Contributing factors to vehicle to vehicle crash frequency and severity under rainfall

• Using several available data sources, microscopic geographic and secondary weather data, including grade, water film depth, and car-following distance, which are known to be significant to highway safety but frequently deficient in datasets, were estimated.
• Weather and the vertical curve were found to affect the injury severities of vehicle to vehicle crashes more than any other type of crash in rainy weather.
• The interaction predictors between the indicator of the posted speed limit change and the existence of an off-ramp or between the indicator of travel lane number change and the indicator of pavement surface material change were found to significantly increase the likelihood of vehicle to vehicle crashes in rainy weather.
• The number of travel lanes, a water film depth equal to or greater than 0.08 mm/h, the type of rear-end collision, and the interaction between the travel lane number and a grade equal to or greater than 0% and less than 0.1% were found to increase the likelihood of fatal, incapacitating, and non-incapacitating injuries of vehicle to vehicle crashes in rainy weather.

Introduction: This study combined vehicle to vehicle crash frequency and severity estimations to examine factor impacts on Wisconsin highway safety in rainy weather. Method: Because of data deficiency, the real-time water film depth, the car-following distance, and the vertical curve grade were estimated with available data sources and a GIS analysis to capture rainy weather conditions at the crash location and time. Using a negative binomial regression for crash frequency estimation, the average annual daily traffic per lane, the interaction between the posted speed limit change and the existence of an off-ramp, and the interaction between the travel lane number change and the pavement surface material change were found to increase the likelihood of vehicle to vehicle crashes under rainfall. Results: However, more average daily rainfall per month and a wider left shoulder were identified as factors that decrease the likelihood of vehicle to vehicle crashes. In the crash severity estimation using the multinomial logit model that outperformed the ordered logit model, the travel lane number, the interaction between the travel lane number and the slow grade, the deep water film, and the rear-end collision type were more likely to increase the likelihood of injury crashes under rainfall compared with crashes involving only property damage. Practical implications: As an exploratory data analysis, this study provides insight into potential strategies for rainy weather highway safety improvement, specifically, the following weather-sensitive strategies: road design and ITS implementation for drivers' safety awareness under rainfall.