Generation of road accident risk maps

Abstract

Knowing the factors that affect the likelihood of an accident occurrence has been increasingly challenging to the researchers given the huge social and financial costs that derive from road accidents. In Portugal, developments in this area have mainly involved interurban roads studies. However, according to ANSRi, about 70% of Portuguese road accidents occur in urban spaces, a trend common to most European countries. The lack of national or local information systems containing geo-referenced road accidents, geometric characteristics of roads, among others, hamper the creation of tools that help to assess the risk of exposure at a micro level, i.e. road intersections. The weaknesses mentioned above led us towards the implementation of models in a GIS-based environment in order to estimate the frequency of accidents for urban areas according to several breakdowns: road element, type of accident and the inclusion of explanatory variables related to road environment. One of the challenges faced by researchers when applying these models is the absence of data or its poor quality. Therefore, it is necessary to cross and analyse information from different sources, such as traffic variables (from model transportation planning), digital cartographic data, and other geometric variables, that may not be obtained in a direct way (e.g. using OpenStreetMap or Google Maps). In a further step, the estimation models will be programmed and applied according to the type of road element (e.g. intersections, roundabouts, segments). Finally, new information will be generated with all inputted data: a digital map with the number of accidents per road element. Such figures will need to be converted into something more meaningful for potential users, such as levels/categories (e.g. high, medium and low risk of exposure), which can be mapped. This paper proposes a methodology for automatic generation of road accident maps with those levels/categories. Thus, maps will serve as a decision support tool not only to insurers (who are likely to tax drivers more effectively, according to their exposure to risk), but also to drivers themselves (through generation of alarms that will allow them to tailor their driving performance), envisaging road safety improvement.