Steve Carver, Steve Frysinger and Rene Reitsma
This paper describes the outcomes from the initial NCGIA I-17 specialist meeting on Collaborative Spatial Decision-making (CSDM) in relation to the role of environmental modeling and GIS within this field. The meeting, held in Santa Barbara 16-19th September 1995, was attended by 32 participants, many of which had interests or experience in the environmental aspects of decision-making and support. Although much of the meeting was taken up by more general issues, such as developing theories of group decision-making, understanding decision-making processes, methods for multiple representations and design of user interfaces, many issues and problems relating directly to environmental modeling and environmental applications were raised. These included: which models are appropriate for use in CSDM environments; how they can be successfully integrated into GIS and GIS- based Spatial Decision Support Systems (SDSS); the availability of suitable data sets; and problems of multiple representations relating to the compatibility of different model outputs. The general consensus at the close of the meeting was that there was much work to be done in the general arena of CSDM of which identifying and formalising the specific role environmental modeling was a part. This paper presents some thoughts and experiences of three of the I-17 participants in environmentally based CSDM research to help illustrate some of the more pertinent issues and problems involved. Specific issues covered by the paper include: the role of simulation models in environmental decision-making and negotiation setting; the use of GIS and multi-criteria evaluation (MCE) techniques for addressing local environmental decision problems; the role of GIS and MCE in exploratory group decision- making; and examination of laboratory and real world results.
Initiative 17 of the NCGIA was recently set up to focus on research issues regarding collaborative spatial decision-making and follows two previous and related NCGIA research initiatives on Use and Value of Geographic Information in Decision-making (I-4) and Spatial Decision Support Systems (I- 6). The setting up of the current initiative by Paul Densham, Marc Armstrong and Karen Kemp had much to do with the realisation that GIS-based Spatial Decision Support Systems (SDSS) form the basis of a potentially powerful set of decision-making tools for problems involving two or more interest groups. It has long been recognised that GIS provides the user with a flexible framework for the development of SDSS (Clarke 1990, Fedra & Reitsma 1990). Work by a number of authors and research teams, including the NCGIA and the Regional Research Laboratory (RRL) Initiative in the UK, has largely identified, and in some cases solved, the main issues and problems regarding GIS-based SDSS for the single user or interest group. The application of such systems in the collaborative decision-making environment brings with it a new set of theoretical, conceptual and methodological problems, thereby opening up a whole new area of research. Particular problems relevant to collaborative spatial decision-making inevitably develop their own distinctive terminology, but include such issues as multiple representations, process intervention, stakeholder empowerment, advocacy, spatial bargaining, group decision-making process, etc. These collaborative aspects of SDSS technology are beginning to appear in a new, but growing literature in the GIS decision-making and support field. The objectives of I-17, as defined by Densham, Armstrong and Kemp, are to: 1. examine the body of theory on the design, implementation and use of computer supported cooperative work (CSCW) environments and evaluate its utility for GIS/GIA; 2. identify impediments to the development of highly interactive, group-based spatial modeling and decision-making environments; 3. develop methods for eliciting, capturing and manipulating knowledge bases that support individual and collective development of alternative solutions to spatial problems; 4. develop methods for supporting collaborative spatial decision-making (CSDM), including methods for managing spatial models; 5. extend capabilities for supporting multi-criteria decision-making in interactive, CSDM environments; and 6. characterise CSDM processes to understand how CSDM technology is and potentially can be used in various CSDM subject domains. To address these issues and problems the NCGIA has set up a working group of 32 people who are actively involved in research allied to this field. A list of I-17 participants and their affiliations is given in Appendix 1. The group, which was drawn mainly from the US but includes representatives from the UK, Switzerland, and Germany, met for the first time in Santa Barbara between September 16-19th 1995. The meeting adopted a flexible structure of small group 'break-out' discussions based around loose topic areas such as tool development, human computer interaction, institutional issues and multiple representations, followed by wider round-table discussions focusing on particular issues identified by the break-out groups. These discussions revealed the presence of two basic interest groups within the I-17 participants: those with an interest in the theoretical and conceptual implications of CSDM; and those whose interests lay more within the technical aspects of enabling the CSDM process. During the meeting it became apparent that a significant proportion of the research issues and problems being raised were being driven from an environmental applications perspective. Questions identified included: which environmental models are appropriate for use in CSDM applications; how can they best be integrated into GIS and GIS-based SDSS; how do restrictions on the availability of suitable environmental data sets affect the utility of these techniques; and how can the problems of multiple representations be solved in relation to the potential incompatibility of model outputs. These and other issues are dealt with below.
