Tracking surface ozone responses to clean air actions under a warming climate in China using machine learning

Surface ozone, a major air pollutant with important implications for air quality, ecosystems, and climate, shows long-term trends shaped by both anthropogenic and climatic drivers. Here, we developed a machine learning-based approach, namely the fixed emission approximation (FEA), to decouple the effects of meteorological variability and anthropogenic emissions on summertime ozone trends in China under the clean air actions. Anthropogenic emissions drove an approximately +23.2 ± 1.1 µg m −3 increase in summer maximum daily 8 h average ozone during 2013–2017, followed by an approximately −4.6 ± 1.5 µg m −3 decrease between 2017 and 2020 in response to strengthened emission controls. In contrast, meteorological anomalies, including heatwaves and rainfall conditions, emerged as substantial drivers of ozone variability during 2020–2023. Satellite-derived formaldehyde-to-nitrogen dioxide ratios revealed widespread urban volatile organic compounds-limited regimes for ozone production, with a shift toward nitrogen oxides-limited sensitivity under influence of heatwaves. Extending the FEA framework to assess long-term climate influences from 1970 to 2023, we find that sustained climate warming has driven a substantial increase in urban summertime ozone in China. These results demonstrate that climate change was increasingly offsetting the benefits of emission reductions and highlight the need for integrated ozone mitigation strategies that jointly address emission controls and climate adaptation in a warming world.