55572 Water Working Notes Water Working Notes Note No. 27, June 2010 Modeling for Watershed ManageMent: a Practitioner's guide Jay R. Lund Susanne M. Scheierling and Grant Milne Water Working notes are published by the Water Sector Board of the Sustainable Development Network of the World Bank Group. Working Notes are lightly edited documents intended to elicit discussion on topical issues in the water sector. Comments should be e-mailed to the authors. Table of ConTenTs Acknowledgements ................................................................................................................................................................................................. iv Disclaimers ................................................................................................................................................................................................................. iv Contact Details ............................................................................................................................................................................................................. iv Acronyms and Abbreviations .............................................................................................................................................................................v Executive Summary.................................................................................................................................................................................................vii 1. Introduction .......................................................................................................................................................................................................1 2. Background on Modeling ........................................................................................................................................................................3 2.1 Purposes of Modeling ..................................................................................................................................................................3 2.2 What is a Model?..............................................................................................................................................................................4 2.3 Organizing Modeling Efforts ....................................................................................................................................................4 2.4 Model Selection and Use ...........................................................................................................................................................5 2.5 Further Issues .....................................................................................................................................................................................5 3. Modeling for Watershed Management.........................................................................................................................................7 3.1 Watershed Management Problems .....................................................................................................................................7 3.2 Modeling Watershed Problems ..............................................................................................................................................8 3.3 Approaches to Integrating Models ...................................................................................................................................12 3.4. California's Central Valley as an Case Study ..................................................................................................................17 4. Conclusions and Practical Lessons ................................................................................................................................................19 References ............................................................................................................................................................................................................... 21 Annex: Overview of Key Watershed Management Models and Developer Links ................................................ 25 Figures Figure 1. Computer Modeling as Extension and Organization of Thinking About Watershed Problems ..................................................................................................................................................................2 Figure 2. Watershed Management Problems and Activities ..................................................................................................7 Figure 3. Spatial and Temporal Scales of Some Watershed Management Problems .............................................8 Figure 4. Example Interactions Among Component Models................................................................................................9 Tables Table 1. Examples of Models of Various Watershed Management Components .....................................................9 Table 2. Examples of Models and Approaches to Integrated Modeling of Watershed Problems ...............13 Table 3. California Central Valley Examples of Models and Approaches to Modeling Watershed...............18 iii aCknowleDgemenTs This working note is a collaborative product of the World Specialist, Africa Environment and Natural Resources Man- Bank's Energy, Transport and Water Department, and the agement Unit, World Bank. The authors also acknowledge Bank's Water Resources and Watershed Management The- the excellent administrative and editorial support of Ryma matic Group. It was prepared by Jay R. Lund, Director, Cen- Pitts, Program Assistant, South Asia Agriculture and Rural ter for Watershed Sciences, and Ray B. Krone Professor of Development Department, World Bank. Environmental Engineering, University of California, Davis; Susanne M. Scheierling, Senior Irrigation Water Economist, Approving Manger: Energy, Transport and Water Department, World Bank; Julia Bucknall, Sector Manager, ETWWA and Grant Milne, Senior Natural Resources Management Specialist, Agriculture and Rural Development, South Asia, World Bank. The authors would like to thank the following ConTaCT informaTion peer-reviewers for their excellent comments: Julia Bucknall, To order additional copies, please contact the Water Help Sector Manager, Energy, Transport and Water Department, Desk at whelpdesk@worldbank.org. This paper is available World Bank; Hanan G. Jacoby, Lead Economist, Develop- online at http://www.worldbank.org/water. ment Research Group, World Bank; Halla Qaddumi, Re- search Analyst, Africa Water Resources Management Unit, Authors may also be contacted through the Water Help World Bank; and N. Harshadeep Rao, Senior Environmental Desk at the email address above. DisClaimer This volume is a product of the staff of the International permission may be a violation of applicable law. The Bank for Reconstruction and Development/The World International Bank for Reconstruction and Develop- Bank. The findings, interpretations, and conclusions ment/The World Bank encourages dissemination of its expressed work and will normally grant permission to reproduce portions of the work promptly. in this paper do not necessarily reflect the views of the Executive Directors of The World Bank or the govern- For permission to photocopy or reprint any part of ments they represent. this work, please send a request with complete infor- mation to the Copyright Clearance Center, Inc., 222 The World Bank does not guarantee the accuracy of the Rosewood Drive, Danvers, MA 01923, USA, telephone data included in this work. The boundaries, colors, de- 978-750-8400, fax 978-750-4470, http://www.copyright. nominations, and other information shown on any map com/. All other queries on rights and licenses, includ- in this work do not imply any judgment on the part of ing subsidiary rights, should be addressed to the Office The World Bank concerning the legal status of any territo- of the Publisher, The World Bank, 1818 H Street NW, ry or the endorsement or acceptance of such boundaries. Washington, DC 20433, USA, fax 202-522-2422, e-mail pubrights@worldbank.org. The material in this publication is copyrighted. Copying and/or transmitting portions or all of this work without iv aCronyms anD abbreviaTions AquiVal Graphic User Interface model for groundwater HEC IFM Hydrologic Engineering Center ­ Integrated resources planning flow model AQUATOOL A modeling suite for reservoir and aquifer HEC PRM Hydrologic Engineering Center ­ Prescriptive systems, common in Spain Reservoir Model BASINS Better Assessment Science for Integrating HEC RAS Hydrologic Engineering Center ­ River Analysis Point and Non-Point Sources System CALSIM California Water Simulation model HEC ResSim Hydrologic Engineering Center ­ Reservoir CALVIN California VALue Integrated Network Simulation Model DAMBRK Hydraulic model for dam break analysis HEC RPT Hydrologic Engineering Center ­ Regime DHI International Engineering Consulting Firm Prescription Tool (Hydrological modeling) HEC SSP Hydrologic Engineering Center ­ Statistical DWOPER Operational Dynamic Wave Model for flow Software Package forecasting in natural rivers IGSM Integrated Groundwater and Surface Water EPANET Model for piped water distribution (US Model Environmental Protection Agency) IWR-MAIN Institute for Water Resources ­ Municipal and ESSA Environmental Science Service Administration industrial water demand FLDWAV Flood Wave model for flood forecasting MashWin Streamflow generating models (natural or dam breaks) MIKE 11 General river modeling system GAMS General Algebraic Modeling System for MIKE 21 Integrated hydrodynamic model mathematical programming MIKE FLOOD Integrated model for river flooding GCM Global Climate Models MIKE SHE Integrated hydrological modeling system GFDL Geophysical Fluid Dynamics Laboratory (for MODFLOW Groundwater model (US Geological Survey) model development) OptiWin Optimization module of the water resources GIS Geographic Information System system management HadCM Hadley Climate Model PCM Parallel Climate Model HEC Hydrologic Engineering Center, US Army Corp SWAP Statewide Agricultural Production model of Engineers (California) HEC-EFM Hydrologic Engineering Center ­ Ecosystem SWMM Storm Water Management Model (US Functions Model Environmental Protection Agency) HEC EFT Hydrologic Engineering Center ­ Ecosystem SimWin