Defining Spatiotemporal Influence Areas in Freeways for Secondary Accident Detection

Current research on secondary crash identification links the likelihood of secondary crashes occurrence to primary incidents using pre-defined spatial and temporal criteria. This paper extends past research on secondary crash detection by defining dynamic thresholds of the influence area of a primary incident using detailed real-time collected freeway traffic data from upstream loop detectors. Results offer influence curves for different characteristics with respect to the prevailing traffic conditions before the occurrence of the primary incident. For accurately detecting secondary incidents, an empirical method based on speed is applied that imprints influence areas with respect to their dissipation pattern. The results of the proposed approach are compared to those of five prevailing static and dynamic methods for detecting secondary accidents and the differences are identified and discussed.