There are three basic questions regarding the role of environmental modeling in CSDM which may be seen as paralleling those affecting the integration of environmental modeling and GIS in general. These are identified above as: 1. model suitability; 2. model integration; and 3. data availability. Although these issues are generic to all environmental modeling applications in GIS, the multi-user aspects of CSDM create certain unique problems of their own, not least of which is the problem of dealing with multiple representations. This is considered to be so significant as to warrant separate discussion in section 3. The three generic questions regarding model suitability, model integration and data availability are discussed briefly in turn below.
Among the wider issues regarding the suitability of existing environmental models for integration into a GIS framework (e.g. spatial and temporal representations, scales of operation, specification of process linkages, compatibility of model structure with the adopted GIS data model, etc.) are a number of distinct issues which relate particularly to the CSDM context. Research into the role of models in group decision-making (e.g. Kraemer 1985) has shown that models helped to: 1. inform participants of problem constraints; 2. inform participants of marginal differences in policy alternatives; 3. involve participants and secure their commitment to outcomes; 4. separate groups and their positions from the policy problem; 5. set agendas by focusing participants on aspects of the problem; and 6. constrain the scope of the conflict. Kraemer's perspective is consistent with an integrative approach to negotiation in which negotiators attempt to reconcile their divergent interests and achieve joint benefit (Carnevale and Isen 1986) rather than try to maximize individual benefits per se. Integrative solutions are desirable because they contribute to long-term stability of relationships and to organizational effectiveness (Pruitt and Carnevale 1982). Walton and McKersie (1965) argue that integrative solutions to a negotiation problem can be supported by the development of a shared definition of the problem and development of shared information on the requirements of others. Modeling of a natural resource's behavior and sharing the results of such modeling among stakeholders or those participating in the actual negotiation process are expected to contribute to this process. As such, the chosen model should adequately reflect and advance the view(s) of the stakeholder(s) in the group decision-making environment. Many environmental models have been developed for non-decision-making purposes with the result that the majority of models that are used in SDSS are, at best second-hand and at worst inappropriate for the task in hand. With second-hand models there is always the risk that the inputs, mechanisms and results do not match their intended purpose and so require considerable restructuring. This is made all the more difficult in the CSDM environment since there are many stakeholders which can, in extreme circumstances, lead to as many model variations as there are stakeholders. Involving the public in CSDM introduces a further complication in regard to model suitability. This is due to the fact that mental models and even cognitive levels can vary widely between stakeholders giving rise to the need for multiple level systems that may use models and/or user interfaces of varying complexity and sophistication (Watson and Wadsworth 1994). On this point, some authors have noted in the GIS literature that confidence in the results of a GIS analysis is closely related to user understanding of the underlying data and models used (Szajna & Scamell 1993, Heywood et al. 1994). Failure to understand the models being used will inevitably lead to a lack in confidence in the results. This is absolutely critical in the context of CSDM where the attainment of a consensus view or universally accepted compromise solution relies heavily on a global understanding of the decision problem and trust in the tools used.
