Methods for Defining Spatiotemporal Influence Areas and Secondary Incident Detection in Freeways

Abstract

The likelihood of the occurrence of secondary incidents is usually linked to primary incidents by using predefined spatial and temporal criteria. This paper extends past research on secondary incident detection by defining the dynamic thresholds of the influence area of a primary incident, using detailed freeway traffic data from upstream loop detectors. The dynamic thresholds are defined by using both analytical and empirical approaches. Results that use analytical approaches to track the spatiotemporal boundaries of the influence upstream of a primary incident offer influence curves of different characteristics (influence duration and maximum spatial extent) with respect to the prevailing traffic conditions. To accurately detect secondary incidents, an empirical method based on spatiotemporal speed evolution is applied that imprints influence areas with respect to their dissipation patterns. The results of the proposed approach are compared to those of five commonly used static and dynamic methods for detecting secondary incidents. Their differences are identified and discussed.