Identification and spatiotemporal evolution analysis of high-risk crash spots in urban roads at the microzone-level: Using the space-time cube method

Abstract

Abstract The problem of urban crashes brings huge challenges and threats to local police and governments, especially in many cities in developing countries such as China. To reduce the frequency and severity of urban crashes, the local government in China has gradually taken interest in conducting detailed actions of traffic safety improvement at the microzone-level. Therefore, the primary goal of this study is to try a new method in spatiotemporal data mining techniques, the space-time cube method, to find high-risk crash spots at the spatiotemporal level and to obtain their spatiotemporal evolution patterns. The cumulative frequency curve method was performed to identify high-risk crash spots, and the contributory factors of forming these spots were analyzed by the latent class analysis method. The results showed that: (1) key parameters’ selection is crucial in the space-time cube construction; (2) the exit ramp gore point in interchanges, intersections, and entrances of neighborhoods were prone to have many high-risk crash spots at the spatiotemporal scale; and (3) locations with consecutive, persistent, and sporadic hotspots patterns need different risk monitoring strategies and traffic safety improvement. The feasibility and advantages of the space-time cube method in hotspots identification at the microzone-level were confirmed.