Looking at model integration, the problems faced in CSDM are very much the same as those faced in any other GIS application, and as such include decisions about model/GIS linkages, shared data structures, spatial and temporal components, etc. They are however, greatly complicated by the inclusion of multiple stakeholders and the need to arrive at consensus decisions. These isues are reflected in the current concern in CSDM research with the practical and theoretical aspects human computer interaction and cross model comparisons. How the stakeholder interacts with the computer and database in a CSDM environment exerts a significant influence over the success of the decision- making process. As above in regard to user confidence in model results, if the level of complexity and detail in the system and its user interface do not match the mental models of the user, then the effectiveness of the group decision-making process is likely to be impaired. Again, multiple level systems may be the answer to such problems, but there is always the risk of over simplification where complex environmental problems are concerned. In the case of cross model comparisons, in any CSDM application different models may be used by different stakeholders to forward their own cause or ideas in relation to a common theme or problem. The question therefore arises of how to ensure compatibility between models when integrating their outputs to identify areas of commonalty or compromise solutions. This aspect of model integration has far reaching implications in regard to the problems of multiple representations which are discussed in detail below. On a practical level, the time delays involved in CSDM operations resulting from sharing, executing and comparing real models can be a significant impediment to collaborative decision making, especially if the stakeholders are working from geographically separate locations. The tendency is often for one or more of the decision groups to lose patience and elect not to use models in their work, thereby throwing the whole process into disarray. Such practical issues were not discussed in much detail at the I-17 meeting, but it is suggested here that they represent strong determining factors in the success or failure of CSDM.
As with any environmental modeling process, lack of suitable data can severely restrict the usefulness of model outputs. This is particularly true in the case of integrating environmental modeling and GIS, in that many environmental models have been developed outside of a GIS framework. As they stand they work well, but problems usually arise when adapting these models to run within or along side a GIS (i.e. difficulties with data formats, file transfers, decisions about tight or loose coupling, etc.). Even when successfully reworked, many environmental models still suffer from the lack of suitable GIS data sets on which to run. Taking a simple lumped hydrological model as a case in point, it is easy enough to adapt the model spatially by placing a fine grid over the catchment and making each cell assume the form of the original lumped model. The outputs in terms of overland flow, through flow and groundwater flow from up slope cells are used as inputs to down slope cells and so on until the accumulated precipitation inputs reach the lowest point in the catchment basin as stream flow. Major problems begin to arise with this model when trying to identify the spatial variations in key model inputs and parameters such as precipitation, infiltration rates, evapo-transpiration, antecedent soil moisture conditions, etc. Any such problems experienced in this context with a single user or decision maker may immediately be increased several fold when working with two or more stakeholders because of differences in model complexities and data requirements. Taking the hydrological example further in the context of a CSDM problem concerning land use planning in a large catchment; the forestry and farming lobby are likely to be interested in soil moisture conditions and so would require a model and data sets which accurately predict this and related variables in order to make informed decisions about yields, planting and harvesting times, etc. Alternatively, water supply agencies are more likely to be interested in runoff and stream discharge when planning supply regimes, bore hole locations, river abstraction points, reservoir storage, etc. and so would require a model and data sets more focused on runoff volumes than the foresters and farmers. Availability of appropriate meta data with which to fashion and temper the use of certain data sets is also a key problem in CSDM. Such meta data may include information about spatial resolution, data sources, spatial variability of error levels with a data set, lineage, etc. In collaborative situations, stakeholders are much more inclined (particularly in the heat of the discussion) to compile data sets which, while they appear compatible, are in fact not because they are based on conflicting modeling assumptions. This is a tricky enough problem when dealing with a single user, but is multiplied further by the number of stakeholders in collaborative decision-making. Meta data is essential in a CSDM environment and must be readily accessible to all stakeholders to allow them to make intelligent decisions regarding data usage.
Representation, in the context of the current discussion, refers to manner in which the views of interest groups and individuals (i.e. the stakeholders) are presented to the decision group as a whole. Representation may be via mental (psychological, social, cultural and cognitive), visual or computational means; all or some of which may be presented via appropriate models within a GIS framework. The problem of multiple representations in CSDM refers simply to the difficulties and uncertainties of trying to represent the different interests of two or more stakeholders in a common data space. At the level of the discussion presented above regarding generic problems of model suitability, model integration and data availability, multiple representation present several difficulties concerning model choice, cognitive ability, mental models, human computer interaction, compatibility of model outputs and variable data requirements. Perhaps the main area of difficulty in relation to multiple representations is that posed by the need to cope with differences in the compatibility of different model outputs. Although no real consensus of opinion was reached at the I-17 meeting, it appears to the authors that there are perhaps two levels to the multiple representations problem concerning model compatibility: those where the objectives of the different stakeholders differ so much as to require recourse to different models; and those where the objectives of different stakeholders are similar enough to use the same model but perhaps using different data and levels of emphasis. These are discussed below.
