We present a comparison of animal-vehicle collision patterns which we have computed for selected roads in several countries (e. g. Poland, Finland, France, Hungary, Italy, Slovakia, Spain, Sweden and the USA). We used the KDE+ method for objective identification of dangerous locations (hotspots, clusters).
Application of the KDE+ method returns AVC divided into two groups: within a cluster and outside the cluster. Collisions within the clusters were predominantly caused by local factors directly connected to the location of the cluster (the parameters of the road and its closest vicinity). Collisions outside of the clusters were more likely induced by global factors (e. g. the time of day, regional weather).
The data from national or local AVC databases differed by a system of localization (GPS, linear stationing, other) and road network data. Various GIS approaches had to be, therefore, used to prepare the input files. The outcomes of these pilot studies with regard to the similarities and differences of the AVC patterns will be discussed.
road kill; GIS; traffic accidents; animal-vehicle collisions; kernel density estimation; hotspots; KDE+