REFERENCE: Grant Ian Thrall, Bert Swanson and D. Nozzi, 1988. "Greenspace acquisition and ranking program (GARP): a computer assisted decision strategy." Computers, Environment, and Urban Systems, Vol. 12, 161- 184.

(c) 1998, Grant Ian Thrall, Bert Swanson, Dom Nozzi, all rights reserved. Permission to reproduce this article in any form must be obtained in writing beforehand from Grant Ian Thrall, Bert Swanson or Dom Nozzi. Placement of this article on the Internet does not alter the copyright nor does it place this article in the public domain.

GREEN-SPACE ACQUISITION RANKING PROGRAM (GARP): A COMPUTER-ASSISTED DECISION STRATEGY

Grant Ian Thrall- Department of Geography, University of Florida

Bert Swanson- Department of Political Science, University of Florida

Dominic Nozzi- Department of Community Development, City of Gainesville, Florida

ABSTRACT. GARP, an acronym for Green-space Acquisition and Ranking Program, is a computer-assisted decision strategy (CADS) that can be the basis of an orderly and rational local government program of acquiring landfor open space and recreation. Fifteen criteria are used to rank parcels. Numerical values are assigned according to how well the parcel conforms to each criterion and the importance of the criterion for "active" and for "passive" recreational use. Active use generally requires capital intensive development such as ball fields. Passive use- is typically low intensity such as bridle and walking paths. The numerical valuesfor each parcel are summed separately for active and for passive use; these scores reveal an ordinal ranking of the properties separatelyfor active and for passive use.

Local governments have long been in the business of acquiring, swapping and regulating land for public facilities -schools, fire stations, roads, parks, and so on. These decisions impact land owners, developers and the general public. Serious questions have been raised about the propriety of these land transactions (Gardiner & Lyman, 1978). How are park and recreation lands presently selected by local governments? Offers of land are presented to planning staff or elected officials by developers, environmental- ists, elected officials, or neighborhood organizations. The respective groups in effect recommend that a parcel be acquired as soon as possible. The parcel is subsequently (and quite subjectively) evaluated on its own merits, without regard to how the parcel fits into a general scheme of recreation or open space, and independent of comparative evaluations. Such opportunistic acquisitions may indeed provide park and recreation space, but in a non-rational and non-strategic manner.

  For example, a municipality with a limited greenspace acquisition budget may purchase a parcel being offered for sale merely because (1) there is enough money to purchase the parcel, (2) the appraisal of the parcel compares favorably to the asking price, and/or (3) the parcel is suitable for development of recreational facilities.

  Unfortunately, this approach usually fails to consider whether there are other available parcels which satisfy a wider range of community interests, which may include:      
   (a) more environmentally significant?;
   (b) within an urban-defining greenbelt area that would control costly urban sprawl?;
   (c) able to support a wider range of recreational or educational interests?:
   (d) in a high-growth area of the community?;
   (e) nearer to concentrations of population?; or
    (f) better able to link public lands to each other?

  With a systematic assessment of vacant parcels available for park acquisition, the municipality is more likely to maximize its return on greenspace investment, perhaps even with the purchase of fewer parcels.

  Rather than an objective and orderly process, the decision for which parcels of land are to be acquired for park development is more often the result of political machinations than that of a rational, orderly process. The acquisition process has historically been opportunistic; that is, decision makers usually respond to local lobbyists rather than a clear, long-term community objective. For example, developers know that positive externalities from nearby public open space increase the value of their land holdings (Thrall, 1987, 1988). Environmentalists seek to preserve sensitive lands they have come to favor for passive recreation. Public service organizations lobby for development of active recreation facilities: ball fields, tennis courts, and so on. Politicians may play one special interest group against the other, for funds are limited and tradeoffs are made. Where high returns or emotions are at stake, charges of graft can be raised, and indeed can occur.
  
     Objectivity in land transactions can contribute to eliminating waste of public revenues and contribute towards the creation of an optimal spatial distribution of publicly owned lands (e.g., U.S. House of Representatives, 1988). A common practice of some local governments is to "buy high and sell low." A city purchases property well above the assessed value and subsequently sells that property at less than the assessed value. Local governments may receive gifts of land from an estate or a party seeking some tax benefits. These properties are seldom located in a manner that would conform to some long-term social criteria of acquisition. Land swapping is a practice that can benefit the community or leave it with an undesirable spatial distribution of public lands.

  The inefficiency in present practices of local government parkland acquisitions has stimulated some local governments to engage in a more proactive land acquisition procedure so that land can be acquired on a timely and needed basis. In this manner, local governments can secure better location and forestall increasing land prices (Shoup & Mack, 1968). It is incumbent upon public officials to gain the confidence of their constituents that land acquisition decisions are made in the public interest. Consequently, there is a need for an explicit system of criteria with standards and principles that can be systematically applied.

  The strategy presented here was developed as a supplement to the Recreation and Environmental Conservation Elements of the Comprehensive Plan of the City of Gainesville, Florida. The goal was to create an objective strategy for the ranking of parcels of land for public acquisition. The system is designed to minimize short-sighted, politically motivated land acquisition decisions, and instead facilitate long-range decision making in the public interest.

