August 2006
                                       External Review Draft
 Climate and Land Use Change Effects on
Ecological Resources in Three Watersheds:
             A Synthesis Report
          External Review Draft Report

          EPA Global Change Research Program
                   Susan H. Julius
                  Britta G. Bierwagen
                  Thomas E. Johnson

                 ICF International, Inc.
                    Randall Freed
                    Susan Asam
                    Sarah Shapiro
                  August 2006
      U.S. Environmental Protection Agency
       Office of Research and Development
  National Center for Environmental Assessment
        Global Change Research Program

      August_2006_...                                                  _ _.External Review Report:. Watershed. Case.S.fypfes
 i                                          Preface

 2    The Environmental Protection Agency's Global Change Research Program (GCRP), located
 3    within the Office of Research and Development, focuses on assessing how potential changes in
 4    climate and other global environmental stressors may impact water quality and aquatic
 5    ecosystems, air quality, and human health in the United States. The Program's focus on aquatic
 6    ecosystems is consistent with the Research Strategy of the U.S. Climate Change Research
 7    Program - the federal umbrella organization for climate change science in the U.S. government -
 8    and is responsive to EPA's mission and responsibilities as defined by the Clean Water Act.
 9    The GCRP's aquatic ecosystem assessments also address an important research gap identified in
10    the 2000 report entitled Water: The Potential Consequences of Climate Variability and Change
11   for the Water Resources of the United States (Gleick and Adams, 2000) that contributed to the
12    2001 U.S. National Assessment. In this document, the authors express the need for methods to
13    represent the underlying mechanisms linking food webs and hydrologic regimes in order to make
14    more useful projections of the effect of climate-induced changes in the hydrologic regime on
15    habitat, species, and the overall health of aquatic ecosystems.
16    Since 1998, the National Center for Environmental Assessment's office of the GCRP has
17    assessed the consequences of global  change on aquatic ecosystems. Through its assessment
18    projects, this Program has provided timely scientific information to stakeholders and policy
19    makers to support them as they decide whether and how to respond to the risks and opportunities
20    presented by global change.
21    Effects of global change drivers differ by place and in scale, necessitating place-specific impacts
22    information to enable stakeholders to respond appropriately. Place and scale also determine
23    appropriate adaptation strategies and expected outcomes. This report is a synthesis of three
24    watershed case-study assessments conducted by GCRP to advance the capability of managers to
25    consider climate and land use change in watershed management decisions. The goal of this
26    report is three-fold: (1) to understand the potential impacts of global changes and the availability
27    of effective responses at the watershed scale across different geographic regions, (2) to learn how
28    GCRP and other research programs might improve the process of conducting future assessments,
29    and (3) to improve GCRP's and other research programs' understanding about effective ways to
30    inject climate change impacts and adaptation information into decision making processes.
31    Toward that end, this report compares and contrasts methods and processes employed by the
32    three case study teams to learn effective analytic, project management, and decision support
33    approaches.
36                                            Peter Preuss, Ph.D.
37                                            Director
38                                            National Center for Environmental  Assessment
39                                            Office of Research and Development
40                                            U.S. Environmental Research Program

     August_2006_...                                                  _ _.External Review Report:. Watershed. Case.S.fypfes


         San Pedro Case Study
The contributions of many people's experiences and insights made this report possible. The
authors would like to thank those individuals whose help was particularly valuable. First and
most important are those individuals who participated on the water shed case study teams funded
by EPA. They devoted their excellent scientific and leadership skills to the conduct of each case
study and made impressive methodological advances in the science of climate change impacts.
They also gave of their time and energy to help us understand how we might improve the process
of doing assessments that produce information befitting the needs of decision makers. These
teams of researchers are:

   Maryland Case Study    Glenn E. Moglen, Principal Investigator, University of Maryland
                           Nancy Bockstael, University of Maryland
                           Karen Nelson, University of Maryland
                           Margaret Palmer, University of Maryland
                           James E. Pizzuto, University of Delaware

                           JeffT. Price, Principal Investigator,  California State Univ, Chico
                           Kate Baird, University of Arizona
                           Mark D. Dixon, Arizona State University
                           Hector Galbraith, Galbraith Environmental Sciences LLC
                           Thomas Maddock, University of Arizona
                           Juliet Stromberg, Arizona State University

