UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

An approach to the management of groundwater pollution Walker, Daniel Arthur 1996

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_1996-148483.pdf [ 14.64MB ]
Metadata
JSON: 831-1.0053327.json
JSON-LD: 831-1.0053327-ld.json
RDF/XML (Pretty): 831-1.0053327-rdf.xml
RDF/JSON: 831-1.0053327-rdf.json
Turtle: 831-1.0053327-turtle.txt
N-Triples: 831-1.0053327-rdf-ntriples.txt
Original Record: 831-1.0053327-source.json
Full Text
831-1.0053327-fulltext.txt
Citation
831-1.0053327.ris

Full Text

AN APPROACH TO THE MANAGEMENT OF GROUNDWATER POLLUTION by DANIEL ARTHUR WALKER B.Sc. (High Honours), University of Regina, 1982 M.Sc. (Applied), McGill University, 1985 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES (Department of Earth and Ocean Sciences) We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA June, 1996 © Daniel Arthur Walker, 1996 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of E d f ^ V x a ^ o O Q b * > O t K SciQflCej The University of British Columbia I Vancouver, Canada ! DE-6 (2/88) 11 Abstract This thesis considers the management of groundwater pollution as a societal and technical problem. It consists of four elements: theoretical foundation, conceptual model, management strategies, and compound model. The theoretical foundation recognizes two distinct approaches to management: the hard and soft paradigms. The hard paradigm deals with hard problems, characterized by well-defined boundaries, goals, and alternatives; quantifiable uncertainty; and a unilateral decision-maker. A hard problem can be addressed with algorithmic logic and linear procedure. The hard paradigm employs condensed conceptual models, and relies on transform models as predictive tools. (A conceptual model synthesizes observations; a transform model maps an input to an output.) The soft paradigm deals with soft problems, characterized by ambiguous boundaries, goals, and alternatives; non-quantifiable uncertainty; and multiple decision-makers. A soft problem requires argumentation and iterative procedure. The soft paradigm relies on detailed conceptual models, and employs transform models as heuristic tools. The two paradigms can be reconciled through the concept of soft/hard complementarity which views a problem as a soft problem in the overall sense, with embedded hard sub-problems. The conceptual model contains two major systems: the pollution system, consisting of groundwater flow and pollutant transport subsystems; and the management system, consisting of technical and decision subsystems. The emphasis is on the decision subsystem, which is described in terms of rules of governance; multiple issues; multiple stakeholders; and various decision processes. The management strategies include both decision and technical strategies. The decision strategies consist of three subsets: strategies for iterative decision-making; strategies for integration of the decision system; and strategies for emphasizing negotiation. The technical strategies deal with iterative technique; integration of the technical system; and design for mitigation. The compound model is called S A M (Simulated Aquifer Management). It consists of a set of transform models, joined by linkages which are either algorithmic or which require the modeler's subjective judgement. The technical system is represented by models of flow, transport, optimization of pumping rates, and costing of technical options. The decision system is represented a game theory model and a role-playing exercise. S A M is used as a heuristic device, and as illustration of both the management strategies and soft/hard complementarity. i i i Table of Contents Abstract ii Table of Contents iii List of Tables vi List of Figures vii List of Boxes viii Acknowledgements ix Dedication x Quotation xi 1. Introduction 1 1.1 The problem of groundwater pollution management 1 1.2 Goal, scope, and intended audience 3 1.3 Preview of thesis 5 2. Soft and hard paradigms: a review and synthesis 9 2.1 Contrasts between the hard and soft paradigms 12 2.1.1 Philosophical concepts 13 2.1.2 Hard problems and soft problems 16 2.1.2.1 Example of a soft problem 18 2.1.3 General methodological concepts 19 2.2 Modeling 24 2.2.1 Conceptual models 25 2.2.2 Algorithmic models 27 2.2.3 Analog models 30 2.2.4 Compound models 31 2.2.5 Virtues and pitfalls in modeling 32 2.2.6 Hard and soft paradigm views of modeling 34 2.3 Complementarity of the soft and hard paradigms 38 2.4 Relevant hydrogeological literature on the management of groundwater pollution 40 2.5 Summary 44 3. A conceptual model of groundwater pollution management 46 3.1 Pollution system 48 3.2 Technical system 49 3.3 Decision system 52 3.3.1 Rules of governance 53 3.3.2 Stakeholders 55 3.3.3 Issues 59 3.3.4 Decision-making 60 3.3.4.1 Negotiation 61 3.3.4.2 Unilateral regulatory decision-making 70 3.3.4.3 Judicial decision-making 71 3.3.4.4 Linear and iterative decision-making 74 3.4 Complexity, uncertainty, and integration 75 3.5 Summary and discussion 77 3.5.1 Summary 78 3.5.2 Relation to the soft and hard paradigms 79 iv 4. Strategies for groundwater pollution management 81 4.1 General concepts 81 4.2 What is good management? 83 4.3 Decision strategies 87 4.3.1 Iteration strategies 87 4.3.1.1 Iterative decision-making 88 4.3.1.2 Contingent decision 91 4.3.1.3 Iterative liability reduction 93 4.3.2 Integration strategies 95 4.3.2.1 Process integration 96 4.3.2.2 Inclusive participation 97 4.3.2.3 Geographic integration 100 4.3.2.4 Example to illustrate importance of integration strategies 102 4.3.3 Negotiation strategies 103 4.3.3.1 Primary negotiation 104 4.3.3.2 Principled negotiation 106 4.3.3.3 Balanced negotiating power 107 4.3.3.4 Mediation 108 4.3.4 Summary of decision strategies 109 4.4 Technical strategies 110 4.4.1 Iterative technique 110 4.4.2 Technical integration 114 4.4.3 Mitigative technique 117 4.5 Summary and discussion 119 4.5.1 Summary 119 4.5.2 The management strategies and soft paradigm thinking 123 5. SAM: a compound model of groundwater pollution management 125 5.1 Overview of S A M 125 5.2 Pollution model 130 5.3 Technical model 132 5.4 Decision model 137 5.4.1 COOP (conflict analysis model) 138 5.4.2 P A Q M A N (role-playing model) 145 5.4.3 Simplifications 146 5.5 Summary and discussion 148 5.5.1 Summary 148 5.5.2 S A M and the management strategies 149 5.5.3 S A M and soft/hard complementarity 152 V 6. A run of SAM 154 6.1 Hypothetical case 155 6.1.1 Pollution system 155 6.1.2 Management system 157 6.1.2.1 Decision system 157 6.1.2.2 Technical system : 159 6.2 S A M using COOP as the decision model 161 6.2.1 Round 1 161 6.2.1.1 Pollution model 161 6.2.1.2 Technical model 166 6.2.1.3 Decision model (COOP) 170 6.2.1.3.1 Preference Variation 1 172 6.2.1.3.2 Preference Variation 2 175 6.2.2 Round 2 178 6.2.2.1 Pollution model 179 6.2.2.2 Technical model 181 6.2.2.3 Decision model (COOP) 184 6.3 S A M using P A Q M A N as the decision model 188 6.3.1 Round 1 198 6.3.2 Round 2 201 6.4 Summary 204 7. Summary and discussion 207 7.1 Summary 207 7.1.1 The theoretical foundation: hard and soft paradigms 209 7.1.2 The conceptual model 210 7.1.3 The management strategies 211 7.1.4 The compound model S A M 213 7.2 Development of the thesis 215 7.3 Contributions and directions for future research 219 7.3.1 Soft paradigm thinking 219 7.3.2 Management strategies 220 7.3.3 Compound model S A M 222 7.3.4 Empirical study of groundwater pollution management 227 7.4 Concluding thought 228 Glossary 229 References 234 Appendix 1. Notes on principled negotiation 242 Appendix 2. Notes on game theory and conflict analysis 250 V I List of Tables Table 4-1. The management strategies 120 Table 5-1. The sub-models of S A M 129 Table 5-2. Simple example of the use of preference masks to develop a preference vector for one stakeholder 141 Table 5-3. List of the proposed management strategies 149 Table 6-1. Summary of the management strategies 157 Table 6-2. Locations, installation dates, and extraction rates of water wells 165 Table 6-3. Round 1 technical options 168 Table 6-4. Cost estimates for Round 1 technical options 170 Table 6-5. The stakeholders' options in the Round 1 negotiations 171 Table 6-6. Preference masks for each stakeholder, for Preference Variation 1 in the Round 1 negotiations 173 Table 6-7. Number of outcomes in each category for Preference Variation 1 in the Round 1 negotiations 174 Table 6-8. The set of Pareto-optimal equilibriums (possible outcomes) for Preference Variation 1 in the Round 1 negotiations 174 Table 6-9. Frequency of technical option acceptance for the 24 Pareto-optimal cooperative equilibriums (possible agreements) for Preference Variation 1 in the Round 1 negotiations 174 Table 6-10. Preference masks for each stakeholder, for Preference Variation 2 in the Round 1 negotiations 176 Table 6-11. Number of outcomes in each category, for Preference Variations 1 and 2 in the Round 1 negotiations 177 Table 6-12. The set of Pareto-optimal equilibriums (possible outcomes) for Preference Variation 2 in the Round 1 negotiations 177 Table 6-13. Frequency of option acceptance in Pareto-optimal equilibriums, for Preference Variations 1 and 2 in the Round 1 negotiations 177 Table 6-14. The randomly-selected agreement from Round 1, as a basis for proceeding to Round 2.... 179 Table 6-15. Round 2 technical options 183 Table 6-16. Cost estimates for Round 2 technical options 184 Table 6-17. The stakeholders' options in the Round 2 negotiations 184 Table 6-18. Preference masks for each stakeholder, for the Round 2 negotiations 186 Table 6-19. Number of outcomes in each category in the Round 2 negotiations 186 Table 6-20. The set of Pareto-optimal equilibriums (possible outcomes) in the Round 2 negotiations.. 187 Table 6-21. Frequency of technical option acceptance in Pareto-optimal cooperative equilibriums (possible agreements) in the Round 2 negotiations 187 Table 6-22. Decisions on the P A Q M A N Round 1 technical options 198 Table 6-23. Comparison between P A Q M A N and COOP Round 2 technical options 201 Table 6-24. Decisions on the P A Q M A N Round 2 technical options 203 Table 6-25. Illustration of the management strategies by the run of S A M described in this chapter 206 Table 7-1. Contrasts between the hard and soft paradigms 209 Table 7-2. Summary of the conceptual model 210 Table 7-3. Summary of the decision strategies 212 Table 7-4. Summary of the technical strategies 213 List of Figures Figure 1-1. A typical case of groundwater pollution 2 Figure 2-1. Linear solution procedure in the hard paradigm 21 Figure 2-2. The experience-action cycle 21 Figure 2-3. The relationship among observations, model, and purpose 24 Figure 2-4. Classification of models used in this thesis 25 Figure 2-5. The relationship between conceptual and transform models 27 Figure 2-6. A compound model ; 31 Figure 2-7. Distinctions between simulation-optimization and hydrogeological decision analysis 43 Figure 3-1. Systems relationships in the general conceptual model of groundwater pollution management 47 Figure 3-2. The pollution system 48 Figure 3-3. The technical system 50 Figure 3-4. The decision system 52 Figure 4-1. Management strategies 82 Figure 4-2. The decision strategies 87 Figure 4-3. The technical strategies 110 Figure 5-1. Comparison of S A M with the conceptual model of Chapter 3 126 Figure 5-2. The flow of S A M 127 Figure 5-3. Sub-models and linkages in S A M 128 Figure 5-4. Representation of the simplifications imposed by M O D F L O W , M O D P A T H , and GSLIB on the conceptual model of the pollution system 132 Figure 5-5. Flowchart of basic conflict analysis 139 Figure 5-6. Example: comparison of preference mask with feasible outcomes 140 Figure 5-7. Flowchart of cooperative conflict analysis 142 Figure 6-1. Location map 155 Figure 6-2. Reality model of the spatial distribution of hydraulic conductivity in the Kaldor Aquifer.. 163 Figure 6-3. Steady-state hydraulic head surface in the Kaldor Aquifer under long-term pumping 164 Figure 6-4. Steady-state hydraulic head surface and chloride plume in the vicinity of the Midnight Site, under long-term pumping 165 Figure 6-5. Monitoring well locations in the vicinity of the Midnight site 169 Figure 6-6. Comparison of reality and perception models for Rounds 1 and 2, in the vicinity of the Midnight Site 179 Figure 7-1. The elements of the thesis 208 Figure 7-2. The elements of this thesis, as developed and as presented 216 Figure 7-3. Transition in the development of this thesis, from thinking in terms of the hard paradigm to thinking in terms of the soft paradigm and soft/hard complementarity 218 Figure 7-4. Possible extension to S A M 224 List of Boxes Box 6-1. Game Manager's Manual P A Q M A N Version 1.0 188 Box 6-2. General Instructions to Players P A Q M A N Version 1.0 189 Box 6-3. Confidential Instructions to Assistant Director of the Pollution Response Panel 199 Box 6-4. Confidential Instructions to District Officer for the Remediation Enforcement Group 199 Box 6-5. Confidential Instructions to Water Suppler Manager for the Village of Cantorbury 200 Box 6-6. Round 2 Options 202 Box 6-7. PRP's Confidential Information for Round 2 203 Box 6-8. R E G Round 2 Confidential Instructions 203 Box 6-9. VIC's Round 2 Confidential Instructions 203 ix Acknowledgements The advisor always gets top billing in the acknowledgements, and this thesis is no exception. In this case, however, matters are complicated because I have had the benefit of three advisors over the course of my studies. I began under Allan Freeze, and finished under Roger Beckie following Allan's retirement from the university. Tony Dorcey acted as co-advisor for a spell in the middle. All three remained to the bitter end, as members of my supervisory committee. I thank them for their guidance, insights, criticism, and patience. An additional thanks is due to Roger for reading so many drafts, and for putting up with my whinging in the final days. I am also grateful to the two other members of my supervisory committee: Jim Atwater and Leslie Smith. I have benefited greatly from the collegiality of my fellow students in the Groundwater Group. I won't attempt a comprehensive list, for fear of omitting anyone. But two stand out as inspiring examples of unselfish cooperation: Bruce James and Tom Clemo. I remember fondly the many assignments in competitive probability which I completed with Tom's assistance. Joanna Zilsel, Colleen Burke, and Sylvia Walker each took the time to critique earlier drafts, on a meticulous line-by-line basis. I am indebted to them for the insights and perspectives which they have lent to this work. Keith Everard, Mark Leir, and Jason Smolensky gave up a day of their time to act out the roles of stakeholders in a role-playing model of a groundwater pollution problem. I am grateful for their time and their critical analysis of the role-playing model. This thesis is very much a product of critical discussions with many people. Garth van der Kamp suggested the idea of studying groundwater pollution over the geographic scope of a small city, as opposed to a single contaminated site. A number of people involved in a real-world groundwater pollution problem gave freely of their time in interviews during the early stages of this thesis; I omit their names for reasons of confidentiality. Rosemary Knight and Derek Atkins provided advice in the early days. Elizabeth Hill inspired me with her progressive approach to real-world groundwater pollution problems. Marina Pantazidou gave me an opportunity to expose my work to critical review by an audience at Carnegie Mellon University. John Wilson and Paul Hofman provided key input at a critical point. I offer a collective thanks to the many helpful people whom I have not named here. The responsibility for errors, omissions, and frivolous comments rests entirely with myself. Finally, I thank Sam, Joanne, and Colleen for their patience and support during the many iterations of this thesis. Dedication To Samantha and to Colleen xi ...In that Empire, the craft of Cartography attained such Perfection that the Map of a Single province covered the space of an entire City, and the Map of the Empire itself an entire Province. In the course of Time, these Extensive maps were found somehow wanting, and so the College of Cartographers evolved a Map of the Empire that was of the same Scale as the Empire and that coincided with it point for point. Less attentive to the Study of Cartography, succeeding Generations came to judge a map of such Magnitude cumbersome, and, not without Irreverence, they abandoned it to the Rigors of sun and Rain. In the western Deserts, tattered Fragments of the Map are still to be found, Sheltering an occasional Beast or beggar; in the whole Nation, no other relic is left of the Discipline of Geography. Jorge Luis Borges and Adolfo Bioy Casares The map is not the territory. attributed to Alfred Korzybski by Gregory Bateson 1 1. Introduction 1.1 The problem of groundwater pollution management The societal and technical response to groundwater pollution is a controversial issue. The controversy is largely driven by conflicts over distribution of the direct and indirect economic costs among various sectors of society, and by uncertainty about the impact on human and ecosystem health. A strong argument can be made for making a determined response to groundwater pollution, on the basis of a number of ethical principles. It is a localized problem which affects some communities more than others (Szasz, 1994); the principle of locational equity argues that society as a whole should respond to the problem. It is a long-term problem which is passed from one generation to the next. The principle of intergenerational equity (Catron et al., 1996) argues that the current generation has an obligation to deal with the problem now. It is a problem with uncertain impacts on human health and ecosystem health. The precautionary principle (e.g., IJC, 1994) argues that this uncertainty is not an excuse for inaction. But these arguments must be tempered by an obvious constraint: groundwater pollution is merely one of a spectrum of environmental problems to which society must apply its limited resources. Furthermore, experience indicates that groundwater pollution has the potential to consume an inordinate share of those resources. One may conclude that there is a need for a bounded response. 2 Technical organization(s) Party(s)with 1 direct legal responsibility for remediation . Party(s) with indirect legal responsibility for remediation Remediation regulator(s) L7 7 Aquatic environmental quality regulator(s) Water supply regulator(s) Water well operator Users of water from water well Figure 1-1. A typical case of groundwater pollution. A bounded response to groundwater pollution depends on an appropriate response at the level of individual cases. Consider a typical case of groundwater pollution, as represented in Figure 1-1. This figure shows both the pollution and the affected parties. There are many variations on this theme, but the central characteristics are: • Subsurface pollution: sources and plumes of pollution in the subsurface, as well as receptors of the pollution such as water wells and aquatic environments. This aspect of the situation is accompanied by a high degree of complexity (NRC, 1994). • Stakeholders: the set of parties who are affected by the pollution and must make decisions on how to cope with it. Typically, the stakeholders have divergent interests and values. The interactions among the stakeholders add another layer of complexity to the situation (cf. Wartenberg, 1988; Mazmanian and Morell, 1992, p. 57-75; Greene, 1994). 3 • The technical organizations: consultants and other organizations who advise the stakeholders and implement their decisions. The complexity of their task reflects the complexity of both the subsurface pollution and the interactions among stakeholders. The combined efforts of the stakeholders and the technical organizations comprise the management of the groundwater pollution case. Unfortunately, groundwater pollution management has generally encountered a high degree of frustration, in both technical actions and decision-making interactions among stakeholders. The consequence is typically a large expenditure of resources without a corresponding benefit. At the extreme, consider the case of the Rocky Mountain Arsenal in Colorado. Groundwater pollution at this site was described as early as 1954 (Todd and McNulty, 1976, p. 27-28), and was cited in Rachel Carson's (1962) Silent Spring. Today, four decades after it was first recognized, and despite the expenditure of many millions of dollars, groundwater pollution at the Rocky Mountain Arsenal is still highly problematic (Greene, 1994). For many cases of groundwater pollution, management requires millions of dollars and extends over decades. 1.2 Goal, scope, and intended audience This thesis has four main goals: 1. To establish a theoretical foundation which encompasses both stakeholder interactions and technical aspects. 2. To construct a conceptual model of groundwater pollution management on this theoretical foundation. 4 3. To develop a set of management strategies for groundwater pollution, at the scale of the individual case, based on the theoretical foundation and guided by the conceptual model. 4. To develop a compound model of groundwater pollution management, guided by the conceptual model and reflecting the management strategies. The theoretical foundation, the conceptual model, and the management strategies all involve general statements. There will naturally be exceptions to these generalizations. Such exceptions are important in themselves, but are not necessary for the immediate purposes of this thesis. There are a number of limitations to the scope of this thesis: 1. It is limited to the issue of groundwater pollution, ignoring the associated issue of land pollution. These two issues are essentially two sides of the same coin: land pollution provides the source for groundwater pollution. 2. It is limited to the issue of existing groundwater pollution. No attempt is made to explore the issue of activities which may result in future groundwater pollution, such as the construction of waste disposal facilities. These two issues have many similarities, but there is a fundamental difference in dynamics between coping with damage done and limiting future damage. 3. It is limited to groundwater pollution management in Canada and the United States. 4. The complicated legal issue of assigning responsibility (liability) to particular parties in a groundwater pollution case is not considered. 5 This thesis is intended for a hydrogeological audience, but touches on a number of topics which may be unfamiliar to such an audience, such as negotiation studies and conflict analysis. Such topics are reviewed in some detail, either in the body of the thesis or in appendices. Terminology is carefully defined, both in the text and in the accompanying glossary, for two reasons. The first reason is that many terms have different shades of meaning, especially in different disciplines. (For example, Horgan (1995) notes that 31 different definitions have been put forth for the term "complexity".) The second reason is that this thesis introduces a number of concepts which require new terms. 1.3 Preview of thesis The theoretical foundation is established in Chapter 2, which contrasts two ways of thinking about the management of problems such as groundwater pollution: the hard paradigm and the soft paradigm. The hard paradigm is rooted in science, mathematics, and engineering. It views a problem as a "hard problem", characterized by: • clear boundaries and linkages with other problems; • clear goals, alternatives, and consequences; • limited uncertainty which can be quantified; and • a linear, unilateral mode of decision-making. Hard paradigm thinking develops linear solutions with fixed endpoints. Modeling is used to constrain arguments and to predict consequences. The classic example of the hard paradigm in 6 groundwater pollution management is the combination of a simulation model of pollutant transport with an optimization model of pumping rates (simulation-optimization). The soft paradigm is rooted in frustration with the hard paradigm. It views a problem as a "soft problem", characterized by: • ambiguous boundaries and complex linkages with other problems; • goals, alternatives, and consequences which are not well-defined or well-understood; • pervasive uncertainty which may not be quantifiable; and • iterative management which involves conflict and negotiation among multiple stakeholders with divergent interests and values. Soft paradigm thinking emphasizes argumentation over algorithms, and iterative coping approaches over linear solutions. Modeling is used as a heuristic tool, to expand arguments and develop understanding. The contrasts between the two paradigms are straightforward, but there is also a soft/hard complementarity: the broader sweep of soft paradigm thinking can be applied to an overall soft problem, defining the appropriate hard sub-problems for the focused tools of hard paradigm thinking. This complementarity provides the basis for a reconciliation of the two paradigms, as well as for synergies in practice. This thesis has a soft paradigm orientation, but is also concerned with soft/hard complementarity. Groundwater pollution management is treated as a soft problem containing embedded hard sub-problems. Chapter 3 presents a conceptual model of groundwater pollution management, based on soft paradigm thinking as described in Chapter 2. This is a generalized picture which attempts to 7 encompass the spectrum of management approaches. It is organized according to a systems framework. The two main systems are the pollution system and the management system. The pollution system encompasses the phenomena of groundwater flow and pollutant transport. The management system encompasses two subsystems: a technical system and a decision system. The technical system is composed of organizations such as consulting firms which observe the pollution system, generate advice, and implement decisions. The decision system includes the various stakeholders; their issues of concern; and their modes and processes of decision-making. The emphases in Chapter 3, and subsequent chapters, are on the management system, on the decision system, and on negotiation as a mode of decision-making. These emphases reflect the soft paradigm orientation of the thesis, and recognition that these areas are perhaps less familiar to a hydrogeological audience. Chapter 4 presents a set of proposed management strategies for groundwater pollution, guided by soft paradigm thinking as described in Chapter 2 and by the descriptive conceptual model of Chapter 3. "Strategy", in this context, refers to a generalized and relatively abstract plan for coping with particular aspects of a generic groundwater pollution case. An example of a strategy is primary negotiation, which holds that negotiation should be both first in importance and first in a series of modes of decision-making. The strategies are divided into two main subsets: decision strategies and technical strategies. The main emphasis is on the decision strategies, consistent with the emphasis in the conceptual model. These are divided into three subsets: iteration, integration, and negotiation. The iteration strategies reflect the need for iterative decision-making to cope with complexity, 8 uncertainty and changing circumstances. The integration strategies reflect a concern that groundwater pollution management is often fragmented in a number of senses, and would benefit from efforts to reduce this fragmentation. The negotiation strategies reflect the importance and potential of negotiation in groundwater pollution management. The discussion of technical strategies is more limited. The strategies considered are: iterative technique, consistent with iterative decision strategies; technical integration, consistent with integrative decision strategies; and mitigative technique, reflecting practical limitations on technical abilities. Chapters 5 and 6 present SAM (Simulated Aquifer Management), a compound model of groundwater pollution management. SAM is both a heuristic tool in the soft paradigm sense, and an illustration of the management strategies and soft/hard complementarity. It is based on the conceptual model of Chapter 3. Chapter 5 provides a general description of SAM, and Chapter 6 presents a specific run. SAM includes two main sub-models: the pollution and management models, representing the pollution and management systems of the conceptual model. The emphases are on the management model and on negotiation as a mode of decision-making. The technical system is represented by the types of algorithmic models which might be employed by a technical consultant: groundwater flow, pollutant transport, optimization of pumping rates, and cost estimation. The decision system is represented by two parallel models of negotiation: an algorithmic model based on game theory, and an analog model using role-playing volunteers. The thesis closes with Chapter 7, which summarizes the thesis and discusses its development, contributions, and directions for future research. 9 2. Soft and hard paradigms: a review and synthesis This chapter develops a theoretical foundation for the thesis, through a review and synthesis of two contrasting yet complementary paradigms for studying the management of real-world problems. These are the hard paradigm and the soft paradigm. The emphasis in this chapter and throughout the thesis is on the soft paradigm, with consideration of its complementarity with the hard paradigm. There are no pejorative connotations associated with either paradigm; rather, the intention is to recognize the distinct powers of each and their potential complementarity. The hard and soft paradigms should not be confused with the colloquial usage of "hard science" for the natural sciences, and "soft science" for the social sciences. Synthesis implies generalization. There will be exceptions to the general statements made in this chapter. Such exceptions are significant, but are not germane to the immediate purpose of the thesis. The hard paradigm is grounded in mathematics, science, and engineering. It involves a high degree of idealization in the conceptual model of a problem, permitting the application of mathematically-rigorous methodologies. It does not claim to be a science, but rather to apply scientific knowledge and methodology to management, in a fashion analogous to engineering. The soft paradigm is a response to perceived inadequacies in the hard paradigm. It deals with less-idealized conceptual models of a problem, and places greater emphasis on argumentation and subjective judgement as opposed to mathematical rigor. It argues that management cannot be handled by a purely scientific methodology. 10 Checkland (1981) and Keys (1991) recognized that these two ways of thinking could be seen as contrasting paradigms. The paradigm concept was developed by Kuhn (19701). A succinct definition is provided by Stewart (1990, p. 4): A paradigm is a constellation of beliefs, values, procedures, and past scientific achievements that is shared within a community of scientists, and is learned during their training or in their common research experiences. These shared group commitments provide a framework that highlights specific scientific problems, restricts possible solution tactics, and establishes criteria for the rapid evaluation of solutions. The classic example of a paradigm in geology is the plate tectonics theory (Stewart, 1990). A new paradigm generally develops in response to perceived deficiencies in an older paradigm. The hard paradigm arose in response to a perceived lack of scientific logic and mathematical rigor in management practices. In particular, its roots can be traced to the application of science and mathematics to military problems during the Second World War. The hard paradigm is the basis for fields such as operations research, systems analysis, decision analysis, management science, and policy science. The soft paradigm began to develop largely in the 1960s and 1970s (Checkland and Scholes, 1990, p. 14-18). Independently, a number of practitioners within the hard paradigm became dissatisfied with its application to their fields of interest. At the same time, the application of the hard paradigm to planning and public policy came under severe criticism (Hoos, 1983 ; Rittel and Webber, 1973). This dissatisfaction and criticism led to the independent development of a number of literatures: 1 First edition published in 1962. 2 Original edition published in 1972. 11 • adaptive management in natural resources (Holling et al., 1978; Walters, 1986; Lee, 1993; Grayson, 1994; Gunderson et al., 1995); • soft systems methodology (SSM) in business management (Checkland, 1981; Wilson, 1984; Checkland and Scholes, 1990); • strategic assumption surfacing and testing (SAST) in business management (Mason and Mitroff, 1981); and • argumentation in policy analysis (Majone, 1989; Throgmorton, 1991; Fischer and Forester, 1993). In the geological sciences, the observational approach in geotechnical engineering (Terzaghi and Peck, 19671; Peck, 1969) and hydrogeological decision analysis (Massman and Freeze, 1987a, 1987b; Freeze, 1988) can be said to exhibit soft paradigm tendencies, although they do not belong in the list of soft paradigm literatures. The hard paradigm has a considerable degree of unity. This reflects both its maturity as a paradigm, and its reliance on a standard toolkit of mathematical methodologies, such as optimization, for a wide variety of situations. In contrast, the soft paradigm shows a high degree of disunity. Although the literatures of the soft paradigm tend to converge on similar concepts, they rarely cross: they have not yet been woven into whole cloth. For example, although adaptive management and SSM are similar in a number of aspects (as noted by Miser and Quade, 1985, p. 12), they do not reference one another. Nor is there any connection between SSM and SAST, although both are in the field of business management. This lack of integration is even more striking when one considers that the 1 First edition published in 1948. 12 philosophical roots of the soft paradigm have been traced back to the work of John Dewey in the 1920s (Mason and Mitroff, 1981, p.19-21; Lee, 1993, p. 91 et seq.\ Parson and Clark, 1995, p. 431). One point where a linkage among literatures does occur is between adaptive management and the work of Majone on argumentation in policy analysis (see Parson and Clark, 1995, p. 446). Disunity in the soft paradigm has a number of sources. It reflects the inherent methodological variability in the soft paradigm (Section 2.1.3, below). There are also disciplinary and geographic boundaries: adaptive management began in the natural resources field in British Columbia, SSM in the business field in Britain, and SAST in the business field in California. This chapter consists of four main sections. Section 2.1 is a description of the contrasts between the hard and soft paradigms. Modeling is an important aspect of both paradigms; this topic is examined in Section 2.2. The two paradigms also have a complementary relationship; this is explored in Section 2.3. The chapter concludes with a consideration of the hydrogeological literature on the management of groundwater pollution, seen in the light of the two paradigms (Section 2.4). 2.1 Contrasts between the hard and soft paradigms This section considers the contrasts between the hard and soft paradigms in terms of philosophical concepts, conceptualization of the problem, and general methodological concepts. 13 2.1.1 Philosophical concepts This section examines a variety of philosophical concepts which underlie the distinctions between the two paradigms. Both hard and soft paradigms make use of systems thinking, which is a way of thinking about real-world situations in terms of interacting systems. In the definition of Checkland (1981, Ch. 3), a system is a complex entity composed of interacting parts, contained within an environment, and characterized by: • Emergent properties: defined as properties which cannot be understood or predicted merely from knowledge of the parts of the system. As a simple example, consider water, which has properties which cannot be understood or predicted merely from knowledge of hydrogen and oxygen. • Hierarchical organization, with distinct subsystems in each layer of the hierarchy. As an example, consider a large surface water drainage system which consists of a set of smaller drainage subsystems. Rapoport (1970) introduced the terms "hard systems" and "soft systems" to distinguish between the hard and soft paradigm approaches to systems thinking. A hard system is a real-world entity with distinct boundaries. A soft system is a mental construct with amorphous boundaries, serving as a way to organize and develop thinking (Checkland, 1981; Pruzan, 1988; Checkland and Scholes, 1990; Keys, 1991, p. 170; Flood and Carson, 1993, p. 98). A key aspect of systems thinking is holism, which is the effort to understand some aspect of the real world as a whole, allowing for emergent properties (Checkland, 1981, p. 13; Simon, 1981, p. 195), linked problems (Mason and Mitroff, 1981, p. 15,20), and complex systemic 14 causality rather than linear causality (Pruzan, 1988). Holism is the antonym of reductionism, a highly successful scientific tactic which attempts to understand some aspect of the real world by understanding its parts. Integrative thinking is a combination of reductionist and holistic thinking (Lewontin, 1991). In practice, both hard and soft paradigms employ integrative thinking, rather than pure holism or reductionism. The difference is in emphasis. The hard paradigm, with its scientific ideals and methodologies, tends towards a reductionist focus (Pruzan, 1988, p. 36). The soft paradigm puts greater emphasis on holistic aspects (Checkland, 1981; Mason and Mitroff, 1981; Pruzan, 1988). Hypothesis-testing and experimentation are important aspects of scientific thinking. They are clearly consistent with the hard paradigm. Some soft paradigm thinkers emphasize an experimental, hypothesis-testing philosophy of management, particularly in the adaptive management literature (Walters, 1986; Lee, 1993). Other soft paradigm writers argue that that real-world problems always include uncontrolled factors which could be used to explain away negative evidence and undermine the basis for hypothesis-testing (Rittel and Webber, 1973; Checkland, 1981; Checkland and Scholes, 1990). The hard paradigm generally adheres to the scientific ideal of objectivity. The soft paradigm, in contrast, embraces subjectivity. Lee (1993) states that "all management is biased" (p. 61), and suggests that what is required is consistency, rather than objectivity. Pruzan (1988, p. 36-37) argues that subjectivity is unavoidable, given that the situations of concern are inherently ambiguous, and that values enter into any perception of a real-world problem. Checkland (1981) and Mason and Mitroff (1981) argue that a corollary of this view is the need to think in terms of multiple subjective perceptions of a problem, rather than a unique and objective reality. Majone 15 (1989) argues that the analyst cannot be a neutral observer, but rather is subject to his or her own subjective biases, and should be seen as an advocate for a proposed course of action. The hard paradigm is founded on the axiom of rationality. Stated loosely, this is the concept that, given a set of possible alternatives to choose from, a decision-maker will select the alternative for which he or she has the highest preference (Fraser and Hipel, 1984, p. 240). Rationality can be defined as the maximizing of utility, where utility is a real-valued function which describes preferences and is typically a surrogate for money. This seemingly straightforward concept is a cornerstone of mathematical simulation of human behaviour, in game theory and economics. (Note that this is a specialized use of the word "rationality", and should not be confused with the more common definition of rationality as pertaining to reason or sensibility.) The soft paradigm modifies this view of rationality. Mautner-Markhof (1989) and Lee (1993) endorse the concept of bounded rationality developed by Simon (1981'). Simon argues that rationality is problematic in its application to the real world, where alternatives and preferences are not clearly-defined, consequences are uncertain, and the ability to process information is limited. Bounded rationality recognizes these limitations, and is the basis for the methodological tool of "satisficing" (Section 2.2.6). Pruzan (1988) notes the idea of collective rationality, involving conflict and consensus among multiple decision-makers. 1 Simon originally proposed this concept in 1955. 16 2.1.2 Hard problems and soft problems A key point of departure between the hard and soft paradigms lies in the conceptual view of the problem. The hard paradigm looks at hard problems, and the soft paradigm looks at soft problems (Checkland, 1981).1 Note that "hard problem" and "soft problem" refer to concepts rather than reality. In general, hard problems are more idealized than soft problems. Specific differences can be seen in terms of: • boundaries and linkages with other problems; • goals, alternatives, and consequences; and • the nature of decision-making. A hard problem has clear boundaries. Its linkages with other problems can be understood, and are sufficiently simple that it can be considered independently of the other problems. A soft problem does not have clear boundaries (Rittel and Webber, 1973, p. 159; Checkland, 1981). It is characterized by complex linkages with other problems (Mason and Mitroff, 1981; Mautner-Markhof, 1989). These linked problems create a state of organized complexity (Weaver, 1948), in which efforts to solve one problem may have unexpected effects on other problems (Rittel and Webber, 1973, p. 159; Mason and Mitroff, 1981). Rittel and Webber (1973, p. 165) envisage a hierarchy in which each soft problem is a symptom of a higher-level soft problem. A hard problem involves clearly-defined goals, alternatives, and consequences which can be determined as the initial steps of analysis. Uncertainty in consequences can be quantified (Section 2.1.3). 1 Rittel and Webber (1973) refer to "tame problems" and "wicked problems"; these are essentially equivalent to hard problems and soft problems. 17 In a soft problem, the goals and alternatives are not well-defined (Rittel and Webber, 1973; Mason and Mitroff, 1981; Checkland, 1981; Checkland and Scholes, 1990). There may be merely a sense of "dis-ease", rather than a crisp understanding of goals and alternatives (Checkland, 1981). The effort to establish goals and alternatives may be the most important and frustrating aspect of coping with the problem. Consequences are often unpredictable, reflecting complex linkages with other problems; it follows that uncertainty is not particularly amenable to quantification (see Section 2.1.3). In a hard problem, decision-making is generally unilateral, consisting of the selection of the "best" alternative by a single decision-maker. "Best" is narrowly defined by a single criterion, commonly an economic criterion (Pruzan, 1988, p. 38). The mathematical expression of this choice of "best" alternative is the objective function (see Section 2.2.2). The actual decision-making process is implicit (Majone, 1989, p. 15). Some hard paradigm methodologies allow for consideration of multiple decision-makers or multiple criteria (e.g., Major, 1977), but the decision-making process remains implicit. In a soft problem, decision-making involves conflict among multiple stakeholders with divergent interests and values (Rittel and Webber, 1973; Mason and Mitroff, 1981; Pruzan, 1988; Majone, 1989; Mautner-Markhof, 1989; Lee, 1993). Goals and alternatives are contentious and not well-defined. The decision-making process must be considered explicitly, as an untidy affair involving conflict and negotiation. 18 2.1.2.1 Example of a soft problem A recent example of a traffic control problem in a Vancouver neighbourhood provides an example of a soft problem. This incident is described in articles and letters to the editor in local newspapers (Francis, 1995; Hughes, 1995; Roberts, 1995; Appelbe, 1996). The situation of concern was a high accident rate on Blenheim Street. The City of Vancouver originally treated this as a hard problem, and formalized the goal as reduction of the accident rate. The chosen alternative was the installation of traffic lights at three intersections on Blenheim Street. The expected consequence was a reduced accident rate. However, implementation of the traffic lights resulted in an unexpected consequence: Blenheim became a more efficient thoroughfare, and attracted more and faster traffic. The City of Vancouver then held a public consultation with residents along Blenheim Street, leading to a proposed alternative of installing traffic circles on Blenheim, with the expected consequence of slowing down traffic. This alternative was popular with the residents of Blenheim Street, but was opposed by the residents of nearby streets. These residents feared that the traffic circles would divert traffic onto their streets, and claimed that they should also have a say in the matter. They argued that the residents of Blenheim knew that they were buying into a busy street, and should not be pushing to divert the traffic onto quieter streets with higher property values. It was further argued that the issue had implications for the residents over a large part of the city (Roberts, 1995). The traffic circle proposal was eventually abandoned, despite the insistence of City of Vancouver representatives that the traffic circles were an experimental measure, to be removed if they did not work as intended. 19 This situation exhibits many characteristics of a soft problem: a lack of clear boundaries, complex linkages with other problems, unexpected consequences, and contentious goals and alternatives arising from conflict among multiple stakeholders with divergent interests. A quote from one of the residents captures the concept of linkages among soft problems: They [the City of Vancouver] seem to have no overall concept of how to solve problems...When they fix one problem, it creates another problem elsewhere (Francis, 1995, p. 1). When a similar problem arose recently in another Vancouver neighbourhood, the City of Vancouver simply stepped aside and gave the local residents two years to sort it out by themselves (Appelbe, 1996). 2.1.3 General methodological concepts The intent of this section is to make broad generalizations about methodological concepts in the hard and soft paradigms. No effort is made to review specific methodologies. This section examines methodological standardization, solution procedures, algorithms and arguments, the handling of uncertainty, and the evaluation of outcomes. The important methodological topic of modeling is covered separately in Section 2.2. The introduction to this chapter has already noted that the degree of methodological standardization is a key distinction between the two paradigms. The hard paradigm relies upon a standardized toolkit of mathematical methodologies such as optimization. This is possible because the high degree of idealization in a hard problem reduces the situation of concern to those elements which fit into a standard methodology. For example, optimization can be applied to the mix of ingredients in chicken feed, to groundwater pumping rates, or to global energy usage. 20 In contrast, soft problems are too diverse to tackle with standardized methodologies. The soft paradigm must generate a specific methodology for each soft problem. For example, despite the underlying similarities in adaptive management and soft systems methodology, their methodological details quickly diverge. The former deals with ecosystems, and the latter deals with business systems. Without applying a high degree of idealization, it is not possible to apply the same methodology to both circumstances. In the hard paradigm, a hard problem is generally viewed as something to be dealt with through a linear solution procedure, with a distinct beginning and endpoint (Figure 2-1). This begins with a definition of the problem, followed by the design of alternatives and prediction of their consequences. The alternatives can then be ranked, and the results communicated to the decision-maker. The decision-maker considers the alternatives and makes a decision. The decision is implemented, and its results evaluated. The soft paradigm views a soft problem as a situation to be managed through iterative cycles of decision, implementation, and learning. This is a strong feature of adaptive management (Ffolling et al., 1978; Walters, 1986; Lee, 1993) and soft systems methodology (Checkland, 1981; Wilson, 1984; Checkland and Scholes, 1990). Checkland (1981) describes the experience-action cycle (Figure 2-2), with action leading to experience, and experience creating knowledge which guides further action. Rittel and Webber (1973) argue that "wicked problems" are never really solved; rather, they are repeatedly re-solved. Each solution generates new and unexpected consequences, which must be dealt with in turn. 21 Formulation (problem definition) Goals; values and criteria; boundaries and constraints Alternatives Identifying, designing, and screening Predicting consequences Building and using models T Ranking alternatives Communicating results Evaluating the analysis Decision and implementation Evaluating the outcome Figure 2-1. Linear solution procedure in the hard paradigm. Adapted from Figure 1.1 in Miser and Quade (1988, p. 23). Experience-based! knowledge Experience Action Figure 2-2. The experience-action cycle. Adapted from Checkland and Scholes, 1990, Figure 1.1, p. 3. An algorithm is a set of rules which transform an input into an output, without the exercise of human judgment. Algorithms are inherently objective (in a limited internal sense), and are commonly executed with computers. Argumentation is methodical reasoning expressed 22 through natural language, and often supported by graphic visualization (see Walton, 1989, for a text on argumentation). Argumentation is inherently subjective. Arguments typically produce counter-arguments, leading to a dialectic which tests the competing arguments. In the hard paradigm, algorithms have primacy. Argumentation has a supporting role, in the assumptions which underlie the structure of the algorithm and its input, and in the interpretation of its output. In the soft paradigm, argumentation has primacy (Mason and Mitroff, 1981, Ch. 9; Majone, 1989; Throgmorton, 1991; Fischer and Forrester, 1993). It the basis for Majone's (1989) concept of an adversarial process of multiple advocacy in public policy analysis, and for Lee's (1993) concept of politics as the stabilizing gyroscope in the adaptive management of large-scale ecosystems. Uncertainty is a concept with different implications in the two paradigms. It is surprisingly difficult to provide general definitions of uncertainty and the associated concept of complexity. These words have many shades of meaning. As noted in the Introduction, Horgan (1995) refers to 31 different definitions of "complexity". The following working definitions of complexity and uncertainty have been used in this thesis. Complexity is an intrinsic property of some aspect of the real world, correlated with the amount of knowledge or information which an observer would need to understand or describe that aspect. Uncertainty is a function of an observer's ignorance about some aspect of the real world. It is not an intrinsic property of that aspect of the real world. Freeze et al. (1990) make a similar distinction in a hydrogeological context, referring to heterogeneity rather than complexity. In the usage of these thesis, complexity and uncertainty may be quantitative or non-quantitative, and may be temporal or spatial. 23 In the hard paradigm view, uncertainty is limited and quantifiable (e.g., Morgan and Henrion, 1990). It can be described with well-behaved probability distributions. These distributions can be translated into dollar costs or statistical lives, and used as a basis for decision-making. It can be reduced by obtaining additional information about the problem; this concept is a keystone of the Bayesian statistical methodology (e.g., Freeze et al., 1990). In the soft paradigm view, uncertainty is pervasive, and not necessarily amenable to quantification (Holling et al., 1978, p. 132-133). Clark (1978) claims that uncertainty cannot be eliminated or minimized, and that efforts to do so are generally counterproductive. He argues that the focus should rather be on developing coping mechanisms based on resilience, on deliberate experimentation, on anticipation of the unexpected, and on learning from the inevitable surprises. This philosophy is one of the direct foundations of adaptive management (Holling et al., 1978, Ch. 10) and is independently reflected in Checkland's (1981) action-experience cycle (Figure 2-2). Hard paradigm thinking views post hoc evaluation of management outcomes as conclusive. The soft paradigm is less sanguine about the potential for post hoc evaluation. Rittel and Webber (1973, p. 163) argue that any implemented solution to a "wicked problem" produces an unending sequence of consequences, often with negative implications; therefore, they claim that there can be no immediate or ultimate evaluations of a solution. Majone (1989, Ch. 8) is less pessimistic, but does caution that there is no such thing as an empirical, objective, neutral evaluation of a policy or plan. Maj one's prescription is a dialectic of multiple evaluations from distinct points of view, echoing his call for multiple advocacy of proposed policies and plans. 24 2.2 Modeling Modeling is an important methodological aspect of both hard and soft paradigms. This section is a rather extended treatment of modeling, and provides important background material for the following chapters. This section describes various types of models, discusses virtues and pitfalls in modeling, and contrasts the hard and soft paradigm views of modeling. A model is an abstraction of some aspect of the real world, based on observations and directed towards some purpose (Figure 2-3). It is a means rather than an end. A model may have any of a wide range of purposes. It may organize knowledge, by fitting it into a particular framework. It may generate knowledge by improving the understanding of some aspect of reality, or by predicting the future behaviour or expected character of some aspect of reality. It may communicate knowledge. It may prescribe a course of action. A single model may be directed towards one or more of these purposes. For example, a model of groundwater flow in a particular aquifer may be intended to organize existing knowledge about the aquifer, to predict the future behaviour of the aquifer, and to communicate both the existing knowledge and the predictions to concerned individuals. Real world •^[Observations^ Model «-J Purpose Figure 2-3. The relationship among observations, model, and purpose. 25 The interactions among reality, observations, model, and purpose are bi-directional (Figure 2-3). Observations are made on the real world, but the real world constrains the observations which can be made. The model is based on the observations, but the requirements and limitations of the model guide the observations. The model serves the purpose, but the purpose is adjusted to reflect the limitations of the model. If the purpose is prescriptive, then the real world itself may be adjusted. In practice, these interactions are also iterative — an ongoing process of adjustment among observations, model, purpose and (perhaps) the real world. The classification of models used in this thesis has two main categories: conceptual and transform (Figure 2-4). A conceptual model is a synthesis of observations. A transform model is a model which transforms an input into an output. Transform models are further subdivided into algorithmic models, analog models, and compound models, as described below. Transform models Algorithmic models Analog models Compound models Figure 2-4. Classification of models used in this thesis. 