A vine copula-based analysis of spatial dependence of traffic conflict risk at highway ramp areas

Abstract

Understanding spatial dependence on real-time safety risk is crucial for developing effective traffic management and control strategies at highway ramp areas. Prevalent statistical models for spatial analysis of safety risk are limited in their capability to capture complex, non-linear, and asymmetric dependence patterns across multiple locations. In this study, a novel vine copula approach is proposed for the spatial analysis of traffic conflict risk at highway ramp area, allowing for more flexible specification and control of heterogeneous spatial dependence patterns. Specifically, spatial correlations in traffic conflict risk, measured by time-integrated time-to-collision (TIT) across road segments and traffic lanes, are estimated using high-resolution vehicle trajectory data collected from unmanned aerial units. The results reveal significant variations in spatial dependence patterns of traffic conflict risk, particularly asymmetric tail dependencies in high-risk scenarios. These findings underscore the need for advanced statistical models to accommodate spatial dependence in traffic safety analysis, thereby supporting the development and implementation of more effective real-time traffic safety management systems.