                           Annette Huber-Lee, Principal Investigator, Tellus Institute
                           Hector Galbraith, Galbraith Environmental Sciences LLC
                           David Pur key, National Heritage Institute
                           Jack Sieber, Tellus Institute
                           David Yates, National Center for Atmospheric Research

We would also like to thank Catriona Rogers who helped develop the Ecosystems Focus Area
aquatic research, and who contributed to the vision and direction of the watershed assessments
discussed in this report. Dr. Rogers also managed the case study conducted by the University of
A number of internal reviewers earned our appreciation for identifying improvements and
corrections that had escaped us. Their suggestions as well as their challenging questions helped
us to prepare a report that will be useful to a larger audience of readers.  They are Naomi
Detenbeck (EPA ORD/NHEERL), Bruce Herbold (EPA Region 9), Chris  Weaver (AAAS at
EPA), and Kate Schofield (EPA ORD/NCEA).
         Sacramento Case Study

     August_2006_...                                                  _ _.External Review Report:. Watershed. Case.S.fypfes
 i                                  Executive Summary

 2    Purpose of the Report
 3    The effects of global change drivers differ by place and in scale, necessitating place-specific
 4    impacts information to enable stakeholders to respond appropriately. Place and scale also
 5    determine appropriate adaptation strategies and expected outcomes. This report is a synthesis of
 6    three watershed case-study assessments conducted by the Global Change Research Program
 7    (GCRP) - a unit of the National Center for Environmental Assessment within the Office of
 8    Research and Development - to advance the capability of managers to consider climate and land
 9    use change in watershed management decisions. The report provides a summary of the scientific
10    findings from those three case studies conducted in the San Pedro River Watershed, the
11    Sacramento River Watershed, and several small watersheds in Maryland.  It also provides
12    insights we gained from a comparison across case studies of the process of conducting watershed
13    assessments to improve our capability to support decisions.

14    Case  Study Results
15    The goal of the Maryland case study was to better understand how the effects of climate
16    variability and change on stream ecosystems depend on land use choices in surrounding areas in
17    order to assist regional planners to adapt to climate change and variability. The team developed
18    and  applied the Forecasted Index Simulation System (FISSh) to model the combined effects of
19    land use change and climate change on stream fish assemblages over the next century. Two
20    scenarios were used in their analyses:  (1) baseline scenario with managed growth and no climate
21    change; and,  (2) urban sprawl with climate change (USCC). Mean annual air temperature
22    increased by  3.2  degrees Celsius and maximum air temperature rose by 10.5 degrees Celsius
23    over the baseline scenario. These temperature changes represent the difference between the mean
24    and  maximum air temperatures from the time period of 1995-2004 to the time period of 2085-
25    2094, as projected by the HadCM3 model. Although total precipitation was relatively
26    unchanged, the number of large storms increased from 4 to  10 per decade. Streamflow increased,
27    sediment delivery increased, and annual mean water temperature rose by 0.9 degrees Celsius.
28    Fish food sources showed significant seasonal fluctuations under the USCC scenario, and the
29    availability of detritus dropped to only 2 percent of the baseline level. Because of the lower
30    levels of detritus, invertebrates had food deficits from April to July and small fish showed
31    deficits for most of the year. Non-native warm water fish species were relatively  advantaged by
32    projected land use change and climate change. Under the combined influence of climate change
33    and  projected land use change, there could be a considerable change in community composition
34    and  a loss of  diversity.
35    The goal of the San Pedro case study was to model the likely effects of climate change,
36    urbanization, and groundwater withdrawals on ecological resources and biodiversity in the San
37    Pedro Riparian National Conservation Area (SPRNCA) in order to aid in managing the area's
38    development and hydrologic conditions. Five climate scenarios were evaluated. The team
39    analyzed species, vegetation, and habitat suitability first, and then used linked models to
40    incorporate vegetation data into the groundwater and surface water models and to tie the
41    fluctuations of groundwater level to evapotranspiration processes. Results  showed that altered
42    hydrology and climate change would fragment existing riparian and wetland communities and
43    lead to their replacement by more mesic or xeric communities (i.e., vegetation more typical of
44    the desert matrix). The influence of climate change on pioneer riparian communities depended

