oa Machine Learning Analysis of PM2.5 Driving Contributions in the Southern Sichuan Basin

The Southern Sichuan Basin in China is frequently plagued by particulate matter (PM2.5) pollution events in winter, owing to its unique topographical and meteorological conditions. This study combined the machine learning method with the receptor model to reveal the significance of driving factors and their impacts on PM2.5 concentrations. The results indicated that three primary wintertime severe PM2.5 pollution episodes in the Southern Sichuan Basin were driven by the combination of high temperatures (> 283 K), high atmospheric pressure (> 980 hPa), high relative humidity (> 80%), weak wind speeds (< 1 m·s-1), and low boundary layer height (< 500 m). Emissions were identified as the dominant factor (76.5%), followed by meteorological conditions (12.8%) and atmospheric chemical reactions (10.7%). Secondary sources (25.5%) and transport-related sources (24.7%) were identified as the main contributors to PM2.5 concentrations. The sensitivity analysis of secondary inorganic aerosols revealed that the most influential factor was ammonium (NH4+), followed by sulphate (SO42-) and nitrate (NO3-). This study advances our understanding of PM2.5 drivers and informs targeted pollution control strategies.