  GARP (Green-space Acquisition and Ranking Program) is a computer-assisted decision strategy (CADS). GARP is composed of a set of clearly defined criteria that the program user applies to each prospective parcel of land. The criteria are comprehensive but sufficiently general to allow GARP to be applied in most geographic regions. Fifteen criteria are used to rank parcels. Numerical values are assigned according to how well the parcel conforms to each criterion and the importance of the criterion for "active" and for "passive" recreational use. Active use generally requires capital intensive development such as ball fields. Passive use is typically low intensity such as bridle and walking paths. The numerical values for each parcel are summed separately for active and for passive use; these resulting scores reveal an ordinal ranking of the properties for each of the two uses.

  Since tools of this type are sufficiently novel, throughout the process of designing GARP consideration was given as to whether a modification would deleteriously affect eventual adoption of the CADS, or make its adoption easier without sacrificing the overall objective of the CADS. For a discussion of the process of adoption of new ideas see Morrill, Gaile and Thrall (1988). GARP's final form was designed to be understandable to politicians, non-elected community leaders, news reporters, administrative heads of the various governmental agencies and to the average citizen. Criteria had to be sufficiently clear so that a layman could understand the mechanisms that drove GARP.

  Ranking parcels purely on the basis of their environmental importance requires an intensive time-consuming evaluation of the specific parcel as well as the regional envi- ronment. Credible environmental assessment should be done by persons with specialized expertise in the subject. Individuals may consider themselves advocates of the environment, but the expertise of local planning staff falls more in those areas that deal with the built environment. Such expertise does not necessarily translate into a skilled assessment of the nonbuilt environment. Since nonbuilt environments differ considerably between regions, the criteria required to rank local ecosystems on the basis of their relative importance then also change between regions. Considerably more investment is then required to create an ecosystem data base. Thus, while it may be logically possible to construct a general CADS for ranking parcels by their environmental importance, resources are usually sufficient only to roughly assess such importance. More impor- tantly, a ranking of lands on the basis of their environmental importance may be entirely different from that based upon their recreational value. We do not consider it feasible to objectively determine which between environmental versus recreational land at this time is more worthy of acquisition; evaluating which is more important and when the two can coexist does require human judgment and management expertise. Therefore, the environment is treated in a very general manner, and then only in the context as to how its preservation will enhance the built environment.

  GARP may be the first CADS for evaluating open space for urban active and passive recreation. Some of the criteria used in GARP are reminiscent of a tabulation scheme presented in the earlier work of Sargent (1976). Tanic (1986) has outlined a general scheme for using artificial intelligence in planning urban systems. Jones and Davidson (1987) describe a computer system for the storage and retrieval of information on land use. Cook and Seiford (1984) present a model for ordinally ranking arteries for highway corridors. Dobson (1979) contributed a screening procedure for energy facilities. Can- ter, Atkinson and Leistritz (1985) detail a checklist for increases in infrastructure as population increases, including that for general expenditures on parks and recreation facilities; but, no criteria are presented for which acreage to acquire. A well-known automated system for ranking land for some form of protection from development is LESA (1983): (National Agricultural) Land Evaluation and Site Assessment.

  LESA was designed to determine the quality of land for agricultural purposes and to assess its agricultural economic viability. The purpose of LESA was to guide decision making by developers, land owners, government staff, and elected officials. LESA consists of land evaluation and site assessment criteria. In the land evaluation mode, soils from the region are rated and ranked from highest to lowest by type of agricultural use: cropland, forest, rangeland, and so on. In the site assessment mode, land is ranked according to other factors, including scores that decline with proximity to urban areas, especially with proximity to urban infrastructure such as sewer lines; it is recognized that once infrastructure of this sort is extended into rural areas, the adjacent land experiences development pressure that often leads to a conversion to nonagricultural use (Thrall, 1987, Chs. 3 & 12). An impartial and automated system such as LESA was a significant improvement over an ad-hoc stratagem for preserving farm land.

HOW GARP OPERATES

The Creation of GARP

  A "modified delphi" technique was used to arrive at the criteria used in GARP. Normally, the delphi technique collects the wisdom of experts located throughout the country and then secures their input on selected topics. The procedure was modified to involve only local experts, several of whom were faculty at the University of Florida. The experts were chosen for their knowledge about human open space and park land needs, their knowledge about flora and fauna behavior and needs. Included were experts in geography, city planning, water district management, parks and recreation programs, wildlife management, engineering, biology, planning, public administration, agricultural economics, and environmentalists. These experts were assembled together for two half-day sessions; during these sessions there was a free exchange of views between all participants. The objective was expressed to the participants in a very general manner along the lines of the introduction to this paper. Several general criteria were suggested by the organizers -and many quickly abandoned or modified by the group.