One example used to illustrate this and facilitate further discussion at the I-17 meeting was that of river pollution. The two different stakeholders in this example were presented as a chemical company wishing to discharge effluent into the river at an upstream location and a city water management authority at a downstream location affected by the planned discharges. In this case the aim of the consensus building problem is defining a mutually acceptable level of discharge from the chemical plant into the river (i.e. what contaminants and how much?). At first glance, the objectives of the two stakeholders seem entirely different. The chemical company wants the cheapest means of disposing of its process effluents (i.e. river discharge), whilst the city water managers want to minimise health risks and maximise water quality by keeping discharges to a minimum. However, seemingly conflicting objectives or 'positions' can often be abstracted into a single type of 'interest' (Fisher & Ury 1981). In this case, both the city and the plant have an interest in the 'flow' of the river. Note that this abstraction of positions into interests does not imply that there is no conflict. On the contrary, the uses of river flow as transporter of drinking water and of chemical effluent are, at least at certain levels, mutually exclusive. What is important here, however, is that two seemingly different perspectives can be abstracted into a single set of interests. As a consequence, a single model or set of models could be used to represent the environmental resource and its policies for utilization, thereby creating a common platform for discussion and negotiation. Obviously the city and the plant management would use different utility functions to evaluate the acceptability of proposed policies, but at least a common modeling platform can be formulated.
Far more complex situations arise when positions cannot be reconciled into one or more common interests. This would happen, for instance, when the downstream city is planning the construction of a reservoir for the storage of drinking water, in between the plant and the city. Although the city still has an interest in the flow of the river in that it replenishes the reservoir and as such does not want the flow to carry the upstream plant's effluent, it now has an additional interest in that pollution will accumulate in the reservoir. Where in the previous example one could conceive of coordinating effluent release schedules with water intake schedules such that both the city and the plant would satisfy their objectives (never mind the cities farther downstream or the aquatic life in the river), now such coordination becomes almost impossible and entirely new policies; e.g. storing the effluent and shipping it periodically, may have to be designed. As a consequence, the same river flow and water quality model will most likely not suffice anymore for representing the salient features of the decision problem. Although this in itself does not prevent meaningful communication between the partners in the negotiation process, it does imply the integration of other, often vary different models. How to develop CSDM environments and models which are flexible enough to accommodate such dynamic integration's of models is a matter for further discussion and research.
Multi-criteria evaluation (MCE) modeling techniques are a good example of how a single model can adequately cope with multiple representations. There are numerous examples in the literature of MCE models being integrated with GIS to solve a number of site search and suitability analysis problems, including regional land use planning in the Netherlands (Janssen & Rietvelt 1990), nuclear waste disposal in the UK (Carver 1991) and industrial location in developing countries (Eastman et al. 1993). In the nuclear waste example the main stakeholders involved are the nuclear industry, the general public and the environmental lobby; each of which can be assumed to have distinctive objectives in mind concerning the search for a suitable site for a nuclear waste disposal facility. The basic objectives of each stakeholder can be summarised as: nuclear industry (minimise costs and maximise safety); general public (maximise distance and minimise health risks); and environmental lobby (minimise environmental impact and minimise health risks). However, with the ultimate and basic objective of all stakeholders being the same (i.e. the identification of a mutually acceptable site) the decision problem can effectively be addressed by all parties using a single MCE siting model with the differences in the stakeholder's specific objectives being expressed through use of different data sets and weighting schemes. The result from such an approach is three different surfaces describing the suitability of different areas of the UK for nuclear waste disposal according to the specific objectives of the stakeholders. Since these have been drawn using the same model, the surfaces are immediately compatible and so can easily be compared to identify commonalties and individual compromise solutions via simple map overlay techniques.
