Divergent drivers of winter PM2.5 pollution in representative cities of the Sichuan Basin: Insights from observation and machine learning.

Fine particulate matter (PM2.5) continues to pose a significant challenge to regional air quality and human health in Chinese urban areas. This study elucidated the chemical compositions, source origins, and secondary transformation mechanisms of wintertime PM2.5 in Chengdu and Zigong, two representative cities within the Sichuan Basin. Observational results showed higher average PM2.5 mass concentrations in Zigong (63.0 ± 28.1 μg/m³) than in Chengdu (52.5 ± 29.2 μg/m³). Although secondary inorganic aerosols contributed 45.5% to PM2.5 in both cities, a distinct chemical divergence was observed: nitrate dominated in Chengdu, whereas sulfate prevailed in Zigong. Positive Matrix Factorization (PMF) identified secondary transformation as the primary driver of aerosol accumulation. Specifically, nitrate enrichment in Chengdu was driven by the synergy of high NOx emissions and stagnant meteorology, while sulfate formation in Zigong was accelerated by enhanced sulfur oxidation under high humidity and heavy metal catalysis. Machine learning model coupled with SHapley Additive exPlanations was developed to quantify the individual and coupled effects of meteorology and emissions. Results indicate that Chengdu is highly sensitive to secondary transformation under humid, stagnant conditions, necessitating urgent NOx reductions. In contrast, Zigong is primarily influenced by local emissions and poor diffusion. Under high humidity (RH >70% for Chengdu; >90% for Zigong), targeted emission control should be prioritized in Chengdu under strong easterly or northerly winds, while stringent emission regulation should be prioritized in Zigong under strong easterly or southerly winds. These findings provide a scientific basis for differentiated, city-specific multi-pollutant management in complex basin topographies.