Speed profile variation as a surrogate measure of road safety based on GPS-equipped vehicle data

Road network screening for potentially high incident locations is the first step in a road safety improvement program. During the screening process, road network crash data are required for the identification of high crash locations, a.k.a., black spots. In situations where historical crash data are limited or not available, surrogate safety measures, such as traffic and roadway characteristics are often considered. A surrogate safety measure is an indirect measure of safety, which attempts to assess the safety of a road facility through means other than crash data. Among speed characteristics measurements speed variation is often used as a surrogate measure of safety. There are a number of studies that attempt to establish a relationship between speed variation and crash risk but the existence form of such a relationship is still hotly debated in the literature. The increasing use of Global Positioning System (GPS) devices for collecting traffic operations data, such as vehicle speed and travel time, was led to interest in using GPS data derived measures as potential indices for roadway safety. As the deployment of GPS-instrumented vehicles becomes more prevalent, we may be able to use this new data streams to better evaluate roadway safety. Our hypothesis is that vehicle speed characteristics may be used to reveal roadways with safety issues such as poorly-designed road geometries, limited sight distance, and high conflict movements from/to side streets.;The primary objective of this research is to explore the use of speed variation over a roadway segment as an indirect means to estimate crash frequency of the facility. This estimated crash frequency can be used as a substitute when historical crash data are unavailable or a proactive means to identify sites that need further engineering studies.;To accomplish this objective, sample operating speed and incident data were collected for corridors in the Metro Atlanta area. To measure operating speeds, second-by-second speed data were obtained from more than 460 GPS-equipped vehicles participating in the Commute Atlanta Study over the 2004 calendar year. Incident data was provided by the Georgia Department of Transportation Office of Traffic Safety and Design. Based on the speed and incident data, several definitions of speed variation are considered as potential surrogate safety measures. The quantified relationships between surrogate measures and crash frequency are developed using Binary Recursive Partitioning methods and a Generalized Linear Modeling (GLM) approach.;This research effort is expected to result in several contributions. First, this study will develop a methodology to determine speed profile under various conditions using vehicle activity data. Second, a speed variation definition suitable for GPS data that can be used as a surrogate safety measure will be recommended. Lastly, the process will provide safety prediction models for identifying high crash locations in the network screening process for urban streets.