2.2.1 Conceptual models A conceptual model is a synthesis of observations. The process of synthesis involves filling gaps in observation with interpolation, extrapolation, and intuition. A conceptual model is presented in words, pictures, data, and mathematical expressions (Wilson, 1984, p. 9). It often Conceptual models 26 includes a combination of quantitative and qualitative elements, and may involve both subjective and objective aspects. Conceptual models may be general or specific. A general conceptual model corresponds to a general set of phenomena, such as regional groundwater flow. A specific conceptual model corresponds to a particular case, such as regional groundwater flow in the Milk River aquifer. Note that specific conceptual models may be generated for hypothetical cases as well as for real-world cases. Also note that a general conceptual model may be largely implicit, particularly if it is already well-known to an intended audience. A conceptual model can be applied directly to a purpose, independently of any other model (Wilson, 1984, p. 9). The most obvious example of a stand-alone conceptual model in geology is the geological map, with its necessary blend of observation, interpolation, extrapolation, and interpretation. Another example is the facies model: a written and graphical synthesis of studies on a particular sedimentary environment. It is used by sedimentologists to provide a norm for comparison, a guide for observation, and a basis for prediction and interpretation (Walker, 1986, p. 4-7). The ore deposit model (e.g., Sheahan and Cherry, 1993) is another example. Conceptual models can also be applied to the purpose indirectly, through the filter of a transform model (algorithmic, analog, or compound), as illustrated in Figure 2-5. A corollary is that conceptual models are necessary precursors for transform models. Simplification of both general and specific conceptual models is necessary to meet the limitations of the transform model, and is achieved by applying a series of assumptions and parameterizations. The simplified version of the general conceptual model corresponds to the structure of the transform 27 model, and the simplified version of the specific conceptual model corresponds to the input for the transform model. Both the output of the transform model and the exercise of developing its structure and input can contribute to the purpose. Figure 2-5. The relationship between conceptual and transform models. 2.2.2 Algorithmic models An algorithmic model is a transform model based on one or more algorithms. Algorithmic models are commonly referred to as "computer models", because they are generally implemented on computers. The term "algorithmic model" is preferred here because it emphasizes a more fundamental aspect of this type of model. Algorithmic models are the most important category of transform model. An algorithmic model is intrinsically objective, because the output represents the logical consequences of the input, according to a formalized recipe. This objectivity applies only in a limited internal sense, because external subjectivity is involved in the observations, the development and simplification of the conceptual models, and the interpretation of the output. But within the algorithmic model itself, all subjectivity is abandoned and the process is given /' Real world '• 28 over to the inexorable execution of a formalized recipe. Algorithmic models derive a special legitimacy from this internal objectivity. Algorithmic models can be applied to hypothetical cases as well as to real-world cases, using generalizations based on real-world observations. The concept of the hypothetical case is discussed in Schwarz (1988, p. 334-339). Examples of the use of hypothetical cases in the hydrogeological literature are provided by Gorelick et al. (1984) and Massman and Freeze (1987a, 1987b). Three types of algorithmic modeling require further mention here: simulation, game theory, and objective-function modeling. Simulation models deal with the behaviour of some part of the real world, such as the flow of groundwater in an aquifer or the flow of electricity in a circuit. Simulation modeling is a standard technique in hydrogeology, and is used in the SAM compound model (Chapter 5). Game theory is a mathematical abstraction of human conflict, competition, and cooperation (Thomas, 1984; Fraser and Hipel, 1984). It is described in more detail in Appendix 2. Briefly, a game is a situation involving two or more decision-makers (players), each of whom makes one or more decisions (moves), subject to the rules of the game. Each set of moves by all players in a game results in an outcome, and each outcome is associated with an individual payoff for each player. It is assumed that each player will behave rationally, that is, each player will select a set of moves with the intent of maximizing their individual payoff. The outcomes which result from each player attempting to maximize their individual payoff represent possible resolutions of the situation. A variant of game theory is used in the SAM compound model (Chapter 5). 29 Objective-function models are formalized abstractions of human decision-making, in which the objective is represented as a mathematical function to be maximized or minimized. Objective-function models play an important role in the hard paradigm. Two important types of objective-function modeling are described here briefly: optimization and risk-cost-benefit analysis. In an optimization model, a problem is reduced to an objective function and a set of constraints. The variables in the constraints and the objective function are called decision variables because they represent quantities about which decisions can be made. The objective function is the mathematical abstraction of the goal; it combines the decision variables with their associated cost factors. The constraints are inequalities and/or equalities which represent limitations; they define a feasible region in the space of the decision variables. The optimal values of the decision variables are the values which maximize (or minimize) the objective function. The optimization algorithm must select the optimal decision variables from the feasible region. It is possible to use multiple objective functions, without necessarily making a priori tradeoffs among goals (Major, 1977; Kaunas and Haimes, 1985; Cohon et al, 1988). Generally, however, only one objective function is used, and any other goals are translated into constraints. Optimization modeling is discussed further in Sections 2.2.6 and 2.4. An optimization model forms part of the SAM compound model (Chapter 5). Risk-cost-benefit analysis involves ranking a discrete set of technical options. An objective function is evaluated separately for each option. The objective function typically represents a time-discounted stream of monetary benefits and costs for the option under evaluation. The costs and benefits may be probabilistic as well as deterministic. A probabilistic 30 cost or benefit is defined as the cost or benefit of an event multiplied by the probability of that event; a probabilistic cost is referred to as a "risk". Risk-cost-benefit analysis in hydrogeological problems is discussed in Section 2.4. 2.2.3 Analog models An analog model is a transform model which represents some aspect of reality by creating an analogous situation which is more accessible to observation. For example, the flow of groundwater in an aquifer can be modeled using a network of electrical resistors, due to the analogy between the flow of water in a porous medium and the flow of electrical current in a network (Freeze and Cherry, 1979, p. 178-181). Other examples would be the use of a centrifuge to model the subsurface flow of immiscible fluids, or the modeling of surface water flow through the construction of scaled-down channels. Role-playing is a type of analog model which can be applied to human situations. In a role-playing model, the human interactions in a situation of interest are simulated by humans acting out the roles of the various parties, subject to both formal rules and the subjective judgement of the role-players. An example would be a simulation of the emergency reaction to an earthquake in a particular region. Such models are widely used for simulating human interactions, with applications in fields such as business, behavioural science, systems analysis, policy analysis, political science, emergency planning, and environmental planning (see Stahl, 1988; Greenblat, 1988; Duke and Kemeny, 1989; Mautner-Markhof, 1989; Dolin and Susskind, 1992). In the literature, role-playing is more commonly referred to as "gaming" or "gaming-31 simulation". The term "role-playing" has been adopted in this thesis to avoid confusion with the distinct topic of game theory. Role-playing is used in the SAM model (Chapter 5). 2.2.4 C o m p o u n d models A compound model (Figure 2-6) is a transform model which consists of a series of linked sub-models. A common hydrogeological example of a compound model is simulation-optimization, which involves the coupling of a simulation model and an optimization model (see Section 2.4). -Input Submodel 1 Submodel 2 Submodel n -Output Figure 2-6. A compound model. This example includes linkages which are algorithmic (solid lines) and subjective (dashed lines). Many different configurations of linkages and sub-models are possible within a compound model. A sub-model in a compound model is commonly algorithmic. Less commonly, an analog model may be included. A linkage in a compound model may be algorithmic or subjective. A linkage is algorithmic if the output from one sub-model becomes the input for another sub-model with no exercise of human judgment. For example, in a typical simulation-optimization model in hydrogeology, output from the simulation model flows directly as input to the optimization model, without any human intervention. A linkage is subjective if it requires an exercise of human judgment to incorporate the output from one sub-model into the input for another sub-model. 32 A compound model may contain compound sub-models, giving rise to a nested hierarchy of compound models. 2.2.5 Virtues and pitfalls in modeling This section considers some of the virtues and potential pitfalls in modeling. Reference is made here to the IIASA energy model, a compound model of the global usage and supply of energy, produced by the International Institute for Applied Systems Analysis (Hafele, 1981). The development and use of this model was analyzed by Keepin (1984) and Wynne (1984). Their analysis provides useful insights into the practice of modeling. Parsimony is a virtue in modeling. An overly-complicated model may reduce understanding at the cost of increased effort. Walters (1986, p. 185-200) describes the concept of "optimum model complexity." Holling (1978) and Walters (1986) discuss the importance of "model compression": simplifying a model without losing its essential features. A lack of parsimony can be demonstrated when the results of a model can be duplicated by a much simpler model. Keepin (1984) demonstrated a lack of parsimony in the IIASA global energy model. McLaughlin and Johnson (1987) demonstrated a lack of parsimony in a set of groundwater modeling studies. A transparent model is one which is straightforward and well-documented, with respect to subjective aspects, simplifying assumptions, and methodological details. Ideally, another modeler should be able to reproduce the model independently and obtain similar output. If a model is not transparent, then it cannot be properly evaluated, applied, or improved. Transparency is particularly important with respect to compound models with internal subjective 33 linkages. In the IIASA energy model, algorithmic sub-models were interwoven with subjective linkages (Hafele, 1981, p. 401) which were undocumented, or obscurely documented (Wynne, 1984). A transform model is expected to perturb its input in a non-trivial way, producing output which was not obvious at the beginning of the exercise. The identity transformation pitfall arises when a transform model simply performs a one-to-one mapping of input to output. Keepin (1984) demonstrated that the algorithmic sub-models in the IIASA model did not significantly perturb their input. Compound models are subject to a particular pitfall: the potential non-transparent dominance of internal subjective linkages over the sub-models themselves. Keepin (1984) and Wynne (1984) argued that the algorithmic sub-models in the IIASA energy model were ineffective, and the internal subjective linkages became dominant. Insidiously, there can be a negative synergy among the pitfalls. For example, in the IIASA global energy model, the identity transformation pitfall and the dominance of internal subjective linkages were hidden by a lack of parsimony and transparency (Keepin, 1984; Wynne, 1984). A good model must be sufficiently realistic relative to its purpose. This is a complicated matter, involving subjectivity, and uncertainty in the conceptual model, uncertainty in simplifying assumptions and parameters, non-uniqueness in model output, methodological limitations, and difficulty in comparing model output with real-world behaviour. The pitfalls discussed above can all work towards a lack of realism. Keepin (1984) and Wynne (1984) felt that the IIASA energy 34 model was not realistic, and further concluded that it had been misused to provide subjective opinions and assumptions with a gloss of "science" and "objectivity". Finally, modeling should be addressed to a purpose. It is not an end in itself. This is a particular concern with complicated algorithmic models. The development of such models often represents a significant intellectual challenge, to the extent that the original purpose may become secondary. A related problem is the Pygmalion pitfall, which occurs when the researcher develops a myopic bias in favour of a model in which he or she has invested time, energy, and money to construct or to understand (Walters, 1986, p. 167). Concerns over matters such as realism have led to a recent debate over the value of algorithmic simulation modeling in hydrogeology, both within the modeling community (McLaughlin and Johnson, 1987; Konikow and Bredehoeft, 1992; Hofmann, 1994; Oreskes et al., 1994; Blair, 1994) and within the legal profession (Scarrow, 1989). The debate is driven by the increasing importance of hydrogeological simulation modeling in specific disputes and in the development of legislation. 2.2.6 Hard and soft paradigm views of modeling The distinction between hard and soft paradigms is brought out in their use of conceptual and transform modeling. In the hard paradigm, conceptual models are streamlined and simplified to meet the requirements of transform models. By contrast, the soft paradigm places great emphasis on detailed conceptual models, as a basis for argumentation. In the hard paradigm, transform models are central, and are valued for their ability to constrain an argument. The output of a transform model is its most important benefit. A 35 transform model is seen as a proxy for the real world, capable of predicting behaviour and prescribing a course of action. This view relies heavily on the realism of the model, and it follows that testing of the model by comparing its output with empirical data is a critical step. For example, simulation models of groundwater flow may be tested against empirical data which has not been used in model calibration. In the soft paradigm, transform models receive less emphasis. They are used as heuristic tools, and are valued for their ability to expand an argument by stimulating new ideas. The understanding that arises through the iterative process of model building is the important benefit of transform modeling. The output, and even the model itself, is less important (see Walters, 1986, Chapter 3, especially Figure 3.1). Oreskes et al. (1994, p. 644), writing in the context of simulation models in hydrogeology, captures this philosophy of modeling: ...the primary value of models is heuristic. Models are representations, useful for guiding further study but not susceptible to proof. The transform model is seen as an aid to thinking, rather than as a proxy of the real world. This view does not rely heavily on the realism of the model, and testing of the model by comparing its output against empirical data is not particularly important. Some transform models are inherently compatible with soft paradigm thinking. An example is Bayesian statistics, with its capacity to accept subjective probabilities and to iteratively update probabilities on the basis of new information (Walters, 1986; Freeze et al., 1990). Other transform models can be adapted to soft paradigm thinking. Optimization modeling provides a good example of such adaptation. 36 In the hard paradigm, optimization modeling is used for prescriptive purposes. At the extreme, it may be presented as a form of automated decision-making. The soft paradigm makes a number of criticisms of prescriptive optimization. Optimization of complex and uncertain real-world problems is always approximate (Simon, 1981, p. 35). Formulating a problem as an optimization model requires reducing it to a set of equations, and may sacrifice important aspects of the real-world situation in exchange for mathematical rigor and computational tractability (Rogers and Fiering, 1986, p. 150S; Pruzan, 1988, p. 39; Cocklin, 1989a). Non-quantifiable factors are automatically excluded (Cocklin, 1989a, p. 135). The results of optimization are very sensitive to the choice of objective function (Casti,1979, p. 31-32). For some problems, unbiased estimation errors in formulation can yield an optimistically-biased solution, i.e., the estimated optimum is consistently better than the true optimum (Hobbs and Hepenstal, 1989). For many complex real-world problems, the response surface is relatively flat (i.e., within the constraints, no decision is particularly better than the others), and locating an optimum is not important (Rogers and Fiering, 1986, p. 150S-15IS). Efforts to inject more reality into an optimization model result in reduced computational tractability. In particular, optimization problems which include nonlinearities, integer decision variables, uncertainty, dynamic factors, or multiple objective functions are relatively difficult to formulate and to solve. It is particularly difficult to define the objective function and constraints in a situation with multiple stakeholders and contentious goals. The formulation of a particular objective function may implicitly favour one group or another. An attempt can be made to use multiple objective functions, but this does not solve the problem of determining tradeoffs between these multiple objectives (Cocklin, 1989a, p. 139), or of turning fuzzily-defined goals 37 into crisp objective functions. Finally, the stakeholders in a multi-stakeholder situation are unlikely to surrender their authority to automated decision-making. A number of soft paradigm adaptations of optimization have been developed: optimization for adaptive management; optimization to generate alternatives; optimization for understanding; satisficing; and feasibility analysis. Note that these adaptations are used in ways which tend to expand an argument, as noted above with respect to the general use of transform modeling in the soft paradigm. Walters (1986) uses optimization in the design of adaptive management policies for fisheries. For example, Walters uses dynamic programming to model sequential series of future decisions contingent on future conditions (feedback policies), encouraging decision-makers to consider possible future tradeoffs in a range of future conditions. Walters also uses dynamic programming to develop a balance between short-term performance and learning (actively adaptive policies). Walters acknowledges the quantitative limitations of this work for complex real-world situations, but sees it is useful for providing qualitative insights and understanding. Rogers and Fiering (1986, p. 151S-152S) and Cocklin (1988b, p. 144-148) proposed that optimization be used to generate a variety of alternatives (a "negotiation set") for submission to a decision-making process involving multiple stakeholders. These alternatives can be generated from the set of near-optimal solutions; from the set of locally optimal solutions; through consideration of the "free play" allowed by uncertainty in the objective function(s) and constraints; or from a Pareto-optimal set generated by multi-objective optimization. Cocklin (1988a, 1988b) argued for the use of optimization to generate understanding and creativity rather than "optimal solutions". In general, the exercise of producing an optimization 38 model demands that assumptions about goals and constraints be made explicit; this process can provide opportunities for creative thinking (Cocklin, 1988a, p. 139). It is also possible to use optimization in a mode that explores tradeoffs and relationships among goals, without seeking an optimal solution (Cocklin, 1988b, p.152-155). Satisficing (Simon, 1981, p. 36) ignores optimality and focuses on constraints. It is a methodological expression of the bounded rationality concept (Section 2.1.1), involving the determination of constraints and a heuristic search for a non-unique alternative which satisfies those constraints. In other words, satisficing is satisfied with any solution which lies within the feasible region. Schwarz et al. (1985, p. 222) describe satisficing as "replacing objectives by constraints". Lee (1993, p. 53) considers satisficing to be highly compatible with adaptive management. Feasibility analysis (Majone, 1989, Ch. 4) also ignores optimality and focuses on constraints. It is concerned with expansion of the feasible region. This is facilitated by distinguishing between fixed, fictitious, and flexible constraints. The fixed constraints must be respected; the fictitious constraints can be discarded; and the flexible constraints can be adjusted to expand the feasible region. 2.3 Complementarity of the soft and hard paradigms The previous sections have emphasized the distinctions between the hard and soft paradigms. Indeed, it is generally the case that competing paradigms in a single field of study are incompatible. For example, a geologist cannot simultaneously operate under both the plate tectonics paradigm and the older paradigm of fixed continents. It is also generally the case that 39 two competing paradigms are incommensurable, meaning that they cannot be measured by the same standards. This stems from difficulty in communication between the two sets of adherents, due to fundamental disagreements and misunderstandings (Kuhn, 1970, p. 198; Stewart, 1990, p. 5). For example, consider the difference in the concept of "uncertainty" between the hard and soft paradigms. However, there is an important sense in which soft and hard paradigms may be complementary. For many management situations, the soft paradigm can supply a global understanding, and the precision tools of the hard paradigm can be applied to specific sub-problems (Pruzan, 1988, p. 35). In other words, the situation can be seen as a soft problem matrix containing embedded hard sub-problems. In the soft paradigm literature, this concept is best-developed within the adaptive management literature (Holling et al., 1978; Walters, 1986; Lee, 1993). Soft/hard complementarity has a further implication: the hard paradigm operates on the hard sub-problems with an awareness of the overall soft problem, adapting its methods to the circumstances of the soft problem. In other words, the hard sub-problems are not tackled independently of the soft problem matrix. At a philosophical level, soft/hard complementarity is seen in Lee's (1993) distinction between the direction-finding "compass" of science and the stabilizing "gyroscope" of politics. Lee advocates a concept of "civic science" which pragmatically remains at arm's-length from the political aspects which it seeks to inform. This has a certain resonance with Weinberg's (1972a, 1972b, 1985) concept of trans-science, referring'to questions which are at the boundary of 40 science, but which cannot be answered by science, and must be addressed through public argumentation. At a practical level, soft/hard complementarity requires techniques for working at the soft/hard interface. Holling et al. (1978) and Walters (1986) describe an interface technique: intensive workshops in which stakeholders interact with ecosystem modelers, using techniques such as dynamic programming to generate alternative scenarios and hypotheses, and to illuminate unavoidable tradeoffs. This approach embodies the soft paradigm view that the model is a heuristic tool, and that value of modeling lies in the understanding gained through the process of model building. The compound model SAM (Chapter 5 and 6) explores soft/hard complementarity, with the hard sub-problems represented by transform sub-models, and the soft matrix represented by subjective linkages between the sub-models. 2.4 Relevant hydrogeological literature on the management of groundwater pollution The hard paradigm is obviously compatible with the perspective, norms, and research tradition of hydrogeology. For several decades, algorithmic models have contributed to the organization, generation, and communication of hydrogeological knowledge. The importance of algorithmic models in hydrogeology is demonstrated by the sophisticated simulation models in current use for understanding and predicting the behaviour of groundwater systems. It is not surprising, therefore, that hydrogeologists would apply hard paradigm thinking to the management of groundwater problems. 41 Beginning with the work of Maddock (1972a, 1972b), hydrogeologists have applied algorithmic models to the management of groundwater problems, including both groundwater supply and groundwater pollution. This field expanded following the publication of a review article by Gorelick (1983), and a textbook by Willis and Yeh (1987). The basic approach is to use a compound model. The groundwater phenomena are represented by one or more simulation models. Management is represented by an objective-function model. Generally, only a single decision-maker is considered. The typical path for the development of these models is to begin by applying them to a generalized hypothetical case, and then to extend them to real-world cases. There have been major advances in mathematical sophistication, although widespread application to real-world problems has lagged behind. The study of complexity and uncertainty has been a major focus of hydrogeological research in the past decade. The approach has been grounded in the hard paradigm, and efforts have concentrated on quantifying the spatial distribution of key parameters used in simulation models, particularly hydraulic conductivity. The study of groundwater management has both incorporated and contributed to advances in this area. Recent work has focused on the management of groundwater pollution, reflecting a public policy emphasis on this problem. In this body of work, the decision-maker is typically the responsible party or the owner-operator of a polluted or potentially-polluted site, with a simple preference for the lowest-cost solution. The regulator is represented as a constraint or a probabilistic cost. Decision-making is treated as an automatic choice of the optimal decision, as determined by an objective-function model. The value to be optimized is generally money, or a 42 proxy for money. There are two main streams to this literature: simulation-optimization and hydrogeological decision analysis. Simulation-optimization for groundwater pollution management requires simulation of the groundwater system (flow or flow-and-transport). The simulation model evaluates the objective function for an optimization model. The decision variables in the optimization model are typically pumping rates at specified well locations; less commonly, the locations of pumping wells are also decision variables. In a few cases, monitoring well locations are used as decision variables. The regulatory party is generally implicitly embodied in the model as a constraint, e.g., a compliance boundary at which pollutant concentrations must not exceed some specified level. Uncertainty is often considered, through either sensitivity analysis or stochastic techniques. A partial list of references for simulation-optimization in groundwater pollution management would include Gorelick (1982), Gorelick et al. (1984), Kaunas and Haimes (1985), Wagner and Gorelick (1987), Ahlfield et al. (1988), Andricevic and Kitanidis (1989), Wagner and Gorelick (1989), Ahlfield and Sawyer (1990), Meyer et al. (1994), McKinney and Lin (1994), and Sawyer etal. (1995). Hydrogeological decision analysis (Massman and Freeze, 1987a, 1987b; Freeze et al, 1990; Massman et al, 1991; Freeze et al, 1992; Sperling et al, 1992; James and Freeze, 1993; James and Gorelick, 1994) makes use of risk-cost-benefit analysis (see Section 2.2.2). Deterministic costs are assigned to various technical options. Uncertainty is embedded in probabilistic costs (risks) which are assigned to regulatory penalties. The probability component of the probabilistic cost is the probability of a regulatory violation, given a particular technical option. This is evaluated using a stochastic simulation model of the groundwater system, and 43 may be updated on the basis of new information with the use of Bayesian statistics. Note that the regulatory party is implicitly embodied in the model as a probabilistic cost. The methodological distinctions between simulation-optimization and risk-cost-benefit analysis are summarized in Figure 2-7. Beneath these methodological distinctions, there are strong similarities between the two approaches. Both belong to the hard paradigm, and treat decision-making as a rational, algorithmic, unilateral process. Both are presented as methodologies which can be applied to the management of real-world cases of groundwater pollution. Simulation-optimization Simulation model ..evaluates objective function for... Optimization model Hydrogeological decision analysis Simulation model ...evaluates probability component of probabilistic cost (risk) for... Risk-cost-benefit model Figure 2-7. Distinctions between simulation-optimization and hydrogeological decision analysis. To some extent, hydrogeological decision analysis goes beyond the hard paradigm, and shows soft paradigm tendencies such as a relatively rich conceptual model and recognition of complex linkages among systems. There is also underlying concern with matters such as societal goals, conflict, adversarial relationships, equity, justice and ethics (see Freeze, 1988). In particular, Massman and Freeze (1987a, 1987b) use risk-cost-benefit analysis as a basis for discussing the regulatory policy problem of potential groundwater pollution. They examine the problem of potential groundwater pollution at a generic waste management facility, using the objective function of the owner-operator of the facility to evaluate how the owner-operator would 44 respond to alternative regulatory policies. Conclusions are based on argumentation as well as algorithmic output, although the argumentation is closely tied to the algorithmic output. 2.5 Summary This chapter has described the hard and soft paradigms of management. The hard paradigm has its roots in mathematics, science, and engineering. It favours reductionism, hypothesis-testing, objectivity, rationality, and algorithms. The soft paradigm has its roots in responses to perceived deficiencies in the hard paradigm. It favours holism, experience-based learning, subjectivity, bounded rationality, and argumentation. The hard paradigm views a situation as a hard problem, characterized by: independence from other problems; clearly-defmed goals, alternatives, and consequences; and a unilateral decision-maker. The soft paradigm views a situation as a soft problem, characterized by: complicated linkages with other problems; goals, alternatives, and consequences which are not well-defined; and conflict among multiple stakeholders with divergent interests and values. Hard problems can be dealt with through a linear solution procedure; soft problems require iterative cycles of learning, action, and experience. Modeling techniques are used in both paradigms, but in different ways. The hard paradigm uses streamlined conceptual models, and emphasizes the use of transform models to constrain an argument and to predict the consequences of a course of action. The soft paradigm emphasizes detailed conceptual models, and uses transform models as heuristic devices for learning through the exercise of model building. 45 These views of the hard and soft paradigms are necessarily generalizations, although they should not be misconstrued as caricatures. There are, of course, specific exceptions; one may easily find examples of research which is primarily within one paradigm, but shows characteristics of the other. Such examples emphasize the need to recognize the two paradigms. This is not to say that a researcher must be in one camp or the other, but rather that research can be conducted and discussed with greater clarity through recognition of the distinctions. This leads into the concept of soft/hard complementarity, with the soft paradigm supplying the global understanding of an overall soft problem, and the hard paradigm focused on appropriate hard sub-problems embedded within the soft problem. Previous hydrogeological studies on the management of groundwater pollution can be seen in the light of the two paradigms. Simulation-optimization research is based in the hard paradigm, and generally does not go beyond it. Risk-cost-benefit analysis research is also based in the hard paradigm, but may show soft paradigm tendencies. The soft paradigm and soft/hard complementarity provide the theoretical framework for the other elements of the thesis: conceptual model (Chapter 3), management strategies (Chapter 4), and compound model (Chapters 5 and 6). The conceptual model describes groundwater pollution management as a problem with complex rules of governance, multiple stakeholders, multiple issues, multiple modes and processes of decision-making, and pervasive non-quantifiable uncertainty. The management strategies prescribe ways of coping with groundwater pollution, based on the conceptual model and supported by argumentation. The compound model SAM is based on the conceptual model. It is both a heuristic tool in the soft paradigm sense, and an illustration of the management strategies and soft/hard complementarity. 46 3. A conceptual model of groundwater pollution management This chapter presents a general conceptual model which attempts to describe the essential characteristics of groundwater pollution management through the lens of the soft paradigm. The purpose of this model is to organize and communicate a picture of groundwater pollution management, as the descriptive basis for the prescriptive management strategies (Chapter 4), for the compound model SAM (Chapter 5), and for the specific conceptual model in the run of SAM (Chapter 6). The model in this chapter presents a composite, generalized picture of the spectrum of groundwater pollution management in Canada and the United States. There are significant geographic variations across this spectrum, between Canada and the United States (Cherry, 1993), among regions of the United States (Church and Nakamura, 1993), and among Canadian provinces (Cherry, 1993; Ford et ah, 1994). There are also significant variations over time. An effort is made here to capture the range of variations. There are inevitably many exceptions to this generalized picture. These are individually important, but are not necessary for the immediate purpose of this thesis. The systems concept is the basis for the organization of the conceptual model. A groundwater pollution situation can be seen as an interaction between a pollution system and a management system (Figure 3-1). The pollution system is contained within the natural environment and consists of two subsystems: a flow system and a transport system, referring to the flow of groundwater and the transport of pollutants. The management system is contained 47 within the human environment. It consists of two subsystems: a technical system and a decision system. The technical system encompasses the scientific and engineering aspects. It observes the pollution system, provides advice to the decision system, and implements its decisions. The decision system encompasses the stakeholders and their interactions. The term "decision system" is used advisedly, in order to emphasize the complex character of decision-making in groundwater pollution management. The main emphasis in this conceptual model is on the management system. Within the management system, the emphasis is on the decision system; and within the decision system, the emphasis is on negotiation as a mode of decision-making. As noted in the Introduction, these emphases are consistent with the soft paradigm orientation of the thesis, and with recognition that these areas are perhaps less familiar to a hydrogeological audience. HUMAN ENVIRONMENT Technical a q v ' c e J Decision system 4 system M A N A G E M E N T S Y S T E M (decisions Figure 3-1. Systems relationships in the general conceptual model of groundwater pollution management. 4 8 3.1 Pollution system The pollution system is described here very briefly. For greater detail, the reader is referred to a standard text, such as Domenico and Schwartz (1990). The pollution system is seen in terms of two subsystems: a flow system and a transport system (Figure 3-2). stream at risk a. Systems representation. b. Block diagram. Figure 3-2. The pollution system. The flow system involves the flow of groundwater in porous and/or fractured media. Flow occurs in both the saturated and unsaturated zones. The geometry of the flow system involves a complex assemblage of hydrogeologic units. The flow system is heterogeneous with respect to the hydrogeologic properties, particularly hydraulic conductivity (K). This heterogeneity occurs within units as well as between units. Boundaries between units are often irregular. There may be high-K connections which cut across low-K units, providing significant pathways for pollutant transport. 49 The transport system consists of source(s), plume(s), and receptor(s). The source can be subdivided into primary and secondary sources (see Mackay et al, 1993). The primary source is the immediate region where chemicals have been dumped or injected. Local redistribution of pollutants from the primary source generates a larger and more diffuse secondary source. The mechanism of this local redistribution is often multi-phase flow. The secondary source is often difficult or impossible to remove (NRC, 1994). Advection is the main mechanism for the transport of pollutants (Freeze et al., 1990), generating a plume of dissolved pollutants which moves away from the source in the direction of groundwater flow. Other mechanisms may also be important; these include dispersion, colloidal transport, and transport as a separate phase. Pollutant transport may be retarded by sorption to the solid media, and pollutant mass may be lost from the transport system due to degradation or volatilization. The receptors may be water wells or streams, where the pollutants can damage human or ecosystem health. Both flow and transport are transient phenomena in general, although a flow system may achieve a steady-state condition if it is not perturbed. The pollution system possesses a high degree of complexity, arising from complicated chemical, biological, and physical interactions in a multi-phase system which is spatially heterogeneous with respect to important characteristics such as hydraulic conductivity (NRC, 1994, Chapter 2). 3.2 Technical system The roles of the technical system are to observe the pollution system, generate technical advice for the decision system, and implement technical decisions (Figure 3-3). This model of 50 the technical system allows for the cycle of observation->advice->decision->implementation to repeat indefinitely. observation of the pollution system r Management system Technical organizations Technical goal Linear or iterative technique Technical uncertainty Technical system technical advice based on observations r \ Decision system implementation of technical decisions technical decisions <• based on technical advice Figure 3-3. The technical system. The elements of the technical system can be described in terms of technical organizations, technical goal, linear or iterative technique, and technical uncertainty. The technical organizations are the organizations which have the scientific and engineering skills to perform the roles of the technical system. The technical organizations may be independent consulting and contracting firms, or in-house bodies within regulatory agencies or large corporations. The technical goal is the ultimate goal to which the management system aspires. It is decided upon by the decision system, under advice from the technical system. The technical goal can be described in terms of two end-members of a spectrum: restoration and mitigation. Restoration is the goal of protecting human and ecological health by achieving a permanent remedy, with a definite endpoint. It requires the extraction or in-situ treatment of pollutants down to concentration standards at which no further action is required. Extraction involves excavation of polluted solid materials and pumping of polluted groundwater, with subsequent treatment or disposal. In-situ treatment involves the introduction of chemical or 51 microbial agents to transform the pollutants into harmless substances, without the need for extraction. Success in restoration is measured in terms of numerical concentration standards for pollutants in water and soil, or mass of pollutants removed. Concentration standards are typically set by statutes or regulations at very low values which apply to all cases. Under some statutes and regulations, site-specific standards may be substituted for the general standards. Pump-and-treat is the classic example of a restoration technique, intended to remove a plume of polluted groundwater. It involves the withdrawal and injection of water through a system of wells and/or trenches, combined with surface treatment of polluted groundwater which is removed by this system. Pump-and-treat has been a major element of restoration efforts, with largely disappointing results (Travis and Doty, 1990; NRC, 1994, Chapter 3). Restoration is associated with linear technique. Linear technique is intended to proceed through a progressive sequence of actions which culminate with the endpoint of achieving the technical goal. Linear technique is associated with linear decision-making (see Section 3.3.4.4). Mitigation is the goal of reducing damage and risk to human and ecological health, without attempting to provide a permanent remedy. It involves actions which must continue for a long and indefinite period. Physical or hydraulic containment of pollutant migration is an important aspect of mitigation. The purpose of mitigation is to reduce damage and risk to human and environmental health. In cases where restoration cannot be achieved, mitigation becomes the technical goal by default. Hydraulic containment is the mitigative equivalent of pump-and-treat. It is similar in implementation to pump-and-treat, but is less aggressive and is designed with the goal of containing the plume of polluted groundwater rather than removing it. 52 Mitigation is associated with iterative technique, which involves a repetitive and open-ended approach to the design and implementation of technical options, moving towards a condition of closure, but with no set endpoint. It is associated with iterative decision-making (Section 3.3.4.4). In many cases, restoration and linear technique fail to achieve their purpose, and change by default to mitigation and iterative technique. There is a growing technical consensus that many cases of groundwater pollution are inherently persistent problems for which restoration is not a feasible goal (Freeze and Cherry, 1989; Travis and Doty, 1990; NRC, 1994). 3.3 Decision system The decision system (Figure 3-4) is described here in terms of rules of governance, stakeholders, issues, decision-making, and complexity and uncertainty. Management system / Pollution ( system Technical system Decision system /'Rules of governance^ /'Stakeholder^ Legislation Statutes Regulations Jurisprudence Government policy Regulatory Responsible Injured Peripheral Decision-making Modes of decision-making Processes of decision-making Linear and iterative decision-making ' Issues Technical decisions Responsibility Compensation Complexity and uncertainty Figure 3-4. The decision system. 53 3.3.1 Rules of governance Groundwater pollution management operates under a set of complex and linked rules of governance, established by various levels of government. The rules of governance include legislation (statutes and regulations), jurisprudence, and government policy. The legal definitions in this section are taken from Yogis (1990) and Castel and Latchman(1994). Legislation includes both statutes and regulations. A statute is a formal set of rules passed into law by the legislative branch. Regulations are detailed rules created by the executive or administrative branches under the authority of a specific statute. Statutes and regulations are applied by the administrative branch of government, and interpreted by the judicial branch. Jurisprudence refers to rules established by previous judicial decisions, and, to a lesser degree, by decisions of administrative tribunals established under specific statutory authority. Common law refers to law in an area that is governed only by previous judicial decisions. Government policy refers to general courses of action established by the executive and administrative branches. For example, government policy might involve rigorous administrative enforcement of a particular set of regulations. Liability and litigation are important aspects of the rules of governance for groundwater pollution. Liability refers to a legal obligation or responsibility. Litigation refers to a contest in which two or more parties go to court for a judicial resolution of their differences. Both statutes and common law may provide legal grounds for liabilities such as a polluter's liability for remediating a polluted site or for damage to human health. These topics are discussed further in Section 3.3.4.3. 54 Rules of governance may be directly or indirectly applicable to groundwater pollution. The classic example of a directly applicable statute is the United States federal Superfund statute1 (see Fogleman, 1992). A Canadian example of a directly applicable statute is the British Columbia provincial Waste Management Act. As an example of indirectly applicable rules, note that Superfund itself has no numerical concentration standards, but refers to other "applicable or relevant and appropriate regulations" (ARARs) (Fogleman, 1992, p.45). In another example of an indirectly applicable statute, the British Columbia Waste Management Act makes important references to the Canadian federal Transportation of Dangerous Goods Act. Rules of governance for groundwater pollution management vary from jurisdiction to jurisdiction. In the United States, groundwater pollution management is both a federal and a state responsibility. In Canada, it is largely a provincial responsibility, and the rules of governance vary widely from province to province (Ford et al., 1994). An important difference between the United States and Canada lies in the distinction between a statute which compels the regulatory body to take enforcement action and a statute which enables the regulatory body to take enforcement action at its own discretion. The United States tends to favour compelling statutes, and Canada tends to favour enabling statutes. The rules of governance for groundwater pollution management vary over time. At the time of writing, the relevant rules of governance are undergoing drastic revision, both in Canada (e.g., Matas, 1996) and in the United States. These changes are accompanied by changes in related factors, particularly a reduction in the funding for regulatory agencies. 1 More precisely: the Comprehensive Environmental Response, Compensation and Liability Act of 1980 (CERCLA), and subsequent re-authorizations. 55 3.3.2 Stakeholders The management of a typical groundwater pollution case involves a large set of stakeholders (cf. Wartenberg, 1988; Mazmanian and Morell, 1992, p. 57-75; Greene, 1994). A stakeholder is any party which has an interest in the management of the problem, and which is sufficiently well-organized to pursue its interests and make decisions about how to cope with the situation. The stakeholders can be classified into four main categories: regulatory, responsible, injured, and peripheral. Regulatory stakeholders are agencies of the administrative branch of government, charged with applying the rules of governance with respect to groundwater pollution. A single case of groundwater pollution involves a set of regulatory stakeholders. This set may include agencies with different responsibilities, such as groundwater supply as opposed to groundwater pollution. It may also include agencies at different levels of government, such as provincial as opposed to federal. In the United States, the federal Environmental Protection Agency has a major role in regulating groundwater pollution. In Canada, the regulatory authority for groundwater pollution lies mainly with provincial agencies. There is a certain nuance to classifying regulatory agencies as stakeholders, rather than as impartial referees (see Hoberg, 1993, p. 319-320). It implies that each regulatory agency has its own set of interests, and that the interests of different regulatory agencies may diverge and even conflict. A further implication is that individual regulatory agencies are not necessarily assumed to represent the interests of society as a whole. Responsible stakeholders are individual citizens, corporations, or government agencies who are responsible for the costs of a groundwater pollution case. The assignment of 56 responsibility is typically problematic, and a source of conflict. In Superfund jargon, stakeholders who may be responsible are referred to as "potentially-responsible parties" (PRPs). A responsible stakeholder may acquire responsibility indirectly, rather than through the direct act of causing the pollution. Depending on the rules of governance, examples of indirectly responsible parties can include: • a government assuming responsibility for an abandoned ("orphan") site; • the vendor or purchaser of a polluted site; • the generator of wastes found at a polluted site; • a financial institution acquiring responsibility through the act of providing a loan to another responsible stakeholder; • an insurance corporation acquiring responsibility through a policy held by another responsible stakeholder. Conversely, the stakeholder which directly caused the pollution may avoid paying, as in the case of a defunct corporation which cannot be brought to task for its action. Injured stakeholders are those who are negatively affected by the pollution itself (other than the responsible stakeholders). Consider the following examples of injured stakeholders in a case of groundwater pollution: • the local community, which might be exposed to the toxicological and psychological consequences of polluted groundwater; • the operator of a water well, who might lose the well as a source of water, and be exposed to litigation for supplying polluted water to consumers; 57 • a local environmental non-governmental organization, which might be concerned with discharge of polluted groundwater to local aquatic ecosystems; • the local municipal government, which might face a decrease in its taxbase as a consequence of the presence of the polluted site. Injurious consequences may be direct or indirect; they may be immediate or eventual; and they may be actual or potential. The major concerns of injured stakeholders are often focused on health impacts. The health impact of chronic low-level chemical pollution on humans and other biota is a controversial issue with high stakes and high uncertainty, and there is neither scientific nor societal consensus (see, for example: Carson, 1962; Colburn et al, 1990; NRC, 1991; Peterle, 1991; BMA, 1991; Francis, 1994; IJC, 1994; Corcoran, 1994). The difficulty in evaluating the health impact of chemical pollution in groundwater is even greater, because toxicological uncertainty is compounded by uncertainty about the pathway of exposure (NRC, 1991; Maughan, 1993). On a national scale in the US, the human health consequences of groundwater pollution do not appear to be high relative to other forms of pollution (EPA, 1987). This conclusion is qualified by large uncertainties in the study, and by the recognition that averaging over a national scale may hide the severity of local consequences. There is a small to moderate increase in risk for specific cancers and birth defects for persons living in proximity to a polluted site and/or consuming polluted water (Johnson, 1993; Lybarger, 1993; Johnson, 1995). There are often negative psychological consequences for individuals and communities which are exposed to groundwater pollution (Edelstein, 1988; Vyner, 1988; Greenberg et al, 58 1994; Hitt, 1995). These impacts include disruption to normal life as well as fear of toxicological consequences, and are exacerbated by the uncertainty of the situation. Peripheral stakeholders are stakeholders who may indirectly assist or hinder management, without being otherwise involved. An example of assistance would be a factory which is able to use polluted groundwater in an industrial process. An example of hindrance would be a new excavation adjacent to a polluted site, interfering with efforts to contain the pollution. Note that some stakeholders may not be recognized as such until they unexpectedly affect the management of a groundwater pollution case. Unrecognized stakeholders have the potential to destabilize existing management, once their involvement becomes apparent. Injured stakeholders may be unrecognized until the sudden revelation of injury, such as the discovery of pollution in a water supply well. Unrecognized responsible stakeholders are often revealed by the chain of litigation which is associated with many cases of groundwater pollution. Peripheral stakeholders may go unrecognized until they undertake an action which affects management. Technical and legal experts are excluded from consideration as stakeholders in this model, although they could be regarded as such. Technical experts are treated as an explicit component of the technical system. Legal experts are treated as an implicit component of the decision system. The localized nature of most cases of groundwater pollution is an important characteristic, strongly influencing the dynamics of management and the identification of stakeholders. Potentially-responsible stakeholders can be readily identified through their association with the polluted site. Injurious consequences are concentrated in a small and distinct 59 area, providing a strong incentive and opportunity for injured stakeholders to organize. In contrast, consider air pollution, which often comes from many widely-distributed and difficult-to-identify polluters, with injurious consequences diffused over a large area. 3.3.3 Issues Groundwater pollution cases typically involve a complex set of issues. These can be classified into the categories of technical decisions, responsibility, compensation, and peripheral issues. Technical decisions are concerned with choices of management goal (restoration or mitigation), and choices of technical options prepared by the technical system. Responsibility (liability) is a critical and contentious issue. A single case of groundwater pollution often involves multiple potentially-responsible stakeholders, each of whom will seek to minimize their share of responsibility. This is a difficult ethical and legal issue with far-reaching repercussions, and this thesis does not attempt to address it directly. Probst and Portney (1992) provide an analysis of this issue in the United States context; Ford et al. (1994) offer a discussion in the Canadian context. Compensation to injured stakeholders is an issue which frequently involves litigation, for both human health injuries (Jorgensen, 1989, p. 106-109) and economic injuries (Kopp and Smith, 1993; Wruck, 1993). As noted above, the localized nature of groundwater pollution results in local concentration of injurious consequences and ready identification of responsible stakeholders, increasing the likelihood that injured stakeholders will seek compensation. 60 Peripheral issues include a variety of matters which arise from groundwater pollution cases. For example, the owner of a polluted site may seek relief from municipal taxes, due to reduction in the value of the land (Sidnell, 1993). 3.3.4 Decision-making Decision-making can be described in terms of the mode of decision-making, the decision-making process, and the distinction between linear and iterative decision-making. The mode of decision-making refers to the distribution of decision-making authority among the stakeholders who are involved in a problem (Dorcey and Riek, 1987). In negotiation, the authority is shared among a subset of at least two stakeholders. In unilateral decision-making, the authority is vested in one stakeholder or an outside party such as a judge. Negotiation is an important mode of decision-making in the management of groundwater pollution; it is a focus of this thesis, and is described in detail in Section 3.3.4.1. Two types of unilateral decision-making are also important in the management of groundwater pollution: unilateral regulatory decision-making (Section 3.3.4.2) and judicial decision-making (Section 3.3.4.3). A decision-making process is an interaction among a subset of the stakeholders who are involved in a problem, using one or more modes of decision-making, with the intent of reaching decisions over a subset of issues (Dorcey and Riek, 1987). For example, a regulatory process may involve negotiation between the regulator and one of the stakeholders on one subset of issues, and unilateral regulatory decision-making on another subset of issues. The subset of stakeholders who interact in a given process are participants in that process. 