Accident risk of road and weather conditions on different road types

Abstract This study was designed to investigate the relative accident risk of different road weather conditions and combinations of conditions. The study applied a recently developed method which is based on the notion of Palm probability, originating in the theory of random point processes, which in this case corresponds to picking a random vehicle from the traffic. The method consists of calculating the Palm distribution of different conditions and comparing it with the distribution of the same conditions as seen by the accidents. The condition affects the accident risk statistically, when these two distributions differ. The study included all police reported single- and multi-vehicle accidents (N = 10,646) occurring on 43 main roads in Finland during the years 2014–2016. A major contribution of this paper is the demonstration of the method on national scale by using estimated hourly traffic volumes on road segments instead of measured ones, which would have been available for few roads only. Accident risks are commonly examined in relation to traffic volume. This paper includes the speed of the traffic and thus, the paper examines accident risk in relation to the time spent on the road segment in certain conditions. The hour-level weather and road condition data per segment were obtained from nearby road weather stations. The relative accident risks were increased for poor road weather conditions; however, they were highest for icy rain and slippery and very slippery road conditions. When comparing the relative accident risk based on road type, the results showed that the risk in poor weather and road conditions was higher on motorways compared to two-lane and multiple-lane roads even though the overall risk was lower on motorways. Furthermore, the corresponding relative accident risks were generally higher for single-vehicle accidents compared to multi-vehicle accidents.