U.S. Department of Transportation
Federal Highway Administration
1200 New Jersey Avenue, SE
Washington, DC 20590

Skip to content
Facebook iconYouTube iconTwitter iconFlickr iconLinkedInInstagram

Federal Highway Administration Research and Technology
Coordinating, Developing, and Delivering Highway Transportation Innovations

This report is an archived publication and may contain dated technical, contact, and link information
Back to Publication List        
Publication Number:  FHWA-HRT-17-107    Date:  March 2018
Publication Number: FHWA-HRT-17-107
Date: March 2018


Identification of High Pedestrian Crash Locations



Researchers contacted representatives from a variety of State, regional, and local agencies to gain insights into how they identify high pedestrian crash locations. The goal was to determine at least the following:


Criteria used to select agencies included the following:

Preference was also given to talking to those agencies that did not participate in the NCHRP 17‑73 survey to avoid contacting the same individuals with similar questions.(14) Table 11 lists the questions used with the government representatives.

Table 11 . Discussion questions for government representatives.



Potential Answers/Prompts/Comments


What are the focus areas within your safety programs?

Lane departure, pedestrians, etc. If pedestrians are not one of their focus areas, thank them for their time.


What is your current method for identifying high pedestrian crash locations?

Reactive/hot spots (based on crashes) or proactive/systemic (based on something else)


How do you identify the locations of concern (where you will develop potential countermeasures)?

Potential criteria: count of pedestrian crashes, rate of pedestrian crashes, severity, etc.


What tools are you using to assist in the identification of high pedestrian crash locations?

Spreadsheets, Safety Analyst, usRAP, ArcGIS®, open-source tools, OpenStreetMapTM.


Do you consider exposure? If so, how?

ADT, vehicle miles traveled, pedestrian counts, etc.


Have you used or considered surrogates?



Do you cluster your pedestrian crashes to identify hot zones?

Rely on a software app.


If yes, what criteria are you using?

Distance between the crashes, severity level for the crashes, density, age of the pedestrian, weights (by distance to school), equity (avoid lower income being overburdened)


What skill sets are needed to conduct these studies? Do you have those skill sets in-house, or do you contract?

If you contract, who usually does the work?


How frequently do you conduct the analysis? How frequently should it be conducted?

Looking to compare what they do and what they think they should do


Can you provide examples of what needs to be done to go from your crash records to a final list of sites needing treatments?

Protocol for assigning crashes (latitude/longitude or street names)


What are your lessons learned or takeaway you would offer from your experience?

Looking for general comments


Do you have any other comments?

This question provides the opportunity to cover items not previously discussed.

The following sections summarize the findings from the interviews.

What Are Focus Areas Within Your Safety Programs?

All the agencies participating in the interviews have a pedestrian focus within their safety efforts. Portland, Austin, and Los Angeles are implementing a Vision Zero policy with very aggressive goals for reducing crashes. Another community with a Vision Zero policy is San Francisco, CA, which maintains a pedestrian–vehicle injuries and high injury corridor interactive map on its website.(58) Los Angeles announced in August 2015 the goal of reducing citywide traffic deaths by 20 percent by 2017 and eliminating all traffic deaths by 2025.(59) Portland also has a goal of eliminating all serious and fatal crashes by 2025.

What Is Your Current Method for Identifying High Pedestrian Crash Locations? How Do You Identify the Locations of Concern (Where You Will Develop Potential Countermeasures)?

All the agencies interviewed are primarily using crashes to identify high pedestrian crash locations. They use GIS tools to identify the HCLs. Some agencies focus more on corridors, while others use intersections. After the crash histories for specific locations are identified, the agencies use a variety of techniques to develop a shorter list of sites where treatment will be identified. The crashes at the priority locations are then investigated further to help define which countermeasures to select.

Several agencies noted that the selection of a site for treatment may be the result of an opportunity to treat a site that is also undergoing needed maintenance, such as an overlay. The occurrence of a high-visibility crash or fatality may also affect whether a site is considered for improvements.

Miami-Dade uses GIS crash density maps to identify locations and segments with clusters of crashes. After identifying the initial list of candidate locations, three factors were used to develop the final list of locations: input from MPO staff, information in the Florida DOT work program, and review of aerials.(60) They also consider anecdotal data for injury crashes that do not generate a crash report.

In addition to the consideration of crashes, Austin would like to have a method to integrate public comments into its procedure.

Phoenix has Global Positioning System coordinates for all crashes and plots within a GIS to identify hot spots. Phoenix also considers the distance between points to identify corridors and noted that plotting makes it easy to identify hot spots. The Arizona DOT identifies HCLs using GIS density tools. The locations are separated into highway segments, intersections, and interchanges, with subsequent visual review to identify appropriate and logical end points to the segments. The Arizona DOT also uses a risk assessment methodology to identify segments and intersections where investment can help to lower the risk of pedestrian crashes. The methodology is modeled after those documented in NCHRP Report 803 and a risk assessment tool used by the Washington State DOT.(12,61) Application of the methodology occurs in the following two steps:

New York City identifies priority lists using the following three categories:

New York City bases its lists on 5 yr of data. When identifying potential locations for treatment, the city identifies areas and corridors for approximately 50 percent of the borough. For intersections, the city limits the locations to about 15 percent of the borough to obtain a manageable list (about 5,500 for the entire city).

