|
|||||||||||
SPATIALLY EXPLICIT PLANNING TOOLSConverted to HTML on February 28, 1995Donald G. Morgan, Rick E. Page, Marvin A. Eng and H. Bruce Enns.Research Branch, British Columbia Forest Service31 Bastion Square, Victoria, BC V8W 3E7
E-mail: {dmorgan rpage meng benns}@galaxy.gov.bc.caPaper presented at DSS 2001, Toronto, Ontario, September, 1994
ABSTRACTThe Commission on Resources and the Environment (CORE) was created by the government of British Columbia in 1992 to provide advice on land use and related resource and environmental issues using community participation and consensus-based decision making. The Research Branch of the BC Forest Service compiled relevant spatial information and developed spatially explicit planning tools for the Cariboo/Chilcotin CORE planning area (approximately 8 million hectares in central BC).A set of generic, object oriented tools were developed that allowed for the generation and evaluation of land use plans. Landscape attributes, (e.g. mule deer winter range, topography and timber volume) were summarized by land use category to assess impacts of land use decisions. As plans are interactively modified, the model summarizes landscape attributes immediately. The areas of intact ecosystems in current and proposed parks can be evaluated to assess the conformity of a given plan with the goals of the Provincial Protected Areas Strategy. Gaps in ecosystem representation in parks can also be determined and the location of required park additions identified. By optimizing landscape features between protected areas, special management corridors can be generated to ensure that protected areas do not become isolated or fragmented. This system provides the analytical tools to evaluate technical information in a manner that is understandable to the public. With enhanced understanding and rapid response to land use proposals, this system is increasing the likelihood that we can bring balance and sustainability to land use decisions in the province.INTRODUCTIONThe public is demanding a more participatory role in the development of public policy than ever before. Nationally, this has resulted in the rejection of the referendum on constitutional change that was found unacceptable to Canada's citizens despite endorsment by government leaders. In British Columbia, the use of the land had come under continuing protest by groups of citizens that felt that government policy had not adequately reflected their views. The uncertainty introduced by valley-by-valley conflicts has hampered appropriate development of the province's resources. As well, land use decisions are now required to be demonstrably sustainable. The Commission on Resources and Environment was created in 1992 to help move British Columbia toward a sustainable society and to develop a world-leading strategy for land use planning and management.As part of the overall provincial land use strategy, regional plans are being developed that involve intense public consultation and multi-party, consensus-based negotiations. A critical component of effective shared decision making is equitable access to technical information. Information provides the technical basis for land use planning but, in this case, is also the basis for negotiation. The information must be easily understood, relevant and, most importantly, acceptable to all parties, to fulfil a useful role in negotiation. In close cooperation with all Ministries of the provincial government, Parks Canada and the Department of Fisheries and Oceans, the Research Branch of the BC Forest Service developed and implemented a set of object-oriented planning tools for the Cariboo/Chilcotin Regional Negotiation Table. These tools transformed obscure technical data into images and summaries that could easily be understood and used as a basis for negotiation.
Figure 1. The Cariboo-Chilcotin Region The Cariboo/Chilcotin is an area of 8 million ha of grasslands and forested plateau in south-central BC (Figure 1). The 61,000 residents of the region include 6,000 registered Indians representing 15 First Nations communities. Primary industries are forestry, tourism, ranching and mining. To initiate the table, CORE held over 100 meetings throughout the region, resulting in the self-formation of 24 sectors at the negotiation table representing the entire range of regional interests (Figure 2). The table attempted to negotiate a total package of proposed protected areas, regional zoning plan and policy recommendations for social and economic transition during 62 days of meetings over a 15 month period.
