Integrating Real-Time Traffic Data in Road Safety Analysis

Abstract

Traffic data aggregation has been a serious factor of inaccuracy in most road safety studies. The Average Annual Daily Traffic (AADT) has been the most commonly used measure to reflect traffic conditions. In this paper, we establish a framework for the integration of real-time traffic data in road safety analysis. To this end, we explore the effects of traffic parameters on type of road crash and on the injury level sustained by vehicle occupants. Univariate and ordered Probit models are specified on 4-years of data from the A4-A86 highway section in the Ile-de-France region, France. Empirical results indicate that multi-vehicle crashes tend to occur under low or very high traffic speeds, while single-vehicle crashes appeared to be largely geometry-dependent. Increasing traffic volume was found to have a consistently positive (i.e. decreasing) effect on injury severity, while speed appears to have a differential effect on severity depending on flow conditions. Also, while in higher traffic volumes higher traffic speeds aggravate severity outcomes, in lower traffic volumes speed does not significantly influence severity in a consistent pattern.