Scholarlink

<p>Rapid urbanization in Riyadh has led to elevated levels of key air pollutants, including aerosol index (AI), methane (CH₄), carbon monoxide (CO), nitrogen dioxide (NO₂), sulphur dioxide (SO₂), formaldehyde (HCHO), and ozone (O₃). However, limited research has quantified the differential impact of specific road hierarchies on these pollutants. This study aims to assess how eight OpenStreetMap road classes primary, secondary, tertiary, trunk road, motorway, residential, living street, and unclassified affect spatial patterns of urban air pollution using interpretable machine learning models. Sentinel-5P Level-3 annual pollutant rasters for 2023 were quality-controlled, downscaled to 30 m, and combined with Euclidean distances to each road type. Random Forest, Gradient Boosting, and XGBoost regressors were trained on 80% of the data with five-fold cross-validation and tested on 20%. SHAP (SHapley Additive exPlanations) was used to interpret model predictions at both global and local scales. Predictive performance ranged from R² = 0.55–0.86. Random Forest achieved the highest accuracy for NO₂ (R² = 0.857), AI (0.76), SO₂ (0.69), and HCHO (0.67), while Gradient Boosting performed best for CO (0.57) and CH₄ (0.55). SHAP analysis revealed that high-capacity roads (primary, trunk, motorway) were key drivers of traffic-related pollutants, whereas secondary, tertiary, and residential roads had stronger influence on diffuse or secondary pollutants. NO₂ and CO concentrations increased substantially near major roads, while CH₄ showed moderate increases near secondary and tertiary roads. O₃ levels were less sensitive to road proximity, reflecting photochemical formation processes. Road hierarchy strongly shapes urban pollutant patterns in Riyadh. Evidence-based interventions include establishing vegetated buffer zones along high-capacity roads, implementing targeted traffic rerouting in hotspots (e.g., Al-Rimal, Al-Naseem), and prioritizing mixed-use areas for localized mitigation. Integrating land use, meteorology, and population density into future models will enhance predictive capability. The framework is scalable to other rapidly urbanizing cities, providing a data-driven approach for sustainable urban planning and air quality management.</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

An explainable artificial intelligence based assessment of differential impacts of road hierarchy on urban emissions in Riyadh City

  • 期刊论文

  • 发表日期：

    2026年1月22日

查看全文
  • Javed Mallick,
  • Saeed Alqadhi

摘要

Rapid urbanization in Riyadh has led to elevated levels of key air pollutants, including aerosol index (AI), methane (CH₄), carbon monoxide (CO), nitrogen dioxide (NO₂), sulphur dioxide (SO₂), formaldehyde (HCHO), and ozone (O₃). However, limited research has quantified the differential impact of specific road hierarchies on these pollutants. This study aims to assess how eight OpenStreetMap road classes primary, secondary, tertiary, trunk road, motorway, residential, living street, and unclassified affect spatial patterns of urban air pollution using interpretable machine learning models. Sentinel-5P Level-3 annual pollutant rasters for 2023 were quality-controlled, downscaled to 30 m, and combined with Euclidean distances to each road type. Random Forest, Gradient Boosting, and XGBoost regressors were trained on 80% of the data with five-fold cross-validation and tested on 20%. SHAP (SHapley Additive exPlanations) was used to interpret model predictions at both global and local scales. Predictive performance ranged from R² = 0.55–0.86. Random Forest achieved the highest accuracy for NO₂ (R² = 0.857), AI (0.76), SO₂ (0.69), and HCHO (0.67), while Gradient Boosting performed best for CO (0.57) and CH₄ (0.55). SHAP analysis revealed that high-capacity roads (primary, trunk, motorway) were key drivers of traffic-related pollutants, whereas secondary, tertiary, and residential roads had stronger influence on diffuse or secondary pollutants. NO₂ and CO concentrations increased substantially near major roads, while CH₄ showed moderate increases near secondary and tertiary roads. O₃ levels were less sensitive to road proximity, reflecting photochemical formation processes. Road hierarchy strongly shapes urban pollutant patterns in Riyadh. Evidence-based interventions include establishing vegetated buffer zones along high-capacity roads, implementing targeted traffic rerouting in hotspots (e.g., Al-Rimal, Al-Naseem), and prioritizing mixed-use areas for localized mitigation. Integrating land use, meteorology, and population density into future models will enhance predictive capability. The framework is scalable to other rapidly urbanizing cities, providing a data-driven approach for sustainable urban planning and air quality management.