61 A groundwater pollution case typically involves multiple modes and processes of decision-making, operating both simultaneously and sequentially. For example, a regulator may be simultaneously involved in negotiation with one responsible stakeholder and litigation with another; the litigation may lead sequentially to another negotiation. The particular combination of modes and processes of decision-making reflects both the relevant rules of governance and the particular circumstances of the case. Linear and iterative decision-making are discussed in Section 3.3.4.4. Linear decision-making passes through a progressive sequence of phases to a set endpoint. Iterative decision-making is repetitive and open-ended. 3.3.4.1 Negotiation This thesis places considerable emphasis on negotiation as a mode of decision-making, and this section provides a detailed treatment of the topic. Negotiation is common in groundwater pollution management, in a number of circumstances: • A regulatory stakeholder and one or more responsible stakeholders may negotiate in order to decide on the technical goal and technical options. This is the dominant method of arriving at such decisions in the United States (Gilbert, 1989; Brennan and Frazier, 1991; Fogleman, 1992), and also plays an important role in Canada (e.g., McKenna, 1993). In the United States, federal statutes tend to be very precise about the nature of negotiations between the regulatory stakeholder and the responsible stakeholder (Brennan and Frazier, 1991; Fogleman, 1992). In Canada, such negotiations are less formal and not as subject to legal constraints. This difference is consistent with the general difference in the style of 62 environmental statutes between the United States and Canada (Harrison, 1993; Hoberg, 1993). • Two or more responsible stakeholders may negotiate in order to determine the sharing of costs (Gilbert, 1989; Brennan and Frazier, 1991). The non-profit organization Clean Sites Inc. has been a leader in facilitating and mediating such negotiations in the United States (Susskind and Cruikshank, 1987, p. 231; Brennan and Frazier, 1991). In Canada, the negotiated approach to sharing of costs among responsible stakeholders has been endorsed by a federal-provincial task force (Ford et al, 1994, pp. 32-33). • Two or more regulatory stakeholders may negotiate in order to clarify the distribution of regulatory authority (e.g., Greene, 1994). • Regulatory stakeholders, responsible stakeholders, and injured stakeholders may negotiate in an effort to agree on management activities (e.g., Pinder et al., 1995). In negotiation, decision-making power is shared among two or more participants who communicate with one another in an effort to achieve some consensus or compromise about issues of mutual or conflicting interest. Consensus is a negotiated decision which is not rejected by any of the participants. Compromise is a negotiated decision which involves the granting of a concession by one or more of the participants, often on a quid pro quo basis. A number of terms are used in the literature to describe negotiated decision-making: "bargaining", "alternative dispute resolution" (ADR), "dispute settlement", "conflict resolution", "conflict management", and "mediation". These terms may all be modified by the adjective "environmental", for cases which involve some interaction between human and environmental systems. This thesis will use only the term "negotiation". 63 A participant in a negotiation consists of a negotiator who is involved in the actual communication, and a constituency which the negotiator represents. Generally, any agreement arising from a negotiation must be ratified by the constituencies. Raiffa (1982) developed a number of useful concepts in negotiation, on the basis of both practical experience and expertise in the field of game theory. Some of Raiffa's most important concepts are summarized in this paragraph. Reservation price is the minimum that a participant is prepared to settle for in a negotiation. Distributive (zero-sum) negotiation involves only a division of the metaphorical pie, with no possibility of tradeoffs and joint gains. Integrative negotiation involves a potential for joint gains (synergies) and for tradeoffs, and consequently tends to be more tractable than distributive negotiation. The potential for joint gains arises when the participants develop a negotiated agreement in which each participant achieves a greater benefit than they would have achieved through purely distributive negotiation, or through unilateral action. The potential for tradeoffs arises through asymmetric interests. Examples of asymmetric interests include different perceptions of matters such as the time value of money or the probabilities of future events. This last example opens the possibility of dynamic compromise, in which future payoffs or actions are contingent upon the occurrence of some event. As a simple example of distributive and integrative negotiation, consider the purchase of an automobile. This can involve a purely distributive negotiation, as a straightforward contest in which buyer and seller attempt to meet and better their mutual reservation prices. It can be transformed into an integrative negotiation by, for example, the inclusion of a warranty, which is 64 a tradeoff reflecting differences in perceptions of buyer and seller regarding the probability that the automobile will require repairs during the warranty period. In a negotiation, there is a complicated relationship among complexity, uncertainty, and tractability. High degrees of complexity and uncertainty obviously have negative implications for the tractability of negotiations. But complexity and uncertainty can also increase tractability (Susskind and Cruikshank, 1987). In general, the more complex and uncertain a situation, the greater the opportunity to develop joint gains and tradeoffs; in other words, to transform distributive negotiation into integrative negotiation. Furthermore, the participants' uncertainty about the consequences of failure to reach an agreement can be an important aspect of negotiations, driving the participants from unilateral action towards the greater certainty of a negotiated outcome (Cormick, 1989). Confidentiality is often an essential element of negotiation (Fisher and Ury, 1981; Raiffa, 1982; Cormick, 1989). One aspect of confidentiality is the need for negotiators to keep key information secret from the other negotiators. Another aspect of confidentiality is the need for negotiators to work in private without continual exposure to their constituencies. Negotiating power is largely a function of the unilateral alternatives which the stakeholders can exercise if they fail to reach agreement. To a large extent, any negotiation is driven by these unilateral alternatives. The topic of negotiating power and unilateral alternatives will be examined here through two complementary concepts: the BATNA and the worst-nightmare. Note that BATNAs and worst-nightmares often involve litigation. BATNA is an acronym for Best-Alternative-To-a-Negotiated-Agreement (Fisher and Ury, 1981; see also Susskind and Cruikshank, 1987). A stakeholder's BATNA is its most-65 preferred unilateral action in case of failure to reach agreement. For example, in a labour negotiation, the union's BATNA would be to declare a strike. For an example from groundwater pollution management, consider a hypothetical negotiation between a regulatory stakeholder and a responsible stakeholder. The BATNA of the regulatory stakeholder would be to launch a lawsuit against the responsible party. The BATNA of the responsible stakeholder would be to defend itself in court against the lawsuit. Fisher and Ury (1981) argue that a participant's power at the negotiating table is based on the strength of its BATNA, and not necessarily on the overall power of that participant away from the table. A participant with a strong BATNA can press its negotiating points, secure in the strength of its alternative. A participant can improve its negotiating position by taking steps to strengthen its BATNA. In a case of failure to reach agreement, a number of possible outcomes arise as the participants exercise their BATNAs. The term worst-nightmare describes the worst possible outcome stemming from failure to reach agreement, from the perspective of any given participant. It embodies the participants' uncertainty about the consequences of failure to reach an agreement. (The term "worst-nightmare" is original to this thesis, but the concept is not.) For example, in the case of a union exercising its BATNA of declaring a strike, there are a number of possible outcomes, ranging from immediate capitulation by the company to the permanent closure of the factory. The union's worst-nightmare would be permanent closure of the factory. In the groundwater pollution example, the worst-nightmare of either stakeholder would be to lose the lawsuit. 66 In theory, no participant will accept a negotiated agreement which they perceive to be worse than their worst-nightmare (Fraser and Hipel, 1984, p. 186). It follows that a participant's negotiating power depends inversely on the severity of its worst-nightmare. A participant with an unpleasant worst-nightmare will be eager to reach agreement, and may accept a relatively unfavourable agreement rather than exercise its BATNA. Conversely, a participant with an innocuous worst-nightmare can hold out for a favourable agreement, and will feel free to exercise its BATNA. Note the contrast between BATNA and worst-nightmare. The BATNA concept applies to unilateral actions which may be taken by a given participant, should negotiations fail. The worst-nightmare concept applies to the consequences, for a given participant, of the unilateral actions of all participants following failure of negotiations. The BATNA carries the positive connotation of empowerment through a credible alternative to negotiation; the worst-nightmare carries the negative connotation of unpleasant consequences if agreement is not reached. Note the linkage between BATNA and worst-nightmare: a BATNA is weakened by a bad worst-nightmare. A participant's BATNA and worst-nightmare are not "initial conditions". They are dynamic, and change with changing circumstances. A participant can take measures to strengthen its BATNA and reduce the severity of its worst-nightmare; but such measures are constrained both by the rules of governance and by the limitations of the participant's financial resources. In some jurisdictions, the regulatory stakeholder has a strong BATNA against responsible stakeholders, based mainly on ability to litigate (see Section 3.3.4.3). The worst-nightmare of a regulator may involve scenarios such as failure to get a site cleaned up, or failure to recover costs. In the United States, the balance of negotiating power has tended to favour the regulatory 67 stakeholder over responsible stakeholders (although this currently appears to be changing). In Canada, the regulatory stakeholder is not generally so powerful; this is more true in some provinces than in others (see Ford et al., 1994). The BATNA of an injured stakeholder depends on its legal standing and financial resources to engage in litigation (see Section 3.3.4.3). There is much greater scope for such injured stakeholders to litigate in the United States, as compared to Canada. In the case of an entire community affected by groundwater pollution, the BATNA will also depend on the degree of organization among its members. The worst-nightmare of an injured stakeholder involves failure to have the injury addressed, and the spending of resources in a futile effort to gain redress. The BATNA of a responsible stakeholder tends to be inversely proportional to that of the regulatory and injured stakeholders. The worst-nightmare of a responsible stakeholder can be extreme, given the very high costs associated with groundwater pollution management. Large financial resources are a mixed blessing for a responsible stakeholder. They permit it to strengthen its BATNA to the fullest extent possible under the relevant rules of governance, but they also increase the severity of the worst-nightmare (the "deep pockets syndrome"). Principled negotiation is a philosophy and style of negotiation developed in a series of works produced by the Harvard Negotiation Project (Fisher and Ury, 1981; Fisher and Brown, 1988; Ury, 1991; Fisher et al, 1991; Fisher et al, 1994; see also Raiffa, 1982), and advocated by Dorcey (1986) in the context of natural resources management in British Columbia. It is summarized briefly here, and described in greater detail in Appendix 1. 68 Principled negotiation is cooperative, integrative negotiation in which participants generate joint gains through tradeoffs and synergies; it is summed up by the catch phrase "getting to yes." The essence of principled negotiation is the desire to arrive at a consensual agreement which genuinely satisfies the interests of all stakeholders. It represents an opportunity for the participants to arrive at a decision that integrates their interests and values, or at least identifies compatible interests and values. This opportunity arises through the constructive exchange of information and arguments. Principled negotiation is an alternative to the competitive or positional style of negotiation, in which the negotiator's aim is to obtain concessions from the other participants. Competitive negotiation is zero-sum, distributive negotiation — any gain for one stakeholder is a loss for another. The consensual aspect of negotiation may be more apparent than real, as a stakeholder may be misled or coerced into accepting a decision against its interests. In practical terms, a principled negotiator will attempt to: • Negotiate so as to respond to the participants' underlying interests, rather than their stated positions. • Maintain good relationships among the participants, while simultaneously taking a hard-nosed approach to the problem ("separate the people from the problem"). • Generate a number of options for joint gains and tradeoffs ("win-win options") before reaching a decision. • Use objective criteria to cope with contentious issues. Mediation refers to assistance with procedural and substantive matters in a negotiation, rendered by a neutral party. Folger and Bush (1994) see mediation as having the potential to 69 transform, in a positive sense, both the stakeholders in a conflict and society as a whole. Fisher and Ury (1981) see mediation as a means of changing positional negotiation into principled negotiation. Susskind and Cruikshank (1987) advocate mediation as a means of converting a distributive negotiation into an integrative negotiation. Mediation lies within a spectrum of neutral-party assistance (Raiffa, 1982; Susskind and Cruikshank, 1987), ranging from purely procedural assistance (facilitation) to actual decision-making (arbitration). It encompasses a variety of tasks (see Susskind and Cruiksank, 1987, Table 5.1): • getting the representatives of the stakeholders together; • helping to draft procedural protocols and substantive agenda; • helping stakeholders to recognize their BATNAs; • arranging linkages with the technical system; • helping to invent and package options into a draft agreement; • helping to get the constituencies to ratify the draft agreement; • monitoring implementation; and • initiating a new iteration of negotiation. The mediator's success is dependent on the stakeholders' perception of the mediator's neutrality. This is a complex issue, depending on the reputation of the mediator and the source of funding for the mediator (Bingham, 1986; Susskind and Cruikshank, 1987). There is a wide spectrum of opinions on the success and legitimacy of mediation in environmental matters. Susskind and Cruikshank (1987) and Cormick (1988, 1989) advocate mediation as a successful and legitimate mechanism in environmental disputes, conditional on a 70 number of circumstances. Most importantly, they emphasize that disputes over fundamental principles are unlikely to be resolved by mediation, or by negotiation in general. Amy (1987) delivers a strong criticism of environmental mediation, arguing that it works to the advantage of the more powerful stakeholders and to the detriment of other stakeholders, particularly local communities and environmental organizations. Amy is in agreement that disputes over fundamental principles are not tractable to mediation or negotiation in general; but he further feels that most environmental disputes fall into this category. 3.3.4.2 Unilateral regulatory decision-making Unilateral regulatory decision-making is carried out by a regulatory stakeholder, subject to the rules of governance. It generally involves some communication of information and arguments from other stakeholders to the regulatory stakeholder. In fact, there is a continuum leading from purely unilateral regulatory decision-making, through consultation where the regulator is obligated to hear the views of a subset of stakeholders but retains unilateral decision-making authority, to true negotiation in which decision-making authority is shared with a subset of stakeholders (Dorcey and Riek, 1987). In a negotiation, the regulatory stakeholder's ability to invoke unilateral regulatory decision-making may provide it with a strong BATNA. Unilateral regulatory decisions may be appealed to a judicial body, transforming regulatory decision-making into judicial decision-making. The rules of governance may also include provision for an intermediate step of appeal to a quasi-judicial administrative body, such as the environmental appeal boards which are found in a number of Canadian provinces. The decisions of such a body are subject to judicial review. 71 A regulatory stakeholder may have the authority to decide unilaterally on technical matters, as well as issues such as the distribution of costs among the responsible stakeholders. Such decisions must be supported by some form of power to enforce. Regulatory decisions can be enforced by mechanisms such as fines, injunctions, liens, litigation, and criminal prosecution. In the US, the regulator generally has a great deal of enforcement power, particularly under the Superfund statute (see the following section). In Canada, the regulator generally has much less power to enforce. This point is underlined by a recent judicial decision in Ontario, which upheld a decision of the Ontario Environmental Appeal Board that a stakeholder should not be liable for substantial costs as a consequence of some indirect action such as purchasing a polluted site (Matas, 1995). Regulatory stakeholders are answerable to elected officials. It follows that a unilateral regulatory decision may be sensitive to political pressure exerted by various stakeholders. Injured stakeholders may apply pressure through activities such as participation in hearings, lobbying, petitioning, protesting, and even civil disobedience (Jorgensen, 1989; Szasz, 1994). Responsible stakeholders may also make use of political pressure to influence regulatory decisions. 3.3.4.3 Judicial decision-making Judicial decision-making occurs through a formal process in which decisions are made by a judge or jury, subject to the rules of governance. For groundwater pollution cases, this generally occurs through litigation, in which a set of one or more participants brings a lawsuit against another set of one or more participants, with a judge or jury deciding the outcome. Other judicial processes also play a role in groundwater pollution cases, particularly judicial appeal of a unilateral regulatory decision. The emphasis in this section is on litigation. 72 There is a close relationship between litigation and negotiation. The threat of a lawsuit can provide a BATNA for one or more of the participants in a negotiation, and may also constrict the flow of information among the participants. Once initiated, a lawsuit often leads to negotiations among the litigants. This may be encouraged or even compelled by the judge. The scope for litigation in groundwater pollution management depends on the specific jurisdiction. The situations in the United States and in Canada are discussed here separately. Litigation in environmental disputes plays an important role in the United States. Susskind and Cruikshank (1987) are highly critical of this role for litigation, arguing that it leads to protracted impasse and amplifies adversarial relationships. Amy (1987) is very supportive of this role for litigation, arguing that it empowers the less-powerful stakeholders and provides a formal process which is less susceptible to manipulation by the powerful stakeholders. The United States Superfund law gives the federal Environmental Protection Agency strong powers to sue a responsible stakeholder in an effort to force that stakeholder to implement technical actions, or to recover technical and other costs incurred by the regulatory stakeholder (Fogleman, 1992). The responsible stakeholder is subject to liability which is strict, joint and several, unlimited, and retroactive (retrospective). Strict liability (in United States legal jargon) is liability which applies whether or not the defendant was negligent; this is a very strong standard of liability (Barnett, 1994, p. 59). Joint and several liability allocates responsibility to all stakeholders collectively, and any one stakeholder can be assigned all of the responsibility, regardless of that stakeholder's actions. (This is distinct from several liability, in which responsibility is apportioned to each responsible stakeholder on the basis of that stakeholder's actions.) Unlimited liability implies that there is virtually no limit, other than ability to pay, on 73 the amount that a responsible stakeholder may be compelled to pay. Retroactive (retrospective) liability covers past acts which may not have been illegal at the time of commission, and which may even have been encouraged by government agencies. In addition, Superfund law provides a treble-damages provision, which permits the Environmental Protection Agency to unilaterally undertake technical actions, and then sue a responsible stakeholder for three times the costs of those actions. Several other types of lawsuit related to groundwater pollution management can be brought to court in the United States: • A responsible stakeholder may sue another responsible stakeholder in order to recover a share of technical costs (contribution lawsuit) (Fogleman, 1992). • An injured stakeholder may launch a citizen suit against a regulatory stakeholder in an effort to force it to undertake certain regulatory actions, or a responsible stakeholder for violating federal environmental regulations; provision for such lawsuits is included in most US federal environmental laws (Hoberg, 1992, p.50,78; Jorgensen, 1989). • An injured stakeholder may launch a citizen suit against a potentially responsible stakeholder (Jorgensen, 1989). • A legally-designated resource trustee may sue a responsible stakeholder for natural resource damages, such as damages to groundwater supply (Kopp and Smith, 1993). • An injured stakeholder may sue a responsible stakeholder for damage to health ("toxic tort"), loss of property value, or even medical monitoring. Such lawsuits may be massive, involving thousands of plaintiffs (Whiting, 1994; Hitt, 1995). 74 • A responsible stakeholder could sue an injured stakeholder in an effort to discourage the injured stakeholder from participating in management of the problem, on a variety of grounds (e.g., interference with commercial interests). Such lawsuits are known as SLAPP suits (Strategic Lawsuits Against Public Participation), and are the subject of anti-SLAPP statutes in a number of states in the United States (California Anti-SLAPP Project, 1996). There is less scope for litigation in Canada than in the United States, both generally and for groundwater pollution management in particular. The power of a regulatory stakeholder to sue a responsible stakeholder is generally much more restricted in Canada than in the US. In British Columbia, a statute with Superfund-style provisions was passed by the provincial legislative body in 1993, but was never proclaimed into law. Canadian laws tend to explicitly protect the government from lawsuits issued by private citizens, in contrast to the US concept of the citizen suit. SLAPP suits have appeared in Canada, but no anti-SLAPP statutes have been enacted. Note that Canadian legal jargon differs somewhat from US legal jargon (Ford et al, 1994, p. 32). In particular, there is a distinction between absolute liability and strict liability. Absolute liability applies to activities regardless of whether or not they were regulated or in compliance at the time; it corresponds to retroactive liability. Strict liability is less severe than absolute liability, and allows for a defence based on due diligence or reasonable care in cases where an activity was unregulated, or regulated and in compliance. Note that "strict liability" is weaker under the Canadian definition than under the US definition. 3.3.4.4 Linear and iterative decision-making Decision-making may be either linear or iterative. Linear decision-making passes through 75 a progressive sequence of phases to a set endpoint. Iterative decision-making is repetitive and open-ended. It may move towards a condition of closure, but there is no set endpoint. Formally linear decision-making may become ad hoc iterative decision-making, under the pressure of unexpected events, unintended consequences, and unachieved goals. Decision-making in groundwater pollution management tends to be formally linear, passing through a sequence of phases from initial investigation through to a final decision. The prime example of this is the United States Superfund law, with its sequence of remedial investigation/feasibility study, record of decision, and remedial design/remedial action (D. Dzombak, pers. comm., 1994; J. Wilson, pers. comm., 1994; Hoffman, 1994; Church and Nakamura, 1993, p. 6-7). In practice, formal linear decision-making often fails to cope with the problem, leading to ad hoc iterative decision-making. For example, this has been the case at the Iron Mountain Superfund site in California (Nordstrom and Alpers, 1995); and at the Sydney Tar Ponds in Nova Scotia (MacLeod, 1996; Maclntyre, 1996; Underhill, 1996, McNamara, 1996). 3.4 Complexity, uncertainty, and integration Complexity, uncertainty and integration are important aspects of the management system. Technical uncertainty is an important aspect of the technical system, reflecting the high degree of complexity in the pollution system (Section 3.1), and the expense and difficulty of subsurface investigations. The degree of technical uncertainty is sufficiently large that every case of groundwater pollution can be considered as a new and distinct technical experiment (Freeze and Cherry, 1989; NRC, 1994). Technical uncertainty is partially quantifiable, but always retains a large component which cannot be quantified. 76 The technical system itself may exhibit a high degree of complexity, particularly if it is linked to a complex decision system. The degree of complexity and uncertainty in the decision system is generally high, depending on: • the complexity of the relevant rules of governance, and uncertainty in their application; • the number of stakeholders, the complexity of their relationships, and the uncertainty of their behaviour; • the number and complexity of issues and their linkages; • the number of decision-making processes, their linkages, and the uncertainty of their outcomes; • the degree of complexity and uncertainty in the technical system. The uncertainty in the decision system is largely non-quantifiable. Uncertainty in the management system also applies at a personal level. Individuals involved in the technical or decision systems will be legitimately concerned about the uncertainties in their lives stemming from their personal actions. The management system in a given case varies in the degree of integration. This concept applies to both the decision systems and the technical systems, as well as to the management system as a whole. An integrated decision system consists of a small number of closely-linked decision-making processes. It is associated with an integrated technical system. One style of technical integration is coordinated technical support, which involves coordination and cooperation among the organizations which provide separate technical support for individual stakeholders. A stronger style of technical integration is joint technical support, in which the 77 stakeholders jointly retain one or more technical organizations, and jointly receive the technical advice. In practice, integration is difficult to implement, and is not typical in groundwater pollution management. Integration may occur by accident, by necessity, or by design. An example of integration-by-accident is the case of the former Expo Lands (Pacific Place) in Vancouver, British Columbia. This large package of urban land was assembled and sold by the provincial government. The lands were then discovered to be polluted, and the provincial government assumed responsibility. Consequently, the technical work over this large geographic area was carried out in an integrated fashion. An example of integration-by-necessity is the case of the San Gabriel basin in Los Angeles, California. As a consequence of the very large number of individual polluted sites, this basin was treated in an integrated fashion by the United States Environmental Protection Agency (CH2M Hill, 1990a). An example of integration-by-design is the Remedial Action Plan (RAP) process which has been developed to deal with polluted Areas of Concerns (AOCs) around the Great Lakes, under the 1987 Great Lakes Water Quality Agreement between Canada and the US (Hartig and Zarull, 1992). A RAP process brings a wide spectrum of problems and stakeholders under a single umbrella for each Area of Concern. 3.5 Summary and discussion This chapter has presented a general conceptual model of groundwater pollution. This is a generalized composite picture, based on soft paradigm thinking. It provides a basis for the 78 management strategies proposed in Chapter 4, and for the compound model SAM as described in Chapters 5 and 6. This section summarizes the conceptual model, and discusses its relation to soft paradigm and hard paradigm thinking. 3.5.1 Summary The conceptual model is organized around two main systems: the pollution system and the management system. The pollution system contains groundwater flow and pollutant transport subsystems. The management system contains technical and decision subsystems. The emphasis in the conceptual model is on the management system, and particularly on the decision system. The functions of the technical system are to observe the pollution system, to generate technical advice, and to implement technical decisions. The elements of the technical system are technical organizations, technical goal (mitigation versus restoration), linear versus iterative technique, and technical uncertainty. The function of the decision system is to make decisions, especially technical decisions based on advice from the technical system. The decision system is described in terms of a number of elements. Rules of governance encompass statutes, regulations, common law, and government policy. Stakeholders include regulatory agencies, responsible parties, injured stakeholders (such as the local community), and peripheral stakeholders. Issues include technical decisions, responsibility (liability), and compensation to injured stakeholders. Modes of decision-making include negotiation, unilateral regulatory decision-making, and judicial decision-making. 79 The emphasis in this thesis is on negotiation; the other modes of decision-making are seen chiefly in their role as alternatives to negotiation. Decision-making may be linear or iterative. The decision system is associated with a high degree of complexity and uncertainty, exacerbated by the complexity of the pollution system and by uncertainty in the technical system. 3.5.2 Relation to the soft and hard paradigms An important aspect of the management system is the degree of integration. An integrated decision system consists of a small number of closely-linked decision-making processes. An integrated technical system involves a small number of technical organizations, reporting to the decision system in a coordinated or joint fashion. The relation of the conceptual model to soft and hard paradigm thinking can be discussed in two senses. The first sense is the soft paradigm orientation of the conceptual model. The second sense is the recognition of coupled elements representing the distinction between soft and hard paradigm thinking. The soft paradigm orientation of the conceptual model is brought out by: • the use of systems as organizing concepts; • the emphases on the management system, on the decision subsystem, and on negotiation as a mode of decision-making; • the structure of the model, which explicitly allows for iterative cycles of observation^  advice-»decision-»implementation (Figure 3-1; compare with Figure 2-2); and • the recognition of non-quantifiable uncertainty in both the technical and decision systems. 80 The conceptual model contains couplets of opposed hard paradigm and soft paradigm elements: • restoration and mitigation; • linear technique and iterative technique; • linear decision-making and iterative decision-making; • unilateral decision-making and negotiation. The hard paradigm elements are restoration, linear technique, linear decision-making, and unilateral decision-making. These resonate with the hard problem concepts: clearly-defined goals, alternatives and consequences; bounded and quantifiable uncertainty; a linear solution procedure; and a unilateral decision-maker. The soft paradigm elements are mitigation, iterative technique, iterative decision-making, and negotiation. These resonate with the soft problem concepts: a lack of clearly-defined goals, alternatives, and consequences; pervasive and non-quantifiable uncertainty; multiple stakeholders with divergent interests and values; and iterative cycles of learning, action, and experience. 81 4. Strategies for groundwater pollution management This chapter contains a set of proposed strategies for the management of groundwater pollution. The strategies are based on soft paradigm thinking as discussed in Chapter 2. They are structured according to the conceptual model of Chapter 3, and guided by the qualities of good management discussed below in Section 4.2. The strategies are illustrated by the compound model SAM, as discussed in Section 5.5.2. In contrast to Chapter 3, this chapter is largely prescriptive, but includes some descriptive material presented as supporting evidence. As with Chapters 2 and 3, the material in this chapter includes a number of general statements. Specific exceptions to these generalizations are important, but are not necessary to the immediate purposes of this thesis. 4.1 General concepts A strategy, as used in this thesis, is a generalized and relatively abstract plan for coping with particular aspects of a generic groundwater pollution case. The entire set of strategies presented in this thesis will be referred to as management strategies, because they apply to the management system. The set of management strategies is divided into two subsets (Figure 4-1): decision strategies which apply to the decision system (Section 4.3), and technical strategies which apply to the technical system (Section 4.4). As with the conceptual model of Chapter 3, emphasis is placed on the decision strategies and on negotiation, reflecting the soft paradigm orientation of the thesis and recognition that these areas are perhaps less familiar to a hydrogeological audience. 82 For clarity, the names of the strategies w i l l be set in a different font, e.g., principled negotiation. Figure 4-1. Management strategies. The strategies apply to groundwater pollution management in a generic sense, without reference to a specific jurisdiction. This is the approach taken by Massman and Freeze (1987a, 1987b) in their treatment of the regulation of potential groundwater pollution. Generic strategies have the advantage that they deal with the issues on a more fundamental level, and have relevance over a wider span in space and time. In contrast, strategies which are targeted on a specific jurisdiction are limited to that jurisdiction, and lose validity over time as conditions change significantly in that jurisdiction. The individual strategies are presented in a modular fashion. This has been done so that a reader wishing to extend this work to a specific jurisdiction or case can select those strategies which are appropriate for that jurisdiction or case. Note, however, that there are typically synergistic effects from the combined implementation of two or more strategies. For this reason, and to counteract the reductionism which is inherent in a modular presentation, frequent reference w i l l be made to the linkages among strategies. Not all aspects of the problem of groundwater pollution management have been considered, and the strategies do not add up to a complete framework for managing groundwater 83 pollution. The rationale for this lack of completeness is that each jurisdiction and each case has its own specific circumstances, and it would not be possible to develop a complete set of generic strategies applicable to all jurisdictions and all cases. It is ultimately more productive to aim for the key issues, and to accept a degree of incompleteness. This is consistent with soft paradigm thinking, particularly with the recognition that soft problems do not have clear boundaries. Many of the strategies discussed below would be enhanced by a source of neutral funding. This is particularly applicable to the strategies of balanced negotiating power (Section 4.3.3.3), mediation (Section 4.3.3.4), inclusive participation (Section 4.3.2.2), and technical integration (Section 4.4.1). The mechanism proposed in this thesis for disbursing neutral funding is a quasi-governmental board, established by statute. The legitimacy of the board would stem from its arm's-length relationship with any of the stakeholders in any given problem, including the regulatory stakeholder. Clear rules would have to be established regarding acceptable expenditures and accounting requirements, as well as conflict-of-interest rules for the board's personnel. The neutral funds could be derived from percentages of liability reduction fees (see Section 4.3.1.3), fines, and costs recovered from responsible stakeholders by regulatory stakeholders. It is appropriate that the responsible stakeholders should be the source of the neutral funds, on the grounds that they have a duty to promote good management of the problem for which they bear responsibility. 4.2 What is good management? This section discusses the question of how to characterize good management. The discussion has been stimulated by a number of writers: Raiffa (1982); Dorcey (1986); Nyhart and 84 Dauer (1986); Susskind and Cruikshank (1987); and by works in the principled negotiation tradition, beginning with Fisher and Ury (1981). Four qualities are used here to characterize the goodness of management: goal-setting, efficacy, efficiency, and fairness. Goal-setting refers to the quality of knowing the desired management goals. Efficacy refers to the quality of achieving the desired management goals. Efficiency refers to the quality of producing the desired management goals with minimum effort. Fairness applies to both the decision-making process and the outcome of decision-making. Goal-setting has to do with ends; efficacy and efficiency have to do with means; and fairness applies to both ends and means. In the hard paradigm context, goal-setting, efficacy, and efficiency can be considered objectively, independently, quantitatively, and sequentially. Goal-setting is the starting point — a hard problem begins with a definition of goals. The next step involves efficacy: defining alternatives which will achieve those goals. The final step involves efficiency: determining the "best" alternative; this is the focus of hard paradigm thinking. The quality of fairness is generally not considered explicitly. In the soft paradigm context, all four qualities must be considered subjectively, interdependently, qualitatively, and simultaneously. Goal-setting is difficult to evaluate because the goals are difficult to define and contentious among multiple stakeholders. Efficacy and efficiency are difficult to evaluate under conditions of unclear goals, unquantifiable uncertainty and bounded rationality. Fairness invokes thorny value questions. The following points are suggested as characteristics of good management in the soft paradigm context. They should be seen as ideals rather than as criteria. 85 Good management respects the interests and values of all stakeholders, in both the decision-making process and the substantive outcome. This directly supports the quality of fairness, and indirectly supports the other qualities. The indirect support arises because stakeholders whose interests and values are not reflected in the setting of goals are likely to interfere with the achievement of those goals, with negative implications for efficacy and efficiency. Good management is based on an appropriate balance of decision-making power among the stakeholders. The term "appropriate balance" implies that a stakeholder's decision-making power is weighted according to that stakeholders' legitimate interest in the situation. This directly supports the quality of fairness, and indirectly supports the other qualities. Good management is Pareto-optimal, i.e., it takes advantage of all potential benefits which can accrue to each party at no disbenefit to the others. This supports the qualities of fairness and efficiency. Good management maintains good working relationships among the stakeholders. This supports the qualities of goal-setting, efficacy, and efficiency because good working relationships enhance the ability to determine what must be done, how to do it, and how to do it with minimal waste. This also supports the quality of fairness, because fairness is most likely to be achieved when the stakeholders have a good working relationship, and wish to maintain that relationship. Good management makes the best use of available information. In particular, this implies that information is shared among the stakeholders, rather than withheld to gain some advantage in an adversarial contest. . This supports goal-setting, efficacy and efficiency because information is essential to establishing goals, determining which alternatives are likely to 86 achieve those goals, and determining which alternatives are likely to achieve the goals with the least effort. It further supports efficiency because it reduces wasted and duplicated effort in obtaining information. Finally, it supports fairness, because good information is required in the debate among stakeholders as to what is fair. Good management sets realistic goals. This is a critical point in the context of groundwater pollution management, given the general infeasibility of restoration. Realistic goal-setting obviously supports the quality of efficacy. It further supports efficiency, because efficacy is a prerequisite for efficiency. Finally, it supports fairness: it is unfair for the responsible stakeholder to be forced to work towards an unrealistic goal, and it is unfair for the injured stakeholder to be given an unrealistic perception of what can be achieved. Good management adequately allows for complexity and uncertainty. This supports all of the qualities, since any of them can be undermined by a failure to appreciate the pervasive and dominant roles of complexity and uncertainty in groundwater pollution management. Good management makes as much use of objective criteria as possible. This is an important element of principled negotiation, and can be extended to unilateral decisions. This begs a soft paradigm definition of "objective criteria". In this context, objective criteria should depend on ranges of values rather than single values. It is also useful to distinguish between objective criteria which apply to outcomes and objective criteria which apply to actions. For example, in a case of hydraulic containment, the criteria might be either the desired hydraulic gradient around the zone of containment (outcome), or the pumping rate which is expected to achieve the desired hydraulic gradient (action). Clearly, the most useful criteria are those which apply to outcomes. However, because outcomes are much less controllable and 87 predictable than actions, it may often be appropriate to apply criteria to actions rather than outcomes. 4.3 Decision strategies This section presents the decision strategies and the linkages among them. There are three major subsets in the set of decision strategies: iteration strategies, negotiation strategies, and integration strategies (Figure 4-2). Management strategies Technical strategies (see Section 4.4) Iteration Negotiation strategies r. strategies Integration strategies Iterative decision-making Contingent decision Iterative liability adjustment Primary negotiation Principled negotiation j Balanced negotiating power [ Mediation j ^ Process integration Inclusive participation j Geographic integration j Figure 4-2. The decision strategies. 4.3.1 Iteration strategies The iteration strategies are based on the concept of iterative cycles of decision-making as a means of coping with complexity and uncertainty. There are three iteration strategies: iterative decision-making, contingent decision, and iterative liability reduction. 88 4.3.1.1 Iterative decision-making The strategy of deliberate iterative decision-making is proposed as an alternative to linear decision-making, and to the ad hoc iterative decision-making which arises from the breakdown of linear decision-making. Iterative decision-making is analogous to iterative numerical solution, a technique used in computer programs to solve equations. In an iterative numerical solution, a series of attempted solutions converge towards a final value; the iterations cease when the magnitude of the fluctuations from one iteration to the next is within a closure criterion. In iterative decision-making, a series of decisions converge towards a resolution of the situation; closure is reached and the iterations cease when there is little or no disagreement among stakeholders, and technical actions have achieved the desired results. This analogy carries a number of corollaries: • The initial management iterations are generally more important than the later iterations, and have a greater effect on the situation. This implies that the initial iterations are particularly important, and tend to lock the management into a particular pattern. • Uncertainty about the pollution system and its response to technical actions is reduced over time, as the technical system acquires a better knowledge base. • The magnitude of the problem tends to decrease over time, as pollutant migration is successfully contained and pollutant mass is removed. Iterative decision-making is consistent with and largely dependent on the parallel technical strategy of iterative technique (Section 4.4.3). The arguments in support of iterative decision-making and iterative technique are as follows: 89 • They give the management system an opportunity to cope with complexity and uncertainty by allowing planned periodic responses to new information, new understanding, and changing circumstances. • They give the management system an opportunity to take full advantage of the growth in time of the knowledge base about the pollution system, as technical options are implemented and the consequences observed. • They give the management system an opportunity to take full advantage of the growth over time in the knowledge base about groundwater pollution in general, and particularly the growth in new technologies. • In contrast to ad hoc iterative decision-making arising from a failed linear decision-making and linear technique, deliberate iterative decision-making and iterative technique avoid both wasted effort directed towards an unattainable endpoint, and false optimism about the ability to reach that endpoint. • They are consistent with the technically-realistic goal of mitigation. • They allow technical actions to commence and proceed while the decision system evolves and sharpens its goal-setting. • They have a high tolerance for procedural and technical failure, because subsequent iterations provide an opportunity to recover. Any iteration of decision-making should include a provision and schedule for subsequent rounds of decision-making. A geometric progression for subsequent rounds of decision-making might be appropriate, with the period between rounds increasing over time as the fluctuations settle down. 90 Any iteration of decision-making should also include a provision for triggering a re-examination of the decision if certain circumstances occur or certain objective criteria are not met by a given date. This is actually a combination of two strategies: iterative decision-making and contingent decision (Section 4.3.1.2). Certain circumstances cannot be predicted but may be anticipated in general. For example, provision might be made for re-examining a decision in the case of major disturbances to the groundwater flow system, such as a new excavation in proximity to the polluted site. The criteria for re-examining a decision should be objective, quantitative, comprehensive, and realistic, e.g.: • The provision of a replacement water supply at a given capacity and water quality to be provided by a given date. • Hydraulic head at specified points to match specified values by a given date. • A well of a given capacity and depth to be constructed and operated at a given pumping rate. Note that the first two criteria apply to outcomes, and the last criterion applies to an action. Iterative decision-making requires the development of criteria for closure, i.e., for recognizing that the problem has been dealt with and further decision-making is not required. As with criteria for re-examining a decision, criteria for closure should be objective, quantitative, comprehensive, and realistic. Developing criteria for closure is not a trivial exercise, because it depends on the difficult task of goal-setting. Note that criteria for closure must apply to outcomes rather than actions. Closure in iterative decision-making may also be achieved by default, when the stakeholders lose interest in participating. This should be regarded with caution, however, because stakeholders may stop participating for reasons other than the satisfaction of their 91 interests. They may stop participating due to an exhaustion of resources, or through a cynical dismissal of the legitimacy of the decision-making process. This would have negative implications for the goodness of management, particularly with regard to fairness. Of course, iterative decision-making is not guaranteed to produce good management. Relationships among stakeholders may become strained over time, or the pollution system may exhibit unpleasant and unexpected behaviour. Here, too, the analogy with iterative numerical solution is appropriate, as numerical iterations may fail to converge, and show increasing fluctuations away from a solution. It is therefore important for iterative decision-making to include provision for recognizing a failure to converge, and for making the necessary adjustments to correct such a problem. Criteria for recognizing failure of iterative decision-making should be developed, just as an iterative numerical solver has criteria for recognizing when a numerical solution fails to converge. Again, these criteria should be objective, quantitative, comprehensive, and realistic. As with criteria for closure, this is a non-trivial exercise in goal-setting. One must know the desired outcome in order to recognize when it is not being approached. As with criteria for closure, criteria for failure-to-converge apply to outcomes rather than actions. Recognition of failure-to-converge is only the first step. There should also be an explicit plan for responding to failure-to-converge. 4.3.1.2 Contingent decision A contingent decision is a decision of the form "If Event X occurs, then Action Y will be undertaken." It is a strategy for coping with uncertainty, and is particularly important in long-term management which carries uncertainty about the future behaviour of both the pollution 92 system and the decision system. It is a variant on the theme of iteration, with a new iteration being triggered automatically by the occurrence of the specified event. There are significant difficulties with contingent decision. In general, not all contingencies can be anticipated, and sequential contingencies cannot be realistically projected beyond a few steps. The important combination of contingent decision with iterative decision-making has been noted above. This is a decision of the form "If Event X occurs, then another iteration of decision-making will be triggered to deal with the consequences of Event X". The new iteration may be preferable to the automatic action implied by contingent decision on its own. The contingency should be clearly expressed according to objective and quantifiable criteria, e.g., specified pollutant concentrations or hydraulic head in a specified monitoring well. The rationale is obvious: to avoid any future dispute as to whether a contingency has indeed arisen. Such criteria may apply to outcomes or to actions, as appropriate. Contingent decision is a useful mechanism for resolving sticking points in negotiations, in cases where different stakeholders have different perceptions of how future events will unfold, or where one or more stakeholders require a hedge against unpleasant surprises (Lax and Sebenius, 1986, p.97-98; Susskind and Cruikshank, 1987, p. 124-125). Contingent decision is a key aspect of the observational approach in geotechnical engineering (Terzaghi and Peck, 1967; Peck 1969), and the feedback policy in adaptive management (Walters, 1986). Contingent decision often involves financial instruments. One such instrument is the posting of a bond by the responsible stakeholder. This was recommended by Massman and Freeze (1987b) in the context of waste facility siting. The concept is that a bond is posted by the 93 responsible stakeholder, and returned with interest at some future time if objective criteria have been met, in terms of either outcomes or actions. An example is the negotiated settlement between the waste disposal firm Laidlaw and the state of South Carolina over the possibility of future problems with Laidlaw's Pinewood hazardous waste landfill. Laidlaw agreed to put up a combination of an accumulating bond and a corporate guarantee, to the sum of $100 million (US), as provision against problems occurring over a 100 year time horizon (Mahood, 1992). Property value guarantees, proposed by Zeiss and Atwater (1989a, 1989b) in the context of facility siting, are another type of contingent financial instrument which may be applicable in groundwater contamination cases. A contingent financial instrument can itself be subject to iterative decision-making. The initial financial instrument could be established on a conservative, worst-case basis. It could then be gradually reduced, as justified by new information, during subsequent iterations. 4.3.1.3 Iterative liability reduction Iterative liability reduction is a strategy in which a responsible stakeholder obtains an iterative reduction in its liability in exchange for: • acceptable progress towards the technical goal; and • a series of fees paid by the responsible stakeholder, beginning with a first fee which would cap its liability at a relatively high level. Ultimately, depending on the success of technical actions, the liability might be reduced to zero. This strategy would have to take place in conjunction with iterative decision-making (Section 4.3.1.1). 