Traffic volume and road network structure: Revealing transportation-related factors on PM2.5 concentrations

• Published in: Transportation research. Part D, Transport and environment Vol. 124; p. 103935
• Main Authors: Yu, Chengcheng, Deng, Yongqi, Qin, Zhengtao, Yang, Chao, Yuan, Quan
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.11.2023
• Subjects: Bayesian machine learning; Environmental impact; PM2.5 emission; SHAP theorem; Transportation system; PM2.5 emission; Transportation system; Bayesian machine learning; Environmental impact; SHAP theorem
• ISSN: 1361-9209; 1879-2340
• DOI: 10.1016/j.trd.2023.103935

Understanding the mechanisms by which urban transportation systems affect air pollution can provide guidance for developing a sustainable transportation system. Existing research has revealed the impacts of traffic volume on the concentration of PM2.5, and proposed strategies for reducing emissions and mitigating exposure accordingly. However, there is limited research that links road network structure to the spatial distribution of PM2.5. This study uses Bayesian neural networks to model how PM2.5 concentration is subject to a collection of transportation-wise factors and introduces SHAP models to explain the modeling results. The results show that (1) road network structure and traffic volume matter more than demographics, with respective contributions of 19.8% and 11.6%, to the concentration of PM2.5; (2) the improvement of road network structure has a diminishing marginal benefit in promoting the reduction of PM2.5 concentration. These findings can provide references for the improvement of air quality from the perspective of transportation planning.