Citation

Zeng Q, Wang X, Wen H, Yuan Q. J. Transp. Saf. Secur. 2022; 14(2): 323-337.

Copyright

(Copyright © 2022, Southeastern Transportation Center, and Beijing Jiaotong University, Publisher Informa - Taylor and Francis Group)

DOI

10.1080/19439962.2020.1779422

PMID

unavailable

Abstract

Given the potential difference in drivers' route familiarity, this study aims to investigate the respective factors contributing to local-vehicle and non-local-vehicle crashes on freeway. Crash data from Kaiyang Freeway in Guangdong Province, China in 2014 are collected for the investigation, where the crashes with all involved vehicles registered in the province are labeled as local-vehicle crashes while others are labeled as non-local-vehicle crashes. Bayesian spatial models are developed to build local-vehicle and non-local-vehicle crash frequencies' relationships with traffic characteristics, roadway design, and weather condition. The parameter estimation results indicate that the factors including daily vehicle kilometers traveled, vertical grade, air temperature, and precipitation, impact local-vehicle and non-local-vehicle crash frequencies in the same directions. The proportion of vehicles in Class 1 (e.g., passenger car) impacts the two type crash frequencies in different directions. The proportion of vehicles in Class 3 (e.g., medium-sized bus and truck) and horizontal curvature have significant effects on local-vehicle crashes only. The results are indicative of developing countermeasures from the perspectives of roadway geometry design and driving assistance apps, which aims to reduce freeway crash occurrence.

Language: en

Keywords

Bayesian spatial model; conditional autoregressive prior; Freeway safety; local-vehicle crash; non-local-vehicle crash