Figure 2. Participating Sectors at the Cariboo-Chilcotin Regional Negotiation Table. A crucial element of the regional land use plans and a sticking point in achieving consensus at the round tables, is implementation of the Provincial Protected Areas Strategy (British Columbia 1993). The government is committed to doubling the area of parks in the province to 12% by the year 2000 and have adopted the Strategy to detail how that should be achieved. The first goal of the PAS is "to protect viable, representative examples of the natural diversity of the province, representative of the ... values of each ecosection" (ibid., p. 6). The 110 ecosections of BC are contiguous natural regions that are distinct in landform, hydrology, vegetation and wildlife. They constitute the fundamental planning unit for the evaluation of new parks. To determine which values are inadequately represented in the existing park system, the PAS adopted a technique known as gap analysis (Scott et al. 1992). Through the systematic application of goals and criteria, gaps in the existing representation are identified and areas for inclusion are rated on their contribution to filling those gaps. In conjunction with mapping of ecosections, the Biogeoclimatic Ecosystem Classification system (BEC) (Pojar et al. 1987) divides the landscape based on climate, soils and vegetation into discontinuous climatic units with characteristic biota. Given that it is not possible to map and evaluate all of the ecological values of a province of 93 million ha, a quantitative criterion was established to attempt to distribute the parks system as equitably as possible among the ecosections and among the BEC units within each ecosection. Other information is also used whenever it is available. A regional plan of this magnitude could not have been attempted a few years ago with any degree of technical assistance. In other planning processes, manual drafting produced the resultant maps that lost all of the original detail. For the 3 regional plans that CORE initiated in 1992, sets of 25 paper maps at 1:250,000 were produced as a starting point for negotiations. At a cost of over $250,000 and resulting in hundreds of pounds of cumbersome paper, these maps had limited utility. By early 1993, a digital information set was compiled for Vancouver Island at half the cost. The digital information became fundamental to the subsequent negotiations. A technical working group (TWG) of government experts evaluated all prototype land use plans for the impacts on environmental, economic and social criteria. Manual map overlay required up to one month for these analyses. The regional round tables that meet every 2 to 3 weeks and demanded quicker turnaround. The Vancouver Island TWG had reduced this time to 10 days with 40 critical layers of information using traditional GIS techniques. By the end of the Cariboo/Chilcotin process, we had improved the efficiency of our methods and analysis to the point that a plan developed in the meeting could be evaluated on its technical merits overnight with PAMAP GIS. These analyses were available to negotiators the next day. Using the object-oriented tool kit described in this paper, we reduced the time for analysis sufficiently that feedback on changes to a land use plan is essentially immediate. The analysis is rapid enough that once a question is posed at the negotiation table, the answer is returned in time to affect the decision being made.
METHODOLOGYData AcquisitionSpatial data was obtained from a variety of sources, including forest cover inventory, wildlife habitat mapping, mineral yield potential mapping, sensitive fisheries mapping, cadastral data and many others (total of 40 layers). Data were originally in a variety of formats and scales. It was collated by the Forest Service Research Branch and translated into a common GIS format (PAMAP GIS) maintaining its original level of detail and resolution.In order to move beyond traditional GIS map-based query the fourty available map layers were overlaid generating a single 250 megabyte data set. Even on powerful SUN SPARC workstations the access time into this data base was formidable. As a result, we re-sampled the original 50m resolution data set to 2 km rasters to decrease the data volume to a more manageable 10 megabyte raster and data base file. This resulting file was then used for the interactive decision support planning tool. Despite the reduction in detail the file still contained 25,000 pixels of resolution.
Data StructureThe interactive planning tool was implemented using an object-oriented programming (OOP) technique. Objects were used to describe the various expressions of the original data. An object could be a simple set of procedures that manipulates source data and re- defines it in a new state, such as displaying a raster file and associated data base as a map of forest age classes. Object methods not only contain the methods to describe themselves but also define how they interact with other objects. For example, an ecosection "object" selected for evaluation will interact with other landscape data (other objects) only for the area that the ecosection defines . The same instruction can be sent to different objects and they will respond with the appropriate methods. Two landscape attributes, such as forest age class and timber volume, may require different methods to define themselves, yet when required for land use plan evaluation they can both be invoked with the same instruction, hidden from the user is the more complex nature of their expression. Sub-classes of objects can be defined by aggregating or re-describing simple objects. These sub-classes inherit their parent's data and behaviour. As a result, changes in parent objects, such as a forest cover data base will automatically re-define their children, forest species by age class.Often in traditional GIS application development, original data is modified through a series of procedures and any connection with the original data is lost. The entire procedure must be re-done if there are changes in source data. Using the OOP approach, source data was maintained in its original state (re-sampled GIS data). Other objects are merely different expressions of the same data. Any modifications of the original data would generate automatic update of other related objects. A decision support "toolbox" was used to contain the methods for describing different objects. Invoking one of the tools, for example the land use plan attribute summary, would inherently report, in a graphical form, the area of each landscape attribute in a class defined by the land use plan. The land use plan and specific attributes summarized were previously defined with other tools. Any modifications to tools related to the summary would similarly affect the land use plan attribute summary and re-generate the graphs.