The authors have derived much of the above discussion from direct and shared practical experience in applying GIS-based modeling techniques to both single user and group spatial decision making problems. One of the key problem areas is learning to cope with multiple representations in CSDM. The addition has been made to the single/multiple objective, single/multiple criteria classification of spatial decision problems provided by Eastman et al (1993) of that difficult group of multiple objective, multiple model type decision problems where the compatibility of different environmental models can create such a problem for model comparison and consensus building. Whilst it is hoped that this discussion helps throw some light onto the problems of environmental modeling in CSDM, a number of unresolved issues and new ideas require further attention. These include the role of simulation modeling in negotiation setting and consensus building, exploratory decision support systems and the role of the Internet in creating open systems for mass involvement CSDM.
One of the main problem areas highlighted above is that of model compatibility in multiple objective, multiple model decision problems. It is suggested here that one possible means of coping with the incompatibility problem when trying to arrive at mutually acceptable solutions is to use an automated simulation modeling approach. Looking back at the river pollution problem, the difficulty of comparing the model results from the different models employed by the two stakeholders involved may be overcome using a modeling environment to simulate incremental changes in the type and volume of effluents discharged by the chemical company. By re-modeling the environmental and economic aspects of the problem from the point of view of both stakeholders, it should be possible to identify the point when the objectives of the chemical company and the city water managers begin to coincide. Obviously some compromise will inevitably be required from both parties if a mutually acceptable solution is to be found, but at least a simulation approach of this kind may provide the necessary information to support an amicable group decision. This approach relies very heavily on mutual trust and understanding between stakeholders regarding each other's models and the automated simulation model provided by the CSDM. Experience would suggest however, that when two parties with different models try to collaborate, they often cannot convince each other to use the other's models and they may not trust anyone else's models either. In this case, perhaps the only way forward is to use all of the stakeholder's models and craft the CSDM in such a way that side-by-side comparisons are possible to facilitate the negotiation. This can be termed simple negotiation setting by provision of on-screen comparisons of model results.
Evidence cited in the decision-making, operations research and IT literature points strongly towards the view that effective group decision making requires full hands-on interaction with the stakeholders. It is highly unlikely that the stakeholders in most CSDM problems will be fully conversant with the models and GIS being used. If CSDM environments are to prove truly interactive, then the models and data used must therefore be accessible to the users via appropriately pitched Graphical User Interfaces (GUI). In an ideal world, such GUIs to CSDM tools would be both simple and structured whilst being sophisticated and flexible all at the same time. Obviously these criteria are to some extent mutually exclusive. However, some experimental systems have been designed which use third party authoring software to design and build GUIs onto the front of environmental models and GIS packages. This provides a product that any reasonably intelligent individual with a basic knowledge of computers could happily use. For example, open architecture systems are widely used to 'shield' the user from the complexities of using environmental models and GIS for decision-making (e.g. Frysinger et al. 1993, Frysinger 1995). Watson and Wadsworth (1994) provide an example of a complex group of three inter- linked models for predicting the effect of land use change on the ecology, hydrology and rural economy of the Tyne Basin in NE England. In this example the decision support GUI shields the user from the complexities of the GRASS based GIS engine and the models themselves, but at three different cognitive levels: the academic with a direct knowledge of problem; the practising professional working in an allied field (e.g. agricultural economics, nature conservation, water supply, etc.); and the educated lay-person where no in-depth knowledge of the problem is assumed. Despite the multi-level approach to the GUI, the system is still based rigidly around the three models and their required input data sets. An example of a much more flexible modeling environment is provided by Heywood and Tomlinson (1995) who have designed a flexible 'meta modeling' interface for use in environmental decision making. This allows much greater flexibility in model choice for a specific decision problem by giving the option to make an informed choice between several models. The GUI is set up such as to allow users to construct their own decision trees using a simple and intuitive drag-and-drop approach to building decision paths. The resulting CSDM environment is, to a certain extent, both structured and flexible at the same time in providing the users with an interface, data