Co‐Benefits Between PM 2.5 Control and Carbon Reduction Revealed by Interpretable Machine Learning Methods

作者：Xu Han, Jiahao Huang, Dongheng Zhao, Zhang Zhang, Junbo Huang, Feng Wang, Jin‐Xing Liu, Shuai Jiang, Yinchang Feng, Shaojie Song, Guoliang Shi · 发表于：Geophysical Research Letters · 年份：2025 · DOI：10.1029/2025gl115654 · 被引用次数：4 · 研究领域：Air Quality and Health Impacts、Air Quality Monitoring and Forecasting、Vehicle emissions and performance

Abstract Air pollution and climate change, driven by fine particulate matter (PM 2.5 ) and carbon dioxide (CO 2 ), present critical challenges to human survival. Understanding the interaction between PM 2.5 control and carbon reduction‐specifically, how mitigating PM 2.5 sources impacts CO 2 levels and vice versa‐is essential for effective policy‐making. To address this, we developed an Interpretable machine learning (ML) and source apportionment (IMSA) framework. The framework screens pollutant sources for PM 2.5 and CO 2 , and calculates their contributions, revealing that industrial emissions (IE) (11%, 29%), vehicle exhaust (VE) (13%, 19%), and coal combustion (19%, 15%) are major shared sources. By integrating interpretable ML methods, IMSA uncovers interaction effects, showing that reducing IE significantly lowers CO 2 , while targeting VE more effectively reduces PM 2.5 . The IMSA framework provides critical insights for co‐beneficial strategies to improve air quality and mitigate climate change.