94 The concept of iterative liability reduction is related to the "infeasibility fee" proposed by NRC (1994, p. 270-271); and to the covenant-not-to-sue, which is a mechanism used by US regulatory stakeholders in reaching agreements with responsible stakeholders (Fogleman, 1992, p. 110-111). The rationale for this strategy is as follows. A complete liability release is a major goal of a responsible stakeholder, and the regulatory agency ultimately wants to trade a complete liability release in exchange for satisfactory completion of technical actions by the responsible stakeholder. But a complete liability release for the responsible stakeholder implies a complete assumption of liability by the public. The regulatory stakeholder is understandably reluctant to assign complete liability to the public, and the responsible party is understandably frustrated with the regulatory stakeholder's reluctance. The exchange becomes a source of frustration because it is an all-or-nothing proposition. The concept of iterative liability reduction provides a possible escape from this impasse. The liability reduction fees should be used by regulatory stakeholders to cover public expenditures on groundwater pollution in general, including expenditures resulting from the public assumption of liability, and compensation for injured stakeholders. A percentage of the liability reduction fees should be directed to the neutral funding board discussed in Section 4.1. The details of iterative liability reduction, including its distribution among multiple responsible stakeholders, would be best sorted out through negotiation (see Section 4.3.2). Injured stakeholders would likely be legitimately concerned that iterative liability reduction would be counter to their interests. It would therefore be appropriate for injured stakeholders to participate in the decision-making on this issue, as informed observers if not as 95 actual decision-makers. This proposal is consistent with the strategy of inclusive participation (Section 4.3.2.2). An incentive for injured stakeholders to accept iterative liability reduction would be the availability of compensation funds derived from the liability reduction fee. Iterative liability reduction can be combined with contingent decision (Section 4.3.1.2). This might specify that the liability reduction is only in effect provided that certain contingencies do not occur, e.g., "the liability reduction is void if pollutants reach a given water supply well". This would give particular comfort to a regulatory stakeholder with concerns about the public assumption of excess liability. It would, of course, provide a corresponding discomfort to the responsible stakeholder, and this would presumably be reflected in a reduced liability reduction fee, or some other concession. 4.3.2 Integration strategies In Chapter 3, the concept of integration in the decision and technical systems was introduced. It is argued here that integration of the decision system can contribute to good management in a number of ways. It can improve goal-setting, by forcing a clearer evaluation of the desired effects of management in relation to the interests of multiple stakeholders. It can improve efficiency in decision-making, by reducing redundancy and conflicting decisions. It can improve fairness, by helping to ensure the effective participation of all stakeholders. It can increase tractability, by increasing the number of opportunities for tradeoffs and joint gains. It can improve the flow of information, reducing the overall level of uncertainty and clarifying the existing complexity. It can reduce complexity in one sense, by reducing the number of decision-making processes and improving the linkages among processes. Integration might also tend to 96 increase complexity, because a larger process may be more complex than a number of smaller processes. A broad overview of groundwater pollution management in a given area is the essential starting point for integration. In theory, it should be the regulatory stakeholders who have this broad overview. In practice, this may be otherwise (e.g., Wartenberg, 1988). In some cases a progressive technical organization, a mediator (see Section 4.3.3.4), or the local community might take the lead in fostering integration. Three integration strategies are proposed here: process integration, inclusive participation, and geographic integration. See also the strategy of technical integration, proposed in Section 4.4.1, below. 4.3.2.1 Process integration The most important integration strategy is process integration, which calls for isolated decision-making processes to be merged into a single process, or at least to be strongly linked with one another. Process integration is compatible and synergistic with negotiation, but it is also applicable to unilateral modes of decision-making. Process integration would have the salutary effect of coordinating the efforts of various regulatory stakeholders. Wartenberg (1988) describes a case where lack of coordination among regulatory stakeholders contributed to inadequate management in a case of groundwater pollution. The rules of governance should encourage the merger or linkage of isolated processes. An important corollary is the removal of disincentives to process integration. An example of such 97 a disincentive would be a regulatory policy which rewards regulatory agencies for rapid settlement with individual responsible stakeholders. 4.3.2.2 Inclusive participation Inclusive participation calls for all stakeholders to participate in decision-making. Participation may involve a seat at the negotiating table, standing in a litigation process, or the provision of information and opinions in support of a unilateral regulatory decision. An ethical argument for inclusive participation is that all stakeholders should mutually recognize each others' legitimate right to inclusion in a decision-making process which affects their interests. This clearly improves the fairness of management. In particular, injured stakeholders should be entitled to input in any decision which significantly affects their interests. This input might be as limited as providing information and opinions, or as far-reaching as the right to grant or withhold informed consent on management actions. There are a number of pragmatic arguments for inclusive participation. It improves the quality of goal-setting, because it is extremely difficult to establish the desired goals of multi-stakeholder management without the participation of all stakeholders. It may expand the range of available options for coping with the problem, potentially improving efficacy and efficiency. For example, including a nearby industrial facility as a peripheral stakeholder may provide the opportunity to use treated groundwater as process water in that facility. Another pragmatic argument for inclusive participation is that decision-making should include any stakeholder who could attack the process or the decision from the outside. For example, exclusion of an injured stakeholder can lead to disruptive political pressure and litigation; it can derail carefully-laid plans, compromising efficacy and efficiency. 98 The combination of inclusive participation with the negotiation strategies yields a process in which decisions are made by negotiations in which all stakeholders participate. This is a particularly important type of process, and is increasingly seen as the best approach in complex multi-stakeholder situations. Negotiations involving all stakeholders provide an appropriate setting for decision-making on technical strategies such as mitigative technique and iterative technique (Sections 4.4.3 and 4.4.1). In particular, injured stakeholders may arbitrarily reject and oppose such technical strategies unless they have an opportunity to give their informed consent through negotiations. Shrader-Frechette (1991, p. 206-211) argues for the importance of informed consent in the general context of environmental hazards. Inclusive participation requires that all stakeholders have adequate financial resources to participate. However, injured stakeholders and responsible stakeholders may have limited resources, and it follows that there should be a mechanism to provide needy stakeholders with the necessary resources. This concept is accepted by the US Superfund law, which requires the Environmental Protection Agency to provide Technical Assistance Grants (TAGs) to communities affected by polluted sites and groundwater (Hird, 1994, p. 200). In Canada, there is a general precedent in the concept of intervenor funding, where the government provides funds for concerned stakeholders to make presentations at government hearings. It is important that the disbursement of funding be neutral in character, so that no real or perceived obligation is created. The situation where funds are disbursed by the regulatory stakeholder (as in the case of the TAG grants) is less than ideal, because the regulatory 99 stakeholder is not generally regarded as neutral by the other stakeholders. A better approach would be to disburse funds through a neutral funding board, as discussed in Section 4.1. Breyer (1993) argues for a centralization of decision-making power in groundwater pollution cases — the opposite of inclusive participation. Breyer's argument rests on a claim that the local communities consistently over-estimate the toxicological consequences and demand excessive technical goals. Breyer's argument is challenged here on the grounds that his claim is a very narrow perception of both the consequences to local communities, and the reactions of local communities. His dismissal of the toxicological consequences is at odds with detailed epidemiological evidence (see Lybarger, 1993; Johnson, 1993, Johnson, 1995). Furthermore, he ignores the clear evidence of negative psychological consequences (Edelstein, 1988; Vyner, 1988; Greenberg et al, 1994). His assumption that communities will inevitably demand excessive technical actions is not generally valid — communities may also have an interest in reducing the disruption associated with technical actions, and may in fact demand reduced technical actions (Alston, 1993; Chapell, 1993). A community may even react with complete indifference to a revelation of groundwater pollution. Finally, Breyer's argument can also be challenged on the grounds that to refuse a community any participation in decision-making is to doubly victimize them, by removing their control over a threatening situation. Wartenberg (1988) makes a strong case in favour of inclusion of the local community in management, arguing that this would improve the competency of management, as well as reducing unproductive conflict and psychological stress. The rules of governance should give regulatory stakeholders the mandate to encourage inclusive participation. This does not imply that participation should be only by invitation of 100 the regulatory agency. Participation in the decision-making process should be seen as a stakeholder's right. Some stakeholders will not be aware of the need for their participation until a decision-making process is in progress. Provision should be made to allow unrecognized stakeholders to enter an ongoing process. 4.3.2.3 Geographic integration Management efforts often focus only on the polluted site and its plume. Geographic integration holds that management may be significantly improved by expanding its geographic scope and linking a number of features such as multiple polluted sites and multiple receptors, as well as potential locations for the disposal of polluted/treated soil and groundwater. The following arguments are advanced here in favour of geographic integration: • Geographic integration can encompass possible receptors for the pollution, such as water wells and wetlands. This will help to ensure the participation of injured stakeholders, and help to focus on the actual risk to human and environmental health. • Geographic integration can encompass a number of polluted sites and plumes, bringing a number of responsible stakeholders into a single decision-making process. In particular, consider the case of a complex, multi-source plume in an area with a number of polluted sites. Such a situation almost inevitably results in conflict and litigation among the various responsible stakeholders, each trying to dodge the liability bullet. It is appropriate to treat this as a single large problem, and to seek a solution through a decision-making process which involves all the responsible stakeholders. In situations where responsible stakeholders are not in direct conflict, a larger geographic scope can lead to a number of benefits. In conjunction 101 with technical integration (Section 4.4.1), information and ideas can be shared, and economies of scale can be realized in matters such as the construction of a treatment plant or the delivery of a replacement water supply. Arbitrary discrepancies in regulatory enforcement from site to site can be brought out by direct comparison, contributing to fairness with respect to different responsible parties. Geographic integration can encompass natural hydrogeologic boundaries. These boundaries are an important aspect of any evaluation of a groundwater pollution case, and generally lie well beyond the area of an individual source and plume. Geographic integration may encompass both industrial and agricultural groundwater pollution, in areas where the two types occur in proximity to one another. The industrial pollution is typically point-source, and the agricultural pollution is typically distributed over a wide area. Including the two types of pollution in an integrated management approach would allow a direct comparison of their relative consequences and relative technical tractability, contributing to fair management. Geographic integration may bring in peripheral stakeholders who may assist or hinder management. For example, nearby industrial facilities may be able to assist in the disposal of treated water. It may even be appropriate to encompass a distant area which is intended to receive polluted materials from the site for incineration or landfilling. The community or municipal government in the vicinity of the incinerator or landfill can be considered as injured stakeholders whose concerns should be taken into account. 102 The rules of governance should encourage geographic integration. For example, the Wellhead Protection Program under the United States Safe Drinking Water Act explicitly recognizes that the geographic scope of groundwater pollution extends well beyond the polluted site (EPA, 1989). Encouragement of geographic integration may involve the removal of disincentives, e.g., a policy which rewards regulatory agencies for rapid closure of individual sites. 4.3.2.4 Example to illustrate importance of integration strategies The following true example illustrates the importance of integration strategies. In this case, the regulatory stakeholder agreed to allow the responsible party to dump a large volume of marginally-polluted soil in a landfill in a nearby municipality. Implementation of this decision was attempted. However, the nearby municipality objected to receiving the polluted soil, and promptly passed a bylaw restricting such dumping. Dumptrucks and excavators were idled, at great expense, while an alternative dumping site was sought. In this example, we see management which is deficient in terms of: • Goal-setting: it specified a desired effect which was too narrowly-defined. The goal of disposal of marginal-polluted soil should have been expanded to include disposal in a site which was acceptable to the receiving community. • Efficacy: it did not produce the desired effect of disposing of the marginally-polluted soil. • Efficiency: it led to a waste of expensive machine time. • Fairness: it disregarded the legitimate interests of the receiving community. Moreover, there was an irreversible effect. The municipality developed a mistrust of the regulator. Other municipalities passed similar bylaws, hampering management efforts at many 103 other polluted sites, making waste management generally more expensive and difficult, and ultimately hampering reform of the rules of governance for waste management. The outcome might have been improved by adopting integration strategies. Geographic integration would have linked the polluted site with the receiving landfill. Inclusive participation would have assured that the municipality was recognized as a stakeholder by the regulator and responsible party. Process integration would have linked the negotiation between regulator and responsible party with the municipality's process for deciding whether or not to accept the polluted soil. Technical integration (Section 4.4.1) would have allowed all stakeholders to decide jointly which technical questions needed to be asked, and to examine the situation from the same base of technical knowledge. 4.3.3 Negotiation strategies Negotiation (see Section 3.3.4.1) is the mode of decision-making which delegates management to the collective intellect, the collective rationality, and the collective responsibility of the participants. The purpose of this section is to propose strategies to increase the use, expand the scope, and improve the value of negotiation in groundwater pollution management. This approach follows the lead of Dorcey (1986), who recognized the importance of negotiations in coastal resources management in British Columbia, and proposed ways to increase the value of those negotiations. Four negotiation strategies are proposed here. Primary negotiation and principled negotiation have the potential to produce good management. But negotiation is not a panacea. It is unlikely to produce equitable decisions in the absence of balanced negotiating power. There 104 are practical difficulties associated with negotiation, and mediation is an important mechanism for coping with these difficulties. 4.3.3.1 Primary negotiation Primary negotiation is a strategy which emphasizes negotiation in two senses relative to unilateral modes of decision-making: "first importance" and "first in a series". The first importance of negotiation arises because it is has a high potential to result in good management. Negotiation has the potential to produce fairness in process and substantive outcome, because the individual stakeholders are in the best position to understand and represent their own interests and values. Negotiation has the potential to set appropriate goals, because it permits the stakeholders themselves to collectively translate their values and interests into goals. Negotiation has the potential to result in efficacious and efficient management, through the sharing of information and the development of options for joint gain. Negotiation as first in a series of modes of decision-making implies that management efforts should begin with negotiation, and only proceed to the unilateral modes of decision-making if negotiation fails. Obviously, if negotiation is the most important mode of decision-making, then it should be the first mode of decision-making attempted. This is the long-standing practice in labour disputes, where negotiations come before work stoppages or arbitration. Note that this was not always the case in labour relations. It has developed over many years, and continues to develop, through the changes in the rules of governance, institutions, and the balance of negotiating power among the stakeholders. An important argument for negotiation as first in a series is that the unilateral modes of decision-making have impacts which are partially or completely irreversible. This arises from the 105 adversarial postures which unilateral modes of decision-making impose upon the stakeholders. Positions are hardened, the flow of information is restricted, and personal relationships are damaged. Consequently, a negotiation which follows a unilateral mode of decision-making may be less beneficial than a negotiation which is conducted as first in a series. Consider the common situation in which litigation is the initial mode of decision-making, and negotiation is used to settle the litigation. Such a negotiation will likely be carried out with a restricted flow of information, under an artificial deadline imposed by a judicial schedule, and with a coercive aspect due to the exhaustion of the legal resources of one or more participants. A decision reached under these circumstances is unlikely to be good. This is particularly true with respect to fairness, because a participant whose resources are exhausted will be compelled to accept a less-than-favourable settlement. The rules of governance should require the regulatory stakeholder to make all possible efforts to reach a negotiated agreement before imposing a unilateral decision. In litigation, strong weight should be placed on evidence that a plaintiff or defendant has made serious prior efforts at good-faith negotiation. The personal responsibility of regulatory employees engaged in negotiation should be limited to their own best efforts, and not to the success or failure of the entire negotiation exercise. The point of this proposal is to support the efforts of regulatory employees to engage in negotiation, without placing responsibility for the outcome of the negotiation on their shoulders. This point reflects the distinction between unilateral regulation and negotiation. In unilateral regulation, the regulatory employee carries both the decision-making authority and responsibility 106 for the outcome. In negotiation, the regulatory employee shares the decision-making authority and therefore should not bear complete responsibility for the outcome. Stakeholders with limited resources may be compelled to participate in negotiation without adequate legal and technical advice, giving rise to a problem of fairness. For this reason, primary negotiation should be linked with the strategy of balanced negotiating power (Section 4.3.3.3). Some technical actions may have to be implemented on an emergency basis, e.g., removal of leaking pollutant containers, initial plume control, fencing of the polluted site, provision of emergency water supply, and initial observations and characterization. The regulatory stakeholder should have authority to unilaterally implement a explicitly limited set of emergency actions. This can have beneficial effects for the subsequent negotiation — it can begin with an initial knowledge base, and without the pressure of crisis decision-making on emergency actions. Other stakeholders, particularly the local community, should be fully informed before and during the implementation of emergency actions. 4.3.3.2 Principled negotiation Principled negotiation refers to cooperative and consensual negotiation in which participants generate joint gains through tradeoffs and synergies. It is a keystone in the set of negotiation strategies. It is summarized in Chapter 3, and described in greater detail in Appendix 1. The argument for principled negotiation is that the potential benefits of negotiation cannot be fully realized if negotiation is dominated by a competitive and adversarial style. Under 107 such circumstances, fairness may be illusory, the appropriate goals may not be discovered, joint gains may not be realized, and expediency may be mistaken for efficiency. Principled negotiation cannot be mandated by laws or regulations, or otherwise imposed on stakeholders. It can be encouraged by mediation (Fisher and Ury, 1981) and by training. Regulatory agencies in particular should train their negotiators in the theory and practice of principled negotiation. The implementation of principled negotiation does not require all the stakeholders to follow its tenets. It can be implemented by a subset of the stakeholders, even if other stakeholders insist on using competitive and adversarial negotiation tactics (Ury, 1991). The use of principled negotiation by a subset of stakeholders may induce the others to follow. 4.3.3.3 Balanced negotiating power Cooperative and consensual negotiation requires a balance of negotiating power among the stakeholders. This balance in turn depends on a degree of parity among the BATNAs and worst-nightmares of the stakeholders in a given case. A management agreement reached under an imbalance of negotiating power is unlikely to be fair; the setting of goals and selection of options will be skewed towards the interests of the more powerful party. There may also be a deficit of efficacy and efficiency, particularly if the goals are technically unrealistic. It follows that balanced negotiating power is an important strategy for good management. The unilateral modes of decision-making, even if they are never invoked, play a key role in determining the BATNAs and worst-nightmares of the stakeholders. The most obvious and common imbalance is the weakness of injured stakeholders relative to responsible and regulatory stakeholders. Injured stakeholders should have recourse to 108 litigation, providing them with a strong BATNA, so that they can participate in negotiations on equal grounds with regulatory and responsible stakeholders. Specifically, injured stakeholders should have standing to sue responsible stakeholders for damages, and to sue regulatory stakeholders to enforce actions. (This is largely the case under current United States statutes.) This is not an argument for litigation as a primary mode of decision-making, and the right to litigate should be tempered with a requirement for evidence of prior efforts at good-faith negotiation, as discussed in Section 4.3.3.1. It is important that the regulatory stakeholder have a credible BATNA. The mechanism of the regulatory BATNA may vary depending on the jurisdiction. Possible mechanisms include litigation, imposition of fines, and even criminal charges. The regulatory BATNA should be exercised with restraint. This is consistent with the strategy of primary negotiation, and ultimately reduces the potential for a political backlash which weakens the regulatory BATNA. Balanced negotiating power also depends on adequate provision of technical advice to all stakeholders, including those with limited resources. The provision of technical advice is considered in this thesis under other strategies: inclusive participation (Section 4.3.2.2) and technical integration (Section 4.4.1). 4.3.3.4 Mediation Mediation is the strategy in which a neutral party assists the participants in a negotiation with procedural and substantive matters (see Section 3.3.4.1). Most negotiations over groundwater pollution are sufficiently complicated that mediation can be beneficial, if not essential. This is particularly true if integration strategies (Section 4.3.2) are implemented, increasing the scope and complexity of negotiation. The rules of governance should encourage 109 mediation in groundwater pollution management. It is important that funding for mediation is delivered in a manner that protects the mediator's reputation for neutrality. An appropriate mechanism would be the neutral funding board proposed in Section 4.1. Mediation can help to promote principled negotiation. However, it is unlikely to be successful for cases which lack balanced negotiating power. 4.3.4 Summary of decision strategies This section has proposed a set of strategies intended to improve decision-making in groundwater pollution management: • a set of iteration strategies, intended to cope with complexity and uncertainty through iterative decision-making, taking advantage of the growth of knowledge over time; • a set of integration strategies, intended to bring the fragmented elements of groundwater pollution management into an integrated decision-making system; and • a set of negotiation strategies, intended to improve the value of negotiation in the management of groundwater pollution. These have been described in a modular fashion, but they are broadly consistent with one another, and there could be significant synergistic benefits from applying combinations of these strategies. A number of such combinations have been suggested. The decision strategies require a complementary set of technical strategies. These are discussed in the following section. 110 4.4 Technical strategies This section discusses a set of technical strategies, the linkages among them, and the linkages between the technical strategies and the decision strategies. The intention is to examine technical strategies which are complementary with the decision strategies described in the previous section. This section does not comprise an extensive catalog of technical options for managing groundwater pollution. The technical strategies considered here are iterative technique, technical integration, and mitigative technique (Figure 4-1). C ^ Iterative technique V Technical integration Mitigative technique Figure 4-3. The technical strategies. 4.4.1 Iterative technique The strategy of iterative technique involves a repetitive and open-ended approach to the design and implementation of technical options, moving towards a condition of closure, but with no set endpoint (see Chapter 3). This is a strategy of deliberate iterative technique, as opposed to an ad hoc iterative technique resulting from the failure of a linear technique to achieve its endpoint. A prime requirement for iterative technique is that it be flexible, i.e., capable of adjustment and adaptation in accordance with new information or changing circumstances. Note that iterative technique is complementary with mitigative technique (Section 4.4.3). I l l Iterative technique is the essential technical counterpart to iterative decision-making (Section 4.3.1.1), and is supported by the same list of arguments; these are not repeated here. The following examples serve to illustrate the concept of iterative technique. Iterative technique permits an approach of deliberate, controlled experimentation: making a hypothesis, designing a technical action to test that hypothesis, implementing the technical action, observing the results, and judging the hypothesis accordingly. This usage of "experiment" is distinct from the sense of "innovative" or "untried", or in the sense that every case is a new experiment in groundwater remediation (e.g., Freeze and Cherry, 1989). Deliberate, controlled experimentation is not a general prescription. It is only appropriate for situations which involve predictable and observable results, and few or no confounding variables. Experimentation may also impose short-term problems that need to be balanced against the expected benefit of the expected information to be gained from testing the hypothesis (see Walters, 1986, p. 257-258). Hydraulic containment is an obvious candidate for iterative technique. The first iteration consists of emergency plume control measures. Subsequent iterations can build on this effort, expanding the system of wells and/or trenches as indicated by new information. Subsequent iterations might also involve the construction of physical containment structures such as caps and cutoff walls. Once the containment system is established, the long-term iterations will involve adjustments of withdrawal/injection rates and maintenance of physical structures. One difficulty with iterative hydraulic containment is that design of a central treatment plant requires a good estimate of the discharge of polluted water. This requirement is clearly at odds with the tactic of iteratively expanding and adjusting the hydraulic containment system. 112 This conflict might be resolved by using a modular system of small treatment plants, expanded as necessary along with the rest of the system. This approach would have the advantage that the treatment process could also be designed iteratively, allowing for difficulties in treatment. Hydraulic containment is a technique which can be implemented as a deliberate, controlled experiment. A certain response of the flow system can be hypothesized; withdrawal/injection rates can be adjusted to test that hypothesis; and results can be observed. Modeling of groundwater flow and pollutant transport is also a candidate for iterative technique. The models can be updated on the basis of assessment and intervention from the previous iteration, and used as a guide for assessment and intervention in the next iteration. This approach benefits from clear documentation of the modeling process, and from graphical pre-and post-processors which allow rapid updating of existing models. A side-benefit of iterative modeling is that stakeholders obtain a more accurate perception of the model as an imperfect representation, rather than an oracle. One application of iterative flow modeling is in conjunction with hydraulic containment, and particularly with deliberate, controlled experimentation on hydraulic containment. Reconnaissance surveys use quick and inexpensive methods to obtain preliminary data. This concept is clearly consistent with iterative technique. The purpose is to guide the design of more rigorous and expensive observation procedures, and to quickly detect "hot spots" requiring immediate attention. This would appear to be an obvious first step. However, it is severely restricted in some jurisdictions by detailed regulatory control embodied in strict quality assurance/quality control requirements (QA/QC). This regulatory control reflects concern over false positives and false negatives. Concerns about false positives in a reconnaissance survey can 113 be addressed by rigorous checking of any "hot spots" detected by the reconnaissance. Concerns about false negatives cannot be addressed in this way. Rather, it must be accepted that the reconnaissance survey provided an incomplete picture, to be filled in by more rigorous surveys in later iterations. Innovative approaches typically face high barriers to their acceptance (NRC, 1994). Iterative technique can reduce these barriers. It has a relatively high tolerance for the failure of particular technical options, because future iterations allow for switching to other options. As an example of an innovative approach, consider the possible application of solar and/or wind power as alternatives or complements to grid power for hydraulic containment. The intermittent nature of solar or wind power might be acceptable for this application, particularly for an unconfined aquifer where the water table would take some time to recover during a cessation in pumping. Stand-alone solar or wind power would also be attractive in cases where a polluted site is located at a considerable distance from grid power. Furthermore, solar power panels or wind generators would be an appropriate use of polluted land which might otherwise lie waste. Biological remediation refers to the use of organisms or systems of organisms, rather than mechanical devices, to perform technical tasks. Biological remediation, in general, requires considerable adjustment to achieve satisfactory results, and is therefore consistent with iterative technique. It also generally requires considerable time, and is therefore also consistent with mitigative technique. Biological remediation encompasses the developed technology of microbial remediation (Chappelle, 1993), as well as the emerging technologies of phytoremediation and constructed wetlands. Phytoremediation (Shimp et al, 1993) involves the use of plants to accomplish or assist remedial tasks such as breakdown or sequestering of 114 pollutants, controlling the flow of groundwater, reducing infiltration, or evaporating excess water. Constructed wetlands are used to treat a wide variety of polluted waters (Knight et al, 1993; Bastian and Hammer, 1993). 4.4.2 Technical integration The concept of integration in the decision and technical systems was introduced in Chapter 3. Strategies for integration of the decision system were discussed in Section 4.3.2. This section presents the complementary strategy of technical integration. Lack of integration tends to be the norm for technical systems in groundwater pollution management. Stakeholders tend to be comfortable with separate technical support, because their technical interests can be considered on an individual and confidential basis by a technical organization which they trust. However, there are a number of disadvantages to technical fragmentation: • The flow of information and ideas is restricted. • Technical effort may be duplicated. • The range of technical options under consideration tends to be limited to those which meet the interests of individual stakeholders, as opposed to those options which can provide joint gains. • Stakeholders with limited resources may be unable to obtain adequate technical support. • Separate technical support can degenerate into advocacy and adversarial attitudes, which is seen as counterproductive and damaging to both technical credibility and professional ethics. This point has been made by observers of environmental problems in general (Cormick and 115 Knaster, 1986; Susskind and Cruikshank, 1987, p. 116) and groundwater pollution management in particular (Freeze and Cherry, 1989). Technical integration attempts to overcome these disadvantages. There are two styles of technical integration: coordinated technical support, and joint technical support. Coordinated technical support involves coordination and cooperation among the technical organizations which provide separate technical support for individual stakeholders. This approach can retain some of the advantages of separate technical support, while helping to overcome some of the disadvantages. Stakeholders receive individual and confidential consideration of their technical interests, but there is a flow of information and ideas among the technical organizations. Duplication of effort can be reduced. The range of technical options can be expanded to those which promise joint gains. The problem of technical advocacy is constrained, and productive debate among technical experts is encouraged. For the sake of fairness, coordinated technical support should involve funding for needy stakeholders to retain their own technical support (see Section 4.3.3.3), preferably through a neutral funding board (see Section 4.1). In joint technical support, the stakeholders jointly retain one or more technical organizations, and jointly receive the technical advice. The underlying concept is that technical support should be undertaken on behalf of the decision-making process as a whole, rather than on behalf of individual stakeholders. Joint technical support is highly consistent with mediation (Ozawa and Susskind, 1985; Susskind and Cruikshank, 1987; see Section 4.3.3.4); in a sense, it is the technical equivalent of mediation. It does not preclude the additional use of separate technical support, although this would obviously reduce some of the advantages. 116 Joint technical support has been used productively in cases such as a conflict between the oil and fishing industries of California (Cormick and Knaster, 1986). Chapell (1993) describes an example of joint technical support in a conflict between local residents and the United States Environmental Protection Agency over proposed technical actions at a polluted site. R.A. Freeze (pers. comm., 1995) describes a groundwater pollution management case where a number of responsible stakeholders used joint technical support in an effort to allocate responsibility for technical costs through non-binding arbitration. As with funding for mediation (Section 4.3.3.4), funding for joint technical support should be organized so as to protect the reputation for neutrality of the technical organization(s). Again, the most appropriate mechanism would be a neutral funding board (see Section 4.1). In some cases, a computer-based algorithmic model can serve as the focus of technical integration, relying on the model's inherent (or apparent) objectivity (see Section 2.1.3.2). This is the basis of the workshop approach in adaptive management (Holling et al., 1978; Walters, 1986). Mercer et al. (1989) describe a case where a groundwater model served this purpose in successful negotiations over technical actions in a groundwater pollution case. Susskind and Cruikshank (1987, p. 146) describe the use of an algorithmic model in a negotiation over a sewage disposal problem. Sebenius (1984) describes the value of an algorithmic model in major international negotiations over the Law of the Sea. Nyhart and Dauer (1986) discuss this concept in theoretical terms. Technical integration may have disadvantages for some stakeholders, including loss of confidentiality and possible weakening of BATNAs. This is a particular concern where litigation is in progress, or future litigation is expected. Another disadvantage is that technical 117 integration, especially joint technical support, might eliminate the legitimate challenging of assumptions and interpretations which takes place among experts representing different stakeholders. A practical problem with joint technical support is that many technical organizations might find it difficult to achieve a reputation for neutrality, as a consequence of established ties to individual stakeholders. If technical integration is encouraged by the rules of governance, technical firms would cultivate reputations for neutrality. Technical integration is compatible and synergistic with the integration strategies for the decision system (Section 4.3.2), particularly process integration. It is also compatible and synergistic with the negotiation strategies (Section 4.3.3), particularly mediation. A form of technical integration can arise in unilateral modes of decision-making, e.g., the judge in a litigation process may appoint a neutral fact-finder (Bevan, 1992). 4.4.3 Mitigative technique Mitigation is defined as reducing damage and risk to human and ecological health, without attempting to provide a permanent remedy (see Section 3.2). The alternative to mitigation is restoration, which is the goal of providing a permanent remedy. The main argument for mitigation lies in the deficiencies of restoration. Considerable well-documented experience indicates that restoration is not generally feasible (NRC, 1994). Restoration as a goal often forces technical actions beyond the point of exponential increase in the marginal effort per unit of pollutant removed, leading to large expenditures in heroic efforts to remove small and perhaps insignificant amounts of pollutants. This has the additional negative effect of discrediting 118 technical expertise for coping with groundwater pollution, and ultimately undermining the legal basis for groundwater pollution management (see Breyer, 1993). Mitigative technique is the strategy which designs and implements technical options on the basis of mitigation as the technical goal (see Chapter 3). This is a strategy of deliberate mitigative technique, as opposed to an ad hoc mitigative technique resulting from failure to achieve an intended goal of restoration. Mitigative technique respects the general infeasibility of restoration. But it is not a panacea, and must address new difficulties. Mitigative technique should not be used or perceived to be used as a mechanism by which responsible stakeholders can dodge their obligations, to the detriment of injured stakeholders. Injured stakeholders should have the opportunity to consider the case for mitigation, and specific mitigative options, as matters of informed consent. For this reason, a number of other strategies are particularly important in conjunction with mitigative technique: the negotiation strategies, inclusive participation, and technical integration. Mitigation does not eliminate the high costs of coping with groundwater pollution. It is a long-term exercise of indefinite length, and the long-term costs can be quite substantial. For example, Ford et al. (1994, p.28) quote an estimate of CDN$1 million/year for mitigation of the Mercier site in Quebec. Mitigation also imposes new costs, such as the cost of conducting risk assessment. In fact, the argument for mitigation as opposed to restoration is not that mitigation necessarily saves money, but rather that it provides greater benefit for the money spent. There is one potential economic benefit which is inherent in the long-term nature of mitigation: mitigative technique can emphasize initial capital expenditures which reduce long-term operating costs. Consider a generic case requiring long-term containment of polluted 119 groundwater. The capital cost of structures providing partial physical containment may be justified on a present-value basis if they provide a reduction in the long-term operating cost of hydraulic containment. The long-term nature of mitigative technique may make biological remediation more attractive as a technical option; this point is discussed further in the following section. The open-ended and long-term nature of mitigation implies that mitigative technique should be developed in conjunction with iterative technique and iterative decision-making. 4.5 Summary and discussion This chapter has proposed a set of strategies for the management of groundwater pollution, based on the conceptual model of Chapter 3 and developed with soft paradigm thinking as discussed in Chapter 2. These strategies are illustrated in the compound model SAM, as described in Chapter 5. In particular, the role of the management strategies in SAM is discussed in Section 5.5.1. This section summarizes the management strategies and discusses their soft paradigm orientation. 4.5.1 Summary The strategies reflect a soft paradigm concept of good management, and respect four qualities of good management: goal-setting, efficacy, efficiency, and fairness. Characteristics of good management, stemming from these four qualities, include: • respecting the interests and values of all stakeholders; • achieving an appropriate balance of decision-making power among the stakeholders; 120 • achieving Pareto optimality; • maintaining good working relationships; • making best use of available information; • setting realistic goals; • allowing for complexity and uncertainty; and • using objective criteria where possible. The management strategies are listed in Table 4-1. They are divided into decision and technical strategies, corresponding to the decision and technical systems of the conceptual model. The decision strategies are further divided into iteration strategies, integration strategies, and negotiation strategies. Decision strategies Technical strategies Iteration strategies Integration strategies Negotiation strategies Iterative decision-making Process integration Primary negotiation Iterative technique Contingent decision Inclusive participation Principled negotiation Technical integration Iterative liability reduction Geographic integration Balanced negotiating power Mitigative technique Mediation Table 4-1. The management strategies. The iteration strategies support iterative cycles of decision-making, as a means of coping with complexity and uncertainty. The main iteration strategy is iterative decision-making: decision-making which is deliberately designed to repeat through a series of iterations, moving towards a condition of closure, but lacking a fixed endpoint. Contingent decision is an iteration strategy which refers to decision of the form "If Event X occurs, then Action Y will be undertaken"; it is a useful device for coping with uncertainty, and may increase the tractability of decision-making by taking advantage of differing perceptions of uncertainty among the 121 stakeholders. Iterative liability reduction is the strategy of providing the responsible stakeholder with an iterative reduction in its liability in exchange for technical progress and a series of fees; it can increase the tractability of decision-making by changing the all-or-nothing nature of the responsible party's liability. The iteration strategies in general are linked with the technical strategies of iterative technique and mitigative technique. Iterative liability reduction should be linked with the integration strategy of inclusive participation, as a safeguard for the interests of injured stakeholders. The integration strategies support integration of the decision system, as a means of increasing the tractability and efficiency of decision-making, and improving the flow of information among stakeholders. Process integration refers to the merger or linkage of isolated processes, in an effort to reduce procedural complexity. Inclusive participation refers to the inclusion of all stakeholders in decision-making, in the interests of improved fairness and improved goal-setting. Geographic integration refers to expansion of geographic scope to cover multiple polluted sites, multiple receptors, and potential disposal locations; this can provide economies of scale and facilitate direct comparison of different sources and their relative threat to health. The integration strategies in general are linked with the strategy of technical integration. Inclusive participation should be linked with strategies in which the interests of certain stakeholders might suffer from their inability to participate in decision-making (e.g., mitigative technique or iterative liability reduction). The negotiation strategies support the delegation of management to the collective intellect, rationality, and responsibility of the stakeholders. Primary negotiation refers to negotiation as both "first in importance" and "first in a series", relative to the unilateral modes of 122 decision-making. This has a high potential to support good management. It should be linked with the other negotiation strategies to achieve this potential. Principled negotiation refers to a cooperative and consensual style of negotiation, in which joint gains are achieved through tradeoffs and synergies. Balanced negotiating power requires a relative balance in the BATNAs and worst-nightmares of the participants; this clearly supports the quality of fairness, and indirectly supports the other three qualities as well. Mediation refers to procedural and substantive assistance rendered by a neutral party; this is beneficial in most negotiations and essential in complicated negotiations. There are three technical strategies: iterative technique, technical integration, and mitigative technique. Iterative technique is a deliberately repetitive and open-ended approach to the design and implementation of technical options, moving towards a condition of closure, but lacking a fixed endpoint. It implies planned periodic responses to new information, new understanding, and new circumstances. It is closely linked with iterative decision-making, and is justified by the common failure of linear technique to achieve its set endpoint. Technical integration refers to the merger or linkage of technical support, so that it is provided to the decision-making process rather than to individual stakeholders. This is linked with process integration. It has the potential to support the qualities of efficacy and efficiency, by improving the flow of information and ideas, avoiding duplication of effort, avoiding the problem of technical advocacy for individual stakeholders, and developing options for joint benefit of the stakeholders. It can also support the quality of fairness, by providing technical support for stakeholders with limited resources. 123 Mitigative technique is the effort to reduce damage and risk to human and ecological health. This supports the qualities of goal-setting, efficacy, efficiency, and fairness: the alternative goal of restoration is generally infeasible, and technical efforts to achieve an infeasible goal cannot be efficacious, efficient, or fair to any of the stakeholders. In the interests of fairness to injured stakeholders, it should be linked with inclusive participation and with the negotiation strategies. It is also linked with iterative technique, as the effort to reduce damage and risk to health must be ongoing. 4.5.2 The management strategies and soft paradigm thinking At the broadest level, the management strategies have been developed with soft paradigm thinking. This point is discussed in Section 7.2. At a more detailed level, the strategies are inherently rooted in soft paradigm thinking. The iterative decision strategies and iterative technique reflect the soft paradigm concept of iterative cycles of decision, implementation, and learning as a response to complexity and pervasive uncertainty. The integrative decision strategies and technical integration reflect the soft paradigm concept of ambiguous boundaries and complicated linkages among problems. The negotiation strategies are consistent with the soft paradigm concept of decision-making as a bounded conflict among multiple stakeholders with divergent interests and values. A more specific example of soft paradigm thinking arises with respect to mitigative technique. Under the goal of mitigation, the question of assessing risks to human and ecological health becomes a key issue, going beyond simple numerical concentration standards. The assessment of risk inevitably demands subjective and value-based assumptions regarding which 124 risks require consideration, how to evaluate those risks, and how to incorporate those evaluations into decision-making. These are tasks for which hard paradigm thinking, on its own, is not well-suited, particularly when multiple stakeholders with divergent values and interests are involved. A version of risk assessment which reflects soft paradigm thinking and soft/hard complementarity should be an integral component of mitigative technique. Considerable research is currently underway in this area (see the volume edited by Cothern, 1996). 125 5. S A M : a compound model of groundwater pollution management SAM (Simulated Aquifer Management) is a compound model, based on the conceptual model of Chapter 3. It is an assemblage of transform sub-models, with subjective and algorithmic linkages. SAM has two roles in this thesis: as a heuristic device to develop understanding (Section 2.2.6) and as a means of illustrating both the management strategies (Chapter 4) and the concept of soft/hard complementarity (Section 2.3). The role of SAM as a heuristic device is discussed in Section 7.2. The illustrative role of SAM is embodied in Chapters 5 and 6, and discussed explicitly in Section 5.5.2. Possible applications and extensions of SAM are discussed in Section 7.3.3. This chapter is a general description of SAM. Section 5.1 is an overview of SAM, followed by detailed descriptions of its sub-models in Sections 5.2 to 5.4. Section 5.5 relates SAM to the management strategies and to the concept of soft/hard complementarity. A run of SAM, applied to a hypothetical case, is described in Chapter 6. 5.1 Overview of SAM SAM consists of a set of sub-models, hierarchically organized to match the subsystems of the general conceptual model presented in Chapter 3 (Figure 5-1). The main sub-models are the pollution model and the management model, representing the pollution system and the management system. 126 Conceptual model (Chapter 3) SAM Figure 5-1. Comparison of S A M with the conceptual model of Chapter 3. The pollution model is a compound model which consists of groundwater flow and advective transport sub-models representing the flow and transport systems, as well as a geostatistical sub-model representing heterogeneity in hydraulic conductivity. The management model is a compound model, comprised of technical and decision sub-models representing the technical and decision systems. The technical model is a compound model which includes algorithmic sub-models of groundwater flow, advective transport, optimization of pumping rates, and costing of options. The decision model consists of either an algorithmic or an analog model of negotiation; the algorithmic and analog models can be used interchangeably. The emphasis in SAM is on the management model, consistent with the conceptual model of Chapter 3. 127 Note that algorithmic sub-models of flow and transport occur in both the pollution model and the technical model. In the pollution model, these are intended to represent the reality of the pollution system in SAM, and are referred to as "reality models". This distinguishes them from the corresponding sub-models in the technical model, which represent the technical system's perception of the pollution system, and are referred to as "perception models". The specific conceptual model which precedes a run of SAM describes a hypothetical groundwater pollution case, matched to the general conceptual model of Chapter 3, and containing information which is used to generate input for the sub-models of SAM. Section 6.1 provides an example of a specific conceptual model for SAM. implementation Figure 5-2. The now of S A M . The flow of SAM is summarized in Figure 5-2. It begins with an initial run of the pollution model. A limited sampling of the output from this run of the pollution model is transferred to the technical model as "observations". The technical model generates a set of technical options ("advice") based on these observations. The options are passed to the decision model, which produces "decisions" by choosing among these technical options. The "decisions" are then "implemented" on the pollution model. 128 SAM is designed to operate in an iterative fashion. The pollution model can be re-run to reflect the implemented decisions. The effect of the implemented decisions on the pollution model can be observed by the technical model, as a basis for further advice to the decision model, leading to further decisions, and so on. The sub-models and linkages of SAM are shown in Figure 5-3 and Table 5-1. The following three sections describe the pollution model, the technical model, and the decision model. In each section, brief note is made of the simplifications used to derive SAM from the general conceptual model of Chapter 3. HMODFLOW (flow) BMODPXTHr BA($MOD- • BMINOS (transport) 1 (simulation-optimization) \ MANAGEMENT MODEL Technical model tjbfjRA (costing) -KEY B algorithmic submodel © analog submodel decisions i technical i options r V Decision model ^ subjective linkage ^ algorithmic linkage Figure 5-3. Sub-models and linkages in S A M . 129 System and modeling task Specific model in SAM P O L L U T I O N S Y S T E M P O L L U T I O N M O D E L Graphical input/output B COAST, SURFER™ Flow system Flow model Heterogeneity in hydraulic conductivity H GSLIB Groundwater flow H M O D F L O W (reality model) Transport system Transport model Pollutant transport H M O D P A T H (reality model) M A N A G E M E N T S Y S T E M M A N A G E M E N T M O D E L Technical system Technical model Model of groundwater flow a M O D F L O W (perception model) Model of pollutant transport a M O D P A T H (perception model) Simulation-optimization of pumping rates a A Q M O D and MINOS Cost estimation for technical options a C O R A Decision system Decision model Negotiation a COOP or ©PAQMAN Table 5-1. The sub-models of S A M . Algorithmic sub-models are indicated with a B symbol, and the analog sub-model is indicated with a © symbol. The specific software packages used in the thesis are listed in Table 5-1. Some of these packages are off-the-shelf software that was used without modification: SURFER™, GSLIB, MODPATH, MINOS, and CORA. Minor changes were made to MODFLOW, and COAST was adapted for compatibility with GSLIB. AQMOD was developed for this thesis by rewriting an existing program (AQMAN; Lefkoff and Gorelick, 1987) for compatibility with MODFLOW. COOP was written in FORTRAN for this thesis, on the basis of algorithms in Fraser and Hipel (1984), with one modification. The role-playing exercise PAQMAN was developed for this thesis. 