Simulation model for water resources system Flow Tool management HEC-FDA Hydrologic Engineering Center ­ Flood TOPMODEL A model for predicting catchment water Damage Assessment discharge HEC GeoRas Hydrologic Engineering Center ­ Geographic USEPA US Environmental Protection Agency River Analysis System WAS Water Analysis System HEC HMS Hydrologic Engineering Center ­ Model for WEAP Water Evaluation and Planning model precipitation-runoff modeling v exeCuTive summary Watershed management problems are usually quite diverse, The Working Note seeks to show that computer modeling and involve a wide range of biological, geological, chemi- allows us to better organize, test, and refine our thinking cal, and physical processes with complex human, social, about watershed management problems and potential and economic contexts. More complexity is added because solutions. Typically, following the flow of water leads model- watershed management also commonly involves land and ing to be organized into the following areas: (i) precipita- water integration across spatial and temporal scales. Effec- tion and climate models; (ii) precipitation-runoff models; tively addressing these wide-ranging factors often requires (iii) stream and aquifer models; (iv) infrastructure operations the application of suitable analytical tools. Practitioners models; (v) economic, agronomic, social, environmental and academics have worked on these kinds of problems demand and performance models; and (vi) decision-mak- for thousands of years using different approaches. Simple ing models. Selecting the right model to apply to specific mathematical modeling of such problems began about 200 problems requires that several factors be considered along years ago. Computer modeling has been applied to specific with the objectives for modeling in the context of the field water and watershed management problems for about 50 decision problem. Key factors include understandability, years. It continues to rapidly advance towards more de- development and application time, resources required, tailed and extensive mathematical representations of wa- transferability and maintenance. tershed management issues, coupled with remote sensing and Geographic Information Systems (GIS). Good modeling is common-sense and understanding re- duced to calculation for the purposes of gaining insights Despite these advances in modeling, a recent review of into a real problem. Modeling should aid discussions, help World Bank-supported projects that aim to address wa- thinking and provide insights to problems where individu- tershed management problems indicated that modeling als and interests struggle to understand the problem and and other related tools have so far been applied only to a struggle to work together to address a problem. To aid limited extent. This Working Note is part of the ongoing model development and the interpretation and communi- efforts to improve this situation. It aims to provide a brief cation of modeling and model results and insights, simplic- guide on the usefulness of modeling, the role of models for ity is a great virtue. While complex problems sometimes watershed management, and an overview of various mod- require complex models, shedding of unneeded complexity eling approaches and modeling issues. The intended audi- is important. Local and in-house expertise is preferred when ence is academic practitioners, including Task Team Leaders developing and applying watershed models because of of World Bank-supported projects, but also policy makers, better familiarity with the problems assessed. Model integra- sectoral planners and program managers involved in water- tion is a growing trend but requires as much expertise and shed management. resources as development of any single model component. vii 1. inTroDuCTion "All models are wrong, but some are useful." Box (1979) puter modeling are available from Geoffrion (1976, 1997) and Gass (1983) generally, and into environmental and A watershed is a geographic area that supplies surface or water modeling from Jakeman et al. (2006) and Beck (2002). subsurface water flows to a drainage system or body of Recent advances in modeling, coupled with more acces- water (Smyle et al., 2009). Watersheds can vary in size from a sible remote sensing from a range of satellites and use of few hectares to a basin scale of thousands of square kilome- low-cost geographic information systems (GIS) can increase ters. Watershed management is the integrated use of land, our understanding of the complex relationships between vegetation, and water in a specified watershed to conserve water, land, people, and proposed watershed management land and hydrological services, and reduce downstream or interventions. underground impacts to be compatible with fundamental societal objectives. As watershed management became Despite these developments, a recent review of fifteen an accepted development approach in the 1970s, projects years of World Bank supported projects involving watershed generally applied a soil and water conservation approach management indicated insufficient attention to basic hy- that emphasized engineering and civil works. However, by drology and little application of modeling and other tools in the end of the 1980s, it was clear that new approaches were many projects (Darghouth et al. 2008). This need not be the needed. Since the 1990s, watershed management programs case. As this Working Note indicates, new advances in mod- have tended to integrate broader livelihood improvements eling, remote sensing and GIS, as well as improved data sets and poverty reduction objectives with more participatory can help increase our understanding of the relationships soil and water conservation operations. Best-practice in between water, land, and proposed watershed manage- watershed management now integrates land and water ment interventions. Analyses at the basin or sub-basin lev- management, and for implementation at a micro-watershed els can now be conducted to support more effective water- scale in particular, works closely with local stakeholders. shed planning, for example by estimating the hydrological impacts of watershed management practices and tech- Watershed management problems and programs tend to nologies. Even broader environmental impacts can now be quite complex and diverse, but have some common be more easily quantified using available data and models, attributes. They all involve the flow of water tying the local where time and resources permit. Modeling can also help hydrology to larger scale climate and they all involve some address the demand for more transparency in decision human problem, be it economic, social, or environmental making. But among watershed management practitioners, involving the management of these water flows. Through- an overarching question is "what are these tools and mod- out history, every civilized people has managed water for els, and how can I decide on which ones to use"? To help their purposes at least at a local scale and often at a regional address this problem, a number of Bank contributions have scale (Frontinus 97AD; Evenari et al. 1982; Mencius 1970). been made in the last couple of years to expand this knowl- Beginning in the 1700s, with the French schools of engi- edge base, including the preparation of briefing notes1 and neering, mathematical representations of these problems the organization of technical seminars2 on the topic. began to be used to provide more precise insights into how to manage water problems (Ekelund and Hebert 1999). And in the last 50 years computer modeling has been employed 1 For example, as part of a larger study on integrated river basin management, a briefing note was prepared on system modeling to manage much more detailed and extensive mathemati- in river basin management (World Bank Institute, 2006). cal representations of watershed problems (Loucks and van 2 During SDN Week 2008, hosted by the World Bank's Sustainable Beek 2005). Excellent insights into the application of com- Development Network in Washington DC, a workshop was 1 Modeling for Watershed Management: A Practitioner's Guide This Working Note is part of these ongoing efforts. It exam- and some common issues that may arise. Chapter 3 reviews ines how to employ computer models to aid in managing watershed management problems, and proceeds to discuss watershed problems. Modeling should be closely tied to the the roles of models for watershed management. Some conception of the management problem, the purpose of the approaches to employing and organizing modeling activi- modeling exercise (Figure 1). In some cases the modeling ties for watershed management are discussed, and several exercise will lead to fundamental or detailed changes in the examples are presented. Chapter 4 provides conclusions conceptualization of the problem. In other cases, the model- and practical lessons. Readers who are interested in under- ing will merely confirm or help provide detail to the original standing the use of models in a general sense are referred conceptualization. Overall, computer modeling allows us to to Chapter 2. Those who are searching for specific types of better organize, test, and refine our thinking about water- models to help them with analyzing particular watershed shed management problems and potential solutions. management issues may turn directly to Chapter 3 where the most relevant information is contained. Chapter 2 provides a background of why modeling is use- ful, the conceptual basis, the selection and use of models, figure 1. computer Modeling as extension and organization of thinking about Watershed Problems Modeling thinking Source: Authors. organized by the Water Anchor on "Pushing the Frontiers of Integrated Water Resources Management at River Basin Scales Using a Modeling Approach". During the World Bank's Water Week 2009, J.R. Lund, the lead author of this Working Note, was the key presenter in a seminar on "Modeling for Watershed Management" organized by the co-authors, G. Milne and S.M. Scheierling; and as part of a session on modeling tools for agricultural water man- agement co-organized by the Water Anchor and the Agriculture and Rural Development Department, S.M. Scheierling gave a presentation on "Modeling Approaches Incorporating Hydrologic, Agronomic and Economic Aspects." In 2010, several seminars were organized by the joint Water Resources and Watershed Manage- ment Thematic Group that focused on modeling issues in water management. 