      August_2006_...                                                  _ _.External Review Report:. Watershed. Case.S.fypfes
 1    on the magnitude and direction of precipitation changes: less winter precipitation would result in
 2    fewer winter floods, lower rates of channel migration, and much lower cottonwood and willow
 3    recruitment rates; increased winter precipitation would result in larger and more frequent winter
 4    floods, higher channel migration rates, and higher cottonwood and willow recruitment rates.
 5    Avian biodiversity would be affected, with some of the most abundant bird species being the
 6    most adversely affected by changes in the vegetative community. Results from the three driest
 7    climate scenarios suggested that the gallery forest would be fragmented or non-existent and
 8    would result in biodiversity loss and a likely drop in ecotourism. However, results from the
 9    warm and very wet scenario suggested that the water supply to the ecosystem would be adequate
10    enough to maintain ecosystem services and ecotourism.
11    The goal of the Sacramento case study was to assess how global change-related alterations in
12    water supply and water demand would affect important freshwater ecosystem services in the
13    Sacramento Basin using an integrated decision support tool that would inform on issues such as
14    reservoir location, Federal Energy Regulatory Commission (FERC) dam re-licensing, and system
15    operations to preserve the ecosystem services of interest or of regulatory necessity. The case
16    study team applied the Water Evaluation and Planning (WEAP21) modeling system to link
17    climate and land use/land cover conditions with watershed conditions, water supply and
18    anticipated demands, ecosystem needs, infrastructure, the regulatory environment, and water
19    management options.  Four downscaled climate scenarios were evaluated and all four showed
20    reductions in water availability — there were large impacts on supply at the end of the 100-year
21    simulations. Reservoir levels were much lower in the late summer and  early fall and groundwater
22    pumping increased, except when adaptation measures were simulated. The Sacramento River's
23    water temperature regime would be altered, leading to further reductions in habitat for Chinook
24    salmon due to exceedences of critical spawning and rearing temperatures. Adaptations, such as
25    cropping practices - e.g., improving irrigation efficiency and changing cropping patterns -
26    resulted in a decline in water supply requirements. Managing the releases of cold water stored in
27    reservoirs alleviated some of the future impacts of climate change on habitat for Chinook
28    salmon.

29    Cross-cutting Findings and Recommendations
30    The three case studies contributed both methodological advances and new results to the body of
31    knowledge on climate change impacts that will be broadly useful in three ways: the results
32    themselves may be  extrapolated to similar systems; the methods used to link process models
33    across disciplines may be used by other assessment teams and in other  geographic regions of the
34    country, and the insights gained about the assessment process will be helpful to any research
35    institution seeking to produce useful climate impacts information for decision makers.
36    One significant methodological advance across the three case studies was the development of
37    integrated, interdisciplinary models. Advantages to linking models included: a reduction in
38    uncertainties associated with ignoring system interactions that could not be modeled without an
39    integrated approach; a reduction in uncertainty for decision makers in terms of forecasting
40    effects; an improved ability to provide new information more quickly and easily about a whole
41    suite of components that address decision-relevant questions; and, an improved ability to analyze
42    multiple scenarios quickly. Results from the case studies also show that investing time up-front
43    to clearly  define inputs, outputs, and interactions among submodels helps facilitate a smooth
44    linkage between submodels.

