Modeling and spatial analysis of heavy-duty truck CO2 using travel activities

Abstract

Heavy-duty trucks (HDTs) are vital components of the freight industry yet have faced criticism for their substantial CO₂ emissions. This study, focusing on Xi'an, a crucial freight hub city in China, aims to investigate the factors influencing CO₂ emission from HDTs. A unique aspect of this study is using a Latent Dirichlet Allocation (LDA) model to evaluate the potential impact of different travel activities on CO₂ emissions using travel activities of HDTs extracted from extensive GPS data. Subsequently, the Random Forest (RF) model with a GeoShapley explainer was used to examine both the main and spatial effects of travel activities, road density, land use, and freight hub accessibility on CO₂ emissions. The results revealed the existence of fifteen distinct travel activities among HDTs in Xi'an, eight of which clearly influence CO₂ emissions. Considerable variations were observed in the magnitudes of the impact of different variables on CO₂ emissions, as indicated by GeoShapley values. The density of expressways and main roads has the greatest impact on CO₂ emissions, while various types of travel activities also significantly affect CO₂ emissions, with the impact of different travel activities varying to some extent. Additionally, there is evident spatial heterogeneity in the impact of various variables on CO₂ emissions, with larger positive GeoShapley values tending to concentrate around the 3rd Ring and expressways in Xi'an City. These findings, shedding light on the complex interplay of factors influencing CO₂ emissions from HDTs, provide valuable insights for formulating environmentally sustainable management policies concerning HDTs from spatial perspectives.