Crash relative risk estimates (CRREs) during rain, snow, sleet (i.e., ice pellets), and freezing rain are estimated using a matched-pair analysis in Kansas for the years 1995-2014. Variable-length event periods are defined using both crash reports and nearby Automated Surface Observing System (ASOS) and Automated Weather Observing System (AWOS) observations. Newly developed methods to extract precipitation-type beginning and ending times within ASOS/AWOS observations provide the most accurate account of precipitation type to compute CRREs. Crash relative risk is enhanced during any precipitation type, while a hierarchy of risk based on precipitation type is also evident. For property damage only crashes, CRREs during freezing rain are significantly higher than snow, which in turn are significantly higher than rain. CRREs during sleet, though statistically significantly higher than rain, are not statistically significantly different from snow or freezing rain. For casualty (injury and/or fatality) crashes, CRREs during snow and freezing rain are significantly higher than rain, but the increase in risk during sleet is not statistically significantly higher than rain. These results are consistent with expectations, but are quantified here for the first time. These hierarchies are insensitive to the use of all crashes versus only those with crash-reported precipitation or precipitation-induced roadway surface conditions to compute CRREs. The risk of a single-vehicle crash is higher than a multiple-vehicle crash during snow, sleet, and freezing rain, whereas the opposite was found for rain. This hierarchy is most prominent during weekdays before noon.
Keywords: Crash relative risk; Matched-pair; Precipitation; Winter precipitation.
Copyright © 2020 Elsevier Ltd. All rights reserved.