      August_2006_...                                                    _ _.External Review Report:. Watershed. Case.S.fypfes
 1    Across the case studies, several keystone capabilities or practices were identified that would be
 2    useful for EPA to provide to researchers conducting the assessments. Recommendations were to:
 3    •  Develop and apply best practices for converting GCM output to watershed modeling input
 4    •  Provide tools to develop or apply trend analysis of precipitation and hydrology to
 5       complement GCM output
 6    •  Provide consulting services in keystone skills, such as climate scenario development and
 7       habitat suitability analysis
 8    •  Provide guidance and techniques for expressing uncertainty
 9    While each of the case studies was thorough and innovative in assessing climate change impacts,
10    it was found that attaining the objective of providing decision support was more difficult. To
11    better attain this objective, criteria for selecting and planning future assessments should consider
12    (1) where decisions are being made that are sensitive to climate change, and (2) where there are
13    existing relationships with decision makers that would enable the project team to provide
14    relevant decision support products.  To produce good science and provide effective decision
15    support, spatial  and temporal scale should also be considered - the scale at which GCM and
16    watershed-level information is available, the scale at which key endpoints are assessed, and the
17    uncertainty introduced by bridging the gap.
18    No uncertainty bounds were estimated for the  results produced by each case study, making it
19    difficult to assess whether downscaled climate data from  general circulation models can credibly
20    be used at those spatial scales. The inability to address this issue suggests that it is important to
21    develop a set of best practices or tools  appropriate for analyzing uncertainty in assessments and
22    decision support, and that future assessments should include a requirement to develop an
23    uncertainty analysis plan as an initial deliverable that reflects feedback from decision makers on
24    what types of uncertainty analysis would be most relevant to decision-making. Uncertainty
25    should be addressed more comprehensively, and, where possible, quantitatively to ensure
26    scientific rigor and decision relevance.
27    It would be useful to set out an uncertainty analysis plan as one of the initial deliverables in the
28    process, and to design the plan to reflect feedback from decision makers on what types of
29    uncertainty analysis would be most relevant to decision-making. This feedback would help
30    determine whether the plan should focus on developing several alternative scenarios,  or a risk
31    analytic approach that explicitly assigns probabilities to various components, or some other
32    qualitative or quantitative characterization of uncertainty.
34    Finally, because stakeholder engagement is resource and  time intensive,  it was determined that
35    streamlining and focusing stakeholder-related  efforts is necessary and desirable.
36    Recommendations  to improve stakeholder processes include:
37    •  Focus on decision makers rather than a generic definition of stakeholders and develop a
38       working partnership that builds  technical capacity within the decision making body to
39       increase the likelihood that climate change impacts will be considered beyond the particular
40       assessment
41    •  Build on existing stakeholder relationships to  save time and resources and to open doors to
42       collaboration with new stakeholders who may be similarly interested in study findings
43    •  Facilitate interaction of researchers and decision makers through use of common terminology
44       (e.g., ecosystem services), and through developing Agency expertise to manage public inputs
45       to the study itself, rather than relying on the assessment team alone

      August_2006_...                                                     _ _.External Review Report:. Watershed. Case.S.fypfes
 1    •   As part of the assessment process, set aside time to establish credibility with decision makers
 2       to make them more comfortable with the assessment process and findings and how such
 3       findings fit into ongoing decision processes
 4    •   Ensure that capacity for doing assessments is in place before the project starts by establishing
 5       and applying a set of critical success factors, such as: available, high quality data and models,
 6       clear goals, public awareness of the issue, etc.
 7    •   To better support decisions, ensure that assessment methodologies and model results are
 8       transferable and widely applicable
 9    •   Develop a framework and method for identifying and prioritizing research that is decision-
10       driven and where the information will be most useful

      August_2006_...                                                  _ _.External Review Report:. Watershed. Case.S.fypfes
 i                                    Table of Contents
 2                                                                                      Page
 4    1.       Introduction	1
 5      1.1    Purpose of the Report	1
 6      1.2    The Case Studies	2
 7        1.2.1     Motivation for the Watershed Case Studies	2
 8        1.2.2     Criteriafor Selecting Case Studies	4
 9        1.2.3     The Portfolio of Case Studies	4
10    2.       Case Study Results	8
11      2.1    Maryland	8
12        2.1.1     Goals of the Case Study Assessment	8
13        2.1.2     Major Stressors	8
14        2.1.3     Assessment Methods	9
15        2.1.4     Impacts  & Findings	10
16        2.1.5     Methods and Results Applicable to Other Watersheds	11
17      2.2    San Pedro	11
18        2.2.1     Goals of the Case Study Assessment	12
19        2.2.2     Major Stressors	12
20        2.2.3     Assessment Methods	12
21        2.2.4     Impacts  & Findings	14
22        2.2.5     Methods and Results Applicable to Other Watersheds	15
23      2.3    Sacramento	15
24        2.3.1     Goals of the Case Study Assessment	16
25        2.3.2     Major Stressors	16
26        2.3.3     Assessment Methods	17
27        2.3.4     Impacts  & Findings	18
28        2.3.5     Adaptation Options	19
29        2.3.6     Methods and Results Applicable to Other Watersheds	19
30    3.       Cross-cutting Findings	21
31      3.1    Assessment Processes	21
32        3.1.1     Team Composition and Management	21

