Measures of success: measuring park accessibility using GIS.

Serving the public is fundamental to the mandate of municipal park and recreation departments. The ease with which city residents can reach services and facilities provided by such agencies is one key indicator of how successfully they are meeting this mandate.

Determining levels of accessibility to park and recreation services and facilities can be based on the location criteria suggested in NRPA's public park and open space classification scheme. These locational standards--detailed in Park, Recreation, Open Space and Greenway Guidelines, by James Mertes and James Hall--look beyond providing certain acreages of open space per 1,000 residents to locations of facilities relative to population distribution. Accessibility is measured by distance between sites and potential users. For example, neighborhood parks, the type analyzed in this article, are recommended to be located within a walking distance of one-quarter to one-half of one mile, "uninterrupted by non-residential roads or other physical barriers."

In the past, many agencies have measured their adherence to these standards by placing circles with radii equal to the stated distance around the center of each park. Such an approach, however, is inherently inaccurate. It fails to recognize that many facilities are accessible only at designated points, rather than along their entire boundaries, and that users can't travel as the crow flies to reach them. A method that takes into account realities of the landscape, including public rights of way and entrances to parks, would be preferable. Geographic information systems provide one opportunity to implement such a method.

Geographic Information Systems

Geographic information systems (GIS) offer agencies the ability to calculate with relative ease the level of access they're providing users. GIS is a computerized mapping system that enables both visualization and analysis of spatial data and their non-spatial attributes. For example, a map of the geographic distribution of a municipal park system, relative to the surrounding transportation network, may be accompanied by a database listing quantitative (acreage, number of tennis courts, length of trails, etc.) and qualitative (name and type of park, level of development, condition of play equipment, etc.) information about each facility.

In addition, statistics regarding the socioeconomic characteristics of area residents may be incorporated into the system.

A key principle of GIS is the ability to overlay these different types of data for a geographic area such that the spatial relationships between them may be assessed. GIS may therefore be used by park and recreation providers not only to display service areas of their facilities on the basis of walking distances to them, but also to determine the proportion of residents who are served by, or have access to, them.

As an example, let's consider College Station, Texas, a city of approximately 65,000 residents located 165 miles south of Dallas-Fort Worth, 95 miles northwest of Houston and 100 miles east of Austin. The city's Parks and Recreation Department manages more than 40 facilities, including the 23 neighborhood parks analyzed here.

Measuring Accessibility

I performed this ease study of accessibility using ArcView GIS (Version 3.2), a package produced by the Environmental Systems Research Institute. The network analysis technique I used was implemented in Network Analyst, which is an ArcView extension that can be used to solve many common network problems, including identifying service areas. (More information about Environmental Systems Research Institute's GIS products can be found at www.esri.com.)

Using the Network Analyst extension, I determined the service area of the neighborhood park system. A distance of half a mile was measured outwards from each park's specified points of access along the surrounding streets. Because neighborhood parks are supposed to be accessible by walking along local streets, only this type of street was included in the analysis. A service area was created for every point of access; these were then manually joined into a single service area for each park using the Union command in the Edit menu. For comparative purposes, circular buffers were also created for each park using the Create Buffers function in the Theme menu. A distance of half a mile was specified as the radius of each buffer, and was measured outwards from the center of each park.

The Select by Theme function in the Theme menu was then used to identify census blocks lying inside of each service area. To avoid the need for any form of areal interpolation, only blocks with their geographic center inside of each service area were identified as being within it. (That is, I used the Have Their Center In criterion.) I thought this would provide a more accurate representation than either the Intersect or Are Completely Within criteria, although it should be recognized that this technique does provide only an estimate of the true number and characteristics of residents located within the service zone. The populations of the selected blocks were then summed. The proportion of the College Station population with access, within half a mile, to one or more neighborhood parks could then be calculated.

Who Had Access?

Figure 1, at the left, illustrates the half-mile walking distance service area of College Station neighborhood parks. The irregularity of the shape of this area reflects the locations, and lack thereof of low-traffic, pedestrian-friendly, residential streets. It's evident that access to a park is highest ill the central portion of College Station; access is especially poor in the west and southeast of the city.

Figure 2, at the left, emphasizes the differences in size and shape of service area that may be obtained between the street network-based and the traditional circular approaches, using the same recommended distance of half a mile. The service area according to the traditional buffering method is noticeably larger than that based on actual walking distances. It also assumes that park users are willing and able to cross boundaries such as major streets and the railroad to reach a facility.

In Figure 3, at the left, the network-based service area is superimposed upon a map of property lot lines. This figure therefore illustrates the distribution of major residential areas relative to service areas of neighborhood parks. It's evident that central College Station, where housing density is at its highest, is, for all most part, well-served by neighborhood park facilities. In the north, east and southeast, however service is patchier, and in some cases entire subdivisions lack local]y accessible park lands. This map clearly illustrates to park and recreation agencies, as well as homeowners, where new neighborhood-style parks, might be needed.

When the populations of census blocks within the service area were summed, approximately 41 percent of College Station residents were found to have access to one or more neighborhood parks based upon the street network method. This compares to a proportion of 67 percent under the traditional, circular buffer technique. The substantial difference between these two figures highlights the need for GIS users to be aware of how variations in means of measuring key parameters call significantly affect final results.

Expanding Use and Accessibility

In addition to a means of visualizing and measuring levels Or accessibility to a municipal park system, as well as impacts of the use of alternative measures of distance on final results, many other parks and recreation-related applications of GIS exist. In College Station, for example, initial analysis of levels of access to all mini, neighborhood and community parks throughout the city has been enhanced by further analysis of levels of accessibility within each of 17 park zones. This has allowed not only calculation of the proportion of residents within a specified distance of park for each zone, but also, visual identification of underserved areas. When combined with current and projected population data, the accessibility maps enable the department to identify, areas where new parks are most needed to better serve city residents, as well as to prioritize need by zone.

Other applications include analyzing the distribution of facilities relative to certain socioeconomic and demographic groups (such as the young, elderly and minorities) to ascertain how fairly resources are distributed, as well as interactive, Web-based maps that enable residents to locate and obtain directions to facilities within their community. Providing this sort of information to the public in an easily understood form can only improve the profile of park and recreation agencies among their constituents.

Sarah Nicholls, Ph.D., is an assistant professor in take Park, Recreation and Tourism Resources Department at Michigan State University, in East Lansing. She can be reached at [e-mail].