130 5.2 Pollution model The pollution model uses standard hydrogeological and geostatistical software to generate sub-models which represent the pollution system. The finite-difference program MODFLOW (McDonald and Harbough, 1988; Anderson and Woessner, 1992) is used for the reality model of groundwater flow. The particle-tracking program MODPATH (Pollock, 1989) is used for the reality model of pollutant transport. These programs are sequentially linked — MODPATH input requires MODFLOW output data. MODFLOW is a widely-used code for modeling three-dimensional groundwater flow. It has become a de facto standard for groundwater consulting work. The version of MODFLOW used is MODFLOW PC/EXT Version 1.3 (July, 1992), distributed by the International Ground Water Modeling Center. This version includes the PCG2 solver (Hill, 1990) and the rewetting package (McDonald et al, 1991). The MODFLOW PC/EXT source code has been modified for this thesis, mainly for compatibility with programs which process input and output, and to convert the main storage array to a dynamically allocatable array (FORTRAN 90 standard). The spatial distribution of hydraulic conductivity (K) in the MODFLOW reality model is a complex pattern derived using the GSLIB geostatistical software library (Deutsch and Journel, 1992). The use of GSLIB introduces complexity into the MODFLOW reality model, with a different K value assigned to each node. This has a practical benefit in the development of the MODFLOW perception model. The perception model requires calibration of a simple K pattern, based on limited observations of the pollution model. Empirically, it was found that it is easier to perform this calibration with a complex K pattern in the reality model. 131 The graphical software package COAST (Sperling et al, 1993) is used to create input files for MODFLOW and MODPATH, and to visualize output. COAST was modified for this thesis, mainly for compatibility with GSLIB. SURFER™ (Golden Software, Inc., 1989) is also used to visualize output. Relative to the conceptual model of the pollution system (Section 3.1), the pollution model contains several broad simplifications (Figure 5-4). The hydrogeologic complexities of the subsurface are reduced to a pattern of hydraulic conductivity (K) values generated by GSLIB, with one K value for each node in the MODFLOW finite-difference mesh. MODFLOW is limited to flow in saturated porous media; fracture flow and unsaturated flow cannot be represented. The representation of the geometry of the flow system is somewhat constrained by the rectangular character of the finite difference grid. GSLIB imposes a further restriction of constant grid spacing. It may be necessary to impose arbitrary boundaries in order to constrain the grid to a practical size. MODPATH is limited to advective flow, and assumes that the data which it receives from MODFLOW represents a steady-state flow field. These simplifications are reasonable in the context of SAM, which emphasizes the management system rather than the pollution system. The pollution model is sufficiently detailed to present a challenge to the management model. 132 primary source stream at risk secondary source water well at risk r pollutant source cell MODPATH Simplifications n u n pollutant transport MODPATH sink cell MODFLOW MODPATH constant head boundary MODFLOW Figure 5-4. Representation of the simplifications imposed by MODFLOW, MODPATH, and GSLIB on the conceptual model of the pollution system. 5.3 Technical model The technical model uses a combination of algorithmic models and subjective judgement to represent the observations-to-advice function of the technical system (Figure 5-3, Table 5-1). 133 This represents the work of a technical organization providing integrated technical support for a set of stakeholders. (The implementation function of the technical system is nominally a part of the technical model, for consistency with the general conceptual model. In practice, this function performed by revising the pollution model directly, without reference to the technical model.) The observation linkage between the pollution and technical models consists of limited and imperfect information, subjectively sampled from the output of the pollution model. The information applies to specific point locations in the vicinity of water wells and polluted sites, and consists of hydraulic conductivity values, hydraulic head values, and the presence or absence of a plume of pollution. This is representative of the type of information which would be available to a real-world technical system. The limited and imperfect character of this information represents technical uncertainty. For convenience in calibrating between the reality and perception models, the technical system is assumed to have perfect information with respect to boundary locations, boundary conditions (other than recharge), specific storage, and porosity. The advice linkage between the technical and decision models is represented by a set of costed technical options. The generation of the technical options in SAM involves three tasks: • simulation modeling of flow and transport; • simulation-optimization modeling of pumping rates; and • generation and costing of technical options. MODFLOW and MODPATH (see Section 5.2) are used for the models of flow and transport within the technical model. These are the "perception models", representing the technical system's perception of the pollution system, as distinguished from the "reality models" 134 which represent the actual pollution system. The MODFLOW perception model is calibrated against the MODFLOW reality model by trial-and-error balancing of the spatial distribution of K against recharge to match the limited observations of the hydraulic head field. It may also be necessary to adjust the MODFLOW calibration so that the MODPATH perception model matches the limited observations of the presence or absence of a plume of pollution. Each iteration of SAM provides a new set of incomplete and imperfect information regarding hydraulic conductivity, hydraulic head and plume location. This may necessitate re-calibration of the MODFLOW perception model. Note that the calibration process in SAM is a subjective, trial-and-error exercise bounded by objective information — as are most such calibration processes in the real world. COAST and SURFER™ are used for input preparation and visualization, as with the reality models. The simulation-optimization model assumes that the pollution system contains a set of water supply wells and hydraulic containment wells. The task is to estimate an optimal pumping rate for each well, with the aim of protecting the water supply and containing the spread of pollutants. The simulation-optimization model uses two programs: MINOS and AQMOD. MINOS is a standard optimization program (Murtagh and Saunders, 1987). AQMOD performs the simulation task, and produces files in MPS format, for use as input for linear optimization by MINOS. AQMOD was developed for this thesis, by the author, by revising the existing program AQMAN (Lefkoff and Gorelick, 1987). It was necessary to revise AQMAN because it uses the Trescott code, which is a precursor of MODFLOW. The Trescott code is much more limited than 135 MODFLOW, and uses a completely different input format. AQMOD was developed by splicing MODFLOW code into AQMAN, in place of the Trescott code. The same input files can be used for both the MODFLOW perception model and for AQMOD. A linearization protocol was developed for use with AQMOD and MINOS, in order to deal with the nonlinearity of the response of hydraulic head to pumping stress in an unconfined aquifer. The steps in the protocol are as follows: 1. Establish a trial solution, i.e., guess at a set of approximate pumping rates. 2. Run AQMOD, using the trial solution as the base case. 3. Run MINOS, optimizing the changes in pumping rates from the base case. 4. Use the MINOS solution from Step 3 to adjust the trial solution. 5. Repeat Steps 2, 3, and 4 until the optimal changes in Step 3 are negligible. (Typically, only two or three iterations are required. 6. (optional) Check as to whether or not the feasible region is sufficiently large that the output is not entirely governed by the constraints, by running the optimization with the sign of the objective function reversed (e.g., from maximization to minimization). Typically, optimizing on changes to the pumping rates involves maximizing the negative change at containment wells and the positive change at water supply wells, within reasonable bounds. This assumes that the intention is to pump as little dirty water and as much clean water as possible, within reasonable limits. AQMOD and MINOS are algorithmic models, but they demand an exercise of subjective judgement in matters such as the location of proposed new wells for hydraulic containment. 136 The final task of the technical system is to produce the set of costed technical options as input for the decision model. This task involves considerable subjective judgement. The output of the flow, transport, and simulation-optimization models must be translated into options such as partial or complete source removal, various hydraulic containment schemes, monitoring at certain locations, and so forth. The technical options are developed and costed with the help of CORA Version 3.0 (Cost of Remedial Action), a program developed to produce order-of-magnitude cost estimates for polluted sites in the United States (CH2M Hill, 1990b). Many of the options involve continuing costs over a period of years, with the consequent need to calculate present values. The cost estimates are compiled and reduced to present values with a standard spreadsheet program (EXCEL™.) The technical model is relatively complicated, and is sufficiently detailed to provide a challenging set of technical options to the decision model. However, it does contain a number of simplifications relative to the technical system described in the conceptual model of Chapter 3. It is treated as a single technical organization. Its observations of the pollution system are simplified to a limited sampling of output from the pollution model. The technical advice which it provides is limited to a small number of technical options, with limited information on the costs and benefits of these options. (The number of technical options is constrained by the practical limitations on determining preferences in the decision model; see Section 5.4). Scientific and engineering skills are reduced to a small set of models, spliced together with subjective judgement. The models themselves embody further simplifications. For example, the MODFLOW perception model embodies not only the simplifications which apply to the MODFLOW reality model in the pollution system, but a further simplification: subsurface 137 complexity is reduced to a few large zones of uniform hydraulic conductivity, rather than the more detailed GSLIB geostatistical representation. The MODPATH perception model considers only advective transport, in common with the MODPATH reality model in the pollution model. Technical uncertainty is simplified to the difference between the pollution model and the limited sampling of the pollution model which is provided to the technical model. 5.4 Decision model The decision model in SAM uses either the algorithmic model COOP or the analog model PAQMAN (Polluted Aquifer Management) to represent negotiations among a set of stakeholders in the decision system. COOP is based on conflict analysis, which is a variant of game theory. PAQMAN is a role-playing exercise in which volunteers act out the roles of negotiations among stakeholders, providing an analog of real-world negotiations. The two models are interchangeable and complementary. COOP is the primary model. PAQMAN is used in a secondary fashion, providing human insights which are absent from COOP. (Note that PAQMAN is not software, but rather a role-playing exercise involving human participants.) The set of stakeholders consists of a regulatory stakeholder, a responsible stakeholder, and an injured stakeholder. The input to the decision model consists of a set of options and preference information for each stakeholder. A set of options for a given stakeholder consists of the technical options (the output of the technical model), plus a BATNA option as an alternative to negotiation (e.g., the regulatory stakeholder might have the BATNA of breaking off negotiations and suing the responsible stakeholder). 138 Preparing the preference information for each stakeholder requires subjectivity and creativity on the part of the modeler. It is also an iterative process, requiring a number of attempts to eliminate inconsistencies. The task becomes increasingly complicated as the number of options increases. It is guided by the costs and perceived benefits of the technical options for the given stakeholder. In general, the costs weigh most heavily for the responsible stakeholder, and benefits in terms of damage and risk reduction weigh most heavily for the injured stakeholder. 5.4.1 COOP (conflict analysis model) The conflict analysis methodology of Fraser and Hipel (1984) is a form of game theory. SAM uses cooperative conflict analysis, an extension of basic conflict analysis which can be applied to negotiations. This section provides a brief summary of conflict analysis as used in SAM. General notes on game theory and conflict analysis are provided in Appendix 2. Basic conflict analysis is summarized in Figure 5-5. A conflict situation is modeled as a set of players, each of whom has a set of options. Each option must be either accepted or rejected; an option of this type is a binary option. The combined set of all players' options contains n options, and these can be combined to create a set of 2" mathematically-possible outcomes. Not all of these outcomes are feasible, e.g., some outcomes may contain two or more mutually exclusive options. The infeasible outcomes are rejected by a feasibility analysis, on the basis of feasibility information supplied as input. Preference analysis is then applied to the set of feasible outcomes, generating a preference vector for each player, on the basis of preference information supplied as input. A preference vector is a list of all feasible outcomes, ordinally-ranked in order of preference for a given player. Finally, the set of feasible outcomes is subjected 139 to a stability analysis, which uses the preference vectors to generate a set of outcomes which are stable and therefore likely to occur. These stable outcomes are called equilibriums. In general, conflict analysis produces a set of non-unique equilibriums, and further distinction among these equilibriums requires subjective judgement. INPUT Binary options for each player Set of mathematically-possible outcomes INPUT Feasibility information for the set of outcomes Feasibility analysis Set of feasible outcomes INPUT Preference Preference masks for analysis each player OUTPUT Set of stable outcomes (equilibriums) Set of unstable outcomes Figure 5-5. Flowchart of basic conflict analysis. The preference analysis step requires additional description. The preference information supplied as input consists of an ordinally-ranked list of preference masks for a given player. A preference mask is a list of the binary options, with each binary option assigned a status of 140 accepted, rejected, or free. A given preference mask has a mask weight equal to 2(nummask"rank)) where nummask in the number of preference masks for the given player, and rank is the ordinal rank of the preference mask in the ranked list of preference masks (l=highest). Each feasible outcome is compared to each preference mask for a given player (Figure 5-6). If those binary options which are accepted in the preference mask are also accepted in the outcome, and if those binary options which are rejected in the preference mask are also rejected in the outcome, then the preference mask is said to match the outcome. The binary options which have a free status in the preference mask do not affect the comparison. Each outcome is assigned an outcome weight equal to the sum of the mask weights of all the preference masks which it matches. The outcomes in the given player's preference vector are ordinally ranked according to the outcome weights. A simple example of the use of preference masks is provided in Table 5-2. Equal preferences among sets of two or more preference masks are permitted; equally-preferred preference masks are assigned the same mask weight. Equal preferences among outcomes in a preference vector are also permitted. Preference mask Option 1 Option 2 Option 3 Accepted Rejected Free Feasible outcome 1 Feasible outcome 2 Feasible outcome 3 Option 1 Accepted Option 2 Rejected Option 3 Accepted Option 1 Accepted Option 2 Rejected Option 3 Rejected Option 1 Accepted Option 2 Accepted Option 3 Accepted Figure 5-6. Example: comparison of preference mask with feasible outcomes. This example compares a preference mask with three feasible outcomes. Feasible outcomes 1 and 2 match the preference mask; feasible outcome 3 does not. Note that the free option does not affect the comparison. 141 Preference masks Mask weights Option 1 Option 2 Option 3 Preference mask 1 accepted free free 4 Preference mask 2 free accepted free 2 Preference mask 3 free free accepted 1 Preference vector Outcome weights Preference ranking Option 1 Option 2 Option 3 1 accepted accepted accepted 4+2+1=7 2 accepted accepted rejected 4+2+0=6 3 accepted rejected accepted 4+0+1=5 4 accepted rejected accepted 4+0+0=4 5 rejected accepted accepted 0+2+1=3 6 rejected accepted rejected 0+2+0=2 7 rejected rejected accepted 0+0+1=1 8 rejected rejected rejected 0+0+0=0 Table 5-2. Simple example of the use of preference masks to develop a preference vector for one stakeholder. This example involves a conflict with three options and eight feasible outcomes. The stakeholder has three preference masks. Each of the eight feasible outcomes is compared against these three preference masks. For example, consider the feasible outcome with a preference ranking of 1 in the preference vector. In this outcome, all three options are accepted. This outcome matches all three preference masks, and is given an outcome weight equal to the sum of all three mask weights: 4+2+1=7. Now consider the feasible outcome with a preference ranking of 5. This outcome matches only preference masks 2 and 3, and is given an outcome weight equal to the sum of the mask weights of these two preference masks: 0+2+1=3. Cooperative conflict analysis distinguishes among cooperative and non-cooperative options, and among cooperative and non-cooperative outcomes. A cooperative option is a shared option which must be either jointly selected or jointly rejected by all parties who share that option, and a cooperative outcome is any outcome in which at least two players jointly select at least one cooperative option. 142 INPUT for each player 'Cooperative binary options »Non-ooperative binary options Set of mathematically-possible outcomes INPUT Feasibility information for the set of outcomes Feasibility analysis Set of feasible outcomes INPUT Preference Preference masks for analysis each player f Preference vectors 1 Non-cooperative stability analysis Cooperative stability analysis Pareto-optimality analysis OUTPUT Set of stable outcomes (equilibriums) • Pareto-optimal non-cooperative • Pareto-optimal cooperative • Pareto-suboptimal non-cooperative • Pareto-suboptimal cooperative Set of unstable outcomes • Cooperative Non-cooperative Figure 5-7. Flowchart of cooperative conflict analysis. A flowchart of cooperative conflict analysis is provided in Figure 5-7. The stability analysis is divided into two parts: non-cooperative and cooperative. The non-cooperative stability analysis determines non-cooperative equilibriums (stable outcomes which do not contain any 143 cooperative options). The cooperative stability analysis determines cooperative equilibriums (stable outcomes which contain at least one cooperative option). The stability analysis is followed by Pareto-optimality analysis, which establishes whether the equilibriums are Pareto-optimal or Pareto-suboptimal. In the context of conflict analysis, a given equilibrium is Pareto-optimal if there is no cooperative equilibrium which all players prefer to the given equilibrium, and Pareto-suboptimal otherwise. The set of equilibriums to consider can be safely limited to the Pareto-optimal equilibriums. In negotiation, a Pareto-optimal cooperative equilibrium would be an agreement in which there are no further gains that any of the parties could make at no loss to any of the other parties, i.e., "nothing is left on the table." In summary, cooperative conflict analysis divides the feasible outcomes into six categories: • Pareto-optimal non-cooperative equilibriums, • Pareto-optimal cooperative equilibriums, • Pareto-suboptimal non-cooperative equilibriums, • Pareto-suboptimal cooperative equilibriums, • unstable non-cooperative outcomes, • unstable cooperative outcomes. Cooperative conflict analysis in SAM is implemented by COOP, a program written in FORTRAN for this thesis, by the author, on the basis of algorithms in Fraser and Hipel (1984), with one modification which reduces the computational burden. This modification is based on two assumptions: the cooperative options are mutually-exclusive with the non-cooperative options, and all parties have the same set of cooperative options. The first assumption can be 144 restated as the assumption that the parties must either reach a cooperative agreement involving some combination of the cooperative options, or they do not reach agreement and must fall back on unilateral alternatives such as litigation. The second assumption is equivalent to the assumption that all parties must reach agreement together, or no agreement is reached; there is no provision for side deals among a subset of the parties which are not approved by all of the parties. These are reasonable assumptions for the type of negotiation which is under consideration in this thesis. The computational benefit of the COOP assumptions is to sharply reduce the number of feasible outcomes, by automatically rejecting large blocks of mathematically-possible outcomes as infeasible. In general, the computational burden of a conflict analysis problem grows as 2", where n is the total number of options. Under the COOP assumptions, the computational burden grows as 2non+coop, where non is the number of non-cooperative options, and coop is the number of cooperative options divided by the number of parties. As an example, consider a problem in which each of three parties has 1 non-cooperative option and 8 cooperative options. This yields a total of 3 non-cooperative options, and 24 cooperative options. For this case, n=21, non=3, and coop-%. The relative computational burden of this problem would be 211 with the COOP 27 assumptions, and 2 otherwise. On the computer used for this thesis, the computing time for this problem would be about 1 minute with the COOP assumptions, and 39 days otherwise. In practice, 8 is a reasonable upper limit on the value of coop (i.e., each party has an identical set of 8 cooperative options, for a total of 24 cooperative options). This limiting value reflects the need to develop preference masks for each party, rather than computational burden. 145 The development of preference masks becomes increasingly complicated as the number of options increases. COOP is used within SAM to simulate a negotiation among at least two stakeholders. The technical options produced by the technical model form a set of cooperative options which is held in common by all parties. In addition, each party has at least one non-cooperative BATNA option. In general, a run of COOP involving a large number of options and balanced negotiating power tends to produce a relatively large number of Pareto-optimal equilibriums. (In the example in the following chapter, as many as 56 different Pareto-optimal equilibriums arise from a single round of negotiations.) There are no objective means of selecting any one equilibrium as more likely than the others. Either a random or a subjective choice must be made among the equilibriums in order to proceed to another iteration. 5.4.2 PAQMAN (role-playing model) PAQMAN (Polluted Aquifer Management) is a role-playing analog model which was developed for this thesis by the author. Precursors of PAQMAN were developed by the author for educational purposes, and played by students in an undergraduate groundwater contamination course. PAQMAN was developed for research purposes, as an alternative and counterpoint to COOP. PAQMAN contributes intangible human aspects, reveals inconsistencies, provides insights, and brings new ideas into consideration. PAQMAN brings three volunteer participants together to act out the roles of the responsible party, the regulatory party, and the injured party in a simulated negotiation. All participants receive the same set of general instructions which describe the overall situation. 146 Each participant receives a different set of confidential instructions, describing their preferences among both the shared technical options and the unilateral options for breaking off the negotiation. The confidentiality of these instructions represents uncertainty within the decision system. The instructions can be written so as to parallel an equivalent run of COOP. The participants' choices are tightly prescribed by the instructions, and there is only limited scope for them to generate their own options. The researcher must perform as game manager and observer, as well as performing the minor role of technical consultant to the three stakeholders. The final step is a debriefing session, in which the participants comment on the exercise and discuss the decisions that they made. This is probably the most important part of the exercise. It provides interesting feedback, not just on the results of the exercise, but on the overall concept and structure of SAM itself. (COOP, on the other hand, provides no such feedback, and dutifully performs its algorithmic tasks without question!) 5.4.3 Simplifications The decision model in SAM contains a number of important simplifications relative to the conceptual model of the decision system in Chapter 3. Decision-making in SAM is limited to negotiation involving all stakeholders. Unilateral modes of decision-making are considered only insofar as they provide BATNAs for individual stakeholders. SAM does not pursue the consequences of a failure to reach agreement. This restriction to negotiation reflects the importance of negotiation in groundwater pollution management, and the emphasis on negotiation in the proposed management strategies. 147 The set of stakeholders in SAM is limited to three stakeholders: one responsible stakeholder, one regulatory stakeholder, and one injured stakeholder. This is a practical restriction, imposed mainly by the difficult task of working out preferences for each stakeholder. The stakeholders operate under a simple set of procedural rules. They are permitted to negotiate with each other. Each stakeholder has a BATNA. The regulatory stakeholder can order a responsible stakeholder to perform management actions, and impose penalties or launch litigation to enforce such orders. The responsible stakeholder can appeal a regulatory penalty or defend a regulatory lawsuit. The injured stakeholder can launch litigation against the responsible stakeholder (for compensation) and against the regulatory stakeholder (for failure to act according to a statute). PAQMAN involves uncertainty among the stakeholders about the other stakeholders' preferences. COOP assumes that the stakeholders have a perfect understanding of each others' preferences. The conflict analysis methodology is capable of handling misperceptions about other stakeholder's preferences (Fraser and Hipel, 1984), but this has not been implemented in COOP. The decision model in SAM represents a reasonable balance between capturing the complexity of the decision system and respecting the dictate of parsimony in modeling. Most of the simplifications are practical rather than inherent in the model. (For example, another stakeholder could be added, or misperception could be incorporated into COOP.) 148 5.5 S u m m a r y a n d d i s c u s s i o n This chapter has described the SAM compound model in general, laying out the skeleton of its structure. The following chapter (Chapter 6) provides detailed documentation of a full run of SAM, putting flesh on the skeleton. The concluding chapter includes discussions on the heuristic role of SAM in the development of this thesis (Section 7.2), and directions for future research with SAM (Section 7.3). This section summarizes the chapter, and discusses the relationships of SAM to the management strategies and to the concept of soft/hard complementarity. 5.5.1 Summary SAM is a compound model composed of transform sub-models, with algorithmic and subjective linkages (Figure 5-3). The structure of SAM reflects the systems structure of the conceptual model (Figure 5-1): a pollution model with flow and transport sub-models, and a management model with technical and decision sub-models. The specific sub-models and associated software are shown in Table 5-1. SAM is designed to operate in an iterative fashion, cycling from the pollution model to the management model and back to the pollution model again. A cycle of SAM (see Figure 5-2) begins with a run of the pollution model. Limited and subjectively-chosen observations of the output of the pollution model are provided to the technical model. The technical model then develops a set of technical options based on these observations, using subjectively-linked simulations of flow and transport, simulation-optimization of pumping rates, and costing algorithms. These technical options are passed as advice to the decision model, which uses either 149 an algorithmic conflict analysis program or an analog role-playing exercise to represent negotiations among stakeholders. The technical options are incorporated into a set of subjectively-developed preferences for each stakeholder in the decision model. If the stakeholders reach agreement given these preferences, the decision model transmits a set of decisions on the technical options to the technical model for implementation on the pollution model. A new cycle may then begin with another run of the pollution model. 5.5.2 SAM and the management strategies The structure of S A M accommodates all of the management strategies proposed in Chapter 4 (Table 5-3). Some of the strategies are intrinsic to S A M , and are automatically included; excluding them would require structural modifications. The other strategies are extrinsic, and can be included or excluded by modifying the input. Decision strategies Technical strategies Iteration strategies Integration strategies Negotiation strategies Iterative decision-making Process integration Primary negotiation Iterative technique Contingent decision Inclusive participation Principled negotiation Mitigative technique Iterative liability reduction Geographic integration Balanced negotiating power Technical integration Mediation Table 5-3. List of the proposed management strategies. The strategies which are intrinsic to S A M are shown in italics. The strategies of iterative decision-making, iterative technique, primary negotiation, inclusive participation, process integration, and technical integration are intrinsic to S A M . The central concept of S A M involves an iterative process in which all stakeholders negotiate together over technical options developed by a single technical body, with unilateral modes of decision-making in the background as B A T N A s . 150 Two of the strategies are intrinsic with respect to COOP and extrinsic with respect to PAQMAN. These are principled negotiation and mediation. They are intrinsic with respect to COOP because COOP assumes perfect understanding of other stakeholders' options and preferences — a circumstance which is unlikely to arise in the absence of principled negotiation and mediation. Dropping this assumption would require the extension of COOP to allow hypergames (games of misperception and deception; Fraser and Hipel, 1984, Chapter 4). Principled negotiation is extrinsic to PAQMAN, because players could be given confidential instructions to negotiate in a competitive and adversarial manner. Mediation is extrinsic to PAQMAN because it could be represented by providing a fourth role of mediator in addition to the three stakeholder roles. The rest of the strategies are extrinsic to SAM. Balanced negotiating power is extrinsic, because the balance of negotiating power in SAM is varied by adjusting the relative preferences of the stakeholders with respect to the exercise of the various BATNAs, in both COOP and PAQMAN. It is possible to create a strong imbalance in negotiating power in favour of one stakeholder by giving that stakeholder high preferences for the BATNAs of all stakeholders, and giving other stakeholders low preferences for the BATNAs of all stakeholders. However, such an imbalance simply leads to an obvious result: an equilibrium which includes only the technical options preferred by the stakeholder with the excess power. It is generally more interesting to adjust the preferences so as to represent a rough balance of negotiating power. Geographic integration is extrinsic. For example, the scenario in the following chapter is concerned with a relatively large geographic area, providing an example of geographic 151 integration; but this strategy could be excluded by limiting the scenario to a much more restricted area in the immediate vicinity of the pollution source. Contingent decision and iterative liability reduction can easily be introduced in the preference input. They are not represented in the example in Chapter 6. The technical options in SAM can be based on mitigative technique or restorative technique. (A pollution source in MODPATH is easily removed, unlike a pollution source in the real-world!) The example in Chapter 6 is based on mitigative technique. A caution is in order. SAM has been presented as illustration of the management strategies. It also serves as documentation of the thought process that led to their development; this point is discussed in Chapter 7. However, SAM should not be interpreted as a means of testing of the management strategies by applying them to a representative hypothetical case. This limitation arises for a number of reasons: • Some of the strategies are intrinsic to SAM. • A specific conceptual model for a run of SAM contains a significant amount of detail. This detail is necessary for consistency with the general conceptual model of Chapter 3; speaking more broadly, the detail is necessary to address the complexities of groundwater pollution management as conceptualized in this thesis. However, the detail inevitably introduces an element of idiosyncrasy which would weaken any attempt to use SAM to test the general application of a particular strategy. • The output of COOP generally includes a large number of non-unique equilibriums, requiring subjectivity in interpretation. This would further weaken any attempt to use SAM to test a particular strategy. 152 • As noted in Chapter 2, an algorithmic model derives a special legitimacy from its internal objectivity; but compound models such as SAM, with their internal subjective elements, do not share this legitimacy. The subjective, non-algorithmic elements provide an opportunity for the modeler to consciously or unconsciously bias the results. The effect of this is to reduce confidence in the power of SAM to test the management strategies. • Most importantly, the management strategies form a relatively complex and interlocking set of ideas which are inherently not amenable to testing with transform models. Furthermore, such testing efforts would be counter-productive, because they would limit the strategies to those which could be legitimately tested with such models. 5.5.3 SAM and soft/hard complementarity The concept of soft/hard complementarity was introduced in Chapter 2. This is the idea that the soft and hard paradigms can function together. Many problems can be seen as a soft matrix containing embedded hard sub-problems. In such cases, the soft paradigm can supply global understanding and identify hard sub-problems which can be tackled by the hard paradigm. In SAM, the soft problem of groundwater pollution management is represented by the overall structure of the model, and by the subjective linkages. The overall structure of SAM acknowledges many soft problem characteristics, particularly in the explicit recognition of a multi-stakeholder decision system, and in the allowance for iteration and re-solution. The most important subjective linkages are in the development of the stakeholders' options (technical options and BATNAs) and in the development of stakeholder preferences. The fact that these aspects could not be given over to algorithmic sub-models suggests that they are inherently part of the soft matrix. 153 The hard sub-problems are represented by the algorithmic sub-models, particularly in the technical model. Clearly, simulation modeling of flow and transport addresses a hard sub-problem: understanding and predicting the behaviour of the pollution system. Simulation-optimization also addresses a hard sub-problem, namely the efficiency of a technical option. The hard sub-problem in the decision model is less obvious. As noted above, the stakeholders' options and preferences are subjective aspects which form part of the soft matrix of the problem. The implicit assumption behind the use of COOP is that given a set of options and a set of preferences, then determining the set of possible agreements is reduced to a hard sub-problem. This is a legitimate use of an algorithmic model, given a recognition of the surrounding soft matrix. This soft matrix includes not only the determination of input (options and preferences), but the use that is made of the output (the set of possible agreements). The practical applications of COOP in this context are discussed further in Chapter 7. 154 6. A run of S A M This chapter describes a run of SAM, applied to a hypothetical case which embodies typical characteristics of groundwater pollution. The purpose of this chapter is to flesh out the general description of SAM provided in the previous chapter, documenting the roles of SAM as a heuristic device (see Section 7.2) and as illustration of the management strategies (Chapter 5) and the concept of soft/hard complementarity (see Section 5.5). The chapter is necessarily lengthy, because a compound model with non-algorithmic elements demands extensive documentation in the interests of transparency. The chapter begins with a description of the hypothetical case (Section 6.1). This description is a specific conceptual model, corresponding to the general conceptual model of Chapter 3. It is the necessary precursor for SAM, providing the basis for: • the input which goes into the algorithmic sub-models; • the subjective judgements which link these algorithmic sub-models; and • the instructions provided to the volunteers in the role-playing sub-model. SAM is then applied to the hypothetical case using COOP as the decision model (Section 6.2) and using PAQMAN as the decision model (Section 6.3). The emphases are on the management model and the decision model, consistent with the conceptual model of Chapter 3 and the management strategies of Chapter 4. 155 6.1 H y p o t h e t i c a l c a s e The hypothetical case described in this section follows the same pattern as the descriptive conceptual model of Chapter 3: a pollution system containing groundwater flow and pollution transport systems; and a management system containing technical and decision systems. 6.1.1 Pollution system The area of concern is 10,000 m by 10,000 m (Figure 6-1). Locations are given in the text as eastings and northings (in meters), with the origin at the southwest corner of the area shown in Figure 6-1. 5 o o E 2 4 6 8 5 5 5 5 0 0 0 0 0 0 Springs 0 0 8500 N 6500 N ft 4500 N P 2500 N U 500 N P Figure 6-1. Location map. Contours show topography, with a contour interval of 1 meter. The Midnight Site contains the pollution source. 156 The groundwater flow system is within the unconfined Kaldor Aquifer, consisting of glaciofluvial sands and gravels, with minor clay lenses. The Kaldor Aquifer is square in plan view. Its north, south, east, and west boundaries are formed by vertical interfaces with impermeable till deposits, and correspond to the boundaries of the area of concern shown in Figure 6-1. The bottom boundary is a horizontal interface with impermeable till, at 50 m elevation. The upper boundary is the water table, which is a recharge surface. The overlying topographic surface has an overall slope to the north, with a central gully also sloping to the north. The only natural sink is a 1200 m long line of springs along the middle of the north boundary (4400-5600 E/10000 N). These discharge at 70 m elevation. The only natural source is recharge from precipitation. Recharge is 1 x 10"9 m/s; this value is constant in time and uniform in space. The Kaldor Aquifer contains three water wells, installed in 1935. Each well extracts water at a constant rate of 0.02 m /s. The pollution system consists of a source of trichloroethylene and chloride, and plumes of trichloroethylene and chloride. From 1985 to 1994, trichloroethylene and salt-contaminated water were illegally dumped by the Midnight Disposal Corporation at the Midnight Site (4700 E/6300 N). The wastes were dumped into a lagoon and allowed to percolate into the ground. The dumping operation was discovered and halted in May of 1994. Liquid-phase trichloroethylene has spread vertically through the entire thickness of the aquifer below the lagoon, creating a deep secondary source, in addition to the primary source at the lagoon. Lateral migration of liquid-phase trichloroethylene has been limited to <50 m in any direction, due largely to the horizontal nature of the interface with the underlying impermeable till. A conservative plume of chloride is 157 migrating away from the primary source. A retarded plume of dissolved trichloroethylene is migrating away from both the primary and secondary sources. 6.1.2 Management system This section describes the management system. Reference is made throughout this section, and elsewhere in this chapter, to the management strategies from Chapter 4. For reference, these are listed in Table 6-1. Decision strategies Technical strategies Iteration strategies Integration strategies Negotiation strategies Iterative decision-making Process integration Primary negotiation Technical integration Contingent decision Inclusive participation Principled negotiation Mitigative technique iterative liability reduction Geographic integration Balanced negotiating power Iterative technique Mediation Table 6-1. Summary of the management strategies. The strategies which are illustrated in this chapter are shown in bold type. 6.1.2.1 Decision system The Midnight Disposal Corporation has declared bankruptcy, and its principals cannot be located, having removed themselves to other jurisdictions. All written records pertaining to the Midnight site were destroyed in an unfortunate fire. Consequently, it has not been possible to locate any other liable parties. The polluted site is therefore an orphan site, and responsibility has devolved to the provincial and federal governments. The decision system in the Midnight case encompasses three stakeholders: • the Remediation Enforcement Group (REG), a provincial government agency charged with regulating polluted sites and associated groundwater pollution; 158 • the Pollution Response Panel (PRP), a joint federal-provincial government board charged with remediation of orphan sites and associated groundwater pollution; and • the Village of Canterbury (VIC), the local municipality which operates the three water supply wells in the Kaldor Aquifer. REG has the authority to unilaterally enforce technical actions which it considers appropriate, under the Groundwater and Soil Pollution Act (GASP). Its authority is backed by litigation powers, but it also has the authority to negotiate with the various stakeholders. REG is genuinely committed to protecting environmental quality and human health, and to the regulatory philosophy of "polluter-pays". In the past, REG tended to rely on linear decision-making and technique, restoration as the goal of remediation, and litigation as a means of enforcement. These have proved frustrating, and REG is evolving alternative approaches. PRP has conflicting interests. Both provincial and federal governments are under severe fiscal constraints, but both wish to avoid loss of local political support. Consequently, the PRP wishes to limit both costs and controversy as much as possible. VIC's primary interest is to provide a clean, cheap supply of water to the village. It also needs to maintain credibility with the residents of the village, who are concerned about the potential threat to their health. It is therefore in VIC's interest to obtain the highest possible level of remediation, in order to protect its wells. Furthermore, it wants to obtain this at no cost to itself, and to be seen to do so by the villagers. VIC has legal standing to enter into litigation against both PRP and REG. The former can be sued if it does not undertake appropriate technical measures, and the latter if it does not enforce such measures. 159 REG has approached PRP regarding negotiations on the Midnight case, with the first round to take place in December of 1994. This brings in the strategy of primary negotiation. PRP, mindful of the need to reduce controversy, has suggested that VIC be included in the negotiations. REG has agreed to this, mindful of the VIC's ability to litigate. VIC has agreed to participate, fearing that REG and PRP will concoct a deal which is unfavourable to VIC if VIC is not represented at the table. The participation of VIC brings in the strategy of inclusive participation. It also brings in the strategy of geographic integration, because the inclusion of VIC extends the case beyond the boundaries of the Midnight Site, to include VIC's three water wells. The three stakeholders have agreed that the goal of this first round of negotiations is to establish a consensus management plan for the first management period: 1 January to 31 December, 1995. They have further agreed that if this negotiation is successful, then a second round of negotiations will be undertaken in December of 1995, to establish a consensus management plan for the second management period, 1 January to 31 December, 1996. This brings in the strategy of iterative decision-making. 6.1.2.2 Technical system Consolidated Orthodox Numerics, Ltd. (CON) is an experienced local engineering firm which has been retained by the three stakeholder to act as the sole technical organization on the Midnight case. CON will be paid through an account which will be funded by PRP and disbursed through a trust administered jointly by the three stakeholders. This arrangement illustrates the strategy of technical integration. 160 CON has already undertaken emergency measures under the unilateral authority of PRP. The site has been fenced off. All liquids and sludges have been removed from the lagoon, and shipped to a special waste incinerator in another province. A number of exploratory monitoring wells have been drilled in the immediate vicinity of the lagoon, in order to estimate the degree of pollution. The results are discouraging. They indicate that some stringers of trichloroethylene have sunk to the base of the aquifer, 36 m below the surface. The lagoon itself is 50 m long (east-west) by 25 m wide (north-south). The preliminary estimate of the source volume is therefore 50 m x 25 m x 36 m. Data from the monitoring wells suggest that the base of the aquifer is essentially horizontal, and CON has inferred from this that lateral migration of separate-phase trichloroethylene will not be a serious problem. The water produced by well W2 is being sampled on a monthly basis; no pollutants have yet been detected. CON has also prepared a preliminary set of models based on the limited available information. This preliminary set consists of a MODFLOW model of groundwater flow, and a MODPATH model of chloride transport. (Note that these are perception models, representing CON's perception of the pollution system, as opposed to the reality models which represent the actual pollution system.) On the basis of the Round 1 models, which suggest that the plume is migrating north from the site, a fence of 5 monitoring wells (Ml to M5) will be established 300 m north of the Midnight site. An additional fence of 5 monitoring wells (M6 to M10) has been established 300 m south of the site, in order to establish whether any upgradient sources of pollution exist. (Refer to Figure 6-5 for the locations of these monitoring wells.) 161 The three stakeholders have requested that CON prepare a set of technical options to consider in their negotiations for the first management period (1995). These technical options are described in Section 6.2.1.2. If the negotiations for the first management period are successful, then CON will implement the chosen technical options and generate a new set of technical options for the second management period (1996). This approach brings in the strategy of iterative technique. 6.2 S A M u s i n g C O O P as t h e d e c i s i o n m o d e l This section describes the application of SAM to the hypothetical case described in Section 6.1, using COOP as the decision model. It consists of two iterations, corresponding to the 1995 and 1996 management periods. These iterations are referred to as Round 1 and Round 2 (Sections 6.2.1 and 6.2.2). 6.2.1 Round 1 This section describes Round 1, dealing with the first management period (1995). 6.2.1.1 Pollution model The pollution model consists of a GSLIB model of the spatial distribution of hydraulic conductivity, a MODFLOW reality model of groundwater flow, and a MODPATH reality model of the chloride plume. Figure 6-2 shows the reality model of the spatial distribution of hydraulic conductivity (K) in the Kaldor Aquifer. This distribution was developed on a 50 node x 50 node grid, using the SASIM simulated annealing program from the geostatistics library GSLIB (Deutsch and 162 Journel, 1992). It was based on an isotropic exponential variogram with an effective range of 3000 m. It was conditioned on a set of data points which define a planar trend of increasing K, from south to north. A single realization was chosen from a set of ten equally-likely realizations. The distribution of In K was set to a mean of -7.00 (equivalent to 9.1 x 10"4 m/s) and a variance 5 2 of 1.03, yielding a range of -10.22 to -3.68 (3.6 x 10" m/s to 2.5 x 10" m/s). These values are appropriate for a sand and gravel aquifer (see Freeze and Cherry, 1979, Table 2.2). Specific storage (Ss) and porosity were set at uniform values of 0.0005 and 0.2, respectively. The gray-scale values show the natural logarithm of hydraulic conductivity (K) in m/s. Dark areas indicate high values of K, and light areas indicate low values of K. Note the overall trend of increasing K from south to north, the three high-K zones 164 The MODFLOW reality model of groundwater flow generates the steady-state results shown in Figure 6-3a, given the long-term pumping conditions of Table 6-2. This hydraulic head surface represents the state of the aquifer at the beginning of the first management period (1 January, 1995). Overall flow is from south to north, fed by recharge at the water table and discharging at the springs and wells. The heterogeneous distribution of permeability produces numerous local irregularities. (For comparison, the steady-state hydraulic head surface generated by the Round 1 MODFLOW perception model is shown in Figure 6-3b; this is discussed in Section 6.2.1.2, below). 2 4 6 8 2 4 6 . 8 5 5 5 5 5 5 5 5 5 a. MODFLOW reality model b. Round 1 MODFLOW perception model Figure 6-3. Steady-state hydraulic head surface in the Kaldor Aquifer under long-term pumping. (a) The pollution model, generated by the M O D F L O W reality model. (b) The Round 1 M O D F L O W perception model. The contours represent hydraulic head in meters as of 1 January 1995, with a contour interval of 0.2 m. The dashed rectangle indicates the area of detail around the Midnight Site, as shown in Figure 6-4. 165 Well Number Easting Northing Date installed Extraction rate W l 3100 5100 1935 0.02 m3/s W2 3700 7100 1935 0.02 m3/s W3 6100 8100 1935 0.02 m3/s Table 6-2. Locations, installation dates, and extraction rates of water wells. a. M O D F L O W and M O D P A T H reality models b. Round 1 M O D F L O W and M O D P A T H perception models Figure 6-4. Steady-state hydraulic head surface and chloride plume in the vicinity of the Midnight Site, under long-term pumping. (a) The pollution model, generated by the M O D F L O W and M O D P A T H reality models. (b) The Round 1 M O D F L O W and M O D P A T H perception models. The parallel lines emanating from the Midnight Site represent the chloride plume, up to 1 January 1995. The dates mark the position of the head of the chloride plume as of 1 January of the year indicated; the dates marked beyond 1995 assume no change in the hydraulic head field. The contours represent hydraulic head in meters as of 1 January 1995, with a contour interval of 0.1 m. See Figure 6-3 for the location of the detailed area shown here. The dashed rectangle indicates the more detailed area shown in Figure 6-5. The area around the Midnight Site is shown in Figure 6-4. The MODPATH reality model generates the chloride plume shown in Figure 6-4a, assuming conservative advective transport 166 and the steady state hydraulic head surface under constant long-term pumping conditions. As of 1 January, 1995, the head of the Midnight chloride plume was 650 m from water well W2. (For comparison, the chloride plume generated by the Round 1 MODFLOW perception model is shown in Figure 6-4b; this is discussed in Section 6.2.1.2, below.) 6.2.1.2 Technical model The technical model for Round 1 consists of the Round 1 MODFLOW and MODPATH perception models; an AQMOD-MINOS simulation-optimization model based on the Round 1 MODFLOW perception model; a set of technical options based on these models; and cost estimates for these options. CON has limited data on hydraulic head and hydraulic conductivity (K) values from the vicinity of the water wells and the Midnight site, in the northern half of the Kaldor Aquifer. CON has almost no data for the southern half of the aquifer. The Round 1 MODFLOW perception model has been calibrated with a uniform hydraulic conductivity of 9 x 10" m/s (an order of magnitude greater than the K in the reality model) and a uniform recharge rate of 3 x 10"9 m/s (three times greater than the uniform recharge rate in the reality model). The output of the MODFLOW reality and perception models for steady-state flow are compared in Figure 6-3 and Figure 6-4. At the scale of the aquifer (Figure 6-3), the comparison of the perception model with the reality model appears reasonable for the northern half, but rather poor for the southern half. This reflects the fact that CON has almost no data for calibration from the southern half. At the more detailed scale (Figure 6-4), the errors in the perception model become more apparent. In particular, the zones of influence of the water wells are far too small in the perception model, especially for well W2. 167 The output of the MODPATH reality and perception models are compared in Figure 6-4. The perception model erroneously indicates that the chloride plume is heading towards the springs rather than well W2. This is a consequence of the small zone of influence for well W2 in the perception MODFLOW model. In addition, the perception model shows an exaggerated velocity for the plume, reflecting erroneously high K values along its path. CON has set up an AQMOD-MINOS simulation-optimization model, based on the Round 1 MODFLOW perception model. This has been used to estimate the minimum pumping rate needed for a containment well at the source, in order to prevent any further off-site transport of pollutants in groundwater. This approach could also be used to estimate an optimal pumping scheme to capture or contain the plume. However, this possibility has not been considered at this point, because the preliminary perception models have indicated that the plume is already very extensive, and that containment wells are likely to have very limited radii of influence. The technical options developed by CON for the Round 1 negotiations are provided in Table 6-3. Note that the numbering of the technical options begins with Option 4, to allow for three non-cooperative options in the decision model (see below). 