sets and models whilst allowing relatively free rein in constructing their individual decision paths in an exploratory fashion. This ability to 'explore' the effects of spatial decisions in the virtual world of the spatial database is a key aspect of GIS technology. There is nothing new about this; GIS practitioners have been carrying out 'what if?' type analyses for as long as there have been GIS practitioners. What is relatively new is the idea that once the technicalities of GIS and environmental modeling are protected from the user in the kind of easy-to-use GUI described above, then true CSDM through public exploration of the effects and implications of spatial decisions normally made behind closed doors in government and company offices becomes a distinct possibility. Coupled with the information explosion and the widespread popularity of the Internet, then true public involvement in CSDM problems where the 'public' is a major stakeholder may become tomorrow's reality. This may ultimately lead toward a new breed of spatial decision support systems focused on the interactive exploration of spatial ideas rather than using GIS in a more traditionalist constructive or hypothesis testing role. The paper by Heywood and Carver (1994) explores this concept under the banner of Idea Generation Systems (IGS) using the example of a family working together to identify an area in which to buy a new house based on a multiple objective, single model MCE approach. Here spatial data relevant to the problem are combined and 'explored', first as individuals (daughter, son, mother, father, grand-mother, etc.) and then as a single family unit in order to create spatial constructions of 'ideas' regarding where might be a nice area to live in and with which everyone agrees. As this is a siting problem, a single MCE model can be applied across the board without any compatibility related problems, but a similar exploratory IGS approach could equally be adopted to multiple objective, multiple model decision problems via the simulation modeling approach described above.
As pointed out by various researchers of how models are used in environmental management (Kraemer 1985, Dutton & Kraemer 1985, Reitsma 1996), the physical and or spatial aspects of environmental decision-making and negotiation often only represent a small portion of the overall complexity of the problem. For instance, in a study of negotiations of the Colorado River Annual Operating Plan (AOP), Reitsma (1996) points out that whereas physical modeling plays an important role as part of the AOP process, the overwhelming majority of issues of negotiation had nothing or little to do with the physical aspects of the river. Instead, they addressed issues such as the legality of operational plans or future precedence that might be established as a consequence of this year's operational plans. Although from an environmental modeling standpoint these aspects may be rather irrelevant, from the point of view of collaborative decision-making these aspects and how to integrate them with environmental models are of the utmost importance. At this point, it is unclear how traditional environmental modeling and issues of strategic, tactical and legal decision-making can be integrated into a single CSDM environment. Although candidate theoretical frameworks such as coordination theory (Crowston & Mallone 1990) exist, integration of these very different aspects of collaborative problems into a consistent modeling framework remains a big challenge.
Heywood and Carver (1994) extrapolate their work on IGS in the form of a hypothetical (if rather utopian) discussion of the potential for restructuring democracy through direct public involvement in policy formulation and decision-making via CSDM tools on the Internet. This may be considered the extreme end of the CSDM vision, whereby everyone with access to the Internet can be involved in providing stakeholder representatives with direct feedback regarding decisions of local, regional, national or global importance. Charges of elitism can easily be levelled at such a suggestion given the current state of development in the Internet. Such a model of democratic decision-making relies heavily on access to the technology and so despite the meteoric increase in Internet connections, there is still a danger of creating an 'information underclass' of people who, for whatever reason, have no connection to this resource. However, if the Internet continues to develop at the current rate, and providing it does not suddenly self destruct, it may be safely assumed that nearly 100% of the population will have direct home or local public access in the none too distant future. As it is at present, the Internet is insufficiently well developed to allow such radical changes in the way decisions are made. Perhaps a more realistic view of the present and future role of the Internet in CSDM is that of a simple information service regarding important spatial and/or environmental decisions roughly in the mould of suggestions for Local Agenda 21 which came out of the Earth Summit held in Rio de Janeiro in June 1992. Local examples of such applications do exist and they increasingly provide a valuable means of providing information and soliciting feedback from interested parties and individuals. Future roles for Internet based tools in wider public information, simulation, decision support, consensus building, negotiation and decision-making are, however, still a possibility given appropriate political will.