Summary of the 15 Criteria

The outcome of the delphi was the group collectively arrived at a consensus for the following criteria. The Greenspace Acquisition Ranking Program (GARP) is composed of 15 criteria. These criteria are designed to assess:

  It is intended that GARP be used by citizen participants from the local community such as a local government-mandated task force. These evaluators may visit the pro- posed site in person, or examine a report prepared by planning staff which may include a video tape recording of the site. The report would be sufficiently comprehensive so as to cover all those issues relating to the 15 criteria in GARP. The evaluators would then indicate how the parcel conformed to each criterion.

  The response of the evaluator is input into a commercial spreadsheet software program. For each criterion, the evaluator need only select a switch like "high," "medium," "low," and in some instances "no" indicating the degree of conformity of the land parcel to the particular criterion. For example, conformity to the criterion may fall within the "medium" category which carries for that criterion a value of, say, 3. For each land parcel, two tabulations of scores are kept. The first set of tabulations weights the importance of the criterion to active recreation. The second set of tabulations weights the importance of the criterion to passive recreation use. For example, the weight for active recreation use for a criterion may be 4 and for passive recreation use the weight may be 1. In this example, the spreadsheet will assign to this parcel a score for this criterion for active use of 12=3*4, and for passive use of 3=3*1. Numerical values arrived at in a similar manner are calculated for each of the 15 criteria and then summed, keeping separate the tallies for active and for passive use. The resulting two sets of tallies are the ordinal ranking of land parcels for each recreational category.

  The numerical values presented below were arrived at by the delphi team after the 15 criteria were agreed upon. Using a trial set of weights for active and passive use for each criterion, a set of parcels well known by the evaluators (in this case part of the delphi team) was ordinally ranked by the GARP program. The participants adjusted the weights until the ranking conformed to what they would have arrived at by subjectively ranking the property. A larger set of parcels which included parcels known only from reports prepared by local planning staff were then evaluated using the GARP program. In this manner ranking of the larger set of parcels is expected to be consistent with the prejudices and biases that the evaluator applied to the smaller set of parcels well known to him or her. It is advised that a community that applies GARP should first calibrate the program to their local biases and prejudices through several trials using parcels well known to the evaluators; values for "high," "medium," and "low" should be adjusted for each criterion as well as the weights for active and passive use for each criterion (Elazar & Zikmund, 1975).

Explanation of the Criteria

  Each of the following criteria has a graphic associated with it (Figures I through 15). It was our experience that the graphic was worth a "thousand words" in conveying to the professional and the layperson alike the meaning of each criterion. Indeed, the graphics were originally assembled together in a publication directed to the politician and citizen. The graphics were created using an Apple McIntosh and a laser printer. The following are our 15 criteria.

Density
  What is the current and expected population or development density of the area where the site is located? Current and expected population or development density is calculated by drawing a '/4 mile (1320-foot) radius from the boundaries of the site and determining the acreage of this radius area that carries a zoning or land use designation of "multi-family" and/or "business/commercial." The threshold for the multi-family definition is 5.8 units/acre. A site that is in close proximity to high-density commercial and/or residential areas provides more convenient access to a larger number of park users. For further discussion on the relationship between population density and park space see Thrall (1987, pp. 175, 181-182, 194). For discussion of techniques for measuring population density see Haynes and Rube (1973) and Thrall (1988). Figure 1 shows this criterion pictorially.

  A parcel is to receive a score of "high" if more than 31 acres of the radius area is comprised of multi-family and/or business/commercial parcels; "medium," if between 6 acres and 31 acres of the radius area contain such parcels; and "low," if less than 6 acres of the radius area is comprised of such parcels. Parcels with the "high" ranking receive a numerical weight of 5, "medium" receive a numerical weight of 3, and "low" receive a numerical weight of 1. The final tabulation for this criterion is arrived at by the product of the appropriate density numerical value and a weight of 3 for its active recreation versus a weight of I for its passive recreation use. For example, a parcel receiving the "high" ranking is awarded 15 =5*3 for active and 5 =5*1 for passive tabulations.

Proximity
  What is the proximity of the site to other public parks and open spaces? Sites that are relatively remote from existing parks of the same park type enhance the dispersal, and therefore accessibility, of each type of park. Active parks that are too closely spaced may suffer from underuse. Passive sites are scored neutrally for this criterion since proximity to existing passive parks is often preferable. Figure 2 portrays the meaning of this criterion graphically.

  The site is assessed to determine the number of acres that may be developed for active recreation. If this developable acreage is less than five acres in size (mini-park size), a '/2 mile radius is drawn from the boundaries of the site. If developable acreage is between 5 and 20 acres (neighborhood park size), a 1-mile radius is drawn from the boundaries of the site. If developable acreage is greater than 20 acres (community park size), a 3-mile radius is drawn from the boundaries of the site.

  A parcel is to be considered "high(ly)" remote if within its service radius no active park of the same size category exists. A site remote from existing parks/open spaces is assigned a score of 5.

  A parcel "medium" in proximity has no active parks within its service radius but there is a passive-linear park of the same size category, or if the radius contains an active park of the same size category and the recreation plan shows a park deficiency for the planning area of the park category. This designation may also be assigned to parcels whose service radius contains only passive acreage. A site with "medium" proximity receives a score of 3.