2 2. baCkgrounD on moDeling 2.1 Purposes of modeling 2. Explore and compare solutions without costly trial and error. Development of solutions by field experimenta- "The purpose of computing is insights, not numbers." tion alone is very time consuming and costly. The hard Hamming (1962) to control nature of most field experiments also raises questions of the scientific validity of their outcomes Computer modeling integrates and explicitly represents and interpretation. While computer modeling has knowledge in a level of detail, extent, and complexity and limitations, its use before field experimentation and with a computational speed which would be impossible application can make field experiments more likely to for the unaided human mind or the unaided minds of any provide greater insight and improve the likelihood of group of individuals. As with all scientific knowledge, the success in field applications. knowledge embodied in a computer model must have em- 3. Improve communications, negotiations. Real problems pirical and deductive origins. Much of the value of comput- often involve negotiations and compromise, which er modeling lies in using model development to formally requires communications among parties. Computer state and integrate relevant knowledge in the context of a models can be employed as part of a negotiations set- particular problem. ting where solutions and their implications for each party's interests can be explored rapidly. More generally, computer modeling can have the following 4. Ultimately, modeling and other technical studies should uses: provide decision makers with greater confidence in solu- tions to problems. For practical reasons, many solutions · Integrate empirical and deductive knowledge cannot be field tested before they are adopted. Having · Create a complex testable hypothesis modeling and other technical studies estimate the · Improve and test intuitive understanding (education) likely field performance of various solutions provides · Identify gaps in understanding a more scientific basis for selecting a particular solu- · Explore and compare solutions to problems tion than comparative evaluations based on intuition, · Avoid costly trial and error in the field past experience, or political assessments alone. Ideally, · Reduce uncertainty and provide assurances the technical aspects of computer modeling provide "strong invincible arguments" (Disco and van der Ende For more applied problems, this list can be simplified to the 2003) to aid in moving policy discussions towards use of models to: more promising solutions. 5. More cynically, sometimes computer modeling is un- 1. Improve understanding of the problem. Often the work dertaken because models require time to develop, test, of developing computer models helps formalize and and apply (especially complex models). The delay in improve our thinking and understanding about a decision-making required by extensive technical stud- problem in ways that improve the discussions and ies, modeling, and data collection provides respite outcomes regarding solutions to the problem. Often, for those interested in continuing the status quo or this improvement provides benefits even if numerical avoiding controversy while seeming to address a results from the model itself are not employed. This controversial issue. For individuals and interests, pro- educational value applies especially to professionals tracted modeling can provide an opportunity to delay new to the problem. decision-making. 3 Modeling for Watershed Management: A Practitioner's Guide Overall, computer modeling should be undertaken to de- · Expertise of Modeler. All models are simplifications of velop, test, support, and communicate convincing insights reality, so it is important for the model developer and into a problem and its solutions. user to artfully and carefully employ the model and data, and make interpretations of model results with these simplifications and potential errors in mind, as 2.2 what is a model? well as keeping in mind the problem being addressed. Two good and careful modelers using the same model "Probability theory is nothing but common sense reduced and field data sets are likely to have model results to calculation." Laplace (1819) which are at least somewhat different. To some degree these differences represent the limitations we have on Computer models should represent organized human rea- our understanding and representation of the problem. soning. This reasoning can have only two scientific bases, Sometimes these differences will be unimportant for deductive and empirical. A deductive base involves the use decision-making. Sometimes these differences will rep- of logic to derive implications from established knowledge resent opportunities for controversy. There is often a (which might ultimately be empirical). Empirical knowledge gap between the understanding of a watershed man- is based on systematic and consistent observation (ranging agement problem needed for problem-solving, versus from physical laws--conservation of mass, energy, and mo- the understanding of a watershed often pursued by mentum--to patterns observed from regressions, econo- a research scientist. Good science is not always good metrics, or other generalizations from data). All modeling problem-solving; this sometimes requires different steps should have such a formal foundation, using the com- types of modeling expertise. puter to help organize this reasoning and make it explicit. The modeling, in the end, is to reduce extensive application Model development and use often pose a dilemma. Is a of common sense to calculation. sophisticated model by a less expert modeler better than a simple model by a great modeler? In general the better The model itself then has three parts: modeler will employ data, methods, and theory more use- fully to a particular problem than a sophisticated specialized · Theory and Computer Software. This is usually the scientific modeler using a specialized model. This returns us named model or computer code. There is some un- to the importance of how one organizes modeling efforts derlying theory or conceptualization of the problem for a particular problem. or part of the problem, which is implemented by the computer code. Simplifications and numerical repre- sentations and calculations are often needed to imple- 2.3 organizing modeling efforts ment this conceptualization into the form of computer software. Modeling software can range from highly An important aspect of modeling is to have and keep a fo- specialized scientific codes to spreadsheets. cus around a well-defined problem. Often the first contribu- · Data. Data are the form of model inputs and parameter tion of a modeling effort is to lead to a better definition of values, as well as any user-specified options for how the problem. All modeling should fit and be interpreted in the computer code is to be run, such as which numeri- the context of this problem. cal solution method should be used for solving a set of equations. Data involve field data, data estimated out- The definition of the problem is not merely technical. The side the model from field data or outputs from other institutional context is equally important in establishing models or calculations, or expert judgments. Often the decision framework for the application of models to field data alone have substantial errors. Model results practical land and water use issues, or policy challenges. can be erroneous because of errors in data used in the It is important to develop a sound process for stakeholder model, as well as errors in other aspects of the model. participation to identify the key issues and questions that 4 2. Background on Modeling modeling can help address. In other words, the results and ing on how well the effort is managed. Guidance for insights of both model users, and the users of model results the development of models is summarized in Jakeman must be involved. Further, planning must also ensure that et al. (2006), Gass (1983), and Harou et al. (2009). modeling results will be understood and convincing and · Transferability. Model ownership, proprietary software, insightful to the users of the results and decision-makers. required expertise, data, resources, etc. can limit or The decision making and institutional context are both encourage the ability to transfer a model among orga- important considerations in how modeling efforts are nizations. This is important particularly when several organized and presented. Presentation of modeling to de- organizations will be involved in the work or decisions, cision-makers in purely technical terms, divorced from the or if the project envisions transferring the modeling problem-solving context, can be unhelpful. capability to others. · Maintainability. Data, expertise, and resources are Component technical efforts need to be organized into an needed to continue any modeling effort and maintain integrated framework for problem-solving which includes modeling capability. The lifetime of the model can be a strategy for bringing the technical efforts to bear in deci- quite short if it requires the expertise of one person sion-making. It is important to invest significant resources or external organization. Modeling and data manage- and expertise into developing this link between the model- ment efforts should be designed to be maintained and ing efforts and decision-makers, and it is important to invest ultimately replaced. much effort in integrating component modeling efforts to be coherent technically and comprehensible for decision- For many model developers and users the "best" model is making (Jakeman and Letcher 2003; Jakeman et al. 2006). often the model they know best. This is a natural human This will be further discussed below. condition reflecting the greater ability of an experienced user of a particular model to work around normal model limitations and to be comfortable doing so, the lesser 2.4 model selection and use amount of time needed for someone to develop a model in already-familiar software, and sometimes the financial and Which models and approach to integration and use should professional interest of the modeler. be selected? Several factors and objectives for modeling should be considered in the context of the field decision problem. These include: 2.5 further issues · Understandability. For each audience, as well as for A few additional common modeling issues should be those undertaking