Scholarlink

<p>Tropospheric ozone (O<sub>3</sub>) pollution is a pressing concern in China’s Yangtze River Delta (YRD), yet its drivers remain insufficiently understood. We developed an integrated framework combining Positive Matrix Factorization (PMF) source apportionment, meteorological normalization, and explainable machine learning to investigate O<sub>3</sub> pollution during stagnant periods at a suburban site of YRD, focusing on high-O<sub>3</sub> months. O<sub>3</sub> concentrations frequently exceeded air quality standards, and O<sub>3</sub> formation exhibited a shift from annual volatile organic compounds (VOC)-limited to a warm-season transitional regime, indicating the need for coordinated VOCs and nitrogen oxides (NO<sub>x</sub>) control during pollution periods. PMF resolved seven VOC sources, with mixed industrial emissions (MIE, 18.3%), biogenic emissions/biomass burning (BEBB, 24.3%), and industrial solvent use (ISU, 18.0%) as dominant contributors. Meteorological normalization showed that emission-driven O<sub>3</sub> (O<sub>3</sub>_EMI) was consistently and substantially positive, whereas meteorology-driven O<sub>3</sub> (O<sub>3</sub>_MET) was predominantly negative, underscoring the dominant role of local emissions and net O<sub>3</sub>-scavenging effect of meteorology. SHapley Additive exPlanations (SHAP) analysis identified MIE, BEBB, and nitrogen monoxide (NO) as the top three drivers of O<sub>3</sub>_EMI, with industrial MIE exhibiting the strongest positive contribution, likely due to its high loadings of reactive aromatics and oxygenated VOCs. Industrial VOC emissions thus emerge as the primary controllable driver of O<sub>3</sub> pollution under stagnant conditions. For O<sub>3</sub>_MET, relative humidity and boundary layer height were the most influential factors. By focusing on O<sub>3</sub>_EMI and leveraging PMF-resolved source factors, this framework enables targeted identification of controllable precursors, offering a transferable paradigm for diagnosing O<sub>3</sub> drivers in other regions facing similar challenges.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Unraveling O3 formation drivers in a Yangtze River Delta city under stagnant conditions: a PMF, meteorological normalization, and machine learning framework

  • 期刊论文

  • 发表日期：

    2026年7月7日

查看全文
  • Lei Tong,
  • Weicheng Tong,
  • Dan Li,
  • Dingming Xue,
  • Yang Meng,
  • Mengmeng He,
  • Hang Xiao

摘要

Tropospheric ozone (O3) pollution is a pressing concern in China’s Yangtze River Delta (YRD), yet its drivers remain insufficiently understood. We developed an integrated framework combining Positive Matrix Factorization (PMF) source apportionment, meteorological normalization, and explainable machine learning to investigate O3 pollution during stagnant periods at a suburban site of YRD, focusing on high-O3 months. O3 concentrations frequently exceeded air quality standards, and O3 formation exhibited a shift from annual volatile organic compounds (VOC)-limited to a warm-season transitional regime, indicating the need for coordinated VOCs and nitrogen oxides (NOx) control during pollution periods. PMF resolved seven VOC sources, with mixed industrial emissions (MIE, 18.3%), biogenic emissions/biomass burning (BEBB, 24.3%), and industrial solvent use (ISU, 18.0%) as dominant contributors. Meteorological normalization showed that emission-driven O3 (O3_EMI) was consistently and substantially positive, whereas meteorology-driven O3 (O3_MET) was predominantly negative, underscoring the dominant role of local emissions and net O3-scavenging effect of meteorology. SHapley Additive exPlanations (SHAP) analysis identified MIE, BEBB, and nitrogen monoxide (NO) as the top three drivers of O3_EMI, with industrial MIE exhibiting the strongest positive contribution, likely due to its high loadings of reactive aromatics and oxygenated VOCs. Industrial VOC emissions thus emerge as the primary controllable driver of O3 pollution under stagnant conditions. For O3_MET, relative humidity and boundary layer height were the most influential factors. By focusing on O3_EMI and leveraging PMF-resolved source factors, this framework enables targeted identification of controllable precursors, offering a transferable paradigm for diagnosing O3 drivers in other regions facing similar challenges.