Traffic Safety Analysis
Safety analysis typically begins with the examination of historical crash data, usually spanning a period of 3 to 5 years prior to the analysis.
GIS tools provide a suitable platform for analyzing this data, as crashes can be allocated to network objects (road segments or intersections), facilitating more streamlined analysis at each specific point.
The initial step involves determining and prioritizing black spots (hot spots). Various crash criteria are considered in this regard, depending on the objectives. These include frequency, frequency of specific severity, frequency weighted by severity (or crash cost), rate, and density. Crash frequency is defined as the number of crashes per year. Crash rate indicates the potential for crashes given the amount of travel, while crash density represents the potential for crashes over a specific road segment or area. The relationships between these parameters are outlined below:
Crash rate = Crash count / 10^8 vehicle miles traveled (for road segments)
Crash rate = Crash count / 10^6 vehicles entered (for intersections)
Crash density = Crash count / Segment length in miles (for road segments)
Crash density = Crash count / Area in square miles (for areas)
Different methods are available in GIS tools for identifying black spots at intersections (or generally a specific area) and over a segment of the road.
Once the prioritized black spots have been identified, the attributes of the crashes are analyzed to uncover common factors contributing to the crashes at that location. Reviewing collision diagrams and reports, along with conducting field visits, helps to verify these findings.
Based on the identified causal factors of the crashes at a given spot, mitigation options are proposed to reduce the likelihood of future crashes. These options are then compared based on their crash reduction potential and associated implementation or construction costs.
After a selected mitigation is implemented, a before-after study is required to determine the effectiveness of the solution.
Crash prediction models are developed and calibrated based on historical crash databases and transportation data. These models are used to forecast potential crashes and to estimate the crash reduction potential of various mitigation options. Such crash prediction models are extensively discussed in the Highway Safety Manual (HSM) and are implemented in software. Local methods can also be developed and calibrated to gain more insight.