168 Option Description Goal 4. Monitor northwest of the site Establish 4 monitoring wells ( M l 1 to M14) in a northeast-trending fence, about 300 m northwest of the site (see Figure 6-5). To detect possible movement of the plume to the northwest towards Well W2, rather than north towards the spring as suggested by the Round 1 M O D P A T H model. 5. Monitor south of the springs Establish 4 monitoring wells in an east-trending fence, about 200 m south of the line of springs along the northern boundary of the Kaldor Aquifer. To obtain early warning about possible discharge of pollutants to the springs. 6. Sample at springs Take samples directly from the springs. To determine whether the plume has already reached the springs. 7. Partial source removal Excavate the site (50 m x 25 m) to a depth of 7.6 m. The side walls will be held in place with steel sheeting. Depending on the level of pollution, the material will be either incinerated in an out-of-province facility, or dumped in an out-of-province special waste landfill. Mutually exclusive with Option 8. To remove the polluted near-surface material, and to reduce the source volume. (It is recognized that removal of the entire source volume, down to the base of the aquifer, is impossible. It follows that the source will continue to generate a plume even if this option is undertaken.) 8. Cap source Install a multi-layered cap consisting of a synthetic membrane sandwiched between various layers of clay, native soil, and drainage material. Mutually exclusive with Option 7. To isolate the polluted volume from the surface. 9. Shut down water well W2 VIC closes water well W2 and replaces the lost water supply (0.02 m^/s) by increasing pumping at wells W l and W3 by 0.01 m^/s each, to a total rate of 0.03 m^/s each. PRP to cover any increases in capital and operating costs. (Note: the monthly sampling at W2 will continue even if it is shut down, as it provides a useful monitoring point.) To remove the threat of drawing the plume towards W2, in case the plume is moving to the northwest rather than to the north as indicated by the Round 1 M O D P A T H model. 10. Containment well at source Install a containment well (designated CI) and a carbon treatment plant at the site. The well would be designed for a maximum capacity of 0.08 m- /^s, and the treatment plant would be designed for a maximum capacity of 0.04 m-Vs. During 1995, this system would be operated at 0.04 m-Vs. This is the minimum rate needed to obtain a measurable gradient towards the site on all sides, as estimated by the A Q M O D - M I N O S simulation-optimization model. The extra well capacity is to allow for a possible increase in the rate of pumping; it is a hedge against uncertainty. The treatment plant does not have extra capacity because, unlike the well, it can be designed to be modular and easily expanded. The treated water will be discharged to an ephemeral stream. To prevent pollutants from leaving the site via advective transport. 11. Revise perception models To revise the M O D F L O W and M O D P A T H perception models in light of information to be obtained in 1995. Includes a small budget for limited data acquisition (e.g., pump testing the water wells and the containment well). To obtain a better basis for prediction and design for future management. Table 6-3. Round 1 technical options. 4 3 0 0 E - . 1987 M5 M4J M3 M2 M1 © © MI-I e a a M12 © I M13 © I M14 ® I © - 1986 --plume as modelled by the Round 1 MODPATH perception model L 1 1985 Midnight Site © © © M10 M9 M8 © © M7 M6 0 50 100 150 200 metres Figure 6-5. Monitoring well locations in the vicinity of the Midnight site. See Figure 6-4 for the location of the detailed area shown in this figure. Monitoring wells M l to M10 are to be installed in 1995, and M l 1 to M14 are included in Option 4, proposed for 1995. The crosses indicate the position of the head of the plume on 1 January of the year indicated. Cost estimates for the eight technical options, derived mainly from CORA, are provided in Table 6-4. Obligatory costs such as the emergency measures (including monitoring wells Ml to M10) and CON's fees (exclusive of model revision) have not been considered. Capital costs are assumed to be incurred at the beginning of 1995. Each operating cost is presented as both an 169 5 0 0 0 E annual cost, and as a present value for the beginning of 1995, for periods of 5, 10, and 25 years, 170 assuming 10% interest rate, with costs incurred at the end of the year. The range of present values shown (for 5, 10, and 25 years) reflects the considerable uncertainty as to how long a particular option will have to be implemented. The present values are very rough approximations, because they assume that annual costs and the interest rate will remain constant over time. They are conservative in some respects, and non-conservative in others. For example, they are conservative in that they do not allow for reduced monitoring requirements over time, as the plume becomes well-documented. On the other hand, they are non-conservative in that they do not allow for increased maintenance of aging facilities. Option Capital cost OOOS Annual operating cost 000$ Present value of annual operating cost 000$ 5-year 10-year 25-year 4. Monitor northwest of the site 63 103 389 631 932 5. Monitor south of the springs 55 103 389 631 932 6. Sample at springs 103 389 631 932 7. Partial source removal 11,520 8. Cap source 693 23 86 139 206 9. Shut down water well W2 100 25 95 154 227 10. Containment well at source 1,219 880 3,336 5,408 7,988 11. Revise perception models 100 Table 6-4. Cost estimates for lound 1 technical options. Present values of annual operating costs are for the beginning of 1995, and are based on an interest rate of 10% with costs incurred at the end of each year. Note that the cost estimates are not used directly in the decision model, but rather as guidance in the subjective step of developing stakeholder preferences (see the following section). 6.2.1.3 Decision model (COOP) The decision model as described in this section uses the conflict analysis model COOP. The principles and methodology of conflict analysis are described in Section 5.4.1 and in 171 Appendix 2. Refer to Section 6.3 for a description of the decision model using the role-playing model PAQMAN. In negotiations for Round 1, the stakeholders are faced with the binary options shown in Table 6-5. These include the eight cooperative technical options provided by the technical model (Section 6.2.1.2), and three non-cooperative BATNA options (one for each stakeholder). These options can be combined to produce 2048 mathematically-possible outcomes. The feasibility analysis subroutine in COOP reduces this to 195 feasible outcomes. Four of these feasible outcomes are non-cooperative, and 191 are cooperative. Non-cooperative options: REG PRP VIC 1. Regulate unilaterally 2. Stonewall against regulation 3. Launch lawsuit against PRP Cooperative options: 4. Set up fence of monitoring wells northwest of site 5. Set up fence of monitoring wells south of springs 6. Sample at springs 7. Partial source removal 8. Cap source 9. Shut down water well W2 10. Containment well at source (Cl) 11. Revise perception models Table 6-5. The stakeholders' options in the Round 1 negotiations. Two variations on the stakeholders' preferences are considered here. In Preference Variation 1, REG has a very high preference for reaching a negotiated settlement. In Preference Variation 2, both REG and PRP have a high preference for breaking off talks and acting unilaterally (i.e., they both have a high preference for their non-cooperative BATNA options). The stakeholders' preferences are discussed here in terms of preference masks rather than preference vectors, for two reasons. First, the preference masks are key input developed through a critical subjective step in SAM, and should be documented. Second, the preference vectors are 172 too large to display or discuss conveniently (for Round 1, the full listing of preference vectors for all stakeholders would require three 11 x 195 matrices). 6.2.1.3.1 Preference Variation 1 The preference masks for Preference Variation 1 are shown in Table 6-6. The preferences for BATNAs are very important. Note that PRP and VIC are prepared to exercise their BATNAs conditionally: PRP will stonewall if REG imposes a unilateral solution, and VIC will launch a lawsuit against PRP if it stonewalls. REG, on the other hand, assigns its highest preference to avoiding its BATNA (unilateral regulation), because it does not wish to initiate this cycle of stonewalling and litigation. This gives it a relatively weak bargaining position. The outcomes arising from the Round 1 negotiations under Preference Variation 1, as determined by stability analysis, are shown in Table 6-7, Table 6-8, and Table 6-9. Table 6-7 indicates that there are no Pareto-optimal non-cooperative equilibriums. This implies that the stakeholders will not exercise their BATNAs, there will be no breakdown of negotiations, and there will be a negotiated agreement. This reflects REG's very high preference for reaching agreement, and the fact that the other stakeholders' BATNAs were conditional on REG exercising its BATNA. There are 24 Pareto-optimal cooperative equilibriums, representing 24 possible agreements; 167 feasible cooperative outcomes can be rejected as unstable or Pareto-suboptimal. 173 Preference Masks for Round 1, Preference Variation 1 Mask weight REG's preference masks R E G rejects option 1: R E G doesn't regulate unilaterally (i.e., talks don't break down) 1024 PRP rejects option 2: PRP doesn't stonewall (i.e., talks don't break down) 512 VIC rejects option 3: VIC doesn't litigate against PRP (i.e., talks don't break down) 256 All agree to accept option 10: Containment well at source 128 All agree to accept option 8: Cap source 64 A l l agree to accept option 11: Revise model 32 All agree to accept option 4: Monitor northwest of site 16 A l l agree to accept option 6: Sample springs 8 All agree to accept option 5: Monitor south of springs 4 A l l agree to accept option 9: Shut down water well W2 2 A l l agree to accept option 7: Partial source removal 1 PRP's preference masks A l l agree to reject option 7: No to partial source removal 2048 VIC rejects option 3: VIC doesn't litigate (i.e., talks don't break down) 1012 R E G rejects option 1: R E G doesn't regulate (i.e., talks don't break down) 512 PRP accepts option 2 IF R E G accepts option 1: Stonewall IF R E G does regulate (BATNA) 256 All agree to option 9: Shut down water well W2 128 A l l agree to option 4 IF all reject option 9: Monitor northwest of site IF W2 not shut down 64 A l l agree to option 10 IF all reject option 7: Containment well IF no partial source removal 32 All reject option 10 IF all agree to option 7: No containment well IF partial source removal 16 All reject option 7: No to source capping 8 All agree to option 11: Revise perception models 4 A l l agree to option 6: Sample at springs 2 All reject option 5: Not have to monitor south of springs 1 VIC's preference masks A l l agree to option 10: Containment well at source 256 All agree to option 7: Partial source removal 128 VIC accepts option 3 IF PRP accepts option 2: Litigate IF PRP stonewalls (BATNA) 64 A l l agree to option 4: Monitor northwest of site 32 A l l reject option 9: Not shut down W2 16 All accept option 11: Revise perception models 8 A l l accept option 8: Cap source 4 A l l accept option 5: Monitor south of springs ....equally preferred with: 2 All accept option 6: Sample springs 2 Table 6-6. Preference masks for each stakeholder, for Preference Variation 1 in the Round 1 negotiations. 174 Category of outcome Number of outcomes in category Pareto-optimal non-cooperative equilibrium (breakdown of negotiations) 0 Pareto-optimal cooperative equilibrium (agreement) 24 Pareto-suboptimal non-cooperative equilibrium 4 Pareto-suboptimal cooperative equilibrium 87 Unstable non-cooperative outcome 0 Unstable cooperative outcome 80 Total number of outcomes 195 Table 6-7. Number of outcomes in each category for Preference Variation 1 in the Round 1 negotiations. Pareto-optimal equilibriums (possible outcomes) Options 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 I.RI-G: Regulate 2.PRP: Stonewall 3.VIC: Sue PRP 4. Monitor northwest of site 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 5. Monitor south of springs 1 1 1 1 1 1 1 1 1 1 1 1 6. Sample springs 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 7. Partial source removal 8. Cap source 1 1 1 1 1 1 1 1 1 1 1 1 9. Shut down W2 1 1 1 1 1 1 1 1 10. Containment well at source 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 11. Revise perception models 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 Table 6-8. The set of Pareto-optimal equilibriums (possible outcomes) for Preference Variation 1 in the Round 1 negotiations. A "1" indicates an option which is accepted, and a blank space indicates an option which is rejected. The non-cooperative options are shaded. Technical option Frequency of acceptance (out of 24) l . R E G : Regulate unilaterally 0 2.PRP: Stonewall 0 3. VIC: Sue PRP 0 4. Monitor northwest of site 20 5. Monitor south of springs 12 6. Sample at springs 16 7. Partial source removal 0 8. Cap source 12 9. Shut down water well W2 8 lO.Containment well at source 24 11 .Revise perception models 16 Table 6-9. Frequency of technical option acceptance for the 24 Pareto-optimal cooperative equilibriums (possible agreements) for Preference Variation 1 in the Round 1 negotiations. This is a summary of the results in Table 6-8. 175 The distribution of accepted options in the 24 possible agreements is shown in full in Table 6-8, and summarized in Table 6-9. Note that Option 10, for a containment well at the source, is accepted in all possible agreements. Options 4, 6, and 11 were accepted in more than half of the possible agreements. Option 7, for partial source removal, is rejected in all possible agreements. This distribution of option acceptance is generally consistent with REG's preferences. It follows that REG achieved most of its intentions in spite of its weak BATNA. This was largely due to the interplay of opposing preferences between PRP and VIC. Any choice among these Pareto-optimal cooperative equilibriums would require either an exercise of subjective judgement or a random choice. The random choice method is used below, in Section 6.2.2, as a basis for proceeding to Round 2. The consequences of REG and PRP having higher preferences for their BATNAs are explored in the following section. 6.2.1.3.2 Preference Variation 2 In Preference Variation 1, REG was given a very low preference for any non-cooperative outcome. Preference Variation 2 considers the consequences if REG and PRP were to take a very strong negotiating stance, based on the same options as Preference Variation 1. The revised preference masks are shown in Table 6-10. REG's BATNA (to regulate unilaterally) is now its second preference mask, preceded only by its preference for agreement on a containment well. However, this strong position is moderated by its third, fourth, and fifth preference masks, which indicate a desire that none of the stakeholders exercise their BATNA (i.e., that talks do not break down). PRP also has a stronger stance relative to Preference Variation 1: its BATNA (stonewalling) has moved up to second place in its list of preference masks, and is now unconditional. 176 Preference Masks for R o u n d 1, Preference Variation 2 Mask weight R E G ' s preference masks All agree to accept option 10: Containment well at source 2048 R E G accepts option 1: R E G regulates unilaterally (talks break down) (BATNA) 1024 R E G rejects option 1: Not have to regulate unilaterally (i.e., talks don't break down) 512 PRP rejects option 2: PRP doesn't stonewall (i.e., talks don't break down) 256 VIC rejects option 3: VIC doesn't litigate against PRP (i.e., talks don't break down) 128 All agree to accept option 8: Cap source 64 All agree to accept option 11: Revise perception models 32 All agree to accept option 4: Monitor northwest of site 16 All agree to accept option 6: Sample springs 8 A l l agree to accept option 5: Monitor south of springs 4 A l l agree to accept option 9: Shut down water well W2 2 All agree to accept option 7: Partial source removal 1 PRP's preference masks All agree to reject option 7: No to partial source removal 4096 PRP accepts option 2 : Stonewall (talks break down) (BATNA) 2048 VIC rejects option 3: VIC doesn't litigate (i.e., talks don't break down) 1012 R E G rejects option 1: R E G doesn't regulate (i.e., talks don't break down) 512 IF R E G accepts option 1: Stonewall IF R E G does regulate 256 All agree to option 9: Shut down water well W2 128 A l l agree to option 4 IF all reject option 9: Monitor northwest of site IF W2 not shut down 64 All agree to option 10 IF all reject option 7: Containment well IF no partial source removal 32 A l l reject option 10 IF all agree to option 7: No containment well IF partial source removal 16 All reject option 7: No to source capping 8 All agree to option 11: Revise perception models 4 All agree to option 6: Sample at springs 2 All reject option 5: Not have to monitor south of springs 1 VIC's preference masks All agree to option 10: Containment well at source 256 All agree to option 7: Partial source removal 128 VIC accepts option 3 IF PRP accepts option 2: Litigate IF PRP stonewalls (BATNA) 64 A l l agree to option 4: Monitor northwest of site 32 A l l reject option 9: Not shut down W2 16 All accept option 11: Revise perception models 8 All accept option 8: Cap source 4 All accept option 5: Monitor south of springs ....equally preferred with: 2 All accept option 6: Sample springs 2 Table 6-10. Preference masks for each stakeholder, for Preference Variation 2 in the Round 1 negotiations. Compare with Table 6-6. The outcomes arising from the Round 1 negotiations under Preference Variation 2, as determined by stability analysis, are shown in Table 6-11, Table 6-12, and Table 6-13. 177 CATEGORY of OUTCOME NUMBER of OUTCOMES Preference Variation 1 Preference Variation 2 Pareto-optimal non-cooperative equilibrium 0 2 Pareto-optimal cooperative equilibrium 24 24 Unstable non-cooperative outcome 0 0 Unstable cooperative outcome 80 127 Pareto-suboptimal non-cooperative equilibrium 4 2 Pareto-suboptimal cooperative equilibrium 87 40 Total number of outcomes 195 195 Table 6-11. Number of outcomes in each category, for Preference Variations 1 and 2 in the Round 1 negotiations. Pareto-optimal equilibriums (possible outcomes) Options 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 l .REG: Regulate i l l 1 2.PRP: Stonewall 11 111 3.VIC: Sue PRP i l 4. Monitor W of site 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 5. Monitor south of springs 1 1 1 1 1 1 1 1 1 1 1 1 6. Sample springs 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 7. Partial source removal 8. Cap source 1 1 1 1 1 1 1 1 1 1 1 1 9. Shut down W2 1 1 1 1 1 1 1 1 10. Containment well at source 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 11. Revise perception models 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 Table 6-12. The set of Pareto-optimal equilibriums (possible outcomes) for Preference Variation 2 in the Round 1 negotiations. "1" indicates an option which is accepted, and a blank space indicates an option which is rejected. The non-cooperative options are shaded. Compare with Preference Variation 1 (Table 6-8). Option Frequency of acceptance (out of 24) Preference Variation 1 Preference Variation 2 REG—Regulate unilaterally 0 2 PRP—Stonewall 0 2 VIC—Litigate against PRP 0 1 Monitor northwest of site 20 20 Monitor south of springs 12 12 Sample at springs 16 16 Partial source removal 0 0 Cap source 12 12 Shut down water well W2 8 8 Containment well at source 24 24 Revise perception models 16 16 Table 6-13. Frequency of option acceptance in Pareto-optimal equilibriums, for Preference Variations 1 and 2 in the Round 1 negotiations. This is a summary of the results in Table 6-12. 178 Table 6-11 indicates that Preference Variation 2 introduces the possibility of a breakdown in negotiations, as shown by the two Pareto-optimal non-cooperative equilibriums. As shown in Table 6-12, both of these involve REG imposing regulation unilaterally, and PRP stonewalling. In one of the cases, VIC launches a lawsuit against PRP. The set of Pareto-optimal cooperative equilibriums remains the same as with Preference Variation 1, suggesting that this is a relatively stable set of possible agreements. As before, note that any choice among these Pareto-optimal cooperative equilibriums would require either an exercise of subjective judgement or a random choice. 6.2.2 Round 2 Round 2 covers the development of technical tasks and the negotiation among stakeholders for the second management period (1996). The set of 24 possible Pareto-optimal equilibriums from Preference Variation 1 in Round 1 (Table 6-8) has been taken as the starting point. As these are all Pareto-optimal cooperative equilibriums, they all represent possible agreements among the stakeholders. There is no objective, algorithmic means to select any one of them in preference to the others. An exercise of subjective judgement or a random choice is required. In this case, the random choice method is used. The randomly-selected agreement is Pareto-optimal Equilibrium 20 from Table 6-8. The details of this agreement are shown in Table 6-14. 179 Option Decision 4. Monitor northwest of site Accepted 5. Monitor south of springs Accepted 6. Sample at springs Accepted 7. Partial source removal Rejected 8. Cap source Rejected 9. Shut down water well W2 Accepted 10. Containment well at source Accepted 11. Revise perception models Accepted Table 6-14. The randomly-selected agreement from Round 1, as a basis for proceeding to Round 2. The randomly-selected agreement is Pareto-optimal Equilibrium 20 from Table 6-8. 6.2.2.1 Pollution model Figure 6-6b shows the consequences of implementing the decisions of Pareto-optimal Equilibrium 20 on the reality models. Water well W2 has been shut down, and its drawdown cone has disappeared. There is a strong drawdown cone around Cl, the new containment well at the source. This is sufficient to halt the advance of the plume, and draw it back towards Cl. Figure 6-6. (overleaf) Comparison of reality and perception models for Rounds 1 and 2, in the vicinity of the Midnight Site. (a) The Round 1 M O D F L O W and M O D P A T H reality models. Reproduced from Figure 6-4a. Represents hydraulic head field as of 1 January, 1995. (b) The Round 2 M O D F L O W and M O D P A T H reality models. Shows the effect of the Equilibrium 20 decisions: shutting down water well W2 and running containment well C l for one year, beginning with initial conditions as shown in (a). Represents hydraulic head field as of 1 January, 1996. (c) The Round 1 M O D F L O W and M O D P A T H perception models. Reproduced from Figure 6-4b. Represents hydraulic head field as of 1 January, 1995. (d) The Round 2 M O D F L O W and M O D P A T H perception models. Represents hydraulic head field as of 1 January, 1995. Shows the effect of the Equilibrium 20 decisions: shutting down water well W2 and running containment well C l for one year, beginning with initial conditions as shown in (c). Represents hydraulic head field as of 1 January, 1996. The contours represent hydraulic head in meters, with a contour interval of 0.1 m. The parallel lines emanating from the Midnight Site represent the chloride plume. The dates mark the position of the head of the chloride plume as of 1 January of the year indicated. See Figure 6-3 for the location of the area shown here. The dashed rectangle indicates the detailed area shown in Figure 6-5. 180 a. The Round 1 reality models. b. The Round 2 reality models. c. The Round 1 perception models. d. The Round 2 perception models. 181 6.2.2.2 Technical model Pump testing of the containment well and the water wells has given a better picture of hydraulic conductivity, which is generally lower than assumed for Round 1. The plume has been detected in monitoring well M7, in the monitoring fence established northwest of the site (Figure 6-5). On the basis of this information, CON now believes that the plume is traveling northwest from the site, rather than north. The new perception of lower hydraulic conductivities implies that plume velocity is also lower than previously believed. On the basis of the new information, the Round 1 perception models have been updated to produce the Round 2 perception models (Figure 6-6d). Inspection of Figure 6-6, which compares the reality and perception models for both rounds, demonstrates that the Round 2 perception models are much closer to the reality models, particularly with respect to the location of the plume and the radii of influence of pumping wells. The AQMOD-MLNOS simulation-optimization model has also been updated. In Round 1, hydraulic containment was considered only as a means of preventing off-site transport of pollutants. In Round 2, hydraulic containment of the plume is now considered as an option, given lower plume velocity and greater radii of influence for containment wells. In particular, the simulation-optimization model now considers a second containment well (C2), to be installed at 4500E/6100N, 283 m northwest of CI. The technical options developed by CON for the Round 2 negotiations are provided in Table 6-15. Again, note that the numbering of the technical options begins with Option 4, to allow for three non-cooperative options in the decision model (see below). 182 The AQMOD-MINOS simulation-optimization is applied to two different levels of protection, providing a pair of optimized options for the stakeholders to consider. This is a simplistic example of a soft paradigm variation on simulation-optimization, using it to develop a "negotiation set" of options (see Chapter 2). Cost estimates for these options, derived mainly from CORA, are provided in Table 6-16. Obligatory costs such as CON's fees (exclusive of model revision) have not been considered. Capital costs are assumed to be incurred at the beginning of 1996. Each operating cost is presented as both an annual cost, and as a present value for the beginning of 1996, for periods of 5, 10, and 25 years, assuming 10% interest rate, with costs incurred at the end of the year. As noted with regard to Table 6-4, present values are rough approximations, and are shown for three different periods because there is considerable uncertainty as to how long a particular option will have to be implemented. 183 Option Description Goal 4. Weak containment. Install well containment well C2, 283 m northwest of C l , with a maximum capacity of 0.08 m3/s. Set pumping rates for C l at 0.036 m3/s, and for C2 at 0.025 m3/s, for a combined rate of 0.061 m3/s. Expand capacity of treatment plant (built in 1995) by an extra 0.021 m3/s to handle the increased load. The treated water will be discharged to an ephemeral stream. • Mutually exclusive with Option 5. Containment of entire plume with a minimum inward head gradient of 0.0005. The pumping regime provides the minimum total pumping rate needed to obtain the minimum inward gradient towards the entire plume, as estimated by the A Q M O D - M I N O S simulation-optimization model. 5. Strong containment. Install well containment well C2, 283 m northwest of C l , with a maximum capacity of 0.08 m3/s. Set pumping rates for C l at 0.066 m3/s, and for C2 at 0.046 m3/s, for a combined rate of 0.112 m3/s. Expand capacity of treatment plant (built in 1995) by an extra 0.072 m3/s to handle the increased load. The treated water will be discharged to an ephemeral stream. • Mutually exclusive with Option 4. (Note that the default, if neither Option 4 nor Option 5 is selected, is to continue with the 1995 containment system, with C l pumping at 0.040 m3/s on its own.) Containment of entire plume with a minimum inward head gradient of 0.001. The pumping regime provides the minimum total pumping rate needed to obtain the minimum inward gradient towards the entire plume, as estimated by the A Q M O D - M I N O S simulation-optimization model. 6. Restart water well W2. Begin pumping W2 at 0.02 m3/s; reduce W l and W3 to 0.02 m3/s (i.e., restore the water well pumping regime to the status quo which existed prior to 1995). To restore W2 as a source of water, given that the revised estimate of plume velocity suggests that it is not in immediate danger of pollution. 7. Cap source. Install a multi-layered cap consisting of a synthetic membrane sandwiched between various layers of clay, native soil, and drainage material. (Identical to the Option 8 which was rejected in Round 1). To isolate the polluted volume from the surface. 8. Reduce sampling at springs and monitoring wells south of springs. Reduce the rate of sampling from monthly (as in Options 4 and 5 from Round 1) to quarterly. To reduce coasts. Justified by negative results from the 1995 sampling, and by the revised estimate of the plume's velocity and direction. 9. Further revision of perception models. To revise the M O D F L O W and M O D P A T H perception models in light of information which will be obtained during 1996. Includes a small budget for limited data acquisition. To obtain a better basis for prediction and design for future management. 10. Low-cost extension to monitoring network. 4 additional monitoring wells and 4 piezometers. • Mutually exclusive with Option 11. To refine knowledge of the plume location and verify the efficacy of containment. 11. High-cost extension to monitoring network. 8 additional monitoring wells and 8 piezometers. • Mutually exclusive with Option 10. To refine knowledge of the plume location and verify the efficacy of containment. Table 6-15. Round 2 technical options. 184 Capital Annual Present value of annual cost operating operating cost cost 000$ Option 000$ 000$ 5-year 10-year 25-year 4. Weak containment 796 359 1360 2204 3256 5. Strong containment 2,236 1,301 4933 7996 11812 6. Restart water well W2 -25 -95 -154 -227 7. Cap source 693 23 86 139 206 8. Reduce sampling -103 -389 -631 -932 9. Further revision of perception models 100 10. Low-cost monitoring extension 912 103 389 631 932 11. High-cost monitoring extension 1,824 205 778 1262 1864 Table 6-16. Cost estimates for Round 2 technical options. Present values of annual operating costs are for the beginning of 1996, and are based on an interest rate of 10% with costs incurred at the end of each year. A negative cost indicates a cost reduction. 6.2.2.3 Decision model (COOP) The decision model as described in this section uses the conflict analysis model COOP. The decision model using the role-playing model PAQMAN is described in Section 6.3. Non-cooperative options: REG PRP VIC 1. Regulate unilaterally 2. Stonewall against regulation 3. Launch lawsuit against PRP Cooperative options: 4. Weak containment 5. Strong containment 6. Restart water well W2 7. Cap source 8. Reduce sampling 9. Further revision of perception models 10. Low-cost monitoring extension 11. High-cost monitoring extension Table 6-17. The stakeholders' options in the Round 2 negotiations. In negotiations for Round 2, the stakeholders are faced with the binary options shown in Table 6-17. As in Round 1, these include the eight cooperative technical options provided by the technical model (Section 6.2.2.2), and three non-cooperative BATNA options (one for each stakeholder). These options can be combined to produce 2048 mathematically-possible 185 outcomes. The feasibility analysis subroutine in COOP reduces this to 147 feasible outcomes. Four of these feasible outcomes are non-cooperative, and 143 are cooperative. Only one preference variation is considered for Round 2; the preference masks are shown in Table 6-18. As in Preference Variation 1 of Round 1, REG has a high preference for reaching a negotiated settlement, expressed as a high preference for avoiding the use of BATNAs. The other stakeholders have high preferences for exercising their BATNAs, but these are conditional on REG exercising its BATNA. The outcomes arising from the Round 2 negotiations, as determined by stability analysis, are shown in Table 6-19, Table 6-20, and Table 6-21. Table 6-19 indicates that there are no Pareto-optimal non-cooperative equilibriums (i.e., no failures to reach agreement), reflecting REG's high preference for avoiding BATNAs. There are 57 Pareto-optimal cooperative equilibriums, representing 57 possible agreements; 86 feasible cooperative outcomes can be rejected as unstable or Pareto-suboptimal. The distribution of technical options in the 57 possible agreements is shown in Table 6-20, and summarized in Table 6-21. None of the Pareto-optimal equilibriums include Option 5 (strong containment), reflecting PRP's strong preference against this very expensive option. Similarly, very few of the Pareto-optimal equilibriums include Option 11 (high-cost extension to the monitoring network). Option 9 (further revision of the perception models) is relatively favoured: it is accepted in about two-thirds of the possible agreements. The other options are accepted in roughly half of the agreements. As before, note that any choice among these Pareto-optimal cooperative equilibriums would require either an exercise of subjective judgement or a random choice. 186 Preference Masks for Round 2 Mask weight REG's preference masks A l l agree to Option 4: Weak containment 1024 All agree to Option 7: Cap source 512 All agree to Option 10: Low-cost extension to monitoring network 256 R E G rejects Option 1: Not have to regulate unilaterally (i.e., talks don't break down) 128 PRP rejects Option 2: PRP doesn't stonewall (i.e., talks don't break down) 64 VIC rejects Option 3: VIC doesn't litigate against PRP (i.e., talks don't break down) 32 All agree to Option 9: Further refinement of model 16 All agree to Option 5: Strong containment 8 A l l agree to Option 11: High-cost extension to monitoring network 4 All reject Option 6: Water well W2 N O T restarted 2 All agree to Option 8: Reduce sampling at springs and wells south of springs 1 PRP's preference masks All reject Option 5: N O to strong containment. 1024 R E G rejects option 1: R E G doesn't regulate (i.e., talks don't break down) 512 VIC rejects Option 3: VIC doesn't litigate against PRP (i.e., talks don't break down) 256 PRP accepts option 2 IF R E G accepts option 1: Stonewall IF R E G does regulate (BATNA) 128 A l l agree to Option 10: Low-cost extension to monitoring network... equally preferred with: 64 A l l reject Option 11: N O high-cost extension to monitoring network... equally preferred with: 64 A l l agree to Option 9: Further refinement of model 64 All agree to Option 8: Reduce sampling at springs and wells south of springs 8 All reject Option 6: Water well W2 N O T restarted....equally preferred with: 4 All agree to Option 7: Cap source 4 A l l agree to Option 4: Weak containment 1 VIC's preference masks A l l agree to Option 4: Weak containment 256 A l l agree to Option 10: Low-cost extension to monitoring network 128 VIC accepts option 3 IF PRP accepts option 2: Litigate IF PRP stonewalls (BATNA) 64 A l l agree to Option 5: Strong containment... equally preferred with: 32 A l l agree to Option 11: High-cost extension to monitoring network.... equally preferred with: 32 All agree to Option 9: Further refinement of model 32 A l l agree to Option 6: Water well W2 restarted 4 All agree to Option 7: Cap source ....equally preferred with: 2 A l l agree to Option 8: Reduce sampling at springs and wells south of springs 2 Table 6-18. Preference masks for each stakeholder, for the Round 2 negotiations. Category of outcome Number of outcomes in category Pareto-optimal non-cooperative equilibrium (possible breakdown of negotiations) 0 Pareto-optimal cooperative equilibrium (possible agreement) 57 Unstable non-cooperative outcome 0 Unstable cooperative outcome 48 Pareto-suboptimal non-cooperative equilibrium 4 Pareto-suboptimal cooperative equilibrium 38 Total number of outcomes 147 Table 6-19. Number of outcomes in each category in the Round 2 negotiations. 187 Pareto-optimal equilibriums (possible outcomes) Options 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 I.REG: Regulation 2.PRP: Stonewall 3.VIC: Sue PRP 4. Weak containment 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 5. Strong containment 6. Restart W2 1 1 1 1 1 1 1 1 1 1 1 1 1 7. Cap source 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 8. Reduce spring sampling 1 1 1 1 1 1 1 1 1 1 1 1 9. Revise models 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1. Low-cost extension... 1 1 1 1 1 1 1 1 11. High-cost extension... Pareto-optimal equilibriums (possible outcomes) Options 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 I.REG: Regulation 2.PRP: Stonewall 3.VIC: Sue PRP 4. Weak containment 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 5. Strong containment 6. Restart W2 1 1 1 1 1 1 1 1 1 1 1 1 1 7. Cap source 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 8. Reduce spring sampling 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 9. Revise models 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 10. Low-cost extension... 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 11. High-cost extension... 1 1 1 1 1 Table 6-20. The set of Pareto-optimal equilibriums (possible outcomes) in the Round 2 negotiations. "1" indicates an option which is accepted, and a blank space indicates an option which is rejected. The non-cooperative options are shaded. Technical option Frequency of acceptance (out of 57) 4. Weak containment 31 5.Strong containment 0 6.Restarting W2 26 7.Cap source 31 8.Reduce sampling at springs and wells south of springs 32 9.Further revision of perception models 37 lO.Low-cost extension to monitoring network 32 11.High-cost extension to monitoring network 5 Table 6-21. Frequency of technical option acceptance in Pareto-optimal cooperative equilibriums (possible agreements) in the Round 2 negotiations. This is a summary of the results in Table 6-20. 188 6.3 S A M u s i n g P A Q M A N as t h e d e c i s i o n m o d e l This section describes a run of SAM using PAQMAN. The PAQMAN role-playing exercise was run on 2 May, 1994. The participants were three graduate students from the Department of Geological Sciences at the University of British Columbia. The run took 7 hours to complete. The documentation for this run consists of the game manager's manual; general instructions, provided to all participants; and confidential instructions, specific to individual participants. This documentation is provided in a series of boxes within this section (Boxes 6-1 to 6-9). The general information for the run is provided in Boxes 6-1 and 6-2. Box 6-1. Game Manager's Manual PAQMAN Version 1.0 Materials The players are provided with a set of general instructions (common to all players), and with a set of confidential stakeholder instructions (unique for each player). The general instructions include an overview of S A M , a description of the specific scenario, and instructions on how to play. The confidential instructions contain information about the stakeholder's values, interests, and preferences. The manager has the 486 computer system complete with all the S A M software, plus two funny hats: one for the technical consultant, one for the mediator (if necessary). Duties of the manager The manager has a triple task: he/she must manage the game, play the role of technical consultant, and play the role of procedural mediator (if necessary). The manager must organize the game, keep it flowing, answer questions, prepare new options, determine the consequences of agreements, and debrief the players. Flow of the game The game begins with the assumption that the initial negotiations necessary to begin with Round 1 have already taken place, and that technical options for Round 1 have been prepared. The players will be asked to keep legible journals describing in a free-form fashion their impressions about the process, their reasons for taking certain positions, etc. These journals can be written up during the breaks. They will be carried over into the debriefing process. The breaks can be also be used for get fresh air or coffee, read the paper, etc. Briefing Prior to beginning the game, a briefing session will take place to sort out any questions, inconsistencies, etc. Journal-keeping should be emphasized at this point. 189 Round 1 In Round 1, the players attempt to sort out the various options into an agreement. If they fail in this, then they may suggest new options. They will then take a break while the game manager (as consultant) evaluates these new options using the S A M software. This may be a fairly lengthy process (up to 1 hour), so the players may be advised to take a break. They will then attempt to reach a new agreement with consideration of these new options. If they are successful in reaching agreement, then they take a break while the game manager (as consultant) determines the consequences of this agreement, and develops the Round 2 options. If they cannot reach agreement, even with these new options, then they must fall back upon their B A T N A s , and the game terminates. Development of Round 2 options While the players take a break, the game manager (as consultant) must determine the consequences of this agreement, and develop a set of appropriate Round 2 options. He/she must also develop additional confidential instructions. This break may take up to 1 hour. Round 2 Round 2 will proceed in the same fashion as Round 1. Debriefing A debriefing session, with the aid of the written journals, will be held at the end of the game. Topics can include why certain players adopted certain positions, why certain agreements were reached (or not reached), and so on. Confidential instructions can be revealed to the other players, if they choose. The reality model will be uncovered for inspection, including where the plume was really going. Box 6-2. General Instructions to Players PAQMAN Version 1.0 [table and figures follow text] 1. Introduction P A Q M A N (Polluted Aquifer Management) is a role-playing game developed for the purpose of researching the use of negotiation in the management of groundwater pollution. It is a major component of a larger simulation called S A M (Simulated Aquifer Management), which attempts to represent both the natural and human systems in a hypothetical groundwater pollution problem. S A M is the basis for my PhD research, which has the goal of developing and advocating a policy proposal for the extended use of negotiation in groundwater pollution management, with the broader aim of increasing efficiency, equity and response to uncertainty. This document will briefly review the conceptual model and components of S A M , in order to provide an understanding of how P A Q M A N fits into the broader picture. This will be followed by a detailed description of the specific scenario to be used in this version of P A Q M A N . The final section explains how to play P A Q M A N . A glossary of technical terms is appended. Other information will be provided in a set of confidential instructions to each player. This is the first version of P A Q M A N , and is used for debugging the game. Any suggestions for improvements are welcome. 2. Conceptual model of SAM The underlying conceptual model of S A M summarizes the conception of how negotiated management of groundwater pollution would operate in the real world. It has the following essential components (see Figure 1): 1. A groundwater pollution problem. The groundwater pollution problem involves a contaminated site and a plume of groundwater pollution emanating from that site. 190 2. Negotiated management of this problem. The stakeholders in a typical problem include: a government agency charged with regulating groundwater pollution; a party who is legally responsible for the pollution; and a victim or potential victim of the pollution. In the negotiation, these stakeholders attempt to achieve a consensual agreement on management of the problem over a specified period of time. Each stakeholder is constrained by a set of interests and values, and has one or more unilateral options to abandon the negotiation and pursue alternatives such as litigation. 3. The provision of neutral technical consulting to the negotiation process. The stakeholders jointly retain a technical consultant to advise the negotiation process. "Technical" is used in a broad sense here, and can be considered to be interchangeable with "scientific" or "engineering". 4. The importance of uncertainty in the management of groundwater pollution problems. Uncertainty regarding technical aspects of the problem (such as the location of the pollution plume) is a critical factor in the management of groundwater pollution. 5. The management process is iterative, over a period of years. The process consists of a series of iterative management periods. Each management period requires developing technical options, negotiating an agreement, implementing that agreement, and observing the consequences. With each iteration, uncertainty is reduced and the technical aspects of the problem are better understood. 3. Components of SAM The components of S A M , matched to the components of the conceptual model, are listed below, illustrated in Figure 2, and described in greater detail in subsequent sections. 1. Simulation of the groundwater pollution problem. The groundwater pollution problem is simulated with computer-based modeling (the "reality model"). This simulation is done with the computer programs M O D F L O W (groundwater flow) and M O D P A T H (pollutant tracking). You won't see the reality model until the game is over. 2. Simulations of negotiated management. The stakeholders and their negotiations are represented by complementary simulations: a computer-based approach (COOP) and gaming with human volunteers (PAQMAN) 3. Simulation of neutral technical consulting. The simulated consultant can only observe very limited aspects of the reality model. On the basis of these observations, she/he develops a M O D F L O W / M O D P A T H "perception model" which is essentially a simplified and erroneous version of the reality model. The consultant also makes use of the cost estimation program C O R A , and the optimization software A Q M O D - M I N O S . 4. Representation of uncertainty. Uncertainty is represented by the difference between the reality model and the consultant's perception model. 5. Iterative cycles. S A M cycles through a number of iterative management periods (Figure 3). Each management period contains the following steps, in sequence: a) developing technical options through the simulation of technical consulting; b) reaching an agreement through simulated negotiation; c) simulating the implementation of that agreement with the reality model; d) observing the consequences (in the output of the reality model). 4..The scenario A map of the area is provided in Figure 4. This map also illustrates the coordinate system used in this description. Locations are given in meters east and meters north of the bottom left-hand corner of this map. 4.1. The problem From 1985 to 1994, a variety of liquid hazardous wastes were illegally dumped at a site owned by the Midnight Disposal Corporation (4700E,6300N). The wastes, which included DNAPLs, were dumped into a lagoon and allowed to percolate into the ground. The lagoon is in direct contact with the Kaldor aquifer, an unconfined aquifer used for water supply. The owner of the Midnight Disposal Corporation is Mr. Skip. Since the discovery of the site in June of 1994, Mr. Skip has been on an extended vacation in a Caribbean nation noted for the strictness of its banking secrecy laws. The provincial authorities are anxious to speak with him, but he has not returned their calls. It has not been possible to assign responsibility to any other private parties, due to a paucity of written records and the lack of cooperation on the part of Mr. Skip The Midnight site has therefore been declared an orphan site, and responsibility for remediation has passed to the provincial government. 191 4.2. The stakeholders The provincial agency charged with the task of remediating all orphan sites, including the Midnight site, is the Pollution Response Panel (PRP, pronounced "perp"). The PRP is now legally responsible for all costs associated with the remediation of the Midnight site. The provincial agency charged with regulating the remediation of polluted sites is the Remediation Enforcement Group (REG). Under the Groundwater and Soil Pollution Act (GASPA), R E G has the authority to unilaterally enforce remediation actions which it considers appropriate, and this authority is backed by litigation powers. R E G also has the authority to obtain remediation through negotiation, and its ability to negotiate with the various stakeholders is not tightly constrained by GASPA. The Village of Canterbury (VIC) operates a number of water wells within the Kaldor aquifer. These are Wells W l , Well W2 , and W3. Well W2 is 1000 m east and 800 m E of the Midnight site, and is at risk from the Midnight plume. Well Number Easting Northing W l 3100 5100 W2 3700 7100 W3 6100 8100 4.3 Emergency measures PRP has undertaken emergency measures at the Midnight site. The site has been fenced off. Al l liquids and sludges have been removed from the lagoon, and shipped to a special waste incinerator in another province. A number of holes have been drilled on the site in order to estimate the degree of contamination. The results are discouraging: they suggest that some stringers of D N A P L have sunk to the base of the aquifer, 36 m below the surface. The lagoon itself is 50 m long (east-west) by 25 m wide (north-south). The preliminary estimate of the source volume is therefore 50 m x 25 m x 36 m. High levels of chlorine are present. In the plume, this acts both as a conservative tracer and a non-retarded pollutant. The preliminary investigations suggest that the base of the aquifer is essentially horizontal, and it has been inferred from this that lateral migration of DNAPLs will not be a serious problem. PRP has retained Consolidated Orthodox Numerics, Ltd. (CON) to develop a simple model of groundwater flow and pollutant transport in the Kaldor. This model is referred to as K A L I . Figure 5 illustrates the output of the K A L I model, which suggests that groundwater is flowing north from the site. On this basis, a fence of 5 monitoring wells ( M l to M5) will be established 300 m north of the Midnight site. A n additional fence of 5 monitoring wells (M6 to M10) will be established 300 m south of the site, in order to establish whether any upgradient sources of pollution exist. (Refer to Figure 6 for detailed locations of monitoring wells relative to the Midnight lagoon.) Unfortunately, the monitoring results will not be available in time for the Round 1 negotiations. In addition, the water produced by well W2 is being sampled on a monthly basis; no pollutants have been yet been detected. The K A L I model also suggests that the pollution plume from the Midnight site is traveling north at a disturbing rate, and might be expected to begin discharging at springs along the northern boundary of the Kaldor aquifer within 2.4 years. These springs are ecologically significant because they sustain a protected wetland habitat for the rare Whyah Duck. 4.4. Initial negotiations R E G has proposed a trilateral negotiation to PRP and VIC, to take place in December of 1994. The goal of this negotiation (Round 1) is to establish a consensus management plan for the year 1995. It is further agreed that if this negotiation is successful, then a second round of negotiations (Round 2) will be undertaken in December of 1995, to establish a consensus management plan for the year 1996. The three stakeholders have agreed to rely upon C O N to provide neutral technical advice. C O N will be paid through a trust fund which will be paid for by PRP and disbursed through a trust administered jointly by the three stakeholders. C O N has developed a number of options for consideration in the management plan, as described in the following section. 4.5. Technical options The options developed by C O N for the Round 1 negotiations are described below, and shown schematically in Figure 7. 192 Cost estimates for these options are provided in Table 1. (Obligatory costs such as the emergency measures and CON's fees are not provided.) These cost estimates entail a number of assumptions which should be mentioned. Capital costs are assumed to be incurred at the beginning of 1995. Operating costs are presented as both an annual operating cost; and as the present value (at 10% discount rate) of the annual operating cost incurred over a 25-year period, with each annual cost being incurred at the end of the year. Please keep in mind that although these options are reasonably realistic, and are internally consistent with the rest of S A M , there is no claim that these are the "technically best" set of options. The purpose of these options is to provide fodder for the negotiation. If you do think of better technical options, make a note in your journal and bring them up in the debriefing. 4. Monitor northwest of the site. This option involves the establishment of 4 monitoring wells ( M i l to M l 4) in a northeast-trending fence, about 300 m northwest of the site. (See Figures 5 and 6). The reasons behind this option are: 1) the failure to detect the plume in the northern fence of monitoring wells; 2) the fear that the plume may be heading northwest towards well W2, rather than north towards the spring as suggested by model K A L I . 5. Monitor south of the springs. This option involves the establishment of 4 monitoring wells in an east-trending fence, about 200 m south of the line of springs along the northern boundary of the Kaldor aquifer. The reason behind this option is to obtain early warning about possible discharge of pollutants to the springs. 6. Sample at springs. This option involves taking samples directly from the springs, in order to determine whether the plume has already reached them. 7. Partial source removal. This option involves excavating the site (50 m x 25 m) to a depth of 7.6 m. The side walls will be held in place with steel sheeting. Depending on the level of pollution, the material will be either incinerated in an out-of-province facility, or dumped in an out-of-province special waste landfill. The goal is to remove the polluted near-surface material, and to reduce the source volume. It is recognized that removal of the entire source volume, down to the base of the aquifer, is impossible. It follows that the source will continue to generate a plume even if this option is undertaken. This option is mutually exclusive with source capping (below). 8. Cap source. This option involves a multi-layered cap consisting of a synthetic membrane sandwiched between various layers of clay, native soil, and drainage material. The goal is to isolate the polluted volume from the surface. This option is mutually exclusive with the partial source removal option, because source capping is unnecessary if the near-surface polluted material has been removed. 9. Shut down water well W2. This option requires VIC to close the threatened water well W2. VIC would replace the lost water supply (0.02 m^/s) by increasing pumping at wells W l and W3 by 0.01 m-Vs each, to a total rate of j 0.03 m-Vs each. This change involves certain capital and operating costs for VIC, and these would be covered by PRP. (Note: the monthly sampling at W2 will continue even if it is shut down, as it provides a useful monitoring point.) 10. Containment well at source. The K A L I model suggests that pumping for containment at the source will have a very limited radius of effectiveness, and will not be capable of capturing the plume. The goal of this option is simply to prevent pollutants from leaving the site, with no intention of capturing the plume. It requires the installation of a well (designated C l ) and a carbon treatment plant at the site. The well would be designed for a maximum capacity of 0.08 m^/s, and the treatment plant would be designed for a maximum capacity of 0.04 m^/s. During 1995, this system would be operated at only 0.04 m^/s; according to K A L I , this is the minimum rate needed to obtain a measurable gradient towards the site on all sides. The extra well capacity is to allow for a possible increase in the rate of pumping; it is a hedge against uncertainty. The treatment plant does not have extra capacity because, unlike the well, it can be designed to be modular and easily expanded. The treated water will be discharged to an ephemeral stream which drains into the Whyah Duck wetland. It is expected that the treated water will partially recharge through the streambed and then discharge to the springs, and partially flow directly into the wetland. The intention is to maintain flow into the wetland, given that the spring flow may be reduced due to the containment pumping. 11. Refine model. The K A L I model is recognized to be very preliminary in nature. This option provides a small budget for limited data acquisition (e.g., pump testing the water wells and the containment well), and refining the model according to this data. The goal is to obtain a better basis for prediction and design during the Round 2 negotiations. 