  A parcel is considered "low" in remoteness if within its service radius there exists an active park, and the recreation plan does not show a park deficiency for the planning area of the park category. A site falling in this category receives a score of 1. The final tabulation is arrived at by the product of the above score and a score of 3 for its active recreational value and I for its passive recreational value. For example, a site classified as "high" will receive a score for proximity of 15 = 5 *3 for active and 5 5 *I for passive recreational use.

Resource Access
  Does the site provide access to a natural resource with limited public access? Many environmentally significant resources are not accessible or visible to citizens. As a result, it is likely that many citizens will be unaware of such resources, become apathetic about such resources, and possibly be less able to understand the value of such resources (Gould & White, 1986). Those sites that offer physical or visual access to environmental resources that currently provide little or no such access are therefore considered most valuable by this criterion. Figure 3 illustrates this criterion.

  A site that offers a significant amount of access to a resource that otherwise would be restricted to the public is "high" in access benefits. Without the proposed site, no publicly owned parks presently access the resource in such a way as to permit exploration of the majority of the resource. An example is a parcel of land on a lake where the lake is otherwise completely circumscribed by private land ownership.

  A site that if acquired by the public sector would offer improved access by the public to environs where the public already has access is classified as having "moderate" access benefits. A site is classified as having "low" access benefit if there already is significant public access or if the site does not provide access to a resource.

  A parcel classified as having high access benefits is allocated a score of 3; moderate access benefits receives a score of 2; low access benefits receives a score of 1. This score is weighted by values of 2 for active recreation use and I for passive recreation use.

Trail Access
  Is the site serviced by an existing or potential abandoned railroad or utility easement right-of-way corridor? Rail or utility right-of-way easements are generally wide enough to accommodate some forms of recreational travel (bicycling, hiking, jogging). A limited form of wildlife corridor capability may also be provided by access to such easements. Because rail rights-of-way occasionally are purchased by the public, their potential usefulness as public travel corridors is significant. This is particularly true in regard to their use for recreation, since they are capable of providing pleasurable and safe transportation to locations that are now only accessible by auto. Sites which are traversed by, or in close proximity to, such corridors are therefore valued most highly by this criterion. This criterion is illustrated in Figure 4.

  A site is defined as having "high" direct linkage if a right-of-way either traverses, runs adjacent to, or terminates at the parcel.   A site is defined as "moderate" linkage for pedestrians if a right-of-way passes within 1,000 feet of the site and currently provides sidewalk access which does not cross an arterial roadway. "Moderate" linkage can also come from bicycle access if the right-ofway passes within 3,000 feet of the site and currently provides on-road bicycle lane access meeting state standards.   A designation of "low" linkage is assigned if the right-of-way passes within 3,000 feet of the site but does not currently either provide bicycle or pedestrian facilities as specified above.

  Finally, a designation of "no" linkage is assigned where the site is greater than 3000 feet from the nearest right-of-way. This category may include active rail lines, or rail lines to which, though abandoned by the rail-road, public use is restricted or the rightof-way is not likely to be developed for public use.

  The numerical values assigned to each category are as follows: "high" linkage = 5; moderate" linkage = 4; "low" linkage = 3; "no" link = 1. The score arrived at by evaluating the linkage is then multiplied by 3 for the active recreation tally and by 2 for the passive recreation tally.

Greenbelt
  Would the site be relatively useful as a component in a greenbelt network? If so, the following benefits are provided: (1) control of urban sprawl; (2) integration of area parks into an interconnected network of open spaces like the "emerald necklace" concept of Cleveland, Ohio; (3) contribution to a community identity as one offering a
high quality of life; (4) enhancement of ecosystem viability (wildlife buffers, corridors, and islands); and (5) enhancement of access to each of the park "gems" of the emerald necklace. For discussions of greenbelts see Thrall (1987, ch. 13), Munton (1983), Lichfield and Darin-Drabkin (1980). This criterion is depicted in Figure 5.

  The location of the greenbelt is defined by the long range land use plan of the community. Before such a plan is acted upon, the greenbelt can be considered as a zone extending 2 miles beyond the furthest extension of water and sewer lines from the local utility.

  Sites are considered as "high" in usefulness that are at least partially within the greenbelt region and are over 200 acres in size. Additionally, sites greater than 500 acres in size receive this classification if the parcel is within '/2 mile of the greenbelt zone.

  A site is to be considered as being "medium" in usefulness if it is within the region but is less than 200 acres in size. Any site greater than 500 acres in size but more than '/2 mile from the region also is to receive this classification.

  The classification of "low" is to be assigned to sites that are within '/2 mile of the region but less than 500 acres in size.

  Sites are classified as being of "no" usefulness if none of the above greenbelt conditions are met.
  The numerical values for each greenbelt classification are: "high" usefulness=5; "medium"usefulness=4; "low"usefulness=3;"no"usefulness=l. Inturn,thesevalues are multiplied by 4 for the active recreation tally, and also by 4 for the passive recreation tally.

