M. An, X. Yan, C. Casper, W. Wu

Pages: 47-58

Abstract
In order to effectively evaluate the potential benefits and costs of transportation investments, transportation planners need to forecast future vehicle crashes. This study describes an approach to develop and apply a planning-level crash prediction model at the traffic analysis zone (TAZ) level for safety benefit estimation associated with planning-level transportation projects. First, different categories of factors from the regional transportation planning model were identified by using geographic information system (GIS) functions. These factors are used to develop a planning-level crash estimation model based on negative binomial regression. Following this, the safety benefits for two transportation planning projects were estimated by applying this model in combination with the transportation planning model from a region-wide perspective. The regression results show that the final zonal crash prediction model (ZCPM) contains vehicle hours traveled as the exposure variable and two predicting variables, the number of low-income households and the number of signalized intersections. The ZCPM shows a reasonable goodness of fit when comparing observed and estimated crashes, and safety benefit estimates for two planning projects in this study appear in general consistent with intuitive expectations. This study indicates that the modeling framework seems a useful approach in estimating future crashes and project safety benefits with certain assumptions, which may fill some gaps in current planning-level project evaluation by explicitly considering future safety.

Keywords: transportation planning; zonal crash prediction; safety assessment