Estimating Incident Detection Time based on Incident Management and Traffic Detector Data

In recent years, traffic management agencies have started maintaining detailed and accurate archives of their incident management operations. The availability of this data has allowed the application of advanced data analysis techniques to identify several critical parameters of incident management operations. However, incident detection time, one of the critical parameters, cannot be identified only based on incident management data, since the first timestamp that the incident appears in the incident management database is the notification time, namely, the time that the TMC operator becomes aware of the incident few minutes after the incident actually occurs. Archived traffic detector measurements provide an additional source of data that can be used in estimating incident detection time and the associated influencing factors. This paper presents a methodology to determine incident detection time based on a combination of detailed traffic detector and incident management databases. The results from applying the developed methodology to a case study show that there is a significant difference between the mean of the incident detection times in daylight conditions and dark conditions, with higher detection time during dark conditions. The paper also demonstrates that combining incident management timestamps with traffic detector data can produce additional valuable information to support traffic management center operations.