Connectivity
  Does the site provide a physical linkage between two or more existing public parks or open spaces, or extend the size of such parks? Refer to Figure 6.

  Sites that are adjacent to, or are in close proximity to existing parks act synergistically to improve and enhance the integrity and recreational/ecological value of both sites. For example, if an existing park is enlarged, a greater range of recreational activities can be accommodated, and existing ecological attributes are more easily preserved. Also, connections between parks serve to improve the accessibility to such parks. This criterion recognizes the value of both habitat islands and wildlife corridors.   A parcel is defined as having "high" connectivity if it (1) physically links two or more existing public parks or open spaces; (2) connects lands that have been previously defined in a local conservation plan as forming a network of natural areas; or (3) the parcel is at least 500 acres in size and extends lands that are part of the natural area network. These sites must be at least 300 feet in width at their narrowest reach.

  A parcel is defined as being of "medium" connectivity if (1) the site offers a linkage separated by a roadway; (2) an extension or connection as defined above exists but is less than 300 feet in width or the site is less than 500 acres, or (3) the site will extend the size of a public park or open space.

  A parcel is defined as being "low" or "no" connectivity if neither of the above two conditions is met.

  The numerical values assigned to these criteria are: "high" connectivity= 5; "medium" connectivity = 3; "low" or "no" connectivity = 1. This criterion was j udged to be equally important for active and passive use; the scores for this criterion are then multiplied by 3 for the active recreation tally and also by 3 for the passive recreation tally.

Multi-Uses
  Is the site suitable for both passive and active recreation? It is usually less costly, and in some cases more attractive to acquire and develop a site that can accommodate a wide range of active and passive recreational activities (Brown, Keane & Kaplan, 1986). Single-use or limited-use parcels require a larger number of sites to accomplish the same range of recreational and conservation activities as can be offered in multiple-use sites. Sites are therefore valued more highly in this instance if they can accommodate both forms of recreation. Refer to Figure 7.

  Passive recreation acreage contains desirable natural attributes and is generally not suitable for intensive recreational development. The proportion of the site that is classified as being passive is derived by dividing the acreage that is floodprone or contains significant environmental resources by the total acreage of the parcel; this yields an estimate of the proportion of total acreage that is inappropriate for active recreation development.

  A parcel is classified as being "high" for both active and passive recreation types if between 50-70% of the site is of one or the other type. Also, a parcel is to receive this classification if the smaller of the two types is greater than 50 acres, irrespective of the calculated proportion.

  A parcel is classified as having "medium" value for multiple use if between 70-85 % of the site is composed of one or the other land type. Also, if site contains more than 85 % of one or the other land type, and the smaller of the two types is greater than or equal to 15 acres. Sites that contain more than 85 % of one type (and the smaller proportion is less than 15 acres in size), but contain site improvements such as a building or facilities, are to receive this classification.

  A parcel is to be classified as having "low" value for multiple use if more than 85 % of its land are devoted to either resource- or capital-based uses. A site receives this designation irrespective of the calculated proportion if less than 15 acres are in either land type.

  The numerical values for this criterion are "high" = 5; "medium" = 3; "low" = 1. In turn, these scores are multiplied by I for the active recreation tally and also by I for the passive recreation tally.

Rarity
  A parcel is to be classified as having a rare environment by considering the attributes of the parcel relative to its occurrence at the national, state, and local level. Attributes can be geological, archeological, or biological. Sites which contain rare and/or significant natural attributes are valued highly since public acquisition or management would improve the likelihood of these attributes being preserved for current and future generations. Such attributes are often uniquely suited to providing aesthetic, ecological, scientific, and recreational benefits to the community. The site is awarded points on the basis of the presence of attributes at the site, such as that illustrated in Figure 8. Except as noted, a site is awarded one point for each occurrence from the following list:

  A site is classified as being "high" in rarity if by tabulating points from the above list, total of 2 or more points are obtained.

  A site is classified as "medium" in rarity if I point is awarded from the above list.

  A site is classified as "low" in rarity if no points are awarded from the above list.

  The numerical values used to score the importance of the particular parcel are 5 if the parcel has been classified as being "high"; 3 if "medium"; and I if "low." These scores are multiplied by 2 to derive the active recreation tally and by 3 for the passive recreation tally.

Diversity
  How much diversity does the site possess in regard to natural attributes? Sites with a relatively high degree of natural species diversity are generally recognized as demonstrating high ecological quality (see Figure 9). In addition, sites which contain a diversity of rare and/or significant natural attributes are valued highly since public acquisition or management would improve the likelihood of these attributes being preserved for current and future generations. Sites that are relatively diverse are often more interesting, aesthetically pleasing, and likely to resist extinction.

  The subcriteria (a-i) used in the 8th or Rarity criterion is also used here for the Diversity Criterion. A parcel that cumulates a score of 4 points or more from the list in the 8th criterion is referred to as being of "high" diversity. One to 3 points allows the site to be referred to as being of "medium" diversity. And the designation of "low" diversity is reserved for those sites that accumulate zero points from the above list.