Modelling EnvironmentWe used an object oriented authoring environment called Facet (Facet Decision Systems Inc., Vancouver, BC) running on a SUN SPARCstation IPX. Facet uses a spreadsheet user interface that allows for the rapid development of models that describe object behaviour, such as spatially based queries or correlations. A cell of the spreadsheet may contain data, references to external data bases, modeling expressions, maps, or another spreadsheet. Objects can be displayed as tables, graphs, or maps. As in traditional spreadsheets, changes in one spreadsheet cell causes an update of all related cells.
Model DesignThe relationships between different functions can be easily described using an object- oriented design (figure 3). The original spatial data base can be displayed by expressing the landscape by some underlying attribute, such as mule-deer winter range. Any attribute can be selected for evaluation or a number of attributes can be combined to generate a combined surface. For example a continuous surface of the landscape could be generated optimizing natural features, such as old growth and endangered species habitat, while constrained by human features, such as settlement and areas of high mineral potential.
Figure 3. Land use planning model design, showing tools and their interaction.
One advantage of this design is the completeness of the model. Any one of the underlying object definitions can be modified without having to manually navigate through a pre-determined set of steps. Those objects that are linked are automatically updated. The model also "wakes up" in its previous state, so that land use planning can continue from its latest condition.
The Land Use Planning Tool KitData Entry: A number of land use proposals were put forward by groups participating in the Cariboo/Chilcotin land use process. The land use categories of the plans were identified for future analysis reference. These plans were more useful starting points for negotiation than the status quo. To help constrain the issues for discussion and to compare the relative merits of the proposals, it was useful to subdivide the region into smaller spatial units. To assist in defining the sub-areas for evaluation, a map of the Cariboo/Chilcotin land base, partitioned by ecosections was presented. The user is prompted to select an ecosection (or the entire region) for analysis (Figure 4).
Figure 4. Ecosections of the Cariboo-Chilcotin.
Map Modification: On screen digitizing was employed to change land use category boundaries. At any time during the analysis this tool could be returned to re- define land use category boundaries based on the information gained to that point. Landscape analysis definitions made with other tools are then automatically re- evaluated. Landscape Attribute Summaries: By selecting attributes from the data base a series of pie charts were generated showing the proportional representation of those attributes by land use category. Any changes made in the map were immediately updated in the pie charts providing feedback on the impacts of those land use changes. By combining attributes where appropriate, such as caribou late and early winter habitat, multiple values could be displayed in one graph (figure 5). A numerical attribute such as cubic metres of timber volume, could also be useful by itself or it could be converted to a social value such as jobs provided.
Figure 5. Land Use planning summaries Ecosystem Representation in Parks: Interactive analysis of gaps in ecosystem representation in protected areas was facilitated by providing bar graphs (figure 6) that displayed all of the ecosystems in the area of interest and their representation in different land use categories. Any changes made to the land use plan were automatically reflected. Administrative boundaries of the Cariboo/Chilcotin are not similar to ecosection boundaries (the landscape units used to evaluate ecosystem representation). As a result, some ecosections extend beyond the boundaries of the region. To evaluate existing representativeness of ecosystems in parks, the abundance of ecosystems and their occurrence in parks in the entire ecosection including areas outside of the Cariboo/Chilcotin must be considered. If an ecosystem in an ecosection exists only outside the region then obviously there is no way to represent it within the region. Only the proportion of an ecosystem within the region needs to be represented. Planning processes in adjacent administrative areas will have the responsibility to protect their portion of the ecosystem. As well, an ecosystem may be un-represented in a land use plan for the region yet it may be represented fully in a park outside of the boundary and therefore no necessary additions are required. To capture this complexity an additional bar was added that displayed the pro-rated proportion that ideally needed to be captured in parks in the land use plan.
Figure 6. Biogeoclimatic unit summarized by land use category. Gap Display: Because the model is spatially explicit, gaps in ecosystem representation can be displayed on the map showing where those under represented ecosystems are located (figure 7). Modifications made to the land use plan could then be made to capture those elements into parks.
Figure 7. Location of un-represented ecosystems within an ecosection.