and explaining the modeling, com- touched on. munication of the modeling and its results is needed (Geoffrion 1997). If modeling and results cannot be Data and Models. Which comes first, the model or the data? understood and communicated, then there is likely to Should modeling occur only after "enough" data has been be less confidence in them, and the modeling will be gathered? Or, should data collection be guided by the use less effective. Good documentation of the model and of preliminary models to assess which data would be the model runs are important aspects of understandability, most important and useful? Clearly data and model devel- which are usually insufficient alone. opment should go together with neither being completely · Development and Application Time and Resources. The subservient to the other. Modeling should represent the time and resources available for the development and larger conceptual framework for collecting data, and field employment of modeling tools is usually important. testing model results. Such a modeling framework is not Integrating and making sense of modeling results also entirely empirical and data-driven, but begins with the logic requires expertise, time, and resources. New model and physics of the problem, which involves theory which development and integration also entails risks depend- exists beyond and before data. 5 Modeling for Watershed Management: A Practitioner's Guide For many applications there is often an uncalled-for confi- in decision-making actively employ the software for this dence in field data. Data can sometimes be of poor quality, purpose. This implies that the human decision-making and overconfidence in field data can lead a good model process is configured to employ and seek out the use of to be over-questioned. (For technical and scientific quality such model-derived information and that the modeling control, one should always question a model, or any other and software have been configured to provide information form of knowledge.) By using a model to organize and mo- and exploration in a form valued by decision-makers. A set tivate data collection, where systematic motivation is often of software can hardly be called a decision support system needed, models can lead to better and more useful data if it is not used by decision-makers in decision-making. and more targeted data development efforts. You will never Decision support systems are more easily designed and have enough data, but you might have sufficient data for employed for routine operational decision-making and a decision. Modeling can tell you if you are likely to already tend to be more successful if designed in collaboration have sufficient data. with decision-makers and supported by higher-level man- agement. In cases where agencies or groups who have data feel threatened by a model or other data development efforts, Why Analysis Fails. Applied modeling and analysis fails when sometimes a new model and an opportunity to become it does not provide insights into real problems. In some cas- involved or embarrassment at not being involved can lead es, this is due to stakeholders failing to ask the right ques- to greater cooperation. tions about why modeling is required and what specific land/water or policy issues need to be addressed. Insights Model Calibration and Testing. Model calibration is the set- can be in the form of improved understanding and thinking ting of parameter values to represent what you know, about a problem merely from clearer problem conceptu- including reasonable fitting of parameter values to field alization and organization as part of model development, data. Models are frequently said to be "validated" as well, even if no model results are used. More conventionally, the when they are tested against an independent set of field model can stimulate insights based on comparison of mod- data. Alas, models cannot be validated or proven to be el results for different conditions. Models also can produce valid (Konikow and Bredehoeft 1992). Models can only be insights through the education of new professionals who invalidated by such comparisons to field data. Where field use and apply the model. data are inaccurate, even a good model can be carelessly invalidated. Models can be tested more broadly against However, modeling analysis often fails to develop insights common sense, expert judgment, field data, more complex to real problems. This can be due to a lack of focus on real models, using sensitivity analysis, tests of individual model problems and a lack of focus on improving human thinking component, and tests of software code and numerical solu- about the problem. Sometimes academic modeling of real tion methods (Gass 1983; Beck 2002). problems becomes diverted towards issues of academic rather than practical or policy interest. Sometimes model- Decision Support Systems. Decision support systems are in- ing deteriorates into a quest for unrealistic or unproductive tended to tailor model interfaces and results for direct use rigor, so-called "rigor mortis." Even when the analysis is done by decision-makers or their staffs. As such, there are two well for the problem, analysts often have trouble commu- sides of a decision support system: (i) the model analysis nicating their results to policy-makers. Just as often, policy and display of information, and (ii) the human decision- makers have trouble listening to or understanding the re- makers. The software, model analysis, and information dis- sults of modeling analysis, sometimes due to a lack of time, play is doable and is the easy part of developing a decision patience, or technical skills and sometimes due to a lack support system. However, organizing human decision- of political interest. At the most cynical level, a request for makers to use a "decision-support" system is a major ac- "more detailed modeling" can just be a tactic for delaying or complishment. The institutions and individuals involved steering controversial decisions. 6 3. moDeling for waTersheD managemenT 3.1 watershed management Problems figure 2. Watershed Management Problems and activities Watershed management problems are diverse and often inter-related. Figure 2 illustrates the variety of problems that Upper watershed can be involved in watershed management. These prob- lems can range from the design of a micro-catchment to Rainfed annual crops Reservoir gather and retain rainwater for agricultural intensification Irrigation and groundwater recharge in an arid region, to the inte- grated management of land use at a basin level involving Ri Drainage reservoirs, aquifers and levees, urban land use for regional pa Groundwater ria flooding, and regional water supply problems. Rainfed orchards n m Flooding an Urban water supply ag em Fish and wildlife Even for the micro-catchment, important physical, biological, and sanitation en and economic processes must be understood and perhaps t manipulated. The probability distribution of annual rainfall, Source: Authors. the ability and preparation of the local soils and embank- ments to retain, absorb, and not lose this rainfall, the prob- ability that rainfall will be insufficient so that water must be to be affected by the management of a watershed. These imported or the tree's death endured, the economic costs issues give rise to commonly-observed conflicts involving and benefits of these water and agricultural management water and watershed management. For these reasons, wa- activities, and the political economy of the people who tershed management decision-making often involves many would undertake these activities are all important. For larger- people and interests, ranging from local individuals to high scale watershed problems, many more physical, biological, levels of government. Such decision-making often involves economic, and human processes will become important. professional technical studies to help provide convincing or One aspect of watershed management problems is that external assurances to those involved regarding the facts of complex physical, chemical, biological, and socio-economic the matter to aid in discussions or negotiations. Computer processes are occurring simultaneously. Typically many more modeling often has an important role in such technical things are going on than an individual mind can keep track studies. These technical studies are more effective if they of with any precision, so mathematical and computer mod- are organized to provide such information and assurances els have become common-place for these problems. within the larger context of decision-making, problem ex- ploration, and problem solving. Harou et al. (2009) review Watershed management problems also often involve the development and application of hydro-economic mod- conflicting purposes, particularly for larger watersheds. At eling involving spatially explicit representation of problems a basin level, millions of people are typically involved in and management. watershed management problems and their outcomes. Each person has different interests in the resolution of the Watershed management problems also commonly involve problem. Indeed, watershed management solutions often integrating across spatial and temporal scales. One example involve considerable long-term infrastructure investments is the spatial effects of management of sub-watersheds on and long-term institutional commitments or precedence, so larger watersheds, and vice versa. In terms of temporal scale, several generations of people, even those unborn, are likely the use of reservoir, aquifer, or soil moisture storage affects 7 Modeling for Watershed Management: A Practitioner's Guide how and if annual precipitation is available for a particular figure 3. spatial and temporal scales of some irrigation application during the dry season. Watershed prob- Watershed Management Problems lems often occur at a variety of spatial and temporal scales Basin, simultaneously, which complicates policy development National Regional, and implementation, as well as modeling. These problems National water policy of scale are further complicated by the interaction of water Seasonal Regional Spatial operations infrastructure management aspects with non-water issues, such as larger planning planning Scale Flood urban land use, economic, social, and environmental policies (Log) operations and processes. This complexity can become conceptually Flood and practically overwhelming; this provides a strong ratio- evacuation