Improved Processes for Meeting the Data Requirements for Implementing the Highway Safety Manual (HSM) and SafetyAnalyst in Florida
Recent research in highway safety has focused on the more advanced and statistically proven techniques of highway safety improvement. The Highway Safety Manual (HSM), SafetyAnalyst, and the Interactive Highway Safety Design Model (IHSDM) are the three major safety analysis tools that have the potential to define a new era in highway safety. This research project focuses on the two most recent tools, the HSM and SafetyAnalyst.
The Highway Safety Manual
The HSM marks a radical shift in the approach of practitioners and administrators toward more robust measures of improving highway safety by moving toward statistically proven quantitative analyses. Almost two years after its release, states and local agencies are still struggling with its implementation. Meeting the data requirements is the most challenging task in the initial stages of the HSM implementation.
The HSM provides analytical tools for quantifying the effects of potential changes at individual sites on rural two-lane roads, multilane rural highways, and urban and suburban arterials; however, the data needs are significant. Very detailed roadway characteristics and crash information is required to conduct the empirical Bayes (EB) analysis recommended by the HSM and to derive the calibration factors that are required to accurately represent the agencyís safety performance. For example, to analyze intersections, data on area type, number of lanes, traffic volume, geographic coordinates, number of legs, traffic control type, intersection skew angle, intersection left-turn lanes, intersection right-turn lanes, intersection sight distance, terrain, lighting, right-turn-on-red, left-turn signal phasing, red-light cameras, bus stops, schools, alcohol sales establishments, pedestrian activity level, and maximum pedestrian lanes crossed are needed. Many of these variables are currently unavailable in Floridaís roadway characteristics inventory (RCI) database. With these data limitations, the two main questions to be answered are which variables to collect and how much data for each variable to collect?
First, collecting and maintaining all the required and recommended data variables on the entire road network for the purpose of the HSM implementation is both costly and unnecessary. Therefore, a process to streamline the data requirements without compromising the quality of analysis could result in major cost savings in meeting the data requirements. Given that not all of the variables are likely to have the same impact on safety predictions, it becomes beneficial to assess and rank the impact of each variable on safety predictions. The ranking will help prioritize the additional data to be collected such that the benefit is greatest.
Second, the manual recommends deriving calibration factors using randomly selected 30-50 roadway sites that experienced a minimum total of 100 crashes per year. Given the fact that the minimum sample size is a function of sample variance, this recommendation is clearly questionable as roadway characteristics of different roadway types are likely to have different levels of homogeneity. For example, the roadway characteristics on freeways are more likely to be more homogenous and would require a smaller sample size to achieve the same level of accuracy than local arterial streets, which are more likely to have more varying roadway characteristics. In other words, the recommended 30-50 sites could be too many for some roadway types yet insufficient for others. Similarly, the minimum total of 100 crashes per year is also questionable given the fact that the number of crashes vary widely across different roadway types. For example, intersection locations would generally experience many more crashes than mid-block segments. Similarly, arterial streets usually experience many more crashes than freeways. Research is needed to determine the appropriate minimum sample sizes for different roadway types. After an appropriate minimum sample size is determined, the next question is how the sample locations should be selected. Again, research is needed to develop a sampling process that would best represent Floridaís conditions.
SafetyAnalyst was developed as a cooperative effort by Federal Highway Administration (FHWA) and participating state and local agencies. Often advertised as a companion to Part B of the HSM, SafetyAnalyst automates all the steps in the roadway safety management process. As one of the 27 state agencies that have been participating in the development of SafetyAnalyst, Florida has been ahead of many states in deploying SafetyAnalyst. Specifically, FDOT has sponsored two projects in its effort to implement SafetyAnalyst. The first of these projects was conducted by the University of South Florida (USF) which developed a program to map and convert FDOTís roadway and crash data into the input data format required by SafetyAnalyst.
A second related project was completed recently by Florida International University (FIU). The project successfully developed Florida-based Safety Performance Functions (SPFs) for different types of segments, ramps, and signalized intersections. These SPFs were then applied to generate high crash locations in SafetyAnalyst. Additionally, the project also developed the first known GIS tool for SafetyAnalyst. However, the project was unable to develop SPFs, nor generate any SafetyAnalyst input files for unsignalized intersections due to the lack of the required data in FDOTís Roadway Inventory Characteristics (RCI). In addition, the SPFs and SafetyAnalyst input data files for signalized intersections could only be developed based on very limited data. This was, again, due to the unavailability of the needed data in the RCI. Given the importance of intersection safety, an extra effort is needed to collect the missing data so that analysis can be performed in SafetyAnalyst for intersections.
When attempting to apply the aforementioned USF program to generate SafetyAnalyst input files for new crash data, the FIU research team found that the program had a number of inaccurate data mapping. Additionally, parts of the program files were found to be available only in the compiled DLL format, making it impossible to make the needed corrections. However, what is more problematic with the program is that it has a number of software design and implementation issues that not only make it difficult to modify in the short run, but also more difficult to maintain in the long run. Considering the number of problems found in the program and the difficulties in correcting these problems, it is believed that a new program can be developed from scratch in less time than attempting to modify the existing program. The new program can also accommodate Floridaís new police report that was started in 2011.
The objectives of this project are to:
- Identify and prioritize influential HSM calibration variables for data collection.
- Develop guidelines on the appropriate sample size and suitable sample selection procedures for each roadway type in HSM.
- Develop a new conversion program for generating the required input files for SafetyAnalyst.
- Develop SPFs for intersections and generate input files for all SafetyAnalyst roadway types