Protected Area Linkages: To prevent the isolation of ecosystems and animal populations in a fragmented parks system special management corridors between protected areas are necessary (Noss 1983, Lindenmayer et al. 1993). To have the most utilitarian linkages between reserves landscape attributes must be identified, such as undisturbed areas, that will best facilitate movement of desired elements. However, certain areas, such as human settlement, in the landscape should be avoided. To capture these concepts, a land use planning tool was designed to determine the best locations for special management corridors. To accomplish this a surface was generated using the desirable and undesirable landscape attributes. The desirable attributes where summed and the undesirable landscape attributes functioned as barriers by negating positive values in the surface. The resulting weighted surface showed the areas where potential linkages could be placed. The areas that had the highest co-occurence of positive attributes were the best candidates for corridors. To minimize the distance between protected areas a linear decay function was also incorporated (Tomlin 1990). In its pure form the function generates a surface that is weighted inversely to the distance from a protected area. However, the combination results in the map shown in figure 8, with the value of the surface decreasing from a protected area in relation to how poorly the area was weighted as a corridor candidate.
Figure 8. Surface of corridor potential between protected areas.
DISCUSSIONBy taking advantage of the inheritance provided by the OOP environment, operator processing time was minimized when evaluating a plan. Summaries of different land use plans were viewed to compare their respective impacts on different landscape attributes. Further insight into the impacts of a land use plan was gained by selecting a series of ecosections and evaluating ecosystem representation in proposed and existing protected areas. To determine where modifications should be made to the land use plan, the gap display tool was used to identify where under represented ecosystems were located. Modifications were then made to the original land use plans and immediate feedback on landscape attribute summaries and ecosystem representation was provided.By invoking the protected areas linkages tool, potential beneficial corridor location was determined. These corridors were then compared with special management zones identified in the land use plan and modifications were made to optimize those land use categories along desired landscape elements. Initial development time in the design and implementation of the land use planning tools was longer than in traditional GIS evaluation for these prototypes. However, GIS evaluation lacks the flexibility and simplicity available in this OOP design. Impacts of the changes to development plans could be quickly evaluated to direct land use decision making. By providing a library of methods, in the form of tools, this environment can be easily used in other land use planning exercises merely by changing the original data objects to another region, greatly reducing the cost of evaluation and scenario gaming in subsequent processes. In addition to requiring a great deal of time to preform, GIS evaluation usually requires a variety of software packages, including GIS, data base, and statistical packages. With the OOP design all of the system development and analysis is done within one system decreasing the amount of time needed for analysis and providing inherent documentation of the steps taken in creating the model, this allows for easier system evaluation, update, extension, and maintenance. Because the original data is retained and only modified by sets of methods to generate abstractions of the data (objects) it proves convenient for a public forum where transparency is critical. Methods for data combination and evaluation can easily be demonstrated to the public alleviating their fear of a "black box" analysis. Technically the model proved successful as an environment for exploring the impacts of land use plans and for providing information to modify plans based on optimizing desired landscape attributes. In this experience, there were limitations in bridging the gap between computer assistance and public participation driven land use decision making. The Cariboo/Chilcotin land use planning table spent the first ten months outlining the planning process and functioning of the table leaving little time for actual negotiations. There were extreme differences between interest groups in what proportion of the region should be set aside in protected areas. Because of the large scope of the land base being considered and the variety of interests at the land use planning table, a lot of discussion centred on local rather than regional issues. It was difficult to convince the participants that the process should consider ecological and objective criteria over emotions. There was a fear to committing to lines on a map as a starting point for negotiations, because it was felt that this would compromise later bargaining positions. Though the tools were available at the land use planning meetings for any interested sectors to use in evaluating or developing land use options, they were seldom used to resolve debate. Towards the end of the process, as the members were more exposed to the available technology they did become more comfortable with the tools. The negotiating table collapsed before a final consensus could be reached on land use in the Cariboo/Chilcotin. Participants differences proved to great to be reconciled. A land use plan did eventually materialize (Figure 9) and was presented to the provincial government of British Columbias Cabinet. However, it was not based on consensus, but at least it was formed from public participation.
CONCLUSIONUsing the object-oriented tool kit described in this paper, we reduced the time for analysis sufficiently that feedback on changes to a land use plan were essentially immediate. The analysis was rapid enough that once a question is posed at the negotiation table, the answer was returned in time to affect the decision being made.Despite providing the most current technology for land use planning, decision processes are still and always will be driven by people and their feelings. By designing systems that are intuitively understandable and capable of evolving with new information, public acceptance of such models will increase.
REFERENCES
Rick Page
|
|
||||||||||
|
|||||||||||
|
|||||||||||
Figure 9. Cariboo-Chilcotin CORE land use
plan.
