The Influence of Emission Sources and Meteorological Factors to Long-Term Changes in PM2.5 over China (1980–2022)

PM2.5 is a major air pollutant characterized by complex sources and strong spatiotemporal heterogeneity. However, accurately quantifying the relative contributions of different factors remains difficult due to the lack of long-term datasets and the strong correlations between meteorological factors and emissions. To address this problem, the study utilizes the China long-term particulate matter (CLPM) dataset developed in previous research to investigate the dominant drivers and regional disparities of PM2.5 concentration variations from 1980 to 2022. The analysis employs Gaussian Convolution (GC) to model pollutant diffusion, Partial Least Squares (PLS) regression to address multicollinearity, and the Lindeman-Merenda-Gold (LMG) method to quantify the relative contributions of each driver. The results reveal that as the convolution scale increased from 0.25° to 10°, dominant PM2.5 sources shifted from local anthropogenic emissions to regional biomass burning and large-scale dust transport, highlighting the scale-dependent transition of pollution drivers. Furthermore, PM2.5 concentrations are predominantly explained by emissions, which account for over 60% of the total variance and exceed 80% in eastern China, while meteorological factors are associated with 12–26%. Among these, total precipitation and downward surface solar radiation have the strongest influences on pollutants. It is important to note that these results reflect the statistical explanatory power of emissions and meteorological variables within the regression model. Overall, this research provides a method for separating the statistical influences of emissions and meteorological factors, offering methods for multi-scale explanatory power of PM2.5 and other atmospheric pollutants.