193 4.6. Alternatives to negotiation If any stakeholder decides to walk from the table, then the negotiation will cease and each stakeholder can (and probably will) implement their alternative to negotiation. These alternatives are as follows: 1. R E G can fall back upon its powers as a regulator and impose a regulated solution. 2. PRP can stonewall by using a variety of legal means to appeal this regulated solution. These efforts could delay action at the site for years, and keep both parties busy with the mechanics of the appeal process. 3. VIC can sue PRP for damages to the water resource. This could result in a large but uncertain financial cost to PRP, as well as a political cost to the provincial government. These alternatives are known as B A T N A s (best-alternative-to-a-negotiated-agreement), because they are the best thing that a stakeholder can do if an acceptable agreement cannot be reached. 5. Playing PAQMAN 5.1. Time required The time required to play P A Q M A N is uncertain. A period of four hours is scheduled. The game may finish before this. It may also go beyond this time, if all players agree. 5.2. The game manager The game will be run by a game manager who will double as the technical consultant, and triple as a mediator should the negotiations run into difficulty. 5.3. The computer A computer will be present to assist the consultant with presenting and developing options, as well as tasks such as adding up the costs of various agreements. 5.4. The stakeholders 5.4.1. Roles You will be assigned the role of one of the three stakeholders — R E G , PRP, and VIC. For fairness, the roles will be assigned at random. 5.4.2. Goals Your goal in this game — how to win, if you want — is to get an agreement that meets your priorities (as indicated in the confidential instructions) as closely as possible. So there is a competitive aspect, but bear in mind that you can win by cooperation, by finding a "win-win-win" agreement. It's not a zero-sum game. Falling back on your B A T N A does N O T mean that you lose. If you perceive that your priorities are not going to be met in an agreement, then it is quite legitimate for you walk away from the table. 5.4.3. Journal-keeping You should keep a journal throughout this process. This is free-form: it can contain information on whatever you find interesting, or you think I might find interesting. Possible topics are your general impressions of the negotiation; your reasons for taking certain positions in the negotiation; and comments, criticisms, and questions regarding the game. The journal can be written up during breaks in the game. You shouldn't let it interfere with the actual negotiation. The journals will be used in debriefing. The game manager will retain them at the end of the game. Please make them legible! 5.5. Briefing Prior to beginning the game, a briefing session with the game manager will take place to sort out general questions, inconsistencies, etc. Questions about the confidential instructions can be resolved in private at this time. 5.6. Stages of the game The game begins with the assumption that the initial negotiations necessary to begin with Round 1 have already taken place, and that technical options for Round 1 have been prepared. The "you" in this section is collective, and refers to all the players. 194 5.6.1. Round 1 Round 1 will begin with a review of the Round 1 options by the consultant. You will then commence the Round 1 negotiations, and attempt to sort out the various options into an agreement. If you are unable to reach an agreement, then you can go into a joint brainstorming session, and try to come up with new option(s) that you think may allow you to reach agreement. The consultant will advise you of the cost of evaluating these new option(s), and ask if you are willing to pay. The consultant may also advise you that the option can't be evaluated in the available time. Assuming that option evaluation does go ahead, you will then take a break while the consultant evaluates the new option(s) using the S A M software. This evaluation may be a fairly lengthy process (up to 1 hour), so you take a break at this point to write up journals, get some fresh air, get some coffee, etc. You will then attempt to reach a new agreement with consideration of these new options. A word of caution is in order regarding the generation of new options. You may immediately think of 50 technically better options, but that's not the point. The reason for developing new options here is to break an impasse in the negotiation, not to achieve a technically better solution. Also remember that over-exuberance in the development of new options may tax the powers of the consultant, given the short deadline for evaluation! If you are successful in reaching agreement, either with or without new options, then you take a break while the consultant determines the consequences of this agreement, and develops the Round 2 options. If you run into procedural problems, then you can call for the mediator. If you cannot reach agreement, even with these new options, then you must fall back upon your BATNAs, and the game comes to an end. 5.6.2. Break (development of Round 2 options) While you take a break, as consultant must determine the consequences of the Round 1 agreement and develop a set of appropriate Round 2 options. Also, the game manager must develop additional confidential instructions for Round 2. This break may take up to 1 hour. Again, you can write up your journal, take a stroll, read the paper, whatever. 5.6.3. Round 2 Round 2 will proceed in the same fashion as Round 1. The game ends when ever you either reach agreement or fall back upon your B A T N A s . 5.6.4. Debriefing A debriefing session, with the aid of the written journals, will be held at the end of the game. We might talk about why certain players adopted certain positions, why certain agreements were reached (or not reached), and so on. You can reveal your confidential instructions to the other players, if you choose. This is also the time to help with the debugging by suggesting improvements (such as better technical options) and pointing out inconsistencies. 5.7. Hints on playing Try to act with "professional intent", in other words, try to act as your role would demand in real life. Avoid getting into debates about responsibility for the pollution. The legal responsibility is clearly assigned to PRP. The moral responsibility belongs to Mr. Skip, who is bearing it quite well, thank you. Bring a funny hat to wear when you are "in character". The game manager will bring two funny hats, one for the consultant and one for the mediator. Finally, enjoy the game. 195 [Box 6-2 continued] Table 1. Cost estimates for Round 1 options. Options Capacity cu.m/s Capital cost Operating cost (annual) Present value of operating cost i=10% t=25 years 4. Set up fence of monitoring wells N W of site 4 wells at 50 m spacing to 37.5 m depth $62,667 Sample 4 wells for full suite every month $102,667 $931,909 5. Set up fence of monitoring wells S of springs 4 wells at 100 m spacing to 30.5 m depth $54,667 Sample 4 wells for full suite every month $102,667 $931,909 6. Sample at springs, monthly at 4 sites $102,667 $931,909 7. Partial source removal Excavation $1,466,667 Off-site incineration $5,786,667 Off-site landfilling $4,266,667 8. Cap source Capital $693,333 O & M $22,667 $205,746 9. Shut down water well W2 Capital cost $100,000 Annual operating cost $25,000 $226,926 10. Containment well at source (CI) Capital costs: Power connection, 500 m $32,000 Well construction 0.08000 $146,667 Discharge to stream 0.08000 $106,667 Carbon treatment plant 0.04000 $933,333 Operating costs 0.04000 Pumping $66,667 $605,136 Treatment $813,333 $7,382,659 11. Improve model (field work + desk work) Field work $75,000 Desk work $25,000 T O T A L S FOR VARIOUS O U T C O M E S 4,5,6,7,9,10,11 $13,056,000 $11,010,450 4,5,6,8,9,10,11 $2,229,333 $11,216,196 196 5 Process is iterative, over a period of years. 5 SAM goes through one or more interations. Figure 1. Conceptual model of SAM Figure 2. Components of SAM. Reach agreement Figure 3. Iterative management periods within SAM. Figure 4. Map of the area. Scale 1:100,000. Topography shown at 1 m contour interval. 197 8 5 S 8 I I i ? f f 8 I S S I I ! i I I s ! I S i 1 S i I 8 3 S 1 I I II I I \i i Plume as modeled by KALI. Dates refer to position of head of plume on 1 Jan of year Indicated, - 1 9 8 7 M5 M4] M3 i (X* (£3 M11 1 ) a? tt? M2 Ml M12 ^ rT\ I MM W T 1 9 8 6 1 9 8 5 Midnight Site (Ti fTi fTi ITI fTi U 7 0 7 tt? tr? M10 MS MS M7 M6 Rgure 5. KALI model of groundwater flow and pollutant transport In the Kaldor aquifer, in fne vicinity of the Midnight Site. Scale 1:40X100. Hydraulic head shown at 0.1 m contour Interval. ROUND 1 Figure 6. Detail of area around the Midnight site. Scale 1:10,000. Monitoring wells M1 to M10 (to be installed in 1995), and M11 to M14 (proposed for 1995; Option 4). ROUND 2 » z o I-D. o UJ > UJ 0. o o o Monitor NW of site Monitor S of springs — Sample springs Partial source removal -Cap source Shut down W2 Containment well Refine model z o REG: regulate -PRP: stonewall • VIC: sue PRP — REG: Regulate • PRP: Stonewall VIC: Sue PRP -Figure 7. Schematic diagram of Round 1 options. 198 6.3.1 Round 1 The Round 1 technical options are described in Box 6-2; they are identical to those described above for Round 1 of SAM using COOP (Table 6-3). The detailed cost information provided to the participants is in Table 1 of Box 6-2. The confidential instructions are described in Boxes 6-3 to 6-5; they closely parallel Preference Variation 1 for Round 1 of COOP (Section 6.2.1.3.1). The participants reached agreement in Round 1. Their decision is shown in Table 6-22. This package of options is consistent with Pareto-optimal equilibrium 9 in Round 1 of COOP (see Table 6-8). The participants also injected a couple of options of their own. They agreed to institute a Public Education Program for the benefit of the residents of the village. This option was developed by the players as a response to constraints placed on them by the residents' perceptions and possible reactions to the situation. (This is consistent with good practice in polluted sites remediation; see Lynch, 1994.) They also decided to increase sampling frequency at W2 , as an alternative to shutting it down. These two options, developed independently by the participants, illustrate both the importance of creativity in negotiations and the value of role-playing as a counterpoint to the strictly algorithmic conflict analysis. Option Decision Monitor northwest of site Accepted Monitor south of springs Rejected Sample at springs Rejected Partial source removal Rejected Cap source Rejected Shut down water well W2 Rejected Containment well at source Accepted Revise perception models Accepted Table 6-22. Decisions on the PAQMAN Round 1 technical options. 199 Box 6-3. Confidential Instructions to Assistant Director of the Pollution Response Panel [Round 1] Your priorities are dominated by a desire to satisfy the needs of the current provincial government. The government is under severe fiscal constraints, but wishes to avoid any political problems which might arise from this unwanted responsibility. Your goal is therefore to simultaneously minimize both the cost of remediation and the controversy over the remediation. You have agreed to negotiate because you see it as a possible way to meet this goal. Round 1 Priorities (highest to lowest) 1. Your first priority is to reject partial source removal (Option 7). The high cost and low effectiveness of this option make it a non-starter as far as you are concerned. 2. Your next priority is to avoid a lawsuit from VIC, which would be politically messy and legally expensive, with a strong chance of a large assessment for damages. 3. You also want to avoid a regulated resolution from R E G , as you fear the imposition of a high-cost remediation. 4. You are prepared to stonewall by mounting an appeal against any regulated resolution. This is basically a threat to keep strengthen your hand at the table, both for the Midnight site and for future problems. 5. You would like to see water well W2 shut down (Option 9), because this will significantly reduce your liability with respect to VIC. You are quite happy with paying the relatively minor cost of getting VIC to shut down W2. 6. If W2 is not shut down, you want to set up a fence of monitoring wells northwest of the site (Option 4). If W2 is threatened by the plume, you want to take action as soon as possible. Again, you are motivated here by the fear of liability with respect to VIC. 7. You would like to install a containment well at the source (Option 10), because you see this as a major way to reduce liability. However, you will argue that this is only economically feasible if partial source removal is not imposed. You are strongly opposed to any package which includes both the containment well and the partial source removal. 8. You are opposed to source capping (Option 8), because you don't consider it to be particularly cost-effective option for reducing either environmental risk or liability. 9. You place a relatively low priority on refining the model (Option 11), because you secretly suspect that C O N is just trying to milk more billing hours. However, you will not oppose this too strongly. 10. You place a relatively low priority on sampling at the springs (Option 6), but you are prepared to do it because it is politically important to be seen to protect the Whyah Duck wetland. 11. Your lowest priority is to argue against the monitoring fence south of the springs (Option 5). You consider this to be technically silly, and will argue against it on those grounds. But you are prepared to give it up, again because it is politically important to be seen to protect the Whyah Duck wetland. Box 6-4. Confidential Instructions to District Officer for the Remediation Enforcement Group [Round 1] Your priorities are determined by a balance among: enforcing the legal and regulatory requirements; looking out for possible political problems; and a genuine concern for environmental and human health. Your objective in starting these talks was to try to avoid having to impose a regulatory resolution. You have learned that an imposed resolution usually triggers appeals and litigation from all sides, leading to long-term frustration and waste of your resources, worsening the political problem without dealing with the environmental problem. Round 1 Priorities (highest to lowest) 1. Your first priority is to avoid having the talks broken off at this early stage of a process which you expect to extend over several years. You want to avoid this if you can, and are willing to compromise at this stage. (You will adopt a tougher stance in later stages.) 2. Your first remedial priority is to get a containment well installed at the source (Option 10), because you place a high priority on stopping off-site migration of pollutants. 200 3. Your next priority is to get a cap built over the source (Option 8). This also stops off-site migration (from blowing dust and runoff). Politically, it looks like progress in getting the problem dealt with. And it helps to forestall demands for source removal (see below)., which you consider to be an absurdly expensive and inefficient option. 4. You would like to see the model refined (Option 11), because you believe that a better model will lead to a better decision in Round 2. 5. You would like to install a monitoring fence northwest of the site (Option 4), as insurance in case the plume is headed towards water well W2 rather than the springs. 6. You would like to begin a sampling program at the springs (Option 6), because you want to know as soon as possible if the head of the plume has begun to discharge into the Whyah Duck wetland. 7. You are not against establishing a fence of monitoring well south of the springs (Option 5). Frankly, you feel that such a fence has a low probability of detecting the plume, given the enormous uncertainty regarding its location. However, it is politically useful to have an "early-warning system" in place. 8. You assign a low priority to shutting down water well W2 (Option 9). You feel that this would be an alarmist reaction at this point. 9. You place partial source removal (Option 7) at the bottom of your list of priorities, because you consider it to be a very expensive option which is ineffective at reducing off-site migration. You can't come out against it, for political reasons, but you simply place it at the bottom of your list and won't actively push for it. Box 6-5. Confidential Instructions to Water Supply Manager for the Village of Cantorbury [Round 1] Your priorities are dominated by a responsibility to provide a clean supply of water, by its need to act frugally with a limited budget, and by the need to maintain credibility with the inhabitants of the village. It is therefore in your interest to obtain a very thorough remediation and to protect your wells, at no cost to your budget, and to be seen to do so by the villagers. You have consented to join the negotiations because you want to have a say in the actual decision-making. Round 1 Priorities (highest to lowest) 1. Your first priority is to get a containment well working at the source (Option 10). You feel that it is critical to stop feeding the plume. 2. Your next priority is partial source removal (Option 7). You are aware that this will not eliminate the source, but you feel that it is justified because it would remove a substantial volume of pollutants from the system 3. If the talks breakdown, and PRP stonewalls against an imposed resolution from R E G , you are prepared to litigate against PRP. 4. You want to establish a fence of monitoring wells N W of the site (Option 4), in order to determine whether or not the plume is moving towards W2. You feel that this will be critical information for Round 2. 5. You are not in favour of shutting down water well W2 (Option 9), even with the proposed side payment from PRP. As far as you are concerned, shutting the well down will alarm the villagers and set a bad precedent. It also will cause you technical water supply problems. You are not sure that the side payment will be enough to cover these problems in the long term. 6. You are in favour of refining the model (Option 11), not least because you secretly think it might be valuable evidence in a future lawsuit. (This is not something you want to admit, though.) 7. You are in favour of capping the source (Option 8), although this is low on your list of priorities, because you are afraid that it will forestall any source removal. 8. You are in favour of sampling the springs (Option 6) and establishing a monitoring fence south of the springs (Option 5), but you place a low priority on this because the Whyah Duck wetland is not really your concern. 201 6.3.2 Round 2 Documentation for the PAQMAN Round 2 is provided in Boxes 6-6 to 6-9. The PAQMAN Round 2 technical options (Box 6-6) do not exactly match the COOP Round 2 technical options (Section 6.2.2.2), because they follow a different package of Round 1 decisions. The comparison between the PAQMAN and COOP Round 2 technical options is shown in Table 6-23. PAQMAN Round 2 technical option Equivalent technical option from COOP Round 2 (Table 6-15) Description and goal 4. Weak containment. Option 4 5. Strong containment. Option 5 6. Keep water well W2 running. None, because W2 was shut down in COOP Round 1. Continue pumping W2 at 0.02 m3/s; (i.e., continue with the status quo which existed prior to 1995). Goal: To continue using W2 as a source of water, given that the revised estimate of plume velocity suggests that it is not in immediate danger of pollution. 7. Cap source. Option 7 8. Further revision of the perception models. Option 9 9. Low-cost extension to monitoring network. Option 10 10. High-cost extension to monitoring network. Option 11 11. Continue Public Education Program. None, because COOP example did not consider a public education program. Continue with program to inform local residents. Goal: a response to constraints placed on by the residents' perceptions and possible reactions to the situation. Table 6-23. Comparison between P A Q M A N and C O O P Round 2 technical options. The confidential instructions (preferences) for PAQMAN Round 2 (Boxes 6-7 to 6-9) were developed by the game manager in response to the PAQMAN Round 1 decision, and do not closely parallel Round 2 preferences in COOP. 202 Box 6-6. Round 2 Options 4. Weak containment. CI @ -0.03593, C2 @ -0.02521. 5. Strong containment. CI @ -0.06582, C2 @ -0.04642. 6. Keep W2 running. 7. Source capping. Same cost as Round 1. 8 Further upgrading the model 9. Low-cost extension to monitoring network. Say, 4 monitoring wells and 4 piezometers. 10. High-cost extension to monitoring network. Say, 8 monitoring wells and 8 piezometers. 11. Continue PEP. Cost estimates for Round 2 options. Options Capacity cu.m/s Capital cost Operating cost (annual) Present value of operating cost i=10% t=25 years Weak containment Capital costs: Power connection to C2, 283 m $22,667 C2 well construction + piping 0.08000 $373,333 Addition to treatment plant 0.06114 $400,000 Operating costs Pumping: CI 0.03593 $61,333 $556,725 Pumping: C2 0.02521 $44,000 $399,390 Treatment 0.06114 $1,133,333 $10,287,312 Minus: supplanted operating costs 0.04000 Pumping ($66,667) ($605,136) Treatment ($813,333) ($7,382,659) 5. Strong containment Capital costs: Power connection to C2, 283 m $22,667 C2 well construction 0.08000 $373,333 Addition to treatment plant 0.11224 $1,733,333 Discharge to stream 0.08000 $106,667 Operating costs Pumping: CI 0.06582 $105,333 $956,115 Pumping: C2 0.04642 $76,000 $689,855 Treatment 0.11224 $2,000,000 $18,154,080 Minus: supplanted operating costs 0.04000 Pumping ($66,667) ($605,136) Treatment ($813,333) ($7,382,659) Cap source Capital $693,333 O & M $22,667 $205,746 Further refinement of model $100,000 Low-cost extension to monitoring network Well installation (8 wells) $125,333 Sampling (4 wells) $102,667 $931,909 High-cost extension to monitoring network Well installation (16 wells) $250,667 Sampling (8 wells) $205,333 $1,863,819 203 Box 6-7. PRP's Confidential Information for Round 2 In order of priority: 1. A big no to strong containment. You would walk from the table and stonewall rather than accept this. 2. You do want the low-cost monitoring, because you want to reduce liability and hope to get evidence that they can reduce containment pumping; but you say N O to high-cost monitoring extension. 3. You are not keen on refinement of the model. 4. You want to shut down W2, because keeping it off is a low-cost liability insurance; and put on source capping: it will reduce the liability and flak due to dirty dust, and also forestall any demands for partial removal. 5. Will accept weak containment, if pushed. Box 6-8. REG Round 2 Confidential Instructions Insist upon at least weak containment; at least source capping, and at least a low-cost extension to the monitoring network. Anything less, you will break off talks and impose these options on PRP. Then, in order of priority: further refinement; strong containment; high-cost monitoring; W2 off. Box 6-9. VIC's Round 2 Confidential Instructions 1. Weak containment and low-cost extension to monitoring, at a minimum; otherwise you will sue. 2. You would like strong containment, high-cost monitoring, and refined mode 3. Want to keep W2 running 4. Source capping : no strong opinion either way. The participants reached an agreement in Round 2, as shown in Table 6-24. Despite the differences in technical options and preferences between PAQMAN Round 2 and COOP Round 2, the PAQMAN Round 2 agreement is consistent with Pareto-optimal Equilibrium 26 in COOP Round 2 (see Table 6-20). Option Decision Weak containment. Accepted Strong containment. Rejected Keep W2 running. Accepted Cap source. Accepted Further upgrading of model. Rejected Low cost extension to monitoring network. Accepted High cost extension to monitoring network. Rejected Continue Public Education Program. Accepted Table 6-24. Decisions on the PAQMAN Round 2 technical options. 204 6.4 Summary This chapter has described a run of the compound model SAM on a hypothetical case, in order to document the workings of SAM as a heuristic device and as an illustration of the management strategies and soft/hard complementarity. The hypothetical case is described in Section 6.1. The pollution system consisted of an unconfined aquifer containing a plume of pollution originating from an illegal dump site. The technical system consisted of a consultant who observes the pollution system, develops technical options for the decision system to consider, and implements the chosen technical options on the pollution system. The decision system consisted of negotiations among three stakeholders: a responsible party, a regulatory party, and an injured party with threatened water supply wells. The technical consultant has been jointly retained by the three stakeholders. Each stakeholder had a BATNA (Best-Alternative-To-A-Negotiated-Decision); if any stakeholder had exercised their BATNA, the negotiations would have broken down. SAM was applied to this hypothetical case, using COOP as the decision model (Section 6.2) and using PAQMAN as the decision model (Section 6.3). In both cases, SAM was run through two successive rounds. In each round, the decision model reached agreement on a package of technical options acceptable to all stakeholders, from a set of options provided by the technical model. Between the two rounds, the pollution model was updated to reflect implementation of the agreement reached in Round 1. Using COOP as the decision model, and given particular sets of stakeholder preferences among the technical options and BATNAs, a large number of possible agreements were identified in both rounds, with no possibility of a breakdown in negotiations. (A variation on the 205 stakeholders' preferences in Round 1, with stronger negotiating stances, did indicate some potential for breakdown in negotiations.) All of the possible agreements in Round 1 included a containment well at the source, and none of the possible agreements included the expensive option of partial source removal. In Round 2, the only consistency among the possible agreements was the rejection of an expensive option for strong hydraulic containment. In Round 1 with PAQMAN as the decision model, the instructions for the participants were based on the Round 1 input to COOP. The participants reached agreement on a package of technical options identical to one of the possible agreements in the COOP output for Round 1. The instructions for Round 2 with PAQMAN diverged somewhat from the input to Round 2 of COOP. However, despite this difference, the PAQMAN participants did reach a Round 2 agreement which was consistent with one of the possible Round 2 agreements identified by COOP. An important role of SAM is the illustration of management strategies. Table 6-25 summarizes the illustration of management strategies (see Section 5.5.2) by the run of SAM described in this chapter. Another important role of SAM is the illustration of soft/hard complementarity (see Section 5.5.3). This can be seen through consideration of the run as a whole. The hard sub-problems are emphasized by the sub-models. The overall soft problem matrix is brought out by recognition that the linkages between the sub-models are not automatic, but rather require considerable subjective judgement. An important example of this is the subjective development of stakeholder preference masks (input for the decision model) on the basis of technical options (output from the technical model). Note that the technical model, through both algorithmic sub-206 models and subjective linkages, takes account of the overall soft problem, by providing a range of options tailored to the needs of the decision-making process as a whole, rather than to the needs of any one stakeholder. Strategy Illustration Iterative decision strategies iterative decision-making Two rounds of iterative decision-making. contingent decision Not illustrated. iterative liability reduction Not illustrated. Integrative decision strategies process integration Management is integrated into a single process of decision-making. inclusive participation The process incorporates all three stakeholders: regulatory, responsible, and injured. geographic integration Management incorporates the polluted site and all potential receptors. Negotiative decision strategies primary negotiation Mode of decision-making is negotiation, with other modes of decision-making reserved as B A T N A s by the stakeholders. principled negotiation Not illustrated. balanced negotiating power There is a general balance of negotiating power among the stakeholders, each of whom has a credible B A T N A . mediation Not illustrated. Technical strategies technical integration A single technical organization provides joint technical support to the decision-making process. mitigative technique The technical options are all mitigative rather than restorative in nature. iterative technique Two rounds of observation and generation of technical options. Table 6-25. Illustration of the management strategies by the run of SAM described in this chapter. The strategies which are intrinsic to S A M are shown in italics. 207 7. Summary and discussion This concluding chapter summarizes the thesis and discusses its development, its contributions, and directions for future research. 7.1 Summary This thesis finds its motivation in the heed to develop an appropriate and bounded response to groundwater pollution, in both societal and technical terms, at the level of individual cases. The thesis has four elements (Figure 7-1), each reflecting one of the goals of the thesis: 1. A theoretical foundation derived from soft paradigm thinking and from complementarity between the soft and hard paradigms (Chapter 2). 2. A conceptual model of groundwater pollution management, based on soft paradigm thinking (Chapter 3). 3. A consistent and linked set of management strategies, based on the conceptual model and developed in a soft paradigm context (Chapter 4). 4. SAM, a compound model, based on the conceptual model and used both as a heuristic device and as illustration of the management strategies and soft/hard complementarity (Chapters 5 and 6). Each of these elements will be discussed in turn in the following four sections. 208 Managemen t strategies Decision strategies Technical strategies S A M compound mode l ^ M A N A G E M E N r M O D E L [Technical model decisions 1 1 a d v i c e •—* Decision model Figure 7-1. The elements of the thesis. 209 7.1.1 The theoretical foundation: hard and soft paradigms The theoretical foundation is established in Chapter 2, which contrasts two ways of thinking about the management of problems such as groundwater pollution: the hard paradigm and the soft paradigm. These two paradigms are contrasted in Table 7-1. Hard paradigm Soft paradigm Roots science, mathematics, and engineering frustration with the hard paradigm Philosophy tends towards reductionism and strives for objectivity tends towards holism and acknowledges subjectivity Problems hard problems soft problems Boundaries and linkages with other problems well-defined ambiguous and complex Goals, alternatives and consequences well-defined and well-understood ambiguous and complex Uncertainty bounded and quantifiable pervasive and non-quantifiable Solution procedure linear; fixed endpoint; standardized methodology iterative; individual procedures for each problem Logic algorithms argumentation System an entity in the real-world a conceptual basis for organizing information Conceptual modeling idealized basis for transform modeling a detailed basis for argumentation Transform modeling a standardized tool for constraining arguments and for predicting consequences a heuristic device for expanding arguments and for developing understanding through the iterative process of model building Table 7-1. Contrasts between the hard and soft paradigms. The contrasts between the two paradigms are straightforward, but there is also a soft/hard complementarity: the broader sweep of soft paradigm thinking can be applied to an overall problem, defining the appropriate sub-problems for the focused tools of hard paradigm thinking. This provides a basis for reconciliation of the two paradigms, as well as for synergies in practice. This thesis takes a soft paradigm perspective, but also recognizes the importance of soft/hard complementarity. Groundwater pollution management is regarded here as a soft problem containing embedded hard sub-problems. 210 7.1.2 T h e conc e p t u a l model The conceptual model developed in Chapter 3 is a composite, generalized picture which represents the spectrum of groundwater pollution management in Canada. It is based on the soft paradigm, and organized according to the systems concept. A summary of the conceptual model is presented in Table 7-2. System Function Elements and Characteristics P O L L U T I O N S Y S T E M groundwater flow and pollutant transport • complexity Flow subsystem groundwater flow • porous and/or fractured media • heterogeneous distribution of hydrogeologic units and parameters Transport subsystem pollutant transport • primary and secondary sources • plumes of pollutants • receptors (water wells, springs, streams, etc.) • complex physical, chemical, and biological processes M A N A G E M E N T S Y S T E M coping with the pollution system • complexity and uncertainty • degree of integration Decision subsystem make decisions about coping with the pollution system • rules of governance: legislation (statutes and regulations), jurisprudence, government policy • stakeholders: regulatory, responsible, injured, peripheral • issues: technical decisions, responsibility, compensation • modes of decision-making: negotiation versus unilateral decision-making • processes of decision-making • linear decision-making versus iterative decision-making • complexity and uncertainty Technical subsystem • observe the pollution system • advise the decision system • implement decisions on the pollution system • technical organizations • technical goal: restoration versus mitigation • linear technique versus iterative technique • technical uncertainty Table 7-2. Summary of the conceptual model. The soft paradigm orientation of the conceptual model is seen in the level of detail, the emphasis on the management and decision systems, the emphasis on negotiation as a mode of decision-making, the recognition of non-quantifiable uncertainty, and the iterative structure of the model. 211 The conceptual model contains couplets of opposed hard paradigm and soft paradigm elements: restoration versus mitigation, linear versus iterative technique, linear versus iterative decision-making, and unilateral decision-making versus negotiation. 7.1.3 The management strategies The management strategies (Chapter 4) are generalized, abstract plans for coping with particular aspects of generic groundwater pollution case. They are based on the conceptual model of Chapter 3, and developed in the soft paradigm context, with recognition of soft/hard complementarity. They are predicated on a concept of good management which is based on four qualities: goal-setting, efficacy, efficiency, and fairness. The set of management strategies consists of two subsets: decision strategies and technical strategies. The emphasis in this thesis is on the decision strategies. The set of decision strategies consists of three subsets: iteration strategies, integration strategies, and negotiation strategies. These are presented in Table 7-3, which provides a brief description and set of rationales for each strategy, as well as listing the other strategies with which it is most-closely linked. The technical strategies are technical integration, mitigative technique, and iterative technique; these are presented in Table 7-4. The soft paradigm orientation of the management strategies can be seen with respect to: • the emphasis on iteration, as a means of coping with complexity and uncertainty; • the emphasis on integration, as a means of coping with ambiguous boundaries and complex linkages among problems; and • the emphasis on negotiation, as a means of coping with multiple stakeholders and their divergent interests and values. 212 Strategy Description Rationales and linkages Iteration strategies iterative cycles of decision-making • a means of coping with complexity and uncertainty • linked with iterative technique and mitigative technique iterative decision-making decision-making which is deliberately designed to repeat through a series of iterations; as opposed to linear decision-making • planned periodic responses to new information, new understanding, and changing circumstances • technical actions can proceed while the decision system evolves • avoiding wasted effort and false optimism associated with an unattainable endpoint contingent decision a decision of the form "If Event X occurs, then Action Y will be undertaken" • resolving sticking points in negotiations • coping with uncertainty iterative liability reduction responsible stakeholder obtains an iterative adjustment in its liability in exchange for technical progress and a series of fees • an escape from the impasse in which release from liability is contingent on complete attainment of technical goal • should be linked with inclusive participation Integration strategies integration of the decision system • increasing tractability • improving flow of information • improving efficiency in decision-making • linked with technical integration process integration merging or linkage of isolated processes • reducing procedural complexity inclusive participation participation of all stakeholders in decision-making • improving fairness • improving goal-setting • should be linked with mitigative technique and iterative liability reduction geographic integration expanding geographic scope to include multiple polluted sites, multiple receptors, and potential disposal locations • focus on receptors and actual risk • economies of scale • encompass hydrogeologic boundaries • direct comparison of different sources and their relative threat to health Negotiation strategies delegation of management to the collective intellect, rationality, and responsibility of the stakeholders primary negotiation negotiation as "first importance" and "first in a series" relative to unilateral modes of decision-making • high potential to result in good management • unilateral modes of decision-making have irreversible impacts principled negotiation cooperative and consensual decision-making; joint gains achieved through tradeoffs and synergies • potential benefits of negotiation cannot be fully realized if it is dominated by a competitive and adversarial style balanced negotiating power relative balance in the B A T N A s and worst-nightmares of the participants • goal-setting, fairness, efficacy, and efficiency may all suffer in the absence of balanced negotiating power mediation procedural and substantive assistance rendered by a neutral party • beneficial and perhaps essential in complicated negotiations Table 7-3. Summary of the decision strategies. 213 Strategy Description Rationales and linkages technical integration merging or linking technical support, so that it is provided to the decision-making process rather than to individual stakeholders • improving flow of information and ideas • avoiding duplication of effect • developing technical options for joint gain • providing technical support for stakeholders with limited resources • avoiding technical advocacy • linked with integrative decision strategies mitigative technique reducing damage and risk to human and ecological health, without attempting to provide a permanent remedy or complete restoration • restoration and permanent remedies are generally infeasible and tend to force unnecessary expenditures • linked with negotiation strategies, inclusive participation, process integration, and technical integration iterative technique a deliberately repetitive and open-ended approach to the design and implementation of technical options, moving towards a condition of closure, but with no set endpoint; as opposed to linear technique • linear technique often fails to achieve its set endpoint, and becomes ad hoc iterative technique • planned periodic responses to new information, new understanding, and changing circumstances • linked with iterative decision strategies and with mitigative technique Table 7-4. Summary of the technical strategies. 7.1.4 The compound model SAM The compound model SAM is described in Chapters 5 and 6. It is based on the conceptual model of Chapter 3. It illustrates the management strategies (Chapter 4) and the concept of soft/hard complementarity (Chapter 2). As discussed in the following section, SAM also served as a heuristic tool for the development of understanding through the iterative process of model building. Chapter 5 provides a general description of SAM. The sub-models are algorithmic, with the exception of one analog sub-model (a role-playing exercise). Most of the linkages between sub-models are subjective. 214 The organization of SAM corresponds to the organization of the conceptual model (see Figure 7-1). It contains two main sub-models, the pollution model and the management model. The pollution model represents the pollution system. It consists of algorithmic models of heterogeneity, flow, and transport. The management model represents the management system. It consists of a technical model and a decision model, representing the technical and decision systems. The technical model includes of a numoer of algorithmic models which represent various tasks of the technical system: modeling flow and transport in the pollution system; simulation-optimization of pumping rates; and costing of technical options. The decision model is an algorithmic or analog (role-playing) model of negotiation. SAM is designed to operate in an iterative fashion (see Figure 7-1). A cycle begins with limited observation of the pollution model by the technical model. These observations are used to produce technical advice, using subjective judgement guided by the algorithmic sub-models of the technical model. The advice, again using subjective judgement, is transformed into preferences for the stakeholders in the decision model. The decision model produces a decision, which is implemented on the pollution model. The cycle can then repeat, beginning with a fresh set of observations of the pollution model. Soft/hard complementarity is illustrated by the subjective linkages, which represent the overall soft problem; and by the algorithmic sub-models, which represent the embedded hard sub-problems. Chapter 6 presents an example of SAM applied to a hypothetical case. The role of this chapter is to document the workings of SAM, as both a heuristic device and an illustration of the management strategies and soft/hard complementarity. The pollution system consists of an 215 unconfined glaciofluvial aquifer, an illegal waste lagoon which is a source of trichloroethylene and chloride, and a plume associated with this source. The plume is a potential threat to nearby water wells and to a group of springs. The decision system involves three stakeholders: a regulatory agency, a responsible party, and the operator of the water wells. These stakeholders are engaged in negotiation over management of the pollution system. The technical system consists of a single consulting firm which provides joint technical support to the negotiation process. The example of SAM in Chapter 6 passes through two iterations. The decision model produces decisions which reflect a relative balance of negotiating power among the stakeholders. The technical advice evolves as the technical model improves its observations and understanding of the pollution model. 7.2 Development of the thesis This section discusses the development of the thesis, and the heuristic value of model building. The elements of this thesis have been presented in a linear fashion, beginning with the underlying understanding of hard and soft paradigms, proceeding through the conceptual model and the management strategies, and culminating in the compound model SAM. In practice, the development of this thesis was iterative, involving interactions among these elements (Figure 7-2). The research began as an attempt to apply simulation-optimization modeling to the real-world case of a small Canadian city with a number of pollution sources and threatened water supply wells.1 This line of inquiry represented a logical extension of the 1 The city will remain anonymous, for reasons of confidentiality. 216 existing hydrogeological literature on management of groundwater pollution problems (see Chapter 2). Consistent with this, the initial conceptual model was relatively simple, and involved a single decision-maker dealing with a well-understood pollution system. As developed Conceptual model Hard and soft paradigms Management strategies Compound model As presented Hard and soft paradigms I Conceptual model Management strategies Compound model Figure 7-2. The elements of this thesis, as developed and as presented. In developing the initial conceptual model, a number of interviews were conducted with people involved in this real-world case. These interviews, and review of relevant technical reports, brought out the complexities of the situation. The groundwater flow system involved a complex glacial stratigraphy, and there was a paucity of subsurface information. The management system involved multiple stakeholders with divergent interests and values, and did not exhibit a high degree of integration. In sum, the case had the classic characteristics of a soft problem. The initial conceptual model was progressively revised to reflect these realities, and it became increasingly apparent that a simulation-optimization model could only provide part of 217 what was needed to address this revised conceptual model. In particular, the optimization sub-model was problematic, because it was not clear how to define decision variables, objective function(s) and constraints in a fashion that was both representative of the situation and mathematically tractable. A revised and more-complicated compound model was gradually developed to match the revised conceptual model; this was the precursor of SAM. The simulation-optimization model remained as a sub-model of the revised compound model. At the same time, it became apparent that the revised compound model had to be developed around a simplified hypothetical case; consequently, the effort to model the original case study was set aside. Model development based on a simplified hypothetical case is a common approach in models of groundwater problem management; e.g., Gorelick et al. (1984); Massman and Freeze (1987a, 1987b). The revised conceptual and compound models subsequently went through a number of iterations, becoming increasingly appropriate but complicated. The final version of the conceptual model is given in Chapter 3. SAM, as the final version of the compound model, is presented in Chapters 5 and 6. But of more fundamental importance, this iterative process involved a transition from hard paradigm thinking to soft paradigm thinking. Initially, this was an unconscious transition. The questions arising created a tension which forced the author to recognize and research issues and literature from other fields of inquiry, ultimately leading to the distinction between hard and soft paradigms, as described in Chapter 2. Figure 7-3 illustrates the transition from an initial state of thinking in terms of the hard paradigm to a final state of thinking in terms of the soft paradigm and soft/hard complementarity. 218 1 • Time Figure 7-3. Transition in the development of this thesis, from thinking in terms of the hard paradigm to thinking in terms of the soft paradigm and soft/hard complementarity. In the context of this new understanding, the role of the compound model changed from a hard paradigm emphasis on output to a soft paradigm emphasis on the heuristic value of model building. This heuristic value can be seen on a number of levels. At the highest level, it is shown by the iterative co-development of the compound model and the management strategies. The strategies did not arise from any specific output of the compound model. Rather, they took shape from ideas that were stimulated by the exercise of building the compound model. As discussed in Chapter 5, many of the management strategies became intrinsic elements of the compound model. At an intermediate level, the heuristic value of model building is shown by the co-development of SAM and the concept of soft/hard complementarity. As SAM developed, it became apparent that a considerable amount of subjective judgement was required to link the algorithmic sub-models together. This provided a starting point for thinking about soft/hard complementarity. 219 At a detailed level, an example of the heuristic value of model building is shown by the concepts of BATNA and worst-nightmare (see Section 3.3.4.1). The distinction between these two concepts, and their dual importance in the balance of negotiating power, became apparent during the coding and implementation of COOP. 7.3 Contributions and directions for future research This section discusses the contributions of this thesis and directions for future research, with respect to soft paradigm thinking, management strategies, and the compound model SAM. 7.3.1 Soft paradigm thinking Thinking in a number of disparate fields has independently converged on similar soft paradigm ideas. The synthesis in Chapter 2 recognizes the commonality in this thinking, brings the ideas together, and provides a basis for applying them to groundwater pollution management. Soft/hard complementarity is a useful concept which acknowledges the strengths of both hard and soft paradigms. The soft paradigm provides a context for the application of hard paradigm methods to hard sub-problems. A corollary is that soft paradigm thinking can help to avoid inappropriate uses of hard paradigm methods. There is considerable scope and rationale for further research on the topic of soft and hard paradigms. One direction of research would be a detailed comparative study of the soft paradigm literatures, building on their similarities and explaining their differences. This study could also encompass the literatures which show soft paradigm tendencies, such as the observational approach in geotechnical engineering. The rationale for this line of research would be to build a 220 clearer picture of the soft paradigm, assessing which characteristics are consistent and which vary between different problems with different dynamics. Such knowledge would assist in extending the soft paradigm to new problems. For example, compare the adaptive management of natural resources (Holling et al., 1978; Walters, 1986) and the observational approach in geotechnical engineering (Terzaghi and Peck, 1967; Peck, 1969). Natural resources management and geotechnical engineering are very distinct areas, with very different dynamics. Management of a natural resource involves an ongoing ecological problem. A geotechnical engineering project involves a temporary geological problem, with a definite endpoint at the conclusion of the project. Many of the details of adaptive management and the observational approach are obviously different. But despite the differences, a similar core philosophy of contingent decision-making can be detected in both literatures. There is also scope for further exploration of soft/hard complementarity. The rationale for this lies in the potential strength of an approach based on a fusion of the two paradigms. The work of Walters (1986) is particularly interesting in this respect. 7.3.2 Management strategies Many of the individual ideas in the management strategies of Chapter 4 are simply "in the air of the times", having developed in recent years through hard-earned experience. For example, mitigation becomes an obvious prescription once a scientific consensus has been reached on the general infeasibility of restoration. This thesis incorporates these ideas into a consistent and linked set of strategies, based on the theoretical foundation of the soft paradigm. 221 How can these strategies be implemented in the real world? Some strategies may be simply imposed by circumstances. Others would require an initial push in the right direction ("activation energy"). For example, judicial decision-making may be transformed into negotiation by the judicial imposition of a "cooling-off period", accompanied by an expression of judicial preference that the stakeholders address their conflict through negotiation. (A recent native land claims litigation in British Columbia was transformed into a negotiation by this mechanism.) Another example would be the case of a technical consultant initiating integration strategies by convincing various stakeholders to contribute to a joint management effort (E. Hill, pers. comm., 1993). In some jurisdictions, there may be legislative barriers to the application of the management strategies. For example, detailed generic regulation of technical activities is not compatible with the heuristic spirit of iterative technique; this was noted in Chapter 4 with reference to the concept of reconnaissance surveys. At the least, regulations should have "escape clauses" to allow innovative iterative technique. Freeze and Cherry (1989) expressed a similar concern, to which Hodge and Roman (1989) responded that "flexible regulation" was an oxymoron. There is some truth to this point, but it misses the larger truth that the overall goal is to manage the problem, not to regulate the problem. The concern about "flexible regulation" can be addressed through the use of formal and transparent escape clauses within a decision-making process which embodies inclusive participation. This leads into a crucial point. It must be emphasized that strategies such as primary negotiation, iterative technique, and mitigative technique can lose public credibility in the absence of inclusive participation and balanced negotiating power. Among environmental 222 and community groups in the United States, phrases such as "voluntary cleanup" and "brownfields development" are becoming synonymous with "preferential deals for responsible stakeholders" and "second-rate cleanup". "Negotiation" acquires a negative connotation when it applies to confidential negotiations between a regulator and a regulatee. These perceptions, regardless of merits, have the potential to generate bitter conflict, to the detriment of all stakeholders. The management strategies have been deliberately presented in a generic and modular fashion. This is intended to accommodate their adoption in specific jurisdictions. It is understood that the incorporation of the strategies in specific jurisdictions and specific cases will entail considerable adaptation. 7.3.3 Compound model SAM The development of the compound model SAM played an important heuristic role in the iterative development of the other elements of the thesis (Section 7.2). SAM also illustrates the management strategies and the concept of groundwater pollution management as a soft problem with embedded hard sub-problems. A number of possible extensions to SAM can be suggested, as directions for future research. It would be possible to add additional stakeholders to the decision model. This would not significantly increase the computational burden. The COOP assumptions described in Chapter 5 entail that each additional stakeholder with a single non-cooperative option will only increase the 223 computational burden by a factor of two. The true burden of considering additional stakeholders would lie in the subjective task of determining their preferences. This subjective task might be eased by the Analytical Hierarchy Process (AHP; Saaty, 1980), an algorithmic methodology which can assist with preference setting. AHP attempts to prioritize alternatives in a consistent fashion on the basis of both quantitative measurements and subjective judgements. AHP would be placed as a sub-model within the decision model, preceding COOP or PAQMAN. This does not imply that SAM could or should evolve into a seamlessly-algorithmic version, without internal subjective elements. The internal subjective elements are an important aspect of SAM, because they represent the soft problem matrix surrounding the hard sub-problems. SAM as described in Chapters 5 and 6 has no provision for feedback from the decision model to the technical model, i.e., there is no provision for the stakeholders to request additional technical options. This may tend to give too much weight to the technical model. It also leaves the stakeholders with no alternative but to break off negotiations if they are deadlocked on the technical options. The players in the PAQMAN run described in Chapter 6 chafed about this limitation, both during the game and in the subsequent debriefing session. They would have preferred latitude to generate their own technical options, i.e., they wanted a feedback loop to the decision model. It would be possible to restructure SAM to include a feedback link from the decision system to the technical system (see Figure 7-4). An additional cooperative option of "Request additional technical options" would be provided in COOP, with similar provision in PAQMAN. 