Accident prediction models for high speed intersections (both rural and urban)

Accident occurrence is typically low at rural priority controlled intersections, due to low traffic volumes, compared with priority controlled urban intersections. Unlike many urban intersections, the low accident occurrence makes it difficult, from the accident history alone, to identify accident trends and justify improvement projects. This research project has produced accident prediction models for rural priority-controlled intersections; based on traffic volume, sight distance, approach speed and geometric design. The models can be used to identify the effect of safety deficiencies at an intersection (for example, effect of poor visibility), can be used in economic evaluation of intersection upgrades and can be used to assess the impacts of traffic from new land-use developments on the rural road network. The study also produced flow-only models for high-speed signalised and roundabout controlled intersections in New Zealand and high-speed signalised intersections in Victoria (Melbourne). This enables an analyst to assess the likely changes in accident occurrence and of various accident types of a change in control (from priority to either a roundabout or signals) at higher volume intersections. This paper also outlines some of the more important statistical methods that are used to assess the quality of the models produced.