  The numerical assignments awarded to the parcels are 5 points if the parcel ranks "high," 3 points if "medium," and I point if the parcel ranks "low." In turn, these scores are multiplied by 2 to arrive at the active recreation tally and by 3 for the passive recreation tally.

Ecosystem
  How critical are the natural attributes of the site to the integrity of an ecosystem? Refer to Figure 10. Sites that contain one or more of the following attributes are deemed relatively more meritorious than other sites since public acquisition or management would improve the likelihood of these attributes being preserved for current and future generations. The following attributes are generally considered most critical to the integrity or survivability of area ecosystems:

(a) Near the headwaters of a surface water body;
(b) Within a zone of potentially high groundwater recharge;
(c) Buffering a sensitive ecosystem from incompatible, adjacent land uses;
(d) Part of an intact ecosystem; or
(e) Part of a system in critical need of restoration.

  Information necessary to evaluate if the parcel conforms to any of the above is easily obtained from various government agencies, and from non-government site checks.

  A site is deemed as being of "high" importance to the ecosystem if it conforms to at least two of the above elements.

  A site is referred to as being of "medium" importance to the ecosystem if it conforms to one of the above elements.

  If a site does not conform to any of the above elements, then it is deemed to be of "low" importance to the ecosystem.

  The numerical values assigned to the parcel using this criteria are 5 points if the site receives the classification of being "high"; 3 if "medium"; I if "low." In turn, these scores are multiplied by 2 to derive the active recreation tally for this criterion and by 3 for the passive recreation tally.

Cost
  What is the per acre cost to acquire the site? Local governments have a finite amount of resources at their disposal, and therefore are constrained in their ability to purchase sites. To conform to the budgetary constraints of the government, and to maximize its ability to acquire sites, parcels that are offered for sale at relatively attractive prices are valued more highly than those where the owners attempt to receive unusual windfall returns from the public purchase. Refer to Figure 11.

  The average (or median) per acre land value of comparable properties from sales occurring during a period of six months prior to the assessment is determined. A parcel is referred to as "high" in purchase-value if comparable sites have sold at more than 15 % of the acquisition price of the parcel. A parcel is referred to as "medium" in purchasevalue if comparable sites have sold within 15% of its cost of acquisition. A parcel is referred to as "low" in purchase-value if the selling price of comparable sites are more than 15 % below the cost of acquiring the proposed site.

  A score of 5 is assigned to a parcel that rates a "high" in purchase-value; "medium" in purchase-value receives a score of 3; "low" in purchase-value receives a score of 1. In turn, these scores are multiplied by I for the active recreation tally and by 2 for the passive recreation tally.

Availability
  Is the property owner interested in selling the site? Sites that are offered for sale often can be obtained by more agreeable and flexible terms than their counterparts. Conversely, a property owner who at present is unwilling to consider sale of a parcel is unlikely to offer flexible or agreeable terms. Parcels not for sale may require higher cost of acquisition or even purchase using eminent domain. Acquisition by eminent domain is often politically unfeasible, time consuming, and costly. Therefore, parcels whose owner(s) have expressed a high willingness to enter into negotiations will be ranked by this criterion higher than their counterparts. Refer to Figure 12.

  The numerical values for this criterion are: A "high" score is given to properties that are officially for sale; the numerical value is 5. A "medium" score is assigned to parcels not officially for sale, although there is strong evidence that the owner will be eager to sell; the numerical score is 3. If the availability of the property is unknown, it receives a "low" score of 2. If the owner is known to be neither willing nor eager to sell, "no" sale is likely to occur in the short term so a numerical score of I is assigned. In turn, these scores are multiplied by I for the active recreation tally and by 2 for the passive recreation tally.

Pressure
  When is the site likely to be developed? Sites that are being considered for development are highly susceptible to deterioration or loss of their ecological or recreational value; this deterioration can be significant and permanent. If the site is already developed, or has development constraints such as from deed restrictions or easements, then the property shall be evaluated as having lower development pressure. See Figure 13.

  When a site is in an area likely to be experiencing rapid land use change, it is not uncommon for the owners to intend to offer to sell the property in the near future to private developers. Development pressure is determined by drawing a 0.5 mile radius about the outer edge of the site. If a minimum of 20 acres of land within the inscribed zone is developed as residential, commercial, or industrial use, then the property is classified as "high" in potential for development in less than 5 years.

  If the parcel is greater than 5 acres and less than 20 acres, or if the parcel is serviced by water and sewer lines but does not have nearby development, then the property is classified as "medium" in potential for development in 5 to 10 years. If 5 acres or less of the area within the '/2 mile radius is developed for residential, commercial or industrial use, and the site is not serviced by both water and sewer, then the parcel is classified as "low" in potential for development in the foreseeable future.

  The numerical value assigned to a parcel classified as likely for development is "high" and receives a score of 5; "medium" in potential receives a score of 3; "low" in potential receives a score of 1. In turn, these scores are multiplied by 2 for the active recreation tally and by 3 for the passive recreation tally.