This paper has provided an insight into the current status of research into CSDM. This discussion arises out of the past experience of the authors in this field and from discussions at the recent NCGIA I-17 meeting on this subject. The paper identifies some of the more pertinent issues relating to environmental modeling in GIS-based CSDM systems, including the standard issues of model suitability, model integration and data availability, but also the more specialist topics such as multiple representations and model compatibility. In attempting to address those I-17 objectives relating to identifying barriers to developing working CSDM systems, developing new methodologies and identify potential areas of application, a personal view of work ongoing and potential further developments of the concepts and methods of CSDM is provided. However, whatever the views of the authors, it is clear that a significant number of difficult problems and exciting applications areas exist for those involved in CSDM research. It is hoped that I-17 will help produce workable solutions and applications. Clearly, environmental modeling and GIS-based decision support systems are mutually important research areas which hold great potential for the development of powerful CSDM tools. As global population increases so inevitably will the conflict for the earth's finite natural resources. Discoveries of new resources have not kept pace with development and so the onus falls squarely on the sustainable use of existing known resources. With demand rising relative to supply, CSDM tools will be increasing required to help solve conflicts of interest in resource exploitation and development. Geography dictates through the unequal distribution of natural resources that local, regional, national, international and global disputes over these resources and their use will occur. Spatial models and GIS will increasingly become a key technology in supporting decisions about resource management. The need for robust and sophisticated CSDM tools for aiding these important decisions is both urgent and obvious. The final comments in the above section regarding the potential future role of the Internet in CSDM may seem naive in respect to current political systems and their control over the reality of democratic ideals, but they are intended to provoke discussion and thought as to the potential of this fledgling media. In the near future the Internet may provide global access to a massive array of spatially referenced data sets and more importantly the tools to analyse and use them. The potential, at least, for wider public involvement and the true democratisation of the decision making process is just on the horizon.
The authors would like to acknowledge the support of the NCGIA in providing financial assistance to attend the I-17 specialist meeting, Santa Barbara, September 16-19, 1995.
Marc Armstrong, University of Iowa
David Bennett, Southern Illinois University
Steve Carver, University of Leeds, UK
David Coleman, University of New Brunswick
Helen Couclelis, University of California, Santa Barbara
Joseph Delotto, NYNEX Science & Technology Inc
Paul Densham, University College London, UK
Brenda Faber, CIESIN/TERRA, Fort Collins
Joseph Ferreira, Massachusetts Institute of Technology
Steve Frysinger, James Madison University
Francois Golay, Swiss Fedral Institute of Technology, CH
Thomas Gordon, GMD, KIT-KI, Germany
Jon Gottsegen, University of California, Santa Barbara
Brit Harris, University of Pennsylvania
Lew Hopkins, University of Illinois
Piotr Jankowski, University of Idaho
Douglas Johnston, University of Illinois
Rachel Jones, Loughborough University, UK
Karen Kemp, University of California, Santa Barbara
David Lemberg, University of California, Santa Barbara
Seymour Mandelbaum, University of Pennsylvania
Timothy Nyerges, University of Washington
Alex Pang, University of California, Santa Cruz
Dimitris Papadias, University of Maine
Thomas Pederson, University of Pennsylvania
James Proctor, University of California, Santa Barbara
Rene Reitsma, University of Colorado
Michael Schiffer, Massachusetts Institute of Technology
Susan Suchocki, Claremont Graduate School
Steve Tomlinson, Manchester Metropolitan University, UK
William Wallace, CIESIN/TERRA, Fort Collins
Jeff Wang, US Environmental Protection Agency
Carnevale, P.J.D., and Isen, A.M. (1986). The Influence of Positive Affect and Visual Access on the Discovery of Integrative Solutions in Bilateral Negotiation. Organizational Behavior and Human Decision Processes 37: 1-13.
Carver, S.J. (1991) Integrating Multicriteria Evaluation with GIS. International Journal of Geographical Information Systems 5(3): 321-339 .
Clarke, M. (1990) Geographical Information Systems and Model Based Analysis: Towards Effective Decision Support Systems. H.J.Scholten and J.C.H.Stillwell (eds) Geographical Information Systems for Urban and Regional Planning. Kluwer Academic Publishers: 165-175 .
