Organic aerosols are an important and highly dynamic component of fine particulate matter, yet their long-term response to emission controls is poorly constrained. We analyze a decade (2013–2023) of wintertime aerosol mass spectrometry data from urban Nanjing, eastern China, using a developed machine learning framework that disentangles anthropogenic emission-driven changes from meteorology-driven changes. The mean organic aerosol concentrations decreased from 24.6 to 16.5 μg m–3 during the period of 2013–2017. After accounting for meteorological influences, emission controls account for ∼94% of the observed decrease. However, the effectiveness of anthropogenic emission controls on the total organic aerosol was weakened by a factor of approximately 2–8 times in the subsequent emission control phases, particularly with more oxidized secondary organic aerosol showing a minimal further decrease. Machine learning-based attribution analysis reveals that reductions in fossil fuel combustion and traffic-related aromatic precursors explain, on average, ∼50–60% of the long-term variability in secondary organic aerosol, while the meteorological influence plays a minor role. These results provide observationally source-resolved evidence that current measures are reaching diminishing returns and that effective future controls must target overlooked precursors and secondary formation pathways.
Zhang et al. (Sun,) studied this question.