Jurisdiction
  Should a municipality purchase property outside its jurisdiction when greater shortterm benefits may accrue to land owners nearby the site but are not contributing taxes for the acquisition of the site? Conversely, should a county government purchase a site with general revenues when the property surrounding the site may in the near future be annexed by a nearby city? Refer to Figure 14.

  Intergovernmental disputes lead to higher costs of acquisition and an underprovision of park lands. For some governmental units, which jurisdiction a proposed site for acquisition falls in can present significant political, legal, and maintenance cost problems. Sites that fall within the jurisdiction offer to that jurisdiction greater management control and greater ease of applying regulatory codes of importance to the jurisdiction. Sites that are within or near the jurisdictional limits are more accessible to the residents paying for the acquisition and thereby increasing the political feasibility of acquiring the parcel. 

  For an incorporated city, if the parcel is within jurisdictional limits it is designated as "high" in governability and assigned a value of 5; if the parcel is within the urban area it is viewed as "medium" in governability and assigned the score of 3; if the parcel is outside the urban area it is viewed as "low" in governability and hence receives a score of 1.

  For a county, if the parcel is within a non-incorporated area it is judged to be "high" in governability and assigned a score of 5. If the parcel is in an area likely to be soon incorporated then it is "medium" in governability and receives a score of 3. Land that is incorporated into a city is "low" in governability by the county and hence receives a score of 1. Multiply by 2 for active and by I for passive recreation.

Environmental Degradation
  This criterion recognizes the interdependence or "nonorthogonality" of the criteria. A site that contains acreage that is environmentally important and sensitive can receive a higher score for passive recreation based upon criterion seven, namely Multi-Use. Because of important environmental attributes, a site may also receive a particularly high score attributable to the criteria of Diversity, Rarity, and Ecosystem. Such sites with extraordinary engineering costs can become capable of supporting active recreation such as human and capital intensive ball fields and boating. However, the expenditure of such extraordinary engineering costs should not be encouraged especially when these monies can be used to purchase additional lands elsewhere. In addition, the nearby presence of people or capital development in the active recreation areas, or the failure of engineering technology, may be sufficient to significantly harm and even destroy the passive worth of the site, like that portrayed in Figure 15.

  To discourage sites from ranking high as active recreation when such development will lead to significant environmental degradation of important features of the site, then a negative score of -50 points is used in the active recreation tally. This criterion does not affect the passive recreation tally.

EXAMPLE

  Suggestions for parcel acquisition can come from a variety of sources such as devel- opers, citizen-activists, non-profit organizations, citizen advisory boards, or government staff; for a general discussion see Pinch (1985) and Catanese (1984). A description and assessment of the site is prepared by the local government staff. A staff program operator or citizen advisory task force of evaluators operate GARP.

  Tallies for each site are derived by evaluators selecting one switch ("high," "medium," "low," or in several instances "no") for each of the 15 criteria. A commercially available spreadsheet is used for the record keeping. The spreadsheet retains separate tallies and ordinal rankings of the sites for active and for passive recreation use. There is no mechanism presented here for subsequently comparing active and passive categories.

  Parcels whose ordinal ranking are highest are identified. A presentation to local government commissioners is given with summary recommendations for parcel acquisition. Following approval of the elected representatives, the staff begins procedures for acquiring the property. This can be accomplished by way of non-profit trusts; purchase from general local government revenues; police action such as zoning; applying for funds from state or federal government; or a combination of these mechanisms.

  For illustrative purposes, GARP was used to construct Table 1. In the table, several land parcels are ranked for Alachua County, Florida; the land parcels were selected from a larger set that has been proposed (or already acquired) for public acquisition. The body of the table is constructed by first evaluating whether the parcel ranked as "high," "medium," "low," or in several cases "no" for each of the 15 criteria. For example, Prairie Creek was evaluated as "high" for Rarity (criterion eight); Prairie Creek was then assigned a criterion value of 5. For the Rarity Criterion, active recreation has a weight of 2 and passive recreation has a weight of 3. Hence, Prairie Creek is assigned a value in the body of Table I of 10=5*2 for active recreation and 15 =5*3 for passive recreation. The final tally reveals that among the seven properties, Prairie Creek ranks low for active recreation but is tied for first for passive recreation. Various environmental organizations have campaigned against both public and private capital intensive development of Prairie Creek. Because of its environmental importance, Prairie Creek appears at present to be a likely candidate for public acquisition using funds from the, State of Florida specifically designated for the preservation of important natural area environmental wetlands.


CONCLUSION

  Local governments make decisions to acquire public open space. These decisions impact land owners, developers and the general public. However, these decisions are generally made in an ad-hoc manner. Hence, it is largely by chance that open space acquisition programs can lead to an optimal spatial pattern of public open space. Moreover, as transfer payments from federal and state governments to local governments diminish, greater competition is placed on local budgets.

  GARP, an acronym for Green-space Acquisition and Ranking Program, was designed as a computer-assisted decision strategy (CADS) for the orderly and rational acquisition of land for public open space. GARP evaluates land parcels on the basis of their human recreation potential. Recreation is defined in GARP as either active or passive. Active recreation criteria score parcels on the basis of their development potential and utility for capital intensive use such as ball fields. Passive recreation criteria rank parcels on the basis of its value for low intensity human use, such as bridle and walking paths. Fifteen criteria are contained in GARP.

