Real-time Identification of Crash-prone Traffic Conditions under Different Weather on Freeways

摘要

Understanding the relationships between traffic flow characteristics and crash risk under adverseweather conditions will help highway agencies develop proactive safety management strategiesto improve traffic safety in adverse weather conditions. The primary objective of this study is todevelop separate crash risk prediction models for different weather conditions. The crash dataand traffic data used in this study were collected on the I-880N freeway in California, UnitedStates in 2008 and 2010. This study considers three different weather conditions: clear weather,rainy weather and reduced visibility weather. The preliminary analysis showed that there wassome heterogeneity in the risk estimate for traffic flow characteristics by weather conditions, andthat the crash risk prediction model for all weather conditions cannot capture the impacts of thetraffic flow variables on crash risk under adverse weather conditions.The Bayesian logistic regressions were applied in this study to link the likelihood of crashoccurrence with various traffic flow characteristics under different weather conditions. Themodel estimation results showed that the traffic flow characteristics contributing to crash riskwere found to be different across different weather conditions. The speed difference betweenupstream and downstream station was found to be significant in each crash risk prediction model.And the large speed difference between upstream and downstream station in reduced visibilityweather has the largest impacts on crash risk, followed by that in rainy weather. The ROC curveswere further developed to evaluate the prediction performance of the crash risk prediction modelunder different weather conditions. It was found that the prediction performance of the crash riskmodel for clear weather was better than that of the crash risk model for adverse weathercondtions.