Modeling road traffic safety based on point patterns of wildlife-vehicle collisions

Wildlife-vehicle collisions represent one of the main coexistence problems that appear between human populations and the environment. In general terms, this affects road safety, wildlife management, and the building of road infrastructures. These accidents are a great danger to the life and safety of car drivers, cause property damage to vehicles, and affect wildlife populations. In this work, we develop a new approach based on algorithms used to obtain minimum paths between vertices in weighted networks to get the optimal (safest) route between two points (departure and destination points) in a road structure based on wildlife-vehicle collision point patterns together with other road variables such as traffic volume (traffic flow information), road speed limits, and vegetation density around roads. For this purpose, we have adapted the road structure into a mathematical linear network as described in the field of Graph Theory and added weights to each linear segment based on the intensity of accidents. Then, the resulting network structure allows us to consider some graph theory methodologies to manipulate and apply different calculations to analyze the network. This new approach has been illustrated with a real data set involving the locations of 491 wildlife-vehicle collisions in a square region (40 km × 40 km) around the city of Lleida, during the period 2010–2014, in the region of Catalonia, North-East of Spain. Our results show the usefulness of our new approach to model road traffic safety based on point patterns of wildlife-vehicle collisions. As such, optimal path selection on linear networks based on wildlife-vehicle collisions can be considered to find the safest path between two pairs of points, avoiding more dangerous routes and even routes containing hotspots of accidents. [Display omitted] •To reduce the impacts of wildlife-vehicle collisions a modeling framework is defined.•Weights are based on the intensity of wildlife-vehicle collisions (point pattern).•The optimal road path between two points ndeparture and destination) is based on a safety criterion.•The model is tested on a real data set containing 491 wildlife-vehicle collisions.