipitation in the north than in the south, and a continuous spatial gradient of per capita GDP from coastal industrial clusters to inland ecological zones. These findings underscore the need for regionspecific emission reduction strategies that account for climatic heterogeneity.