Scholarlink

<p>Traditional air quality monitoring identifies pollution hotspots but fails to disentangle causal drivers from mere correlations, hindering effective policy formulation. This study characterizes the spatial heterogeneity of PM₁₀, PM₂.₅, and PM₁ across 12 diverse land-use zones in Chennai, India, and introduces an Enhanced Causal Ensemble Framework to quantify the causal impact of urban features on air quality. High-resolution 1-hour snapshot data were collected from zones ranging from industrial complexes to eco-sensitive protected areas. Overcoming the limitations of conventional regression, we employed Double Machine Learning (DML) and Meta-Learners to estimate the Average Treatment Effect (ATE) of features such as Industrial Activity, Traffic Intensity, and Eco Protection on PM₂.₅ concentrations. Descriptive analysis revealed extreme spatial disparity, with the Dumpsite Zone recording PM₁₀ levels of 570&#xa0;µg/m³. The causal ensemble identified Eco Protection as the most effective pollution mitigation mechanism (ATE: -172&#xa0;µg/m³). Conversely, Industrial Activity (+ 114&#xa0;µg/m³) and Waste Management (+ 38&#xa0;µg/m³) were confirmed as primary drivers of PM₂.₅ escalation, surpassing the causal effect of Traffic Intensity (+ 17&#xa0;µg/m³). We conclude that unorganized waste combustion and industrial emissions are the dominant drivers of peak pollution events in Chennai. A Targeted High-Impact policy—focusing on dumpsite containment and green buffer expansion—is projected to be the most cost-effective strategy for peak-hour urban air quality improvement.</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

Causal drivers of urban PM₂.₅ in Chennai: a high-resolution assessment across twelve land-use zones using integrated monitoring and machine-learning analysis

  • 期刊论文

  • 发表日期：

    2026年5月26日

查看全文
  • Vignesh Jagajeevan,
  • Vidhya Lakshmi Sivakumar

摘要

Traditional air quality monitoring identifies pollution hotspots but fails to disentangle causal drivers from mere correlations, hindering effective policy formulation. This study characterizes the spatial heterogeneity of PM₁₀, PM₂.₅, and PM₁ across 12 diverse land-use zones in Chennai, India, and introduces an Enhanced Causal Ensemble Framework to quantify the causal impact of urban features on air quality. High-resolution 1-hour snapshot data were collected from zones ranging from industrial complexes to eco-sensitive protected areas. Overcoming the limitations of conventional regression, we employed Double Machine Learning (DML) and Meta-Learners to estimate the Average Treatment Effect (ATE) of features such as Industrial Activity, Traffic Intensity, and Eco Protection on PM₂.₅ concentrations. Descriptive analysis revealed extreme spatial disparity, with the Dumpsite Zone recording PM₁₀ levels of 570 µg/m³. The causal ensemble identified Eco Protection as the most effective pollution mitigation mechanism (ATE: -172 µg/m³). Conversely, Industrial Activity (+ 114 µg/m³) and Waste Management (+ 38 µg/m³) were confirmed as primary drivers of PM₂.₅ escalation, surpassing the causal effect of Traffic Intensity (+ 17 µg/m³). We conclude that unorganized waste combustion and industrial emissions are the dominant drivers of peak pollution events in Chennai. A Targeted High-Impact policy—focusing on dumpsite containment and green buffer expansion—is projected to be the most cost-effective strategy for peak-hour urban air quality improvement.