Micro- Rainfed Rainfed tree nale for modeling. However, it is never possible to model all plantations watershed crops aspects of any major watershed management problem. In some sense, we will never understand it all, and therefore we Temporal Scale (Log) can never model it all. (Interests favoring the status quo will sometimes insist on a high level of detail of modeling and Source: Authors. analysis for this reason.) Modeling cannot be used not to rep- resent everything. Instead, modeling is best used to leverage and test our intuition and concerns as policy and manage- Typically, following the flow of water leads modeling to be ment discussions of the problem progress. In practical terms, organized into the following areas: models can help identify priority sites for investments in soil and water conservation measures; help set the scale of op- · Precipitation and climate models erations and phasing; provide information to guide develop- · Precipitation-runoff models ment of operational policies and regulations for land/water · Stream models use in priority watersheds, etc. Sometimes, a large integrated · Aquifer models modeling framework will be needed; at other times a series · Infrastructure operations models of smaller limited modeling applications will be better with · Economic, agronomic, social, environmental demand more of the integration being conceptual in the minds of and performance models those pondering the problem and making decisions. · Decision-making models. Table 1 provides a necessarily incomplete summary of a few 3.2 modeling watershed Problems of the hundreds of models developed for these processes. Importantly, the set of available models is always changing Computer modeling for watershed management problems as new models are developed, and older models fall into is very common. Indeed, the sheer number of models, of- disuse or go unmaintained. Relevant references for these ten competing for attention, sometimes distracts from the models can be found on the links provided at the end of purposes and artful employment of models to aid in solv- this paper or using an internet search engine. Wurbs (1994) ing problems. The organization of modeling for watershed provides a nice review of models existing at that time. Below management is one of the first important technical issues the different modeling areas are discussed in more detail. to be resolved. It is usually best to organize modeling to fol- low the water through the system, using the physics of the Figure 4 presents examples of flows of information that hydrologic cycle to support the logic of modeling. Perhaps are common between these types of model components. more importantly, since it follows the flow of water, this ap- Of course, the arrangement and presence of these com- proach is also relatively easy for decision-makers and others ponents varies with the specific problem. For example, in to understand. some cases, the agronomic system (model) might be sup- 8 3. Modeling for Watershed Management table 1. examples of Models of Various Watershed Management components Model area examples* Precipitation and climate GFDL, PCM, HadCM, local weather forecasting Precipitation-runoff HEC-HMS, WEAP, SWAT, SWAP, SWMM, TOPMODEL, MIKE, local flood forecasting Stream HEC-RAS, DWOPER, FLDWAV, DAMBRK, MIKE 11, MIKE 21 Aquifer MODFLOW, MIKE SHE, IGSM, many others Infrastructure operations HEC-ResSim, CALSIM, local system models Economic, agronomic, social, environ- IWR-MAIN, HEC-FDA, various local and academic models mental demand and performance Decision-making ESSA, Shared Vision Modeling, HEC-RPT, various hydro-economic models * Web links to these and other models are provided in the Annex. Source: Authors. figure 4. example interactions among component Models Climate and weather conditions Precipitation and climate models Precipitation and weather conditions Rainfall -runoff models Runoff Infiltration Stream models Stream-aquifer Aquifer models interactions Flow, water Flow, storage, quality water quality Infrastructure operations Water and water quality deliveries Operating rules Economic, agronomic, social, environmental demand and performance Performance estimates Decision-making Source: Authors. 9 Modeling for Watershed Management: A Practitioner's Guide plied directly by precipitation, without stream, aquifer, or ed model, use of the existing model can be cost-effective. infrastructure. In other cases, a new model might be needed. Precipitation and climate models represent the original Modeling of flows into streams and into aquifers is usually un- source of water to the watershed. Such models can include dertaken separately, but recent years have seen greater use General Circulation Models (GCMs), which represent the of combined modeling of these processes where stream- entire world's climate (albeit simplified) with indications for aquifer interactions might be important. For flooding prob- what might happen in a regional watershed or, more likely, lems, aquifers are usually not important. Stream and aquifer what might happen in the part of the globe where the modeling is quite common professionally, with a wide watershed is located. Downscaling of GCM precipitation, range of expertise available from government agencies, temperature, and other results is usually needed for the consulting firms, and universities. One problem is that these watershed scale. Alternatively, field records of climate can models are often constructed for one purpose and then be used to develop empirical models of precipitation, tem- re-applied for different purposes. For example, a stream perature, etc. To represent climate change, it is often more model developed for flood flows might be less accurate convenient and comprehensible to adjust historical data for low flows encountered for water supply applications. for changed conditions than to rely on downscaled GCM Aquifer models developed for water quality applications results directly. GCM results often provide a basis for adjust- might need to be revisited or expanded spatially before be- ing historical data. Adjusting historical data based on GCM ing applied to water supply problems. Again, it is important and other information on potential climate changes can be that the model be tailored or interpreted for the problem at done with less specialized expertise. Alternatively, for many hand. The US Geological Survey's MODFLOW model is (pri- areas, regional numerical weather forecasting models are marily) an example of an aquifer model, while HEC-RAS is an often available and used over short time scales and might example of a stream hydraulics model. be more suitable for some short-term operations problems. (Giorgi and Mearns 1991; Wilby, R.L. and T.M.L. Wigley 1997) Infrastructure operations models represent operational and infrastructure planning decisions involved in watershed Precipitation-runoff models can range from highly disaggre- management. Operational decisions include reservoir gated representations of detailed flow physics on hill-slopes releases, irrigation diversions, operations of flood bypass and groundwater to simple regression equations based on gates, and aquifer pumping and recharge quantities. For ur- field data, and a wide variety of methods that mix empirical ban water supply and hydropower, such modeling is espe- and physically-based approaches. Hundreds of precipita- cially common. For other applications, infrastructure opera- tion-runoff models have been developed and applied since tions modeling expertise is less widespread, although most the 1970s, supported by widespread technology and exper- large water systems have their own models and in-house tise. Considerable debate exists over the relative merits and modeling expertise. The US Army Corps of Engineers' HEC problems of so-called physics-based versus empirical pre- ResSim model is moderately general simulation software for cipitation-runoff models (Loague and Vander Kwaak 2004). larger watershed reservoir systems. While many precipitation-runoff models are quite complex, rather simple models calibrated on the historical record of- Economic, agronomic, social, environmental demand and ten can be quite effective (Jakeman and Hornberger 1993). performance models represent aspects of the managed and Because these problems involve local geology and climate, unmanaged flow of water that are important for operation- it is often useful to have local engineers or hydrologists in- al, planning, and policy decisions. Agronomic models might volved in developing these models. HEC-HMS is an example indicate how water deliveries would affect the growth, of such a model; but many such models exist, often based maturity, and yields of crops. Economic models might be on local and university software. Existing local models of employed to indicate the economic benefits of urban, runoff from precipitation often can be found. Where the agricultural, or hydropower uses of water, the economic type of runoff problem is similar to that of an existing trust- damages of flooding, or the costs of pumping, treatment, 10 3. Modeling for Watershed Management or construction (Ward and Pulido 2008). Social performance aspects of water resources systems at a regional scale. These modeling might indicate changes in employment or flood models have emerged as a privileged tool for conducting evacuation with water management. Environmental per- integrated water resources management. An overview of formance models might address how water management the concepts and designs as well as the wide applications of affects fish populations or water quality. The US Army Corps hydro-economic models is provided in Harou et al. (2009). In of Engineers model for flood damage reduction analysis hydro-economic models, water allocation is driven or evalu- (HEC-FDA) is one common example for flood management. ated by the economic values it generates. All major spatially Their Ecosystem Functions Model (HEC-EFM) is applied to distributed hydrologic and engineering parts of the system various environmental flow management problems. IWR- are represented, including water balance components such MAIN is a model of urban water demands. as river flows, evaporation from surface water bodies, natural groundwater recharge and discharge, and return flows. Rele- Decision-making models try to integrate some of the above vant water supply infrastructure and operations may include types of models into a software and institutional setting