Analysis of spatial distributions of road accidents
Thesis DisciplineTransportation Engineering
Degree GrantorUniversity of Canterbury
Degree NameDoctor of Philosophy
Traffic accidents result in life and financial loss to the society. In developing countries traffic fatalities are comparable to other leading causes of death. The need for the analysis of the spatial distribution of traffic accidents, as an aid to select the most appropriate type of accident reduction programme (e.g. site, route and area plans) and assessing the effectiveness of such plans after implementation, is very important. The current practice (e.g. visual examination) for assessing the spatial distribution of accidents is reviewed. In this thesis, various methods for the statistical analysis of spatial distributions of accidents (including quadrat and nearest - neighbour methods) are reviewed and further improvements are described. Accidents are random events subject to both temporal and spatial variation. The basic variables for accident analysis are; distance and direction of accident locations in terms of North and East co-ordinates, azimuth, and the year of the accident. A new method for analysing the spatial pattern is proposed, whereby detection of a particular pattern will indicate which type of accident reduction programme is most appropriate. The method distinguishes the spatial distribution (point cluster, line cluster, area cluster or a completely spatially random distribution) of accidents in different types of road networks (regular or irregular and dense or sparse). The method can also help assessment of the changes in spatial distributions of accidents.