Los Angeles uses a score that also considers the number of injured, the number of fatalities (multiplied by a 1.5 factor), the age of the pedestrian (an additional point is assigned for youth and older pedestrians), and a health and equity index.(62) Having the ability to also integrate public comments, work requests such as sidewalk repairs, and trauma data was suggested as another method to identify locations of concern that are not currently integrated into the method but could be considered in the future. The locations are generally identified as either intersections, corridors, or areas.

Portland identified the top 20 high crash streets for driving, bicycling, and walking. The combination of those sites resulted in 30 streets being on their high crash network. The motor vehicle network includes the streets with the highest number of people killed or seriously injured between 2004 and 2013. The bicycle and pedestrian networks include streets with the highest number of crashes, regardless of severity, for people on bikes and walking, respectively, in recognition that the difference between a minor injury and a serious injury for pedestrians or bicyclists is “often random and circumstantial.”(63) The number of 20 high crash streets per mode (30 total) was selected as representing a balance between having enough streets to capture most of the fatal crashes and not having so many as to make the subsequent evaluation and countermeasure implementation steps exceed available resources.

What Tools Are You Using to Assist in the Identification of High Pedestrian Crash Locations?

All the agencies are using a GIS to identify high pedestrian crash locations. Some of the agencies are also using database software to help manage the data. The Arizona DOT imports the crashes into a Google® Fusion™ table/map that assists with the visual review of locations. Using this tool gives the analyst the capability to customize a data table with key crash attributes and to spatially evaluate these attributes associated with the pedestrian crash by running queries or filters. Another advantage is the ability to visually inspect the specific crash location with Google® Street View™ to determine roadway characteristics that cannot be identified with satellite imagery.

Do You Consider Exposure? If So, How?

Several agencies commented that they are not considering pedestrian volume exposure because of the lack of good data or because they are focusing on reducing the overall number of crashes rather than reducing a crash rate. They may consider distance (e.g., per mile) when comparing corridors of different lengths. Several agencies noted that they may collect pedestrian volume data on a project basis when needed during countermeasure identification.

Have You Used or Considered Surrogates?

Some agencies have considered surrogates, including the following:

In most cities, the number of crashes involving pedestrians is the key metric.

The Arizona DOT is using the FHWA Pedestrian Bicycle Crash Analysis Tool (PBCAT) to crash-type each crash on the State highway system. PBCAT identifies the events leading up to the crash.

Do You Cluster Your Pedestrian Crashes to Identify Hot Zones? If Yes, What Criteria Are You Using?

The interviewed agencies do review their data to identify clusters of crashes. Typically, they generate corridors that include multiple intersections. The Miami-Dade MPO noted that it has used age-group crash clusters to show areas where crashes involving juveniles are prevalent.

Los Angeles considers corridors rather than just intersections (similar to New York City and San Francisco). Visual review is used to find the clustering of intersections because the locations are obvious after the plotting of the data.

The Arizona DOT noted that it identifies locations based on GIS density/cluster analysis. The DOT does not have defined criteria (e.g., a certain number of crashes within a given distance) but is discussing potentially developing such criteria. Because of the dispersed nature of the State highway system and the random pattern of crashes, identifying clusters according to specific set criteria is difficult. A visual confirmation of the density locations identified in the GIS has been its approach, which, the DOT noted, would be difficult in a denser urban environment with a denser street network.

What Skill Sets Are Needed to Conduct These Studies? Do You Have Those Skill Sets In-House, or Do You Contract?

The skill sets needed to work with crash data include familiarity with a GIS, the ability to work with attribute tables, and programming skills. Understanding how to identify appropriate crash types and interpreting crash data are also valuable skills.

Many of those interviewed commented that they have gone outside of their immediate department to obtain the needed skill sets.

How Frequently Do You Conduct the Analysis? How Frequently Should It Be Conducted?

The analysis period ranged between 1 and 3 yr. Those locations with a higher number of pedestrian crashes (e.g., New York City and Los Angeles) tend to use a longer update cycle. Those agencies commented that minimal change occurs with a 1-yr increment for their region. Some of the agencies noted that the analysis period can be affected by the availability of crash data because they have to wait until the State releases the data to them.

Can You Provide Examples of What Needs to Be Done to Go From Your Crash Records to a Final List of Sites Needing Treatments?

Each of the agencies noted that problems and/or mistakes exist within the crash datasets they use, which requires them to clean the data. Having the geocodes for each crash results in the ability to identify problem locations faster and with greater confidence. In previous decades, the agencies would pull data for select intersections to determine crash patterns. The availability of geocodes allows the consideration of the entire city or State. There are concerns about the accuracy of the geocodes, especially when trying to determine if a specific crosswalk at an intersection needs attention. The Miami-Dade MPO noted that the Florida DOT has begun to supply local agencies with crash data shapefiles, which eliminates the labor-intensive process of address matching and data entry. New York City commented that local police are experimenting with new equipment that will permit geocoding of the crash location in the field.

What Are Your Lessons Learned or Takeaway You Would Offer From Your Experience?

Observations on lessons learned from their experiences include the following:

Do You Have Any Other Comments?

An additional comment made at the conclusion of the interview by several agencies is the desire to be able to better identify unreported collisions.

New York City noted that Los Angeles and San Francisco are also including an equity focus to ensure that locations in poor or predominately one-race locations are being considered for treatment upgrades.



Federal Highway Administration | 1200 New Jersey Avenue, SE | Washington, DC 20590 | 202-366-4000
Turner-Fairbank Highway Research Center | 6300 Georgetown Pike | McLean, VA | 22101