canals, reservoirs, desalination plants, water and waste-water where decisions are made. For routine operational deci- treatment plants, groundwater or pipeline pumping sta- sions, this might take the form of a decision support system tions, artificial recharge basins and other groundwater bank- tailored to a particular water management project and ing infrastructure. These hydrologic and engineering fea- problem, with local users trained in the use of such software tures are included in a node-link network, where economic on a daily basis. This is common for the integrated use of demands have locations (nodes) and costs (or benefits) are hydraulic network modeling and field monitoring for urban incurred on links. The network accommodates both physical water system operations and for the operations of many and economic spatially distributed systems, and integrates hydropower systems and a few large irrigation systems. For all major hydro-economic elements. Hydro-economic mod- planning and policy-making problems, often special inter- els are applied to study instream and offstream intersectoral face software is made to organize and manage the use of allocation and use; water supply, engineering infrastructure models in a particular planning or policy decision-making and capacity expansion; conjunctive use of groundwater setting. For planning and policy purposes, such software is and surface water; institutions, water markets and pricing; usually tailor-made for specific problems (ESSA, Shared Vi- conflict resolution, transboundary management and sus- sion Modeling). More general software has been developed tainability; land-use management, including floods and wa- for some common types of problems, such as the US Army ter quality; and adjustments to drought and climate change. Corps of Engineers' Regime Prescription Tool (HEC-RPT). Most hydro-economic models are custom-built, often using commercial optimization software. These models are commonly simulation models, which allow users to investigate the likely implications and perfor- The above taxonomy of modeling efforts does not imply mance of specified alternatives. Sometimes optimization that each modeling areas can be "outsourced" to a different models are employed which manipulate a simulation rep- group and then conveniently assembled into a more com- resentation of the system to automatically suggest better prehensive and comprehensible understanding of the prob- solutions (defined as a mathematical objective function, lem. Model integration involves many problems. These in- commonly based on cost, net benefits, or water deliveries). volve technical issues of passing information coherently and Optimization methods are particularly common in hydro- consistently between model functions and across temporal power systems (Jacobs et al. 1995), where operating objec- and spatial scales. The existence of feedbacks between these tives are most strictly economic, and have also found use for model functions also poses technical challenges. Model scoping integrated solutions in the early phases of planning integration usually requires as much talent and resources and policy studies (HEC-PRM, CALVIN ­ see Annex). 3 Refer to the web links in the Annex to find more information An important group of decision-making models is hydro- about each model and specific institutions that could provide economic models which link economic and hydrologic further assistance. 11 Modeling for Watershed Management: A Practitioner's Guide as each individual model, perhaps more. Expertise for such 3.3 approaches to integrating models modeling is often available in government agencies, some universities, and some specialized consulting forms3. Several approaches are available for integrating model- ing efforts to address a specific watershed management problem. Each of these is available and suitable to different degrees for different problems and locations, and the ap- proaches are often hybridized. These approaches are sum- marized in Table 2. 12 table 2. examples of Models and approaches to integrated Modeling of Watershed Problems economic, agronomic, social, and/or Precipitation Precipitation- infrastructure environmental decision- Model(s) and climate runoff stream aquifer operations performance making Mega-models WEAP CALVIN/ HEC-PRM WAS ­ Jordan Stella, Excel, GAMS, etc. Modeling suites HEC and IWR HEC-HMS HEC-RAS RES-SIM HEC-IFM, HEC-PRM IWR-MAIN Aquatool MashWin AquiVal SimWin OptiWin Equalizador DHI MIKE-FLOOD MIKE 11, MIKE 21 MIKE SHE PG&E Hydropower Field data Basin regressions Excel basin Energy and Excel basin optimizations service prices optimizations Home-grown integration GIS Spreadsheets HydroPlatform OpenMI Other Continued on next page 13 3. Modeling for Watershed Management 14 table 2. examples of Models and approaches to integrated Modeling of Watershed Problems economic, agronomic, social, and/or Precipitation Precipitation- infrastructure environmental decision- Model(s) and climate runoff stream aquifer operations performance making Modeling for Watershed Management: A Practitioner's Guide System component models MODFLOW GAMS-econ Spreadsheets Scientific Regression Source: Authors. 3. Modeling for Watershed Management Mega-Integrated Model. Many models of watershed man- Integrated Suite of Models. For some routine problems, wa- agement problems are formulated to include a broad range ter resource software developers have found it useful to of human, physical, and management processes in one develop a suite of models which are designed to be easily model. These mega-models, sometimes called "holistic" (Cai integrated for a specific class of problems. Perhaps the most et al. 2003), are often relatively integrated and inclusive, widespread example is software developed by the US Army but they can never be complete. No model or modeling Corps of Engineers' Hydrologic Engineering Center (HEC). framework can address all aspects of most problems. Each This software was designed mostly for flood control and mega-model will tend to have its strengths and weakness- navigation problems commonly encountered by the US es, reflecting the expertise of the model developers and the Army Corps of Engineers. It includes component models kinds of problems the model has been applied to over time. HEC-HMS (rainfall runoff ), HEC-SSP (for estimating flood Some examples of models taking this approach include frequencies), HEC-RAS (river hydraulics), HEC-GeoRAS (GIS CALVIN (based on HEC-PRM) (Draper et al. 2003; Medellin et display of HEC-RAS results), HEC-ResSim (reservoir opera- al. 2007), WEAP (Yates 2005a, 2005b), WAS (Fisher et al. 2005; tions simulation), HEC-FDA (flood damage estimation), and Rosenberg et al. 2008), and various problem and location- HEC-EFT (environmental flow performance). These models, specific systems dynamics models (Palmer et al. 1999), or with their modeling practices and documentation, devel- optimization formulations (Cai et al. 2003; Cai 2008). oped to be assembled for flood problems allow modeling of these problems to be pursued in a fairly efficient way and While the concept of a mega-model implies that all aspects integrated more easily than most other models. Many mod- of the problem are represented, the depth and flexibility eling integration issues have been addressed in the design of how each aspect is represented is often quite variable. and documentation of the models. Some models include more detailed precipitation-runoff modeling capability (WEAP), while others take time series Other organizations (such as Valencia Politecnic University, of hydrologic inflows as given (CALVIN) or merely single an- DHI, Wallingford, Delft, and various hydropower software nual inflows or probability distributions of inflows (WAS). and system firms) have developed suites of models for vari- Some models have more complete representations of ous problems. AQUATOOL (Andreu et al. 1996) is widely household and farm water management decisions (WAS), a used throughout Spain for water supply and management summarized economic water demand schedules (CALVIN), problems. In California, the Pacific Gas and Electric Corpora- or merely prioritized delivery targets (WEAP). Models simi- tion runs dozens of hydropower reservoirs which require larly vary in their abilities to represent internal operational operational decisions for short and long time horizons. They decisions and uncertainties. have developed a suite of spreadsheet and programming- language based software to support these decisions, which While mega-model software is often convenient, for well- includes hydrologic, operations, and economic models focused idiosyncratic problems it is common to have tailor- (Jacobs et al. 1995). Flood warning system suites also are made mega-models implemented in common systems common (David Ford and Associates, see Annex). Many spe- dynamics software (such as Stella, Extend, or Goldsim) or cialized suites of models are proprietary overall or in their spreadsheet software (such as Excel). Such models can be components. A group of Australian partners is currently well-tailored to the specifics and idiosyncrasies of a problem. developing several suites of models for urban, rural, ecosys- However, many applications of systems dynamics software tem, and river management problems (eWater, see Annex). seem to have difficulty in being maintained over a long time, perhaps due to the more proprietary nature of the software One advantage of integrated suites of models over mega- platform and the expertise needed for continued use and models is that they allow a greater degree of modular- development of such models. WAS (written in GAMS) is quite ization for local conditions and expertise. If a different flexible, but requires GAMS modeling expertise. It seems precipitation-runoff or flood damage model is desired, it is more common for spreadsheet-based models to have longer relatively easy to swap out the suite-designed model for the longevity and use by a broader range of interests. other model, provides the spatial and temporal resolution 15 Modeling for Watershed Management: A Practitioner's Guide of the new component model can be made compatible. GIS. Many other modeling efforts (including the HEC suite) Alternatively, each component of the suite can be used are often designed to interface well with GIS. For problems independently of the suite and integrated in a more novel which are highly spatial, GIS might be a promising software way with other models or