Dutton, W., and Kraemer, K.L. (1985) Modeling as Negotiation: The Political Dynamics of Computer Models in the Policy Process. Norwood, NJ: Ablex.
Eastman, R., Kyem, P., Toledano, J., and Jin, W. (1993) GIS and Decision Making: Explorations in Geographic Information Systems Technology 4. Switzerland: United Nations Institute for Training and Technology.
Fedra, K., and Reitsma, R.F. (1990) Decision Support and Geographical Information Systems. H.J.Scholten and J.C.H.Stillwell (eds) Geographical Information Systems for Urban and Regional Planning. Kluwer Academic Publishers: 177-188.
Fisher, R., and Ury, W. (1981) Getting to Yes; Negotiating Agreement Without Giving In. New York: Houghton Mifflin.
Frysinger, S.P., Copperman, D.A., and Levantino, J.P. (1993) Environmental Decision Support Systems: An Open Architecture Integrating Modeling and GIS. Proceedings of the 2nd International Conference on Integrating GIS and Environmental Modeling. NCGIA: September 1993.
Frysinger, S.P. (1995) An Open Architecture for Environmental Decision Support. International Journal of Microcomputers in Civil Engineering 10(2): 123-130.
Heywood, D.I., and Carver, S.J (1994) Decision Support or Idea Generation: The Role for GIS in Policy Formulation. Proceedings Symposium für Angewante Geographische Informationsverarbeitung (AGIT'94) Salzburg: 259-266.
Heywood, D.I., Oliver, J., and Tomlinson, S.J. (1994) Building an Exploratory Multi-criteria Modelling Environment for Spatial Decision Support. P.Fisher (ed) Innovations in GIS 2. London: Taylor & Francis.
Janssen, R., and Rietvelt, P. (1990) Multi-criteria Analysis and GIS: An Application to Agricultural Land Use in the Netherlands. H.J.Scholten and J.C.H.Stillwell (eds) Geographical Information Systems for Urban and Regional Planning. Kluwer Academic Publishers.
Kraemer, K.L. (1985) Modeling as Negotiating: The Political Dynamics of Computer Models in Policy Making Advances in Information Processing in Organizations 2: 275-307.
Malone, T.W., and Crowston, K (1990) What is Coordination Theory and How Can it Help Design Cooperative Work Systems. ACM CSCW 90 Proceedings: 375-388.
Pruitt, D.G., and Carnevale, P.J.D. (1982) The Development of Integrative Agreements. V. Derlega and J. Grzelak (eds) Cooperation and Helping Behavior: Theories and Research. New York: Academic Press.
Reitsma, R.F. (1996) Structure and Support of Water Resources Management and Decision Making. Journal of Hydrology (in press).
Reitsma, R.F., Zigurs, I., Lewis, C., Sloane, A.M., and Wilson, E.V. (1996) Experiment with Simulation Models in Water Resources Negotiations. ASCE Journal of Water Resources Planning and Management (in press).
Szajna, B., and Scamell, R. (1993) The Effects of Information System User Expectations on Their Performance and Perceptions. MIS Quarterly: 493-516
Walton, R.E., and McKersie, R.B. (1965) A Behavioral Theory of Labor Negotiations. New York: McGraw-Hill.
Watson, P., and Wadsworth, R. (1994) The Construction of a Spatial Decision Support System for Land Use Planning. Proceedings of the 2nd GIS Research UK Conference. Leicester: 337-348.
Stephen J Carver, Dr
School of Geography University of Leeds, LS2 9JT, UK
Tel: +44 113 2333318 Fax: +44 113 2333308
Email: steve@geog.leeds.ac.uk
Steven P Frysinger, Dr
College of Integrated Science and Technology Integrated Science and Technology Program, James Madison University Harrisonburg VA 22807, USA
Tel: +1 540 5682710 Fax: +1 540 5682761
Email: frysinsp@jmu.edu
Rene F Reitsma, Dr
CADSWES University of Colorado Campus Box 421, Boulder CO 80309, USA
Tel: +1 303 4924828 Fax: +1 303 4921347
Email: reitsma@cadswes.colorado.edu