Spatial Analysis and Modeling of Traffic Incidents for Proactive Incident Management and Strategic Planning

Traffic events involving secondary incidents can be particularly problematic for the public and for incident managers. This paper explores the associations of spatial characteristics, including geometric and land use factors, with secondary and nonsecondary incidents. The data used in this study are 2006 incident records from Hampton Roads in Virginia and roadway inventory data, enhanced through geographic information systems to include detailed spatial information. Secondary incidents in the same and opposite directions were identified by using a queue-based method. Such incidents represented nearly 2% of total recorded incidents but showed longer durations than other incidents. The study found statistically significant differences between the distributions of secondary and nonsecondary incidents, implying that higher risks of secondary incidents in certain roadway segments are not necessarily correlated with relatively high risk of nonsecondary incidents. Poisson, zero-inflated Poisson, and negative binomial regression models were estimated by combining traffic exposure, road segment characteristics, and spatial–land use information to explore factors associated with secondary incidents. The models provided helpful information for effective assignment of incident management resources and for support of regionally based strategic planning.