A novel method for quantifying the contribution of regional transport to PM _2.5 in Beijing (2013–2020): combining machine learning with concentration-weighted trajectory analysis

Abstract. Fine particulate matter (PM2.5) is closely linked to human health, with its sources generally divided into local emissions and regional transport. This study combined concentration-weighted trajectory (CWT) analysis with the HYSPLIT trajectory ensemble to obtain hourly resolution pollutant source results. The Extreme Gradient Boosting (XGBoost) model was then employed to simulate local emissions and ambient PM2.5 in Beijing from 2013 to 2020. The results revealed that clean air masses influencing the Beijing area mainly originated from the north and east regions, exhibiting a strong winter and weak summer pattern. Following the implementation of the Air Pollution Prevention and Control Action Plan (Action Plan) by the Chinese government in 2017, pollution in Beijing decreased significantly, with the most substantial reduction in regional transport pollution events occurring in the west region during summer. Regional transport pollution events were most frequent in spring, up to 1.8 times higher than in winter. Pollutants mainly originated from the west and south regions, while polluted air masses from the east showed the least reduction, and the proportion of pollution sources from this region was gradually increasing. The COVID-19 restrictions might have reduced PM2.5 concentrations in 2020. From 2013 to 2020, local emissions were the main contributors to pollution events in Beijing. The Action Plan has more effectively reduced pollution caused by regional transport, particularly during autumn and winter. This finding underscores the importance of Beijing prioritizing local emission reduction while also considering potential contributions from the east region to effectively mitigate pollution events.