A geographically weighted regression approach to investigate the effects of traffic conditions and road characteristics on air pollutant emissions. (1st December 2019)

• Record Type: Journal Article
• Authors: Xu, Chengcheng; Zhao, Jingya; Liu, Pan
• Is Part Of: Journal of cleaner production. Volume 239(2019)
• Publication Date: 2019-12-01
• Subjects: Traffic emissions -- Traffic analysis zone -- Air pollutant concentration -- Spatial heterogeneity -- Geographically weighted regression; Factory and trade waste -- Management -- Periodicals; Manufactures -- Environmental aspects -- Periodicals; Déchets industriels -- Gestion -- Périodiques; Usines -- Aspect de l'environnement -- Périodiques; 628.5
• DOI: 10.1016/j.jclepro.2019.118084
• Languages: English
• ISSNs: 0959-6526

Abstract

Abstract: This study aims to investigate the effects of traffic, road network, employment and social-demographic characteristics on air pollutant emissions at the level of traffic analysis zone (TAZ). Three air pollutants are considered, including carbon monoxide (CO), nitrogen oxides (NO2 ) and particulate matter (PM10 ). The emission, socio-demographic, employment, road network and traffic data are collected from Los Angeles County in 2010. The geographically weighted regressions (GWRs) are used to link the three pollutant concentrations with various contributory variables. Comparison results suggest that GWRs provide better fitness than conventional generalized linear regression model by considering spatial heterogeneity. The estimation results of the three GWRs indicate that the traffic, road network, social-demographic and employment characteristics affect CO, NO2 and PM10 concentrations at TAZ level. The visualization results indicate that the GWR models capture the varying relationship between air pollutant concentrations and traffic-related contributing factors over TAZs. These results can help urban planners to incorporate environmental considerations at the traffic and urban planning stage. The support vector machine model (SVM) is further developed for each air pollutant concentration. Although the 10-fold cross-validation results suggest that SVM provides better prediction accuracy than does the GWR, GWR can provide better inferences by capturing the varying relationship between input and outcome variables across TAZs. Highlights: This study analyzed effects of traffic and road characteristics on air pollutants at traffic analysis zone (TAZ) level. Three air pollutants were considered, including CO, NO2 and PM10 . Geographically weighted regressions link pollutants with traffic-related variables. GWRs considering spatial heterogeneity have better fitness than linear regressions. Traffic, road and employment characteristics affect air pollutant concentrations at TAZ level.