Defining Zones

Zoning is really quite simple. The first step involves selecting the crash problem on which the zones will be based and ensuring the availability of any needed data. The measure is then mapped, and zones are defined.

Step 1. Select the crash problem

The first step is to pick the pedestrian crash problem that the community wants to address. For example, are there excessive pedestrian crashes involving young children, working adults or, maybe, the elderly? Is the problem one of alcohol or drug use by the pedestrian?

As good as zoning is, not every problem can benefit from it. A zone approach is appropriate when all of the following conditions exist:

• Crash data needed to define the zones are available.
  
  
• There are sufficient data to produce a stable map.
  
  
• The pedestrian crashes cluster in some way.

Pedestrian programs are often directed toward specific age groups or toward a specific status of the victim (for example, using alcohol or drugs). A community using zones, therefore, would have to ensure that age and any necessary status data on the victim are present in the available state or local crash files. If not, the needed data must be obtained--likely from police crash reports.

In order to ensure a reasonably stable measure, experience has shown that a minimum of one year's data or at least 100 crash records should be available for establishing pedestrian safety zones. If 100 records for the crashes of interest are not available for a given year, additional whole years of data should be added until at least the recommended minimum number is reached (data for full years are desirable in case there are any seasonal effects). Obviously, the more crash records that are available for any given year, the more stable the zone definition is likely to be.

Step 2. Map the pedestrian crashes

The pedestrian crashes of interest can be mapped either manually or by a computerized mapping system, often called a geographic information system (GIS). The choice depends on the number of crashes available and/or the availability of a GIS program. There are several commercially available GIS programs that are designed to run on a personal computer.

Computerized mapping is obviously more efficient in establishing the correct location of each event. It also permits flexible analysis of the created zones by any number of other variables available in the data. Thus, for example, the data can be analyzed by the full range of variables maintained in a police report (time of day, day of week, etc.) if these records are entered into the computer and maintained with the location information. Maps can be created to show the crashes not only of several target groups (for example, children, working adults and elderly pedestrians) but also of various subgroups of victims or crash circumstances (for example, elderly males or nighttime crashes).

In general, mapping of up to 150 crashes, although tedious, can be accomplished manually. A large map of the area is required, and entry of any data subsets of interest must be planned prior to the start of the mapping. For example, if crash types are of interest, some method such as color coding would be needed to differentiate different crash types of interest. Separate maps might be needed to display different subsets of data. If the residence of the crash victim is also plotted, a means of associating the residence location with the specific crash location is needed.

Step 3. Define zones

A visual examination of the resulting map will show whether the pedestrian crashes of interest cluster in any areas of the city. If no clustering is apparent, that is, the map shows that the incidence of crashes for any given target group is spread essentially randomly, the problem may not be "zonable" for that community using the definitions suggested here.

Most pedestrian crash targets will show some clustering and some dispersion throughout a land area. Therefore, a systematic approach to identifying zones is required. The first steps are to select an initial shape for the zones and to define the target rate, that is, the number of events that must fall in an area for it to be defined as a zone. The approach suggested is to search first for circular zones, then to search for linear zones, then to examine the zones to determine if their shapes need refinement.

Search for circular zones. A circle is an easy area to work with when defining a zone. Since research has shown that most pedestrian crashes occur within one mile of the victim's home or work place, a circular zone with a radius of one mile was adopted for initial "zoning" studies involving pedestrians. A land area of one-mile radius (just over three square miles) is a manageable area in which to concentrate program activities.

A target rate for the events of interest must also be defined before zones can be selected. This rate is the minimum number of crashes that must occur in the circular area in order for the area to be considered a zone. The rate chosen must be high enough to make it worthwhile to mount program activities in the defined area. This rate depends on the size of the sample of crash reports available, how much the crashes cluster and the overall land area of the community.

It is recommended that the analysis start with a target rate of 10 crashes per zone as a minimum. This is a good value if the total annual size of the crash problem being zoned is on the order of 200 or less. If the annual number of crashes is higher, the target rate should be adjusted upward as necessary. For example, if the total number of crashes mapped was 400, a target rate of 20 crashes per zone might be more reasonable. Since the aim of zoning is to select a small land area that will encompass a large number of the target measures, the actual value is less important than the process of looking for clusters.

The individual zone definition process can be initiated for manual maps by simply creating an acetate with a one-mile radius circle and moving it over the map, examining clusters inside the circle and counting the events in the clusters. The same process can be accomplished with a GIS program by defining a circle with a one-mile radius and dragging it over the computerized map with a mouse.

Search for linear zones. Most programs targeted at crash reduction will include activities that can be applied to road segments as well as circular areas. Therefore, the map should also be examined for high frequencies of pedestrian crashes that occur along a single strip of roadway. For an annual crash rate on the order of 200, those roadway segments where six or more crashes occur in a two-mile segment should be identified as linear zones. Again, this rate can be adjusted as necessary if the annual crash rate being examined is higher.

Create final zone shape. The defined circular and linear zones should be examined to determine if efficiency might be improved if they were merged or their shapes changed. For example, circular zones that are contiguous may be more useful if they overlap or are combined into a rectangular or polygon shape if many crashes occur just outside the zones. It may be wise to reduce the size of a circular zone or change its shape if most of the events within it cluster near the center. It may be better to add or delete a block or two to keep a neighborhood intact within a zone. It may also be necessary to revise the zone shape for a specific program activity. For example, it is easier to define the postal addresses of the zone boundaries of a square or rectangular zone than a circular one. This might be necessary if part of the program is to mail or deliver materials to each residence in a zone.

Step 4. Calculate efficiency measure and select final zones

Finally, for all the zones combined, the percentages of both crashes and land area covered should be calculated in order to determine program coverage efficiency. If the ratio of the percent of the problem addressed to the percent of the land area covered in the zones is much less than three, the zones may need to be reexamined to try to improve their efficiency. An efficiency ratio of three to one or higher (for example, 60% of the crashes of interest in 20% of the jurisdiction's land area) will almost certainly permit the application of countermeasures locally within the zones that would be prohibitively expensive if deployed jurisdiction-wide. Ultimately, if this ratio cannot be made greater than two, it may be wise to declare the particular problem under study as "non-zonable."

In defining pedestrian safety zones, therefore, the following two questions are addressed:

• Do the crashes of interest cluster in some way?
  
  
• Does the use of the defined clusters provide enough concentration of the problem to yield increased efficiency and make it possible to use locally deployed countermeasures?

If the answer to either of these questions is negative, the program as planned may not be zonable for that community.