Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
35 In this step, the goal is to identify candidate sites across the entire roadway network with the greatest potential for future crashes that an agency would want to address. These should bear relation to the crash types or locations of interest (from Step 1) and have key risk factors present (identified in Step 3). As previously indicated, this step relies heavily on data about the roadway and related charac- teristics that can be used in screening for the presence of key risk factors. Agencies with incomplete data may find the need to gather additional information for particular sites, potentially through aerial imagery or field visits. Consider Eliminating Low Crash Potential Sites An initial task may be to determine whether any locations within the focus network should be eliminated from consideration, particularly if the sites have low expected or predicted crashes, where feasible countermeasures do not exist, or where there are pending or planned projects. It may require more in-depth knowledge of the network or looking at sites more closely to make this determination, which can also be done during Step 6. C H A P T E R 5 Step 4: Identify Potential Treatment Sites Noteworthy Practice Arizona DOT staff screened their network for high risk locations based on risk factors identified from prior research by an expert team. They did not have pedestrian volume estimates. So to help prioritize which sites with risk factors may benefit most from treatment, they conducted additional reviews using street and aerial view online mapping tools to examine characteristics of the site that would help determine pedestrian activity or exposure. Arizona DOT subsequently eliminated from further consideration some high apparent risk locations (but with zero prior crashes) that were determined to not be in areas where pedestrians would likely walk. See Case Example 3 for more information. Generate Initial List of Sites Site Identification Based on Risk Factor Presence One approach, which can be used regardless of how risk factors were identified in Step 3, is to filter the list of sites to identify those with similar risk factors that can be further evaluated for
36 Systemic Pedestrian Safety Analysis treatment options. Filtering and sorting tools are easy to use if the data are in a spreadsheet-type database (similar to the one shown in Table 2) or in a spatial format that allows querying. Chapter 4 identified a list of treatable roadway factors associated with pedestrian crashes at segments (i.e., midblock locations). The risk factors shown in Table 10 are used to narrow a list of potential treatment sites from 196 (where there was a midblock crossing) to 12 sites that had multiple treatable risk factors present, including four or more through lanes and on-street park- ing. Observed prior crashes and SPF-predicted crashes from the model are summed for each subset of sites and shown in the table, along with the relative SPF-predicted rankings of the sites in each group. These data will come in handy for economic analyses and ultimate site prioritization. This is also an opportunity to consider relevant treatments and whether a treatment or a set of treatments related to the risk factors has been identified. Then the data can be further filtered to identify sites that have the appropriate context (such as number of lanes or traffic volumes) for countermeasure application. Consider a second example in Table 11, in which an agency has identified that streets having four or more through lanes had a higher risk of crashes. The agency wants to take a systemic approach to identify high crash potential road segments suitable for road diets, which might address the risk to pedestrians in crossing multiple lanes, from their network of 23,000 road segments. An initial list of potential sites, filtered only by the number of through lanes, still identifies 1,405 potential sites, far too many to treat, and likely including many with inappropriate vol- umes for the treatment of interest. The agency has a 25,000 ADT treatment threshold for con- sidering road diets, which further filters the potential sites. The presence of two-way center-turn lanes was an additional risk factor identified, and when this filter is applied to the previous two filters, it reduces the candidate site list down to a more feasible 131 segments, which is about 13 miles of roadway. Of course, the adjacency of road segments along a corridor would also be a Risk Factor Number of Sites Total Prior Observed Total SPF- Predicted Range of SPF Rankings (of ~23,000 segments) 1. Presence of midblock crosswalk (1 or more) 196 50 50.1 4 â20,870 2. Plus, 4 or 5+ through lanes 26 24 19.2 4 â2,228 3. Plus, on-street parking 12 10 14.6 9 â2,228 Crashes (8 years) Crashes (8 years) Table 10. Example 1: Identification of potential sites using risk factors. Risk Factor Number of Sites Range of SPF Rankings (of ~23,000 segments) 1. Presence of 4 or 5+ through lanes 1,405 1â9,409 2. Plus, < 25,000 ADT 939 1â9,409 3. Plus, TWLTL present 131 Total Prior Observed Crashes (8 years) 302 171 27 Total SPF- Predicted Crashes (8 years) 296.8 175.2 27.5 7â3,866 Table 11. Example 2: Identification of potential sites using risk factors and countermeasure context.
Step 4: Identify Potential Treatment Sites 37 consideration. In this example and in the previous example, the sites can be further sorted and ranked or prioritized using other data available, whether prior or predicted crashes or other considerations, to refine a treatment plan. Looking Ahead Step 5 describes the process of identifying potentially appropriate counterÂ measures and provides an initial list of countermeasures that may be suitable for systemic application. It is relevant here, as sites identified for systemic treatment must have some relationship to countermeasures that can be used systemically. Refer to Step 5 while performing this screening step, particularly Tables 15 through 17. In Step 6, there will be additional work to prioritize locations and treatments, considering community priorities as well as costÂeffectiveness data. Site Identification Based on Estimated Crash Rankings Another way to identify potential sites suitable for agencies that have taken the approach in Step 3 to determine risk factors using a model-based approach is to make use of SPF or empirical Bayes predictions from the model to identify highest predicted crash sites first. Then perform the filtering techniques described in the previous section. This method may be well suited for an agency wanting to focus on the sites with the highest potential for crashes regardless of the specific treatable risk factors identified. Table 12 provides an illustration of this approach. Combinations of risk factors can be pre- sented and considered in terms of the numbers of predicted crashes or prior observed crashes and relevant sites for locations that are first ranked by overall crash predictions. In this example, the top 200 locations are identified. Identifying sites for systemic treatment requires careful consideration of risk factors, over- all crash potential, and the existence of relevant countermeasures for treatment. Steps 5 and 6 Risk Factor or Combinations Among Top 200 Ranked by SPF Prediction Total Predicted Crashes for all Sites Total Prior Crashes Number of for all Sites Relevant Sites 4 or 5+ through lanes 106 93 146 Presence of on-street, striped parking 72 50 85 TWLTL present 35 39 56 Combination of 4 to 5+ through lanes and TWLTL 32 31 51 Right-turn lane present at adjacent intersection 31 32 40 Combination of 4 to 5+ through lanes and right- turn lane present at adjacent intersection 18 26 25 Table 12. Example 3: Identification of potential sites using predicted crashes and risk factors.
38 Systemic Pedestrian Safety Analysis provide additional opportunities for revision and refinement of sites for potential treatment based on countermeasure availability and prioritization considerations. There are many prag- matic considerations as well, such as just how easily different countermeasures may in fact be implemented, what staff and other departments or agencies are involved, and others. Looking Ahead In Step 6, the list of sites and treatments will be further refined based on costÂeffectiveness and other considerations. As agencies develop their list of sites, they will want to include associated crash prediction estimates (if developed), prior crash history, and risk variables and any other site characteristics that can aid in future decision making, regardless of whether these factors were identified as risk factors or used in initial screening. Additional Resources The technical report has additional background relevant to this step, especially in Chapter 4. In the guidebook see the additional resources listed under Steps 3 and 5 and the case examples for different screening and ranking approaches.