224 Figure 7-4. Possible extension to S A M . A n internal feedback link could be added to the management model, to allow the decision model to request more technical options from the technical model. Along similar lines, it may also be appropriate to provide the decision model with one or more cooperative options which would allow adjustment of certain limiting preferences. The PAQMAN role-players in the example in Chapter 6 insisted on such an option: a "Public Education Program" to inform the local community of the rationale behind certain technical options. SAM has potential alternative applications as an educational method, and as a tool for assisting with management of a real-world groundwater pollution case. In either case, SAM could be freely modified to suit specific requirements. As an educational tool, SAM could be used to teach both the dichotomy and the complementarity of the hard and soft paradigms. PAQMAN is the key sub-model of SAM with respect to education. (As noted above, PAQMAN is based on educational games which were developed by the author for undergraduate classes in groundwater pollution.) The following is a sample set of instructions for implementing SAM with a class of hydrogeology students: 225 • The instructor sets up a run of SAM: the specific conceptual model, the pollution model and the technical model. (The COOP sub-model could be dispensed with, but might be useful for comparison with the PAQMAN results.) • The class would be divided into groups of three. • Each group would be provided with a set of PAQMAN instructions for a first round of negotiations, with technical options as developed by the technical model. • This first round could be carried out during a three- or four-hour laboratory session. Each group would then submit their first agreement to the instructor. • The instructor would then implement each group's agreement on the pollution model, and use the technical model to generate further technical options for each group, for a second round of negotiations. • Each group would be provided with a set of PAQMAN instructions for a second round of negotiations, including the technical options based on the consequences of their first round of negotiations. • The second round could be carried out during a second three- or four-hour laboratory session, one week after the first session. • The instructor would then implement each group's second agreement on the pollution model, and present the results to the class for discussion. As a variation on this theme, and depending on their level of technical sophistication, the students could use the technical model themselves to generate their own technical options. This would be a much lengthier exercise, requiring student access to computers with appropriate 226 software. The sophistication and emphasis of the technical model could be adapted to the students' skills and the instructors' goals for the class. As a tool for assisting with the management of a real-world groundwater pollution case, SAM has a wide range of possibilities, depending on the goal(s) of its user(s). For example, stakeholders in a particular case might be hesitant about certain management strategies, but wish to explore them as possibilities. Under the guidance of a mediator, they could work through a hypothetical case in SAM, acting out the roles in PAQMAN. (The rationale for using a hypothetical case is that stakeholders might not wish to prejudice their positions in their real-world case.) The procedure would essentially follow the same recipe suggested above for SAM as an educational tool. As another example, a mediator with a set of proposed technical options in hand, and some understanding of the stakeholders' preferences, could invoke the decision sub-model of SAM in an effort to develop packages of options which could be agreed upon by all stakeholders. COOP could be used as a "negotiation spreadsheet"1 to define the space of possible agreements. This would build on a suggestion made by Dagnino et al. (1989) for the use of conflict analysis as a practical tool for mediation in environmental problems. PAQMAN could be run with a set of volunteer role-players. (The actual stakeholders might not wish to prejudice their positions by taking part in the role-playing.) This approach could be applied in a forward sense, to help the mediator to sort out possible agreements from complicated sets of options and preferences. It could also be applied in an inverse sense: the stakeholders' reactions to the possible agreements 1 Thanks to Jim Atwater for suggesting the phrase and concept of a "negotiation spreadsheet." 227 suggested by SAM could be used by the mediator to work backwards, critically evaluating the options and the mediator's understanding of the stakeholders' preferences. As a third example, SAM could be used to combine a range of decision system scenarios with alternative models of the pollution system, in order to produce a range of projections of the future course of management. This would be analogous to the approach taken in modeling of climate change, which combines a range of emission scenarios with a range of climate system sensitivities to produce a range of projections of future climate (IPCC, 1996). 7.3.4 Empirical study of groundwater pollution management This thesis provides a soft paradigm basis for empirical research into the real-world management of groundwater pollution. This could be pursued further by research into an active problem, by interviewing stakeholders and observing their interactions. This is likely to be a difficult task, given that many of the stakeholders will experience a fear of liability and desire for confidentiality. An alternative would be forensic research into a problem which has reached closure, by interviewing past stakeholders and reviewing available documents. There are drawbacks here as well: the persistence of groundwater pollution ensures that many problems are never really closed, and the fear of liability and desire for confidentiality may outlive the problem itself. 7.4 Concluding thought 228 In an overall sense, this thesis is a contribution to the ongoing debate about how to cope with the societal and environmental problem of groundwater pollution. The ideas presented here will not emerge unscathed from that debate. This is consistent with the soft paradigm, which views an iterative dialectic of diverse opinions and ideas as the appropriate response to a complex problem. 229 Glossary Term Definition (Italics indicate a cross-reference to another glossary entry.) algorithm A set of rules which transform an input into an output, without the exercise of human judgement. algorithmic model A type of transform model based on one or more algorithms. Always preceded by a conceptual model. analog model A type of transform model which represents some aspect of reality by creating an analogous situation; includes role-playing models. argumentation Methodical reasoning expressed through natural language, often supported by graphic visualization; informal logic. binary option In conflict analysis, a binary (yes-no) choice which is available to a given player. bounded rationality A concept which recognizes the real-world limitations on the axiom of rationality. case-specific conceptual model A conceptual model which corresponds to a to a particular example of a general set of phenomena. The particular example may be hypothetical or actual. common law Refers to law in an area that is governed only be previous judicial decisions. complexity Complexity is an intrinsic property of some aspect of the real world, correlated with the amount of knowledge or information which an observer would need to understand or describe that aspect. In the usage of this thesis, complexity may be quantitative or non-quantitative, and may be temporal or spatial. See uncertainty. compound model A type of transform model which consists of a set of linked sub-models; may include a mix of algorithmic and analog sub-models. conceptual model A synthesis of observations taken from the real world; presented in words, pictures, data, and mathematical expressions. Compare with transform model. conflict analysis A variant of game theory, used in this thesis. cooperative equilibrium In conflict analysis, an equilibrium is cooperative if it contains at least one cooperative option, and non-cooperative otherwise. cooperative option In conflict analysis, a shared binary option which must be either jointly selected or jointly rejected by all parties who share that option. decision strategy A strategy which applies to the decision system. decision system The subsystem of the management system which encompasses stakeholders, issues, and modes of decision-making. decision-making process An interaction among a subset of the stakeholders in a problem, using one or more modes of decision-making, with the intent of reaching decisions over a subset of issues. efficacy The quality of producing the desired management outcome. efficiency The quality of producing the desired management outcome with minimum waste. 230 emergent property A property of a system which cannot be understood or predicted merely from knowledge of the elements of the system. equilibrium A possible stable outcome of a game in conflict analysis, given the players' preferences. game theory A type of algorithmic model; a mathematical abstraction of multilateral decision-making, in which players (decision-makers) participate in a game by selecting moves, subject to the rules of the game and according to their preferences, resulting in outcomes which have associated payoffs for each player. goal-setting The quality of knowing the desired management goals. government policy General courses of action established by the executive and administrative branches of government. groundwater pollution The anthropogenic impairment of the quality of subsurface water. hard paradigm A paradigm for the study of decision-making and management which is grounded in mathematics, engineering, and the natural sciences, and is the basis for fields such as operations research, systems analysis, decision analysis, policy analysis, and policy science. Embraces concepts such as objectivity, reductionism, rationality, and unitary decision-making. A hard paradigm approach is willing to accept a high degree of idealization in the conceptual model, in exchange for the ability to apply mathematically-rigorous algorithmic models, especially algorithmic models of decision-making such as optimization. See hard problem; soft paradigm. hard problem A problem as viewed in the hard paradigm, characterized by a relatively high degree of idealization. Characterized by: characterized by: clear boundaries and linkages with other problems; clear goals, alternatives, and consequences; limited uncertainty which can be quantified; and a linear, unilateral mode of decision-making. heuristic Adjective describing something which allows or assists in discovery or learning; also carries a connotation of learning by trial-and-error. holism The effort to understand some aspect of the real-world as a whole, allowing for emergent properties, inter-related problems, and complex systemic causality (as opposed to linear causality). See reductionism. iterative decision-making Decision-making which is repetitive and open-ended, moving towards a condition of closure, but with no set endpoint. judicial decision-making A formal decision-making process in which decisions are made by a judge or jury. See litigation. jurisprudence Rules established by previous judicial decisions, and, to a lesser degree, by decisions of quasi-judicial administrative tribunals established under specific statutory authority. legislation A term which includes both statutes and regulations; rules established by the legislative branch of government, or under authority delegated by the legislative branch of government. Legislation is applied by the administrative branch of government, and interpreted by the judicial branch. liability A legal obligation or responsibility. litigation Judicial decision-making in which a set of one or more parties brings a lawsuit against another set of one or more parties, with a judge or jury deciding the outcome. 231 management In this thesis, a broad concept involving multiple stakeholders who must make and implement decisions to cope with a persistent and frustrating problem. The relative efficiency of various technical options is just one aspect of management. management system The human system which makes and implements decisions to cope with a groundwater pollution problem. mediation The provision of assistance to a negotiation process by a neutral party, in matters of both process and substance. mitigation The reduction of harmful impacts, by partial removal and/or containment of pollutants, as a technical goal. Compare to restoration. mode of decision-making A classification of decision-making according to the distribution of decision-making authority. In negotiation, decision-making authority is shared among a subset of at least two stakeholders. In unilateral decision-making, authority is vested in one stakeholder or outside party. model A n abstraction of some aspect of reality; a means of achieving some purpose (e.g., organizing knowledge, generating knowledge, communicating knowledge, or prescribing a course of action). negotiation A mode of decision-making in which decision-making authority is shared among stakeholders. objective criteria A set of unambiguous standards forjudging the success of a negotiated or unilateral decision; may apply to outcomes or to actions. In the context of high complexity and uncertainty, objective criteria should apply to ranges rather than single values. objective-function model A type of algorithmic model; a formalized abstractions of human decision-making, in which the objective is represented as a mathematical function to be maximized or minimized. optimization A type of objective-function model, consisting of constraints (a set of inequalities representing real-world limitations) and an objective function (a function to be optimized, representing a real-world goal). The variables in the constraints and objective function are called decision variables. The output consists of the optimal decision variables and the optimal value of the objective function. outcome In conflict analysis, a combination of all players' options in which each option is either chosen or rejected; for a game with n options, there are 2" mathematically possible outcomes. In management, the results of making and implementing decisions. paradigm "...a constellation of beliefs, values, procedures, and past scientific achievements that is shared within a community of scientists, and is learned during their training or in their common research experiences. These shared group commitments provide a framework that highlights specific scientific problems, restricts possible solution tactics, and establishes criteria for the rapid evaluation of solutions." ( Stewart, 1990, p.4) 232 Pareto optimality In conflict analysis, a given equilibrium is Pareto optimal if there is no cooperative equilibrium which all players prefer to the given equilibrium. In negotiation, an agreement could be called Pareto optimal if there are no further gains that any of the parties could make at no loss to any of the other parties, i.e., if "nothing is left on the table." participant The subset of stakeholders in a problem who interact in a given process. process See decision-making process. rationality A n axiom of the hard paradigm: given a set of possible alternatives to choose from, a decision-maker will select the alternative for which he or she has the highest preference. Rationality can be defined as the maximizing of utility. reductionism The effort to understand some aspect of the real-world by understanding its parts. See holism. regulations Detailed rules created by the executive or administrative branches under the authority of a specific statute; the authority is delegated by the legislative branch. unilateral regulatory decision-making Unilateral decision-making by a regulatory stakeholder, subject to the rules of governance; generally involves some communication of information and arguments from other stakeholders to the regulatory stakeholder. restoration The nearly complete removal of pollutants and return to nearly-pristine conditions, as a technical goal. Compare with mitigation. role-playing model A type of analog model which can be applied to human situations; the human interactions in a situation of interest are simulated by humans acting out the roles of the various parties, subject to both the formal rules of the game and the subjective judgment of the role-players. In the literature, commonly referred to as "gaming" or "gaming-simulation". rules of governance A set of complex and linked rules, established by various levels of government, including legislation (statutes and regulations), jurisprudence, and government policy. satisficing An alternative to optimization which ignores optimality and focuses on constraints; a methodological expression of bounded rationality. simulation model A type of algorithmic model; represents the behaviour of some part of the real world, such as the flow of groundwater in an aquifer or the flow of electricity in a circuit. soft paradigm A paradigm for the study of decision-making and management which has arisen as an alternative to the hard paradigm. Moves away from the norms of the hard paradigm, tending towards subjectivity, holism, bounded rationality, and pluralistic decision-making; relies on arguments rather than algorithms. A soft-systems approach does not accept a high degree of idealization in the conceptual model, and is therefore less compatible with mathematically-rigorous algorithmic models. See soft problem; hard paradigm. soft problem A problem as viewed by the soft paradigm; less idealized than a hard problem. Characterized by ambiguous boundaries and complex linkages with other problems; goals, alternatives, and consequences which are not well-defined or well-understood; pervasive uncertainty which may not be quantifiable; and iterative management which involves conflict and negotiation among multiple stakeholders with divergent interests and values. 233 stakeholder A party with an interest in a given situation or problem. M a y be singular or collective; treated as a collective noun in this thesis. statute A formal set of rules passed into law by the legislative branch of government. strategy A generalized, relatively abstract plan for coping with a particular aspect of groundwater pollution management system A complex entity composed of interacting parts and contained within an environment. A system is characterized by emergent properties and hierarchical organization, with distinct subsystems in each layer of the hierarchy. technical strategy A strategy which applies to the technical system. technical system The subsystem of the management system which encompasses the technical aspects of management. transform model A model which transforms or maps an input into an output; in distinction to a conceptual model. Includes algorithmic models, analog models, and compound models. uncertainty Uncertainty is a function of an observer's ignorance about some aspect of the real world. It is not an intrinsic property of that aspect of the real world. In the usage of this thesis, uncertainty may be quantitative or non-quantitative, and may be temporal or spatial. See complexity. unilateral Adjective describing a situation or action involving a single stakeholder, e.g., a stakeholder may have a unilateral alternative of entering into litigation against another stakeholder. unilateral decision-making A mode of decision-making in which decision-making authority is vested in a single stakeholder. utility A real-valued function which describes preferences and is typically a surrogate for money. 234 References Ahlfield, D., J. Mulvey, G. Pinder, and E. Wood, 1988, Contaminated groundwater remediation design using simulation, optimization, and sensitivity theory, 1, Model development, Water Resources Research, 24(3), 431-441. Ahlfield, D.P., and Sawyer, C.S., 1990, Well location in capture zone design using simulation and optimization techniques, Ground Water, 28(4), 507-512. Alston, M . , 1993, Improving Superfund conflicts: views of an EPA Superfund program director, Working Paper 93-2, Conflict Resolution Consortium, University of Colorado, Boulder, CO. Amy, D., 1987, The politics of environmental mediation, Columbia University Press, New York. Anderson, M.P., and W.W. Woessner, 1992, Applied groundwater modeling: simulation of flow and advective transport, Academic Press, San Diego. Andricevic, R., and P.K. Kitanidis, 1989, Stochastic control of contaminant transport problems, p. 271-284 in Sahuquillo, A. , J. Andreu, and T. O'Donnell, Groundwater management: quantity and quality, Proceedings of the Benidorm Symposium, October 1989, IAHS Publication no. 189. Appelbe, A. , 1996, Residents urged to agree over traffic, Vancouver Courier, 3 April, 1996, p. 6. Barnett, H.C. , 1994, Toxic debts and the Superfund dilemma, University of North Carolina Press, Chapel Hill. Bastian, R.K., and D.A. Hammer, 1993, The use of constructed wetlands for wastewater treatment and recycling, p. 59-68 in Moshiri, G.A. , ed., 1993, Constructed wetlands for water quality improvement, Lewis Publishers, Boca Raton. Bevan, A . H . , 1992, Alternative dispute resolution: a lawyer's guide to mediation and other forms of dispute resolution, Sweet and Maxwell, London. Bingham, G. , 1986, Resolving environmental disputes, The Conservation Foundation, Washington, D.C. Blair, E.S., 1994, Model (7«jvalidation — a view from the courtroom, Ground Water 32(4), 530-531. B M A [British Medical Association, 1991], Hazardous waste and human health, Oxford University Press, Oxford. Brennan, M . A . , and J.E. Frazier, 1991, Disposition of environmental disputes without litigation, Ch. 24, p. 263-352 in Carpenter, D.A. , R.F. Cushman, and B.W., Roznowksi, eds., 1991, Environmental dispute handbook: liability and claims, Vol . 2, John Wiley and Sons, New York. Breyer, S., 1993, Breaking the vicious circle: toward effective risk regulation, Harvard University Press, Cambridge, Mass. California Anti-SLAPP Project, 1996, California Anti-SLAPP Project Home Page, http://www.sirius.eom/~casp/intro.html#top, February 5, 1996 edition. Carson, R., 1962, Silent spring, Fawcett, Greenwich, Connecticut. Castel, J.R., and O.K. Latchman, 1994, The practical guide to Canadian legal research, revised edition, Carswell, Scarborough, Ont. Casti, J., 1979, Connectivity, complexity, and catastrophe in large-scale systems, Wiley-Interscience, Chichester. Catron, B.L. , L . G . Boyer, J. Grund, and J. Hartung, 1996, The problem of intergenerational equity: balancing risks, costs and benefits fairly across generations, p. 131-148 in Cothern, C.R., (ed.), 1996, Handbook for environmental risk decision-making: values, perceptions, and ethics, Lewis Publishers, Boca Raton. C H 2 M Hill, 1990a, Public review draft, basinwide technical plan report, San Gabriel Basin, Los Angeles, California, volumes 1 and 2. EPA Contract No. 68-01-7251; EPA Work Assignment No. 14-9L27.1; C H 2 M Hill Project No. LAO62440.TP. C H 2 M Hill, 1990b, Cost of remedial action model: users manual, United States E P A and Office of Solid Waste and Emergency Response Contract No. 68-01-7090. Chapell, W., 1993, The Smuggler Mountain Technical Advisory Committee: a prototype procedure for resolving some Superfund conflicts, Working Paper 93-1, Conflict Resolution Consortium, University of Colorado, Boulder, C O . Chapelle, F .H. , 1993, Ground-water microbiology and geochemistry, John Wiley and Sons, New York. 235 Checkland, P., 1981, Systems thinking, systems practice, John Wiley and Sons, Chichester. Checkland, P., and J. Scholes, 1990, Soft systems methodology in action, John Wiley and Sons, Chichester. Cherry, J.A., 1993, Groundwater cleanup & protection: Canadian commitment lags behind the United States and Mexico, Hazardous Materials Management, June 1993, 6-8. Church, T.W., and R.T. Nakamura, 1993, Cleaning up the mess: implementation strategies in Superfund, The Brookings Institution, Washington, D.C. Clark, W.C., 1978, Managing the unknown: an ecological view of risk assessment, University of British Columbia Institute of Resource Ecology Working Paper 26,. Cocklin, C , 1989a, Mathematical programming and resources planning I: the limitations of traditional optimization, Journal of Environmental Management 28, 127-141. Cocklin, C , 1989b, Mathematical programming and resources planning II: new developments in methodology, Journal of Environmental Management 28, 143-156. Cohon, J., G. Scavone, and R. Solanki, 1988, Multicriterion optimization in resources planning, Ch.5. p. 117-160, in Stadler, W., ed., 1988, Multicriteria optimization in engineering and in the sciences, Plenum Press, New York. Colburn, T.E. , A . Davidson, S.N. Green, R.A. Hodge, C.I. Jackson, and R.A. Liroff, 1990, Great Lakes, great legacy?, The Conservation Foundation, Washington, D . C , and the Institute for Research on Public Policy, Ottawa. Corcoran, T., 1994, Junk science looms over Great Lakes, The Globe and Mail, 18 February, 1994, p. B2. Cormick, G.W., 1988, Where, when and how to use mediated negotiations: a checklist for the potential participant, Canadian Environmental Mediation Newsletter, 5(1), 7-9. Cormick, G.W., 1989, Strategic issues in structuring multi-party public policy negotiations, Negotiation Journal, 5(2), 125-132. Cormick, G.W., and A . Knaster, 1986, Mediation and scientific issues, Environment 25(10), 6-16. Cothern, C.R., (ed.), 1996, Handbook for environmental risk decision-making: values, perceptions, and ethics, Lewis Publishers, Boca Raton. Dagnino, A. , K.W. Hipel, and N . M . Fraser, 1989, Computer assistance in environmental mediation, Systems Practice, 2(2), 181-197. Deutsch, C.V. , and Journel, A. , 1992, GSLIB: geostatistical software library and user's guide, Oxford University Press, New York. Dolin, E.J., and L .E . Susskind, 1992, A role for simulations in public policy disputes: the case of national energy policy, Simulation and Gaming, 23(1), 20-44. Domenico, P.A., and F.W. Schwartz, 1990, Physical and chemical hydrogeology, John Wiley and Sons, New York. Dorcey, A.H.J. , 1986, Bargaining in the governance of Pacific coastal resources: research and reform, Westwater Research Centre, University of British Columbia. Dorcey, A.H.J. , and C L . Riek, 1987, The place of negotiation in environmental assessment, report prepared for the Canadian Environmental Assessment Research Council, Hull, Que. Duke, R.D., and N.K. Kemeny, 1989, Keeping score one score later: two decades of the Simulation & Games journal. Simulation and Games, 20(2), 165-183. Edelstein, M.R., 1988, Contaminated communities : the social and psychological impacts of residential toxic exposure, Westview Press, Boulder, Colo. EPA [U.S. Environmental Protection Agency], 1987, Unfinished business: a comparative assessment of environmental problems, Volume 1: Overview, EPA 230/2-87/025a. EPA [U.S. Environmental Protection Agency], 1989, Wellhead protection programs: tools for local governments, EPA 440/6-89-002. Fischer, F. and Forester, J., eds., 1993, The argumentative turn in policy analysis and planning, Duke University Press, Durham and London. Fisher, R., and S. Brown, 1988, Getting together, Penguin Books, New York. Fisher, R., and W. Ury, 1981, Getting to yes, Penguin Books, New York. 236 Fisher, R., E. Kopelman, and A . K . Schneider, 1994, Beyond Machiavelli: tools for coping with conflict. Harvard University Press, Cambridge, Mass. Fisher, R., W. Ury, and B. Patton, 1991, Getting to yes (Second edition), Penguin Books, New York. Flood, R.L., and Carson, E.R., 1993, Dealing with complexity: an introduction to the theory and application of systems science, 2nd edition. Plenum Press, New York. Fogleman, V . M . , 1992, Hazardous waste cleanup, liability, and litigation: a comprehensive guide to Superfund law, Quorum Books, Westport, Conn. Ford, G. , D. Macdonald, and M . Winfield, 1994, Who pays for past sins? Policy issues surrounding contaminated site remediation in Canada, Alternatives 20(4), 28-34. Francis, B . M . , 1994, Toxic substances in the environment, John Wiley and Sons, New York. Francis, R., 1995, Crossed signals confuse Blenheim traffic woes, The Kitsilano News, July 26, 1995. Fraser, N . M . , and K.W. Hipel, 1984, Conflict analysis: models and resolutions, North-Holland, New York. Freeze, R.A., 1988, Technical analysis and social decision-making (Hagey Lectures; 1987-88), University of Waterloo Press, Waterloo. Freeze, R.A., and Cherry, J.A., 1979, Groundwater, Prentice-Hall, Inc., Englewood Cliffs, NJ. Freeze, R.A., and J.A. Cherry, 1989, What has gone wrong, Ground Water, 27(4), 458-464. Freeze, R.A., B. James, J. Massman, T. Sperling, and L. Smith, 1992, Hydrogeological decision analysis: 4. The concept of data worth and its use in the development of site investigation strategies, Ground Water 30(4), 574-588. Freeze, R.A., J. Massman, L. Smith, T. Sperling, and B. James, 1990, Hydrogeological decision analysis: 1. A framework, Ground Water, 28(5), 738-766. Gilbert, G.C. , Alternative dispute resolution and Superfund: a research guide, Ecology Law Quarterly, 16(3), 803-856, 1989.-Golden Software, Inc., 1989, SURFER Reference Manual, Golden Software, Inc., Golden, Colo. Golden, B.L., E .A. Wasil, and P.T. Harker (eds.), 1989, 77*e analytical hierarchy process: applications and studies, Springer-Verlag, Berlin. Gorelick, S.M., 1982, A model for managing sources of groundwater pollution, Water Resources Research, 18(4), 773-781, 1982. Gorelick, S.M., 1983, A review of distributed parameter groundwater management modeling methods, Water Resources Research, 19(2), 305-319, 1983. Gorelick, S.M., C.I. Voss, P.E. Gill, W. Murray, M . A . Saunders, and M . H . Wright, 1984, Aquifer reclamation design: The use of contaminant transport simulation coupled with nonlinear programming, Water Resources Research, 20(4), 415-427. Grayson, R.B., J .M. Doolan, and T. Blake, 1994, Application of A E A M (Adaptive Environmental Assessment and Management) to water quality in the Latrobe River catchment, Journal of Environmental Management, 41(3), 245-258. Greenberg, M . , D. Schneider, and J. Martell, 1994, Hazardous waste sites, stress, and neighbourhood quality in USA, The Environmentalist 14(2), 93-105. Greenblatt, C.S. 1988, Designing games and simulations: an illustrated handbook. Sage Publications, Newbury Park, Calif. Greene, M . , 1994, The Rocky Mountain Arsenal: states' rights and the cleanup of hazardous waste, Working Paper 94-58, Conflict Resolution Consortium, University of Colorado, Boulder, CO. Gunderson, L . H . , C.S. Holling, and S.S. Light, eds., 1995, Barriers and bridges to the renewal of ecosystems and institutions, Columbia University Press, New York. Hafele, W., 1981, Energy in a finite world: a global systems analysis, Ballinger, Cambridge, Mass. Harrison, K., 1993, Is cooperation the answer? A comparison of Canadian and U.S. enforcement of environmental regulations, talk given to 1993 Annual Meeting of the American Political Science Association, Washington, D . C , 2-5 September, 1993. 237 Hartig, J.H., and M . A . Zarull, 1992, A Great Lakes mission, p. 5-36 in Hartig, J.H., and M . A . Zarull, eds., 1992, Under RAPs: Towards grassroots ecological democracy in the Great Lakes Basin, University of Michigan Press, Ann Arbor. Hill, M . C . 1990, Preconditioned Conjugate-Gradient 2 (PCG2), A computer program for solving ground-water flow equations, Water-Resources Investigations Report 90-4048, U.S. Geological Survey, Denver, Colorado. Hird, J.A., 1994, Superfund: the political economy of environmental risk, The John Hopkins University Press, Baltimore. Hitt, J., 1995, Toxic dreams: a California town finds meaning in an acid pit, Harper's Magazine, 291(1742), 57-66. Hobbs, B.F., and A . Hepenstal, 1989, Is optimization optimistically biased?, Water Resources Research, 25(2), 152-160. Hoberg, G. , 1992, Pluralism by design: environmental policy and the American regulatory state, Praeger, New York. Hoberg, G. , 1993, Environmental policy: alternative styles, p. 307-342 in Atkinson, M . , ed., 1993, Governing Canada: state institutions and public policy, HBJ Holt, Toronto. Hofmann, P., 1994, The case of the immodest modeler: uncertainty and realism in groundwater modeling, New Mexico Conference on the Environment, Journal of Proceedings, Albuquerque, NM, April 24-26, 1994. Holling, C.S., A . Bazykin, P. Bunnell, W.C. Clark, G.C. Gallopin, J. Gross, R. Hilborn, D.D. Jones, R . M . Peterman, J.E. Rabinovich, J.H. Steele, and C.J. Walters, 1978, Adaptive environmental assessment and management, John Wiley and Sons, Chichester. Hoos, I.R., 1983, Systems analysis in public policy: a critique (revised edition), University of California Press, Berkeley. Horgan, J., 1995, From complexity to perplexity, Scientific American, 270(6), 104-109. Hughes, F., 1995, Barriers bedevil Blenheim, Vancouver Courier 86(60), 1. IJC [International Joint Commission], 1994, Seventh biennial report on Great Lakes water quality, Washington, D.C. IPCC (Intergovernmental Panel on Climate Change), 1996, IPCC Climate Change 1995: IPCC Second Assessment Report, http://www.unep.ch/ipcc/ipcc95 .html. James, B. R., and R.A. Freeze, 1993, The worth of data in predicting aquitard continuity in hydrogeological design, Water Resources Research, 29(1), 2049-2065. James, B.R., and S.M. Gorelick, 1994, When enough is enough: The worth of monitoring data in aquifer remediation design, Water Resources Research, 30(12), 3499-3513. Johnson, B.L., 1993, ATSDR: public health actions andfindings, Congressional testimony before the Subcommittee on Superfund, Recycling, and Solid Waste Management of the United States Senate, 6 May, 1993. Johnson, B.L., 1995, ATSDR: public health actions andfindings (Update '95), Congressional testimony before the Subcommittee on Superfund, Waste Control, and Risk Assessment, Committee on Environment and Public Works of the United States Senate, 29 March, 1995. Jorgensen, E.P., ed., 1989, The poisoned well: new strategies for groundwater protection, Island Press, Washington, D C . Kaunas, J.R., and Y . Y . Haimes, 1985, Risk management of groundwater contamination in a multiobjective framework, Water Resources Research 27(11), 1721-1730. Keepin, B., 1984, A technical appraisal of the IIASA energy scenarios, Policy Sciences 17(3), 199-275. Keys, P., 1991, Operational research and systems: the systemic nature of operational research, Plenum Press, New York. T 57.6 K448 1991 Knight, R.L., R.W. Ruble, R.H. Kadlec, and S. Reed, 1993, Wetlands for wastewater treatment: performance database, p. 35-58 in Moshiri, G.A. , ed., 1993, Constructed wetlands for water quality improvement, Lewis Publishers, Boca Raton. Konikow, L.F., and J.D. Bredehoeft, 1992, Groundwater models cannot be validated, Advances in Water Resources, 15(1), 75-83. Kopp, R.J., and V . K . Smith, eds., 1993, Valuing natural assets: the economics of natural resource damage assessment, Resources for the Future, Washington, D.C. 238 Kuhn, T., 1970, The structure of scientific revolutions, 2nd edition, The University of Chicago Press, London. Lax, D.A. , and J.K. Sebenius, 1986, 77?e manager as negotiator: bargaining for cooperation and competitive gain, The Free Press, New York. Lee, K . N . , 1993, Compass and gyroscope: integrating science and politics for the environment, Island Press, Washington, D C . Lefkoff, J.L., and S.M. Gorelick, 1987, AQMAN: linear and quadratic programming matrix generator using two-dimensional ground-water flow simulation for aquifer management modeling. U.S. Geological Survey Water-Resources Investigations Report 87-4061, Menlo Park, California. Lewontin, R.C., 1991, Biology as ideology: the doctrine of DNA. Anansi, Concord, Ont. Lybarger, J., 1993, Superfund sites: adverse health impacts, Congressional testimony before the Subcommittee on Superfund, Recycling, and Solid Waste Management of the United States Senate, 12 April, 1993. Lynch, L., 1994, The value of community outreach, Article V in Insight Press, 1994, Clean-up of contaminated sites: financing, regulatory guidelines and state-of-the-art technologies, Insight Press, Ontario. Maclntyre, M . E . , 1996, Tar-ponds burial slagged: government cleanup plan almost sparks Sydney riot, Halifax /feraWZ/7wte^[http://www.herald.ns.ca/.index.html], 16 January, 1996. Mackay, D . M . , S. Feenstra, and J.A. Cherry, 1993, Alternative goals and approaches for groundwater remediation, p. 35-47 in Neretnieks, I., (ed.), 1993, Proceedings of the Workshop on "Contaminated Soils - Risks and Remedies", Stockholm, 26-28 October, 1993, Dept. of Chemical Engineering and Technology, Royal Institute of Technology, Stockholm, Sweden. MacLeod, S., 1996, Nova Scotia to cover over toxic tar ponds, M P says, Vancouver Sun, 12 January, 1996. Maddock, T. III. 1972a. Algebraic technology function from a simulation model, Water Resources Research, 8(1), 129-134. Maddock, T. III. 1972b. Nonlinear technological functions for aquifers whose transmissivities vary with drawdowns, Water Resources Research,l0(4): 877-881. Mahood, C , 1992, Laidlaw settles landfill spat: $100-million assurance against problems with waste, Globe and Mail, 7 October.) 992. Majone, G. , 1989, Evidence, argument and persuasion in the policy process, Yale University Press, New Haven. Major, D.C. , 1977, Multiobjective water resource planning, American Geophysical Union, Washington, D.C. Mason, R.O., and I.I. Mitroff, 1981, Challenging strategic planning assumptions, John Wiley and Sons, New York. Massman, J., and R.A. Freeze, 1987a, Groundwater contamination from waste management sites: the interaction between risk-based engineering design and regulatory policy 1. Methodology, Water Resources Research, 23(2), 351-367. Massman, J., and R.A. Freeze, 1987b, Groundwater contamination from waste management sites: the interaction between risk-based engineering design and regulatory policy 2. Results, Water Resources Research, 23(2), 368-380. Massmann, J., R.A. Freeze, and L. Smith, 1991, Hydrogeological decision analysis: 2. applications to ground-water contamination, Ground Water 29(4), 536-548. Matas, R., 1995, Waste site cleanup in limbo, Globe and Mail, 28 December, 1995. Matas, R., 1996, Environment pact transfers power to provinces, Globe and Mail, 20 January, 1996, p. A7. Maughan, J.T., ed., 1993, Ecological assessment of hazardous waste sites, Van Nostrand Reinhold, New York. Q H 541.15 R57M38 1993 W D Mautner-Markhof, F., 1989, The reality, management and simulation of complex systems, p. 24-37 in Klabbers, J.H.G., W.J. Sheper, C.A.Th. Takkenberg, and D. Crookall, eds., 1989, Simulation-gaming: on the improvement of competence in dealing with complexity, uncertainty and value conflicts, Pergamon Press, Oxford. Mazmanian, D. , and D. Morell, 1992, Beyond Superfailure: America's toxics policy for the 1990s, Westview Press, Boulder, Colo. McDonald, M . G . , A .W. Harbaugh, B.R. Orr, and D.J. Ackerman, 1991, A method of converting no-flow cells to variable-head cells for the U.S. Geological Survey modular finite-difference ground-water flow model, Open-File Report 91-536, U.S. Geological Survey, Washington, D.C. 239 McDonald, M . G . , and A.W. Harbaugh, 1988, A modular three-dimensional finite-difference ground-water flow model, Techniques of Water-Resources Investigations, Book 6, Chapter A l , U . S. Geological Survey, Washington, D.C. McKenna, B., 1993, Laidlaw wants to let toxins in Mercier lagoon dissolve, The Globe and Mail, 2 April, 1993. McKinney, D . C , and M . - D . Lin, 1994, Genetic algorithm solution of groundwater management models, Water Resources Research, 30(6), 1897-1906. McLaughlin, D., and W.K. Johnson, 1987, Comparison of three groundwater modeling studies, Journal of Water Resources Planning and Management 113(3), 405-421. McNamara, 1996, Tar ponds assessment will put feds on case, Halifax Herald Limited [http://www.herald.ns.ca/.index.html], 1 February, 1996. Mercer, J.W., D.R. Buss, D.S. Ward, and S.S. Hughes, 1989, Use of groundwater models and the settlement process at the Chem-dyne hazardous waste site, p. 555-565 in Jousma, G. , J. Bear, Y .Y . Haimes, and F. Walter, eds., 1989, Groundwater contamination: use of models in decision-making. Kluwer Academic Publishers, Dordrecht. Meyer, P.D., A.J. Valocchi, and J.W. Eheart, 1994, Monitoring network design to provide initial detection of groundwater contamination, Water Resources Research, 30(9), 2647-2659. Miser, H.J., and E.S. Quade, 1985, The context, nature, and use of systems analysis, p. 1-32 in Miser, H.J., and E.S. Quade, eds., 1985, Handbook of systems analysis: overview of uses, procedures, applications, and practice, John Wiley and Sons, Chichester. Miser, H.J., and Quade, E.S., 1988, Introduction: craftsmanship in analysis, p. 1-26 in Miser, H.J., and Quade, E.S., eds., 1988a, Handbook of systems analysis: craft issues and procedural choices, John Wiley and Sons, Chichester. Morgan, M . G . , and M . Henrion, 1990, Uncertainty: a guide to dealing with uncertainty in quantitative risk and policy analysis, Cambridge University Press, Cambridge, U K . Murtagh, B.A. , and M . A . Saunders, 1987, MINOS 5.1 user's guide, Technical Report SOL 83-20R, Systems Optimization Laboratory, Department of Operations Research, Stanford University, Calif. Nordstrom, D.K. , and C N . Alpers, 1995, Remedial investigations, decisions and geochemical consequences at Iron Mountain Mine, California, p. 633-642 in Hyne, T.P., and M . C . Blanchette, (eds.), 1995, Proceedings of Sudbury '95 — Mining and the environment, May 28 - June 1, 1995, Sudbury, Ontario, Canada, C A N M E T , Ottawa. N R C [National Research Council (U.S.), Committee on Environmental Epidemiology], 1991, Environmental epidemiology, Volume 1: Public health and hazardous wastes, National Academy Press, Washington, D.C. N R C [National Research Council (U.S.), Committee on Ground Water Cleanup Alternatives], 1994, Alternatives for ground water cleanup, National Academy Press, Washington, D.C. Nyhart, J.D., and E .A. Dauer, 1986, A preliminary analysis of the uses of scientific models in dispute prevention, management and resolution, Missouri Journal of Dispute Resolution 1986,29-53. Oreskes, N. , K. Schrader-Frechette, and K. Belitz, 1994, Verification, validation, and confirmation of numerical models in the earth sciences, Science 263, 641-646. Ozawa, C P . , and L. Susskind, 1985, Mediating science-intensive policy disputes, Journal of Policy Analysis and Management 5(10), 23-39. Parson, E.A. , and W.C.Clark, 1995, Sustainable development as social learning, p. 428-460 in Gunderson, L . H . , C S . Holling, and S.S. Light, eds., 1995, Barriers and bridges to the renewal of ecosystems and institutions, Columbia University Press, New York. Peck, R.B., 1969, Advantages and limitations of the observational method in applied soil mechanics, Geotechnique 19(2), 171-187. Peterle, T.J., 1991, Wildlife toxicology, Van Nostrand Reinhold, New York. Pinder, G.F., D.P. Ahlfeld, and R.H. Page, 1995, Cleanup solution — conflict resolution, Civil Engineering, March 1995,59-61. Pollock, D.W., 1989, Documentation of computer programs to compute and display pathlines using results from the U.S. Geological Survey modular three-dimensional finite difference ground-water flow model, U.S. Geological Survey Open-File Report 89-381. 240 Probst, K . N . , and P.R. Portney, 1992, Assigning liability for Superfund cleanups: an analysis ofpolicy options, Resources for the Future, Washington, D.C. Pruzan, P., 1988, System OR and operational systems science, European Journal of Operational Research, 37, 34-41. Raiffa, H. , 1982, The art and science of negotiation, Harvard University Press, Cambridge, Mass., 1982. Rapoport, A. , 1970, Modern systems theory — an outlook for coping with change, General Systems Yearbook 15, 15-25. Rittel, H.W.J., and W . M . Webber, 1973, Dilemmas in a general theory of planning, Policy Sciences 4, 155-169. Roberts, S., 1995, Blenheim neighbour feels left out [letter to the editor], Vancouver Courier, 26 August, 1995, p. 4. Rogers, P.P., and M.B . Fiering, 1986, Use of systems analysis in water management, Water Resources Research, 22(9), 146S-158S. Sawyer, C.S., D.P. Ahlfeld, and A.J. King, 1995, Groundwater remediation design using a three-dimensional simulation model and mixed-integer programming, Water Resources Research 31(5), 1373-1385. Scarrow, J.W., 1989, The use of modeling in groundwater contamination cases, Virginia Environmental Law Journal, 9(1), 185-206. Schrader-Frechette, K.S., 1991, Risk and rationality: philosophical foundations for populist reforms, University of California Press, Berkeley. Schwarz, B., 1988, Forecasting and scenarios, p. 327-367 in Miser, H.J., and Quade, E.S., eds., 1988, Handbook of systems analysis: craft issues and procedural choices, John Wiley and Sons, Chichester. Schwarz, B., K . C . Bowen, I. Kiss, and E.S. Quade, 1985, Guidance for decision, p. 219-247 in Miser, H.J., and E.S. Quade, eds., 1985, Handbook of systems analysis: overview of uses, procedures, applications, and practice, John Wiley and Sons, Chichester. Sebenius, J.K., 1984, Negotiating the Law of the Sea, Harvard University Press, Cambridge, Mass. Sheahan, PA., and M . E . Cherry, 1993, Ore deposit models, vol. II, Geoscience Canada Reprint Series 6. Shimp, J.F., J.C. Tracy, L . C . Davis, E. Lee, W. Huang, and L.E . Erikson, 1993, Beneficial effects of plants in the remediation of soil and groundwater contaminated with organic materials, Critical Reviews in Science and Technology, 23(1), 44-77. Sidnell, E.J., 1993, Property assessment of contaminated land in Alberta, Journal of Environmental Law and Practiced, 231-268. Simon, H.A. , 1981, The sciences of the artificial, 2nd edition, MIT Press, Cambridge, Mass. Sperling, T., R.A. Freeze, and J. Massman, 1992, Hydrogeological decision analysis: 3. application to design of a ground-water control system at an open pit mine, Ground Water 30(3), 376-389. Sperling, T., R.A. Freeze, L. Smith, and J. Massman, 1993, COAST: An economic risk-cost-benefit model for assessing groundwater remediation options at contaminated sites. Manual accompanying C O A S T software. Stahl, I., 1988, Using operational gaming, p.122-171 in Miser, H.J., and Quade, E.S., 1988, Handbook of systems analysis: craft issues and procedural choices, John Wiley and Sons, Chichester. Stewart, J.A., 1990, Drifting continents and colliding paradigms: perspectives on the geoscience revolution, Indiana University Press, Bloomington and Indianapolis. Susskind, L., and J. Cruikshank, 1987, Breaking the impasse: consensual approaches to resolving public disputes, Basic Books, New York. Szasz, A. , 1994, EcoPopulism: toxic waste and the movement for environmental justice, University of Minnesota Press, Minneapolis. Terzaghi, K., and R.B. Peck, 1967, Soil mechanics in engineering practice, 2nd edition, John Wiley and Sons, New York. Thomas, L . C , 1984, Games, theory and applications, Ellis Horwood, Chichester. Throgmorton, J.A., 1991, The rhetoric of policy analysis, Policy Sciences 24, 153-159. Todd, D.K. , and D.E.O. McNulty, 1976, Polluted groundwater: a review of the significant literature, Water Information Center, Huntington, N Y . 241 Travis, C.C. and C.B. Doty, 1990, Can contaminated aquifers at Superfund sites be remediated?, Environmental Science and Technology 24(10), 1464-1466. Underhill, B., 1996, Reform calls for federal study of tar ponds plan, Halifax Herald Limited [http://www.herald.ns.ca7.index.html], 18 January, 1996, Ury, W., 1991, Getting past no: negotiating with difficult people, Bantam Books, New York. Vyner, H . M . , 1988, Invisible trauma: the psychosocial effects of invisible environmental contaminants, Lexington Books, Lexington, Mass. Wagner, W.J., and S.M. Gorelick, 1987, Optimal groundwater quality management under parameter uncertainty, Water Resources Research, 23(7), 1162-1174. Wagner, W.J., and S.M. Gorelick, 1989, Reliable aquifer remediation in the presence of spatially variable hydraulic conductivity: from data to design, Water Resources Research, 25(10), 2211-2225. Walker, R.G., 1986, General introduction: facies, facies sequences, and facies models, p.1-9 in Walker, R.G., ed., 1986, Facies models, second edition. Geoscience Canada, Reprint Series 1. Walters, C , 1986, Adaptive management of renewable resources, MacMillan, New York. Walton, D.N. , 1989, Informal logic: a handbook for critical argumentation, Cambridge University Press, Cambridge. Wartenberg, D., 1988, Groundwater contamination by Temik Aldicarb pesticide: the first 8 months, Water Resources Research, 24(2), 185-194. Weaver, W., 1948, Science and complexity, American Scientist, 36, 536-544. Weinberg, A . M . , 1972a, Science and trans-science, Minerva 10, 209-222. Weinberg, A . M . , 1972b, Science and trans-science, Science 177, (4045), 211. Weinberg, A . M . , 1985, The regulators dilemma, Issues in Science and Technology 2(1), 59-72. Whiting, B., 1994, Broad status for suit over groundwater: thousands could seek damage over pollution, Arizona Republic, 26 August, 1994, p. A2. Willis, R., and W.W-G. Yeh, 1987, Groundwater systems planning and management, Prentice-Hall Inc., Englewood Cliffs, N.J. Wilson, B., 1984, Systems: concepts, methodologies and applications, John Wiley and Sons, Chichester. Wilson, J., 1993, Improving toxic materials conflicts through improved public participation, Working Paper 93-3, Conflict Resolution Consortium, University of Colorado, Boulder, CO. Wruck, H.J., 1993, Recovery of environmental damages: a matter of survival, Journal of Environmental Law and Practice, 3, 143-193. Wynne, B., 1984, The institutional context of science, models, and policy: the IIASA energy study, Policy Sciences 77(3), 277-320. Yogis, J. A. , 1990, Canadian law dictionary, 2nd edition, Barron's, Hauppage, New York. Zeiss, C , and J. Atwater, 1989a, Waste facility impacts on residential property values, Journal of Urban Planning and Development 115(2), 64-80. Zeiss, C , and J. Atwater, 1989b, Property value guarantees for waste facilities, Journal of Urban Planning and Development 115(3), 123-134. 242 Appendix 1. Notes on principled negotiation Perhaps the most important recent advance in the field of negotiation studies is the principled negotiation concept ("getting to yes") developed by Roger Fisher, William Ury, and co-workers at the Harvard Negotiation Project (Fisher and Ury, 1981; Fisher and Brown, 1988; Ury, 1991; Fisher et al, 1991; Fisher et al, 1994). Fisher and Ury (1981, p. vii) acknowledge an intellectual debt to Howard Raiffa, who approached negotiation from a game theory perspective (see Raiffa, 1982). Susskind and Cruikshank (1982) extend the principled negotiation concept to multi-stakeholder public policy negotiations. The basics of principled negotiation Fisher and Ury ( 1981) suggest three criteria for judging a method of negotiation: wisdom, efficiency, and amicability. A wise agreement as one that satisfies, insofar as possible, all legitimate interests of the negotiating parties; provides a fair resolution of conflicting interests; is stable over time; and considers the interests of parties not at the table. An efficient agreement is one which is obtained within a reasonable time period. An amicable agreement is one which does not damage, and perhaps improves, the relationships among the parties. The conventional negotiating method is positional: each party states a position, and attempts to maintain that position while pressuring the other party to make concessions. Positional negotiation tends to produce agreements which lack wisdom, efficiency, and amicability. Such agreements are unwise because efforts to satisfy interests are pre-empted by the focus on inflexible positions and the need to reconcile future actions with past positions. The result is typically a suboptimal agreement, for all concerned. Inefficiency arises from the haggling and contests of will which accompany positional negotiation. The lack of amicability stems from the ill-feeling which often accompanies contests of will. Where many parties are involved, positional negotiation is particularly difficult, and perhaps even impossible. The soft negotiating strategy, which emphasizes agreement and amicability over achieving the negotiator's goals, is an inadequate response to the problem of positional negotiation, because it is always dominated by a hard negotiation strategy, which emphasizes winning at all costs. Fisher and Ury (1981) describe three stages of principled negotiation: analysis, planning, and discussion. At each stage, the items to consider are interests, options, objective criteria, and people problems. The analysis and planning stages precede discussion with the other parties. The analysis stage consists of the assembly and analysis of information. In the planning stage, decisions are made on the appropriate course to pursue during negotiations. The discussion stage involves the actual negotiations, following the tenets of principled negotiation. Fisher and Ury propose principled negotiation as an alternative to positional negotiation. Principled negotiation has four prescriptive tenets: 1. Separate the people from the problem. 2. Focus on interests, rather than positions. 3. Generate a variety of options for mutual gain, before making a decision. 4. Use objective criteria to deal with contentious issues. 243 Separate the people from the problem The concept of "separating the people from the problem" is based on the recognition that negotiators are human beings with perceptions and emotions, and that communication between human beings is often imperfect. A negotiator has an interest in both the substance of the negotiation and in the relationship with the other party. Negotiation becomes difficult when these are entangled, and this problem is aggravated by positional negotiation. The solution is to explicitly separate the two interests, and to deal with them independently. A trap to avoid is the attempt to deal with relationship problems by making substantive concessions. Problems of human relationships can be classified as problems of perception, emotion, and communication. Perception is more important than objective reality in human conflict. Fisher and Ury offer various prescriptions for dealing with problems of perception. A key consideration is "saving face", which is defined, non-pejoratively, as the reconciliation of actions, statements, and agreements with past actions and statements, principles, and self-image. The principled negotiator should attempt to understand the perceptions of the other party, and to communicate his or her own perceptions, perhaps through explicit discussion. He or she should attempt to improve the other party's perceptions. This can be done by ensuring that the other party participates fully in the process, and receives credit for good ideas; by taking actions and making statements which are inconsistent with their negative perceptions; and by allowing opportunities for "saving face." Traps in perception include viewing the other side's statements and actions in the worst possible light, and casting blame. Fear and anger are emotions which often have a negative impact on the negotiation process. They are nonetheless legitimate. The principled negotiator should deal with such emotions through explicit discussion with the other party. It may be appropriate to permit emotional outbursts, as a release of tension, but it is important not to react to such outbursts in an escalating fashion. Symbolic gestures, such as an apology, may also be appropriate on occasion. Communication is the essence of negotiation. There are three modes of failure in communication: failure to talk to the other party; failure to listen to the other party; and misunderstanding the other party. The cures are: to speak clearly, purposefully, and persuasively without putting the other side on the defensive; and to listen actively and interactively. Note that some information (such as one's reservation price) may be detrimental to the reaching of a good agreement; it may be prudent to not reveal such information. Good communication often requires an environment of privacy, confidentiality, and small group size. It is possible to prevent relationship problems by developing a friendly relationship with the other side from the beginning, and by attempting to create an atmosphere of mutual, cooperative problem-solving. Fisher and Brown (1988) expand on the concept and importance of working relationships in negotiations. They argue the necessity of uncoupling process from substance ("separating the people from the problem"). They recommend a strategy which is "unconditionally constructive". The "constructive" aspect is that one should undertake to only act in a way which is good for oneself and for the relationship. The "unconditional" aspect is that one should act in this way regardless of how the other party behaves. 244 Fisher and Brown identify six basic elements of a good working relationship, and discuss how to deal with these elements on an unconditionally constructive basis. The six elements are: 1. Rationality (not the rationality of game theory): one requires an appropriate balance of reason and emotion. 2. Understanding: one should attempt to understand one's opponent, with respect to his or her interests, perceptions, and values. 3. Communications: one should strive for active, effective, and clear communications. 4. Reliability: one should be reliable, without depending on the other side to be reliable. 5. Persuasion: one should persuade rather than coerce. 6. Acce