Detection of impact corridors influencing animal accidents along road networks

Animal vehicle collisions (AVC) or accidents involving non-huntable wildlife are a risk for humans and animal biodiversity which should not be underestimated due to the growing number of road kilometres in Europe. Considering huntable wildlife, more than 77,000 AVCs were counted in Austria in the season of 2015/16 (non-huntable wildlife is not included in this statistic). In 2016, 304 people were injured and one person even died as a result of the accident. Recently, several serious AVCs occurred in Austria, which also claimed human lives (ORF 14.11.2017, Standard 12.11.2017, ORF 30.10.2017). In the United States there are up to 1.6 million AVCs per year with 200 people killed and several thousand people injured. The resulting total costs of these accidents are estimated at US $ 4.6 billion (III, 2010). Not only does AnimalProtect target serious accidents for humans, it also addresses the environmental problem of road killed birds or amphibians. Worldwide only very few systematic studies on a national level investigating road-killed animals exists. Since monitoring road-killed animals on this level bears several challenges including large geographic areas and low persistence rates of carcasses resulting in time and personnel intensive monitoring approaches. However, nationwide investigations are necessary in order to be able to predict the danger to humans and animals on a large scale. The aim of the proposed project is to derive hazard zones for AVCs in the vicinity of roads using remote sensing data, expert knowledge and results from previous studies. Therefore, possible impact factors (land cover, terrain, phenological influences, environmental factors, socio-ecological factors, etc.) must be defined based on the behaviour of wild animals in order to subsequently transfer these information to potential data stocks like Copernicus data or Open Street Map (OSM) data. From this knowledge-based approach, risk areas for motorists and various animal species are to be derived. The validation of detected impact areas should be done with present vehicle accident data. In order to do so, interfaces to existing databases of wildlife accidents must be created and the data homogenized. The hazard zones derived from remote sensing and geoinformation will be offered as a cloud based service that returns the risk of an AVC in a particular area (e.g. on a scale from one (lowest accident risk) to five (major accident risk)). Different users (car drivers, insurance companies, navigation services, public authorities, etc.) should draw corresponding conclusions (e.g. when building a new road). The use of the service should also be shown in the context of a software demonstrator, which indicates the corresponding hazard depending on the Global Navigation Satellite Systems (GNSS) based position of the user and allows routing over low risk roads.