The measurement and modelling of proximal safety measures

Safety in the traffic system is often measured by the number and severity of traffic crashes. This long-standing approach has established the use of crash data as an accepted measure of the lack of safety. There is, however, a distinct need for faster, more informative, and more resource-effective methods that yield valid and reliable safety measures and a foundation for predictive modelling. This paper is primarily concerned with traffic safety assessment and prediction based on proximal safety indicators and associated measurement techniques. It reviews the concepts and methods related to three different proximal safety indicators: time-to-accident, time-to-collision and post-encroachment-time. It shows how they can be measured using field data. Similar measures can also be obtained from simulation modelling. The use of simulation models for measuring and predicting intersection safety offers considerable potential for proactive safety analysis.