  Each parcel of land is allocated a value indicating its conformity to each of the 15 criteria. This value is then multiplied by a weight for active and for passive recreation use.

  We recognize that the 15 criteria discussed in detail above are not orthogonal to one anothen Ideally, each criterion would concern a specific and unique concept, each concept being completely independent of one another. We also recognize problems of adding ordinal numerical values such as that attained when evaluating the cost of a parcel (Criterion #1 1) to an ordinal numerical value indicating that the parcel would be useful in a greenbelt system (Criterion #5). Such impurities will arise in the design of decision aids for use by the practitioner (Hepner, 1984). We considered such tradeoffs as necessary for there to be adoption of this computer-assisted decision strategy (CADS) for the orderly and rational acquisition of public open space.

  When implementing GARP it is advised that advantage be taken of the spreadsheet software to produce charts of a variety of cross sections of the data. Such charts can contribute as an important decision aid for they can reveal the consequences of choices made earlier in the evaluation scheme. One's biases and prejudices are revealed from the selection of weights for active and passive recreation and the selection of values for "high," "medium," and "low"; the influence of this upon the ranking of land parcels can be depicted using line graphs of the derived values (from the body of Table 1): Without changing the responses of the evaluators, recalculate the body of the table using different weights and values. Draw a line graph for each parcel across the 15 criteria plus the tally total on the abscissa; depict lines for several parcels on the same figure. Simulations can also be extended to scenarios where selected subsets of the 15 criteria are used, such as evaluating what type of park system would result if only, say, the I Ith criterion, that of cost, is used.

  GARP was used for illustrative purposes to evaluate and rank seven land parcels in Alachua County, Florida. The land parcels were selected from among those that had actually been proposed for public acquisition or had recently been acquired.

REFERENCES

Brown, T., Keane, T., & Kaplan, S. (1986). Aesthetics and management: Bridging the gap. Landscape and Urban Planning, 13, 1-10.

Canter, L. W., Atkinson, S. F.,, & Leistritz, F. L. (I 985). Impact of growth: A guide for socioeconomic impact assessment andplanning. Chelsea, MI: Lewis Publications.

Catanese, A. J. (1984). The politics ofplanning and development. Beverly Hills: Sage Publications. Cook, W. D., & Seiford, L. M. (1984). An ordinal ranking model for the highway corridor selection problem. Computers, Environment and Urban Systems, 9, 271-276.

Dobson, J. (1979). A regional screening procedure for land use suitability analysis. Geographical Review, 69, 224-234.

Elazar, D. J., & Zikmund, J. (1975). The ecology of American political culture. New York: Thomas Y Crowell Company.

Gardiner, J. A., & Lyman, T. (I 978). Decisionsfor sale. New York: Praeger. Gould, P., & White, R. (1986). Mental maps, 2nd ed. Boston: Allen & Unwin.

Haynes, K. E., & Rube, M. 1. (1973). Directional bias in urban population density. Annals of the Association of American Geographers, 63, 40-47.

Hepner, G. F. (1984). Use of value functions as a possible suitability scaling procedure in automated composite mapping. Professional Geographer 36, 468-472.

Jones, G. E., & Davidson, D. A. (1987). A computer-based land resource information system. Town Plan. ning Review, 58, 81-94. LESA. (1983). NationalAgricultural LandEvaluation and Site Assessment (LESA) Handbook. Washington, DC:USGPO.

Lichfield, N., & Darin-Drabkin, H. (1980). Land policy in planning. In Urban and Regional Studies, 8 London: Allen & Unwin.

Munton, R. (1983). London's green belt: Containment in practice. The London Research Series in Geogra- phy, 3. London: Allen & Unwin.

Morrill, R., Gaile, G., & Thrall, G. 1. (1988). Spatial diffusion. Scientific Geography Series, 10. Newbury Park, CA: Sage Publications.

Pinch, S. (I 985). Cities and services: The geography of collective consumption. London: Routledge & Kegan Paul.

Sargent, F. 0. (1976). Rural environmentalplanning. South Burlington, VT.F. 0.

Sargent. Shoup, D. C., & Mack, R. P. (I 968). Advance land acquisition by local government: Benefit-cost analysis as an aid to policy. Washington, DC: US. Department of Housing and Urban Development.

Tanic, E. (1986). Urban planning and artificial intelligence: The URBYS system. Computers, Environment and Urban Systems, 10, 135-46.

Thrall, G. 1. (1987). Land use and urban form: The consumption theory of land rent. London: Routiedge/ Metheun.

Thrall, G. 1. (1988). Statistical and theoretical issues in verifying the population density function. Urban Geography, 9, 518-537.

U.S. House of Representatives, 100th Congress, 2nd Session, June 30, 1988. Providingfor the disposition of certain lands in Arizona under thejurisdiction of the department of the interior by means of an exchange of lands. Washington, DC: GAO.