model components. framework for integrated model development. Home-Grown Integration. It is common for "integrated" System Component Models. Software packages sometimes modeling to consist of a stitched-together set of sequential are available for specific system components of a watershed model runs, with each component model's outputs post- problem. Such packages are often developed by research processed or adapted to provide input data for the next organizations, governments, or universities for scientific model in the sequence. Where several component models purposes and then become applied. MODFLOW (originally have been developed independently without the benefits by the US Geologic Survey for groundwater problems) and of a formal integrated development process, this might be EPANET (by the US Environmental Protection Agency for the only option available in the short term, especially for pipe network hydraulics) are two such models. Packaged one-off problems. models can provide nice, often detailed models of system components. Where local models already exist in such pack- In the longer term, such stitched together modeling can aged software, they might be usefully adapted for larger provide a basis (and motivation) for prototyping more modeling integration or provide a basis for calibrating more integrated suites or mega-models. Since any model of an simplified components of larger system models. Sometimes evolving problem might be useful for only a decade, using local packaged models can be stitched together into a suite home-grown integration for prototyping should be a prom- or with other models, as home-grown integration, usually ising direction. Otherwise, the maintenance of disintegrated for one-off applications. models and their periodic stitching together requires an unusual amount of expertise and resources to maintain Successful modeling packages sometimes evolve or and upgrade. That said, such homegrown integration is broaden to become more like suites of models or mega- best done at home; it is even more difficult for outsiders to models, as applications drive expansion of the model. Both understand the component models and integrate them. MODFLOW and EPANET have expanded over time and can Several groups are working to make it easier to integrate be considered as a suite of models built around a central models in an ad hoc or homegrown way. HydroPlatform is software or mega-models for some applications. The MOD- seeking to develop a common data platform for a variety FLOW farm package, for example, brings in farm processes of models to be compared and employed without the into the MODFLOW groundwater model for irrigation sys- traditional awkwardness of multi-component modeling. tem applications. EPANET now includes various water qual- OpenMI is another research effort to help modular compo- ity modeling capabilities. WEAP began as a regional water nent models communicate with less re-programming or balance model, but now includes a precipitation-runoff and new interfaces. other modeling capabilities. Somewhat to the contrary, GIS often provides an organizing Human Integration. Ultimately, all effective integration and or integrating framework for modeling activities. GIS is es- organization of modeling, no matter the technical process pecially good at providing an explicit spatial framework for and procedures, is human and must be comprehensible to both managing and displaying information and data and be effective. Modeling allows us to work with far larger and integrating land use, spatial processes, and water. However, more complex problems than we can work with individu- GIS integration might not always be as easy as it is promis- ally or as a group. However, to be useful for decision-making ing. While some modeling capability is available within the model has to be done and communicated in such a GIS software, processes will most likely require additional way that it sustains confidence that the work was well done simplification. Maidment (2002) and Maidment and Djokic technically and aids human comprehension of the problem (2000) demonstrate a wide range of capability for the use of and development and evaluation of solutions. 16 3. Modeling for Watershed Management 3.4. California's Central valley as a Case managers requires a wide variety of technical information study which can be provided using models. Each user of model- derived information has different topical and geographic in- Almost every major river system in the developed world terests. So over time, many local, state, and federal agencies has a set of simulation models to aid in water supply and have developed their own models, and have developed flood planning and operations. Simulation models are means to represent the interface of their local or regional also typically available for almost every large urban water models with other parts of the larger Central Valley system. supply system in the developed world. These models re- quire a systematic effort at data collection, organization, The resulting multiplicity of modeling efforts has resulted in distillation, quality control, and storage. The models also some advantages, and some difficulties. Advantages have must be integrated in to operational, planning, and policy included the presence in many agencies and consulting decision-making institutions. For river basins, it is common firms of technically knowledgeable and skilled individuals, to have several models, for different time and spatial scales, able to represent local technical aspects and concerns to for different basin problems and management decisions, higher-level technical and policy discussions and technical such as flood control, hydropower, and water supply. Data efforts. A wide range of modeling and data technologies and institutional activities typically require greater resources also have been employed within the regional water man- than the modeling efforts themselves. Many basins employ agement community. Difficulties also have arisen from this widely-available commercial or public software for model- diversity, in terms of technical and policy disagreements ing, and many basins develop their own. over technical direction, data, representation of different system components, and technical cooperation and inte- California's Central Valley is a rather large and intensive case gration. The variety of models and data also can impede study of how models are applied to watershed problems more systematic studies, as a coherent technical represen- by the wide variety of institutions responsible for managing tation over the broad system must overcome locally and water. A wide variety of national, state, local, private, NGO, topically fragmented representations of the system (Jenkins and scientific institutions are involved in modeling different et al. 2001). aspects of this watershed. Indeed, for our purposes, the ba- sin is so large and complex that examples of most elements It is clear that water users have benefitted tremendously of Table 2 can be found there, as illustrated in Table 3. from the responsiveness of decentralized management and technical work, but have also suffered from the lack of In essence, in a system where water management affects basin-wide technical coherence. millions of water users and thousands of professional water 17 18 table 3. california central Valley examples of Models and approaches to Modeling Watershed Problems economic, agronomic, social, and/or Precipitation Precipitation- infrastructure environmental decision- Model(s) and climate runoff stream aquifer operations performance making developer Mega-models CALSIM State, Federal CALVIN/ Academic HEC-PRM WEAP NGO Modeling suites Modeling for Watershed Management: A Practitioner's Guide HEC HEC-HMS HEC-RAS RES-SIM HEC-IFM Federal National Weather Federal Service PG&E Hydro- Field data Basin Excel basin Energy & Excel basin Private power regressions optimizations service prices optimizations Home-grown integration DWR/USBR Regressions, NWS C2VSIM CALSIM Post-processors, State, Federal NWS SWAP System component models Groundwater C2VSIM, State, Federal MODFLOW Estuary DSM2, State, Federal, other Private Water district Local operations Scientific Various Source: Authors. 4. ConClusions anD PraCTiCal lessons To conclude, the following practical lessons should be kept sometimes require complex models, but the identi- in mind when considering the modeling of watershed man- fication and shedding of unneeded complexity is an agement issues: important part of insight development (Medellin et al. 2009). · Good modeling is common-sense and understanding · Many modeling options exist. The option and ap- reduced to calculation for the purposes of gaining in- proach taken to modeling should reflect the problem sights into a real problem. Modeling is likely to be more being addressed and the human context in which than common sense can manage, but each step in the model results must be understood. modeling process should be supported by common · Identification and use of expertise is important to the sense in the form of empirical and deductive logic. development and use of models. Local and in-house · As such, modeling should aid discussions, help think- expertise is preferred, as this expertise is often closest ing and provide insights to problems where individuals to the problem under study. and interests struggle to understand the problem and · Model integration is a major problem which requires as struggle to work together to address a problem. To do much expertise and resources as development of any so, the modeling should follow the problem, usually model component. The chain of data custody between based on the physics and economics of the problem. model components requires someone who is conver- · To aid model development and the interpretation and sant in both model components, diligent in quality communication of modeling and model results and control, and cognizant of the problem and purpose insights, simplicity is a great virtue. Complex problems motivating the investigation. 19 referenCes Andreu, J., Capilla, J., and E. Sanchis. 