Exploring the relationship between interstate crashes and various speed metrics based upon probe vehicle data
Christopher M. Day
In recent years, a considerable amount of research has been aimed at discerning the determinants of traffic crashes and those circumstances under which crash risk is increased. This is particularly important when considering the potential safety impacts of geometric design or transportation policy decisions. For example, research has consistently demonstrated that crashes are affected by various factors related to the roadway, the surrounding environment, and the involved drivers. This study has two principal goals. The first is to understand how driver speed selection varies with respect to traffic and roadway geometric characteristics. The second goal is to explore the relationship between traffic crashes, operating speeds (i.e., mean speed, 85th percentile speed, speed variance) and other pertinent factors (e.g., traffic, roadway, weather). To achieve these goals, the study utilized traffic, roadway, weather, and speed data (i.e., automatic traffic recorder, INRIX) for Iowa interstates. Simple descriptive statistics are documented to illustrate crash trends on the Iowa interstate network during the study period. A series of regression models were estimated to investigate relationships between various speed metrics and crash rates with respect to traffic, roadway, and weather characteristics. The study suggests that speed measures, such as mean speed, 85th percentile speed, and speed variance, are found to correlate with the roadway geometry. In addition, higher speed variance is associated with more crashes, while the absolute speed of traffic does not necessarily correspond to higher crash occurrences.