Comparing Locally-Calibrated and SafetyAnalyst-Default Safety Performance Functions for Florida’s Urban Freeways

摘要

Freeways carry a large proportion of vehicle miles in the national roadway system. Therefore,assessing their safety performance is of special importance. Although freeway interchanges havesignificantly different crash and traffic flow characteristics compared to basic freeway segments,they are seldom assessed separately in highway safety analysis. This can be attributed to minimalguidance on the definition of interchange areas and lack of network databases that could allowfor easy separation of such areas. Although interchange influence areas are defined in theSafetyAnalyst User Manual, specific method for separating a freeway network into interchangeareas and basic freeway segments is not provided. This paper proposes a method using GIS andspatial manipulation techniques to identify and separate segments within interchange influencearea from the freeway network. Once separated, safety could be evaluated on basic freewaysegments and freeway segments within interchange influence areas individually. With the recentreleases of advanced safety analysis tools including SafetyAnalyst and the Highway SafetyManual (HSM), states are trying to assess the need for developing agency specific safetyperformance functions (SPFs). Even though SafetyAnalyst provides national default SPFscalibrated to local agency data, it is believed that the use of state specific SPFs might betteridentify problematic sites. Using four years of crash data from Florida, state specific SPFs forboth basic freeway segments and freeway segments within interchange influence areas weredeveloped using the negative binomial (NB) model. These state specific SPFs were compared tothe default national SPFs calibrated to Florida data using Freeman-Tukey R square (R~2_(FT))statistic. The results show that Florida specific SPFs produce a better-fitted model than thedefault calibrated SPFs.