1996. "AQUATOOL, a Shamir, and H. Wesseling, 2005. Liquid Assets: An Eco- Generalized Decision-Support System for Water- nomic Approach for Water Management and Conflict Resources Planning and Operational Management." Resolution in the Middle East and Beyond. Washington, Journal of Hydrology 177 (3­4):269­91. DC: Resources for the Future. Beck, M.B. 2002. "Model Evaluation and Performance." In Frontinus, S.J. 97AD, The Two Books on the Water Supply of the Encyclopedia of Environmetrics, ed. A.H. El-Shaarawi City of Rome, Translated by Clemens Herschel (1899); and W.W. Piegorsch, C Volume 3, 1275­79, Chichester: published by New England Water Works Association, John Wiley & Sons, Ltd, Boston, MA 1973. Box, G.E.P. 1979. "Robustness in the Strategy of Scientific Gass, S.I. 1983. "Decision-Aiding Models: Validation, Assess- Model Building." In Robustness in Statistics, ed. R.L. ment, and Related Issues for Policy Analysis." Opera- Launer and G.N. Wilkinson. New York: Academic Press. tions Research 31 (4):603­31. Cai, X., D.C. McKinney, and L.S. Lasdon. 2003. "An Integrated Geoffrion, A.M. 1997. "Maxims for Modelers." http://www. Hydrologic-Agronomic-Economic Model for River Ba- anderson.ucla.edu/faculty/art.geoffrion/home/docs/ sin Management." Journal of Water Resources Planning Gudmdlg2.htm and Management 129 (1):4­17. Geoffrion, A.M. 1976. "The Purpose of Mathematical Pro- Cai, X. M. 2008. "Implementation of Holistic Water Resourc- gramming Is Insight, Not Numbers." Interfaces 7 es-Economic Optimization Models for River Basin (1):81­92. Management ­ Reflective Experiences." Environmental Giorgi, F., and L. O. Mearns (1991), Approaches to the Simu- Modelling & Software 23(1):2­18. lation of Regional Climate Change: A Review, Rev. Darghouth, S., C. Ward, G. Gambarelli, E. Styger, and J. Roux. Geophys., 29(2), 191­216. 2008. Watershed Management Approaches, Policies, and Hamming, R. 1962. Numerical Methods for Scientists and Engi- Operations.: Lessons for Scaling Up. Water Sector Board neers. McGraw-Hill. Discussion Paper Series 11. World Bank, Washington, Harou, J.J., M. Pulido-Velazquez, D.E. Rosenberg, J. Medellin- DC. Azuara, J.R. Lund, and R.E. Howitt. 2009. "Hydro- Disco, C., and J. van der Ende. 2003. "Strong, Invincible Argu- economic Models: Concepts, Design, Applications, ments? Tidal Models as Management Instruments in and Future Prospects." Journal of Hydrology 375 (3­4): Twentieth-Century Dutch Coastal Engineering," Tech- 627­643. nology and Culture, Vol. 44, 502­35. Heinz, I., Pulido-Velazquez, M., Lund, J. R., and Andreu, J. Draper, A.J., M.W. Jenkins, K.W. Kirby, J.R. Lund, and R.E. How- 2007. "Hydro-Economic Modeling in River Basin itt. 2003. "Economic-Engineering Optimization for Cal- Management: Implications and Applications for the ifornia Water Management," Journal of Water Resources European Water Framework Directive." Water Resources Planning and Management 129 (3):155­64. Management 21 (7):1103­25. Ekelund, R. B., and R.F. Hebert. 1999. Secret Origins of Modern Holling, C.S., ed. 1978. Adaptive Environmental Assessment Microeconomics: Dupuit and the Engineers. Chicago: and Management. New York: John Wiley & Sons. University of Chicago Press. Jacobs, J., G. Freeman, J. Grygier, D. Morton, G. Schultz, K. Evenari, M., L. Shanan, and N. Tadmor. 1982. The Negev: The Staschus, and J. Stedinger. 1995. "Socrates--a System Challenge of a Desert. Cambridge, MA: Harvard Univer- for Scheduling Hydroelectric Generation under Un- sity Press. certainty." Annals of Operations Research 59:99­133. Fisher, F. M., A. Huber-Lee, I. Amir, S. Arlosoroff, Z. Eckstein, M. J. Haddadin, S. G. Hamati, A. M. Jarrar, A. F. Jayyousi, U. 21 Modeling for Watershed Management: A Practitioner's Guide Jakeman, A. J. and G. M. Hornberger. 1993. "How Much Palmer, R.N., et al. 1999."Modeling Water Resources Oppor- Complexity Is Warranted in a Rainfall-Runoff Model?" tunities, Challenges and Trade-offs: The Use of Shared Water Resources Research 29 (8):2637­2649. Vision Modeling for Negotiation and Conflict Resolu- Jakeman, A. J., and R.A. Letcher. 2003. "Integrated Assess- tion." In Water Resources Committee of the Water Re- ment and Modelling: Features, Principles and Ex- sources Planning and Management Division Conference amples for Catchment Management." Environmental 1999 ­ Preparing for the 21st Century, Proceedings of Modelling & Software 18 (6):491­501. 29th Annual Water Resources Planning and Manage- Jakeman, A.J., R.A. Letcher, and J.P Norton. 2006. "Ten Itera- ment Conference. tive Steps in Development and Evaluation of Environ- Parker, P., Letcher, R., and A.J. Jakeman. 2002. "Progress in mental Models." Environmental Modelling & Software Integrated Assessment and Modelling." Environmental 21 (5):602­14. Modelling and Software 17:209­17 Jenkins, M.W., A.J. Draper, J.R. Lund, R.E. Howitt, S.K. Tanaka, Peters, C.N., and D.R. Marmorek. 2001. "Application of Deci- R. Ritzema, G.F. Marques, S.M. Msangi, B.D. Newlin, sion Analysis to Evaluate Recovery Actions for Threat- B.J. Van Lienden, M.D. Davis, and K.B. Ward. 2001. "Im- ened Snake River Spring and Summer Chinook Salm- proving California Water Management: Optimizing on (Oncorhynchus tshawytscha)." Canadian Journal of Value and Flexibility," Center for Environmental and Fisheries and Aquatic Sciences 58 (12):2431­46. Water Resources Engineering Report No. 01-1, Dept. Peters, C.N., D.R. Marmorek, and R.B Deriso. 2001. "Applica- of Civil and Environmental Engineering, University tion of Decision Analysis to Evaluate Recovery Actions of California, Davis, CA, http://cee.engr.ucdavis.edu/ for Threatened Snake River Rall Chinook Salmon (On- CALVIN/. corhynchus tshawytscha)." Canadian Journal of Fisher- Konikow, L.F. and J.D. Bredehoeft. 1992. "Ground-Water Mod- ies and Aquatic Sciences 58 (12):2447­58. els Cannot Be Validated." Advances in Water Resources Rosenberg, D., R. Howitt, and J. Lund. 2008. "Water Manage- 15:5­83. ment with Water Conservation, Infrastructure Expan- Laplace, M. 1819. Critique de La Theorie Des Fonctions Genera- sions, and Source Variability in Jordan." Water Resources trices. Research Vol. 44, doi:10.1029/2007WR006519. Loague, K. and J.E. VanderKwaak. 2004. "Physics-Based Hy- Rosenberg, D.E., T. Tarawneh, R. Abdel-Khaleq, and J.R. Lund. drologic Response Simulation: Platinum Bridge, 1958 2007. "Modeling Integrated Water-User Decisions in Edsel, or Useful Tool." Hydrological Process 18:2949­56 Intermittent Supply Systems." Water Resources Research Loucks, D. P. and E. van Beek. 2005. Water Resources Systems Vol. 43, No. 7 Planning and Management: An Introduction to Meth- Smyle, J., G. Milne, and H. Qaddumi. 2009. Rural Watershed ods, Models and Applications. Paris: UNESCO. Management--the Power of Integration. Water Sector Maidment, D.R. 2002. Arc Hydro: GIS for Water Resources. Red- Board Practitioner Notes, Issue 28. World Bank, Wash- lands, CA: ESRI Press. ington, DC. Maidment, D.R., and D. Djokic. 2000. Hydrologic and Hydrau- Ward, F.A., and M. Pulido-Velazquez. 2008. "Water Conserva- lic Modeling Support with GIS. Redlands, CA: ESRI Press. tion in Irrigation Can Increase Water Use." Proceedings Medellín-Azuara, J., L.G. Mendoza-Espinosa, J.R. Lund, J.J. of the National Academy of Sciences, USA (PNAS). Harou, and R. Howitt. 2009. "Virtues of Simple Hydro- Wilby, R.L. and T.M.L. Wigley. 1997. "Downscaling general economic Optimization: Baja California, Mexico," Envi- circulation model output: a review of methods and ronmental Management 90:3470­78. limitations," Progress in Physical Geography, Vol. 21, Medellín-Azuara, J., J.R. Lund, and R.E. Howitt. 2007. "Water No. 4, 530­548. Supply Analysis for Restoring the Colorado River World Bank Institute. 2006. System Modeling in River Basin Delta, Mexico." Journal of Water Resources Planning and Management--Simulating and Optimizing the Re- Management, ASCE, 133 (5):462­71. sponse of a River Basin's Hydrology and Hydraulics to Mencius. 1970. Mencius. Translated by D.C. Lau. London, UK: Changes in Land and Water Use. Briefing Note 6 on Penguin Books. Integrated River Basin Management: From Concepts 22 References to Good Practice. World Bank, Washington, DC. http:// Driven Water Planning Model: Part 1, Model Charac- www-wds.worldbank.org/external/default/WDSCon- teristics." Water International 30:487­500. tentServer/WDSP/IB/2007/10/17/000020953_200710 Yates, D., D.R. Purkey, J. Sieber, A. Huber-Lee, and H. Gal- 17105228/Rendered/PDF/411550River0sy1ling0NOTE braith. 2005. "WEAP21--A Demand-, Priority-, and 1601PUBLIC1.pdf Preference-Driven Water Planning Model: Part 2, Aid- Wurbs, R.A. 1994. Computer Models for Water-Resources Plan- ing Freshwater Ecosystem Service Evaluation." Water ning and Management. IWR Report 94­NDS-7, July. International 30:501­12. Institute for Water Resources, Alexandria, VA. Yates, D., J. Sieber, D.R. Purkey, and A. Huber-Lee. 2005. "WEAP21--A Demand-, Priority-, and Preference- 23 annex: overview of key waTersheD managemenT moDels anD DeveloPer links AQUATOOL: http://www.upv.es/aquatool/index_E.htm MIKE 11, MIKE 21, MIKE FLOOD, MIKE SHE: http://www. mikebydhi.com/en.aspx BASINS: http://www.epa.gov/waterscience/basins MODFLOW : http://water.usgs.gov/software/lists/ground_ CALVIN: http://cee.engr.ucdavis.edu/CALVIN water DAMBRK: http://www.bossintl.com/products/download/ OpenMI: http://www.openmi.org item/DAMBRK OptiWin: http://www.upv.es/aquatool DWOPER: http://www.weather.gov/oh/hrl/nwsrfs/users_ manual/part2/_pdf/24dwoper.pdf PCM: http://www.cgd.ucar.edu/pcm eWater: http://www.ewatercrc.com.au SWAP: http://www.swap.aterra.nl EPANET: http://www.epa.gov/nrmrl/wswrd/dw/epanet.html SWAT: http://swatmodel.tamu.edu ESSA Technologies Ltd.: http://www.essa.com SWMM: http://www.epa.gov/ednnrmrl/models/swmm FLDWAV: http://www.fema.gov/plan/prevent/fhm/dl_fdwv. SimWin: http://www.upv.es/aquatool shtm TOPMODEL: http://www.epa.gov/nrmrl/pubs/600r05149/60 GAMS: http://www.gams.com 0r05149topmodel.pdf GCM: Global Circulation Model (type of model) WAS: http://www.wallingfordsoftware.com GFDL: http://www.gfdl.noaa.gov WEAP: http://www.weap21.org HadCM: http://www.metoffice.gov.uk/climatechange/ Miscellaneous Web Links science/hadleycentre Charles Howard and Associates: http://cddhoward.com HEC Suite of models with US Army Corps of Engineers, Hy- drologic Engineering Center: http://www.hec.usace.army.mil David Ford and Associates: http://www.ford-consulting.com HydroPlatform: http://www.hydroplatform.org Delft Hydraulics: http://www.wldelft.nl/soft/intro/index.html IGSM: http://hydrologicmodels.tamu.edu/PDF/Precipitation- DHI: http://www.dhigroup.com runoff/General/CVGSM.pdf (Largely replaced by IWFM ­ http://baydeltaoffice.water.ca.gov/modeling/hydrology/IWFM) US Army Corps of Engineers, Institute for Water Resources: http://www.svp.iwr.usace.army.mil MashWin: http://www.upv.es/aquatool/manuales/ ManMashwinEsp.pdf 25