      August_2006_...                                                  _ _.External Review Report:. Watershed. Case.S.fypfes
 1        3.1.2     Research Approach	22
 2        3.1.3     Scale	23
 3        3.1.4     Uncertainty	24
 4      3.2    Stakeholder Processes	26
 5        3.2.1     Defining and Identifying Stakeholders	26
 6        3.2.2     Balancing Stakeholder and Researcher Interests	27
 7        3.2.3     Engaging Stakeholder Communities	28
 8        3.2.4     Establishing Credibility	29
 9        3.2.5     Communicating Results	29
10      3.3    Relevance of Impacts and Adaptation to Decision Making	29
11    4.       Conclusions and Recommendations for Future Watershed Assessments	32
12      4.1    Contributions to Body of Knowledge on Climate Change Impacts	32
13        4.1.1     Methodology Development	32
14        4.1.2     Key Results	33
15      4.2    Assessment Process	34
16        4.2.1     Provide Keystone Capabilities and Tools to Assessment Teams	34
17        4.2.2     Emphasize Model Linkages	36
18        4.2.3     Change Selection Criteria for Assessments	36
19        4.2.4     Establish Forum for Researchers to Compare Notes	36
20        4.2.5     Require an Uncertainty Analysis Plan	36
21      4.3    Stakeholder Process	37
22        4.3.1     Focus on Decision Makers	37
23        4.3.2     Build on Existing Stakeholder Relationships	38
24        4.3.3     Facilitate Interaction of Researchers and Decision Makers	38
25        4.3.4     Establish Credibility with Decision Makers	38
26        4.3.5     Communicate Results using Common Language	38
27      4.4    Final Thoughts and Future Directions for GCRP	39

      August_2006_...                                                   _ _.External Review Report:. Watershed. Case.S.fypfes
 i                                        1. Introduction

 2    1.1  Purpose of the Report
 3    The effects of global change drivers differ by place and in scale, necessitating place-specific
 4    impacts information to enable stakeholders to respond appropriately. Place and scale also
 5    determine appropriate adaptation strategies and expected outcomes. This report is a synthesis of
 6    three watershed case-study assessments conducted by the Global Change Research Program
 7    (GCRP) to advance the capability of managers to consider climate and land use change in
 8    watershed management decisions.
 9    The watershed case studies were initiated in 2002 to better understand the  effects of global
10    change on aquatic ecosystems within watersheds and to build capacity at appropriate levels of
11    decision making to respond to these effects. The case studies are now complete, and all three
12    have yielded valuable scientific findings and provided important lessons about assessment and
13    stakeholder processes. This report sets out to document and synthesize the results, findings, and
14    lessons learned by:
15           1.  Summarizing the assessment methods developed by the case studies and  key results,
16              emphasizing those that are applicable elsewhere  and those that support decision
17              making to adapt to climate change;
18           2.  Synthesizing insights on cross-cutting issues related to the assessment and
19              stakeholder processes (e.g., selecting endpoints, identifying and engaging
20              stakeholders);
21           3.  Assessing the relevance of the case studies to decision making  and adaptation to
22              climate change and land use change; and
23           4.  Discussing implications for future directions on watershed analyses.
24    Chapter 1 of this report provides an introduction, Chapter 2  describes the methods and results of
25    the case studies, Chapter 3 provides a discussion of cross-cutting findings, and Chapter 4
26    summarizes conclusions and implications for future watershed assessments.
27    The watershed case-study assessments were conducted by three EPA-funded research teams.
28    GCRP staff provided technical direction to each assessment team, and contributed directly to the
29    synthesis results presented in this report. Additional  support with this synthesis was  provided by
30    ICF International. The locations selected for our case studies are the  San Pedro River watershed
31    led by the American Bird Conservancy, the Sacramento River watershed led by the Tellus
32    Institute, and watersheds in the Washington, D.C. metropolitan area conducted by the University
33    of Maryland. GCRP solicited proposals from these research teams based on criteria established
34    prior to our selection of these locations. Criteria included having a diversity of geographic
35    regions and river ecosystem types represented, different land-use pressures (e.g., agricultural
36    pressures, urban growth pressures), different future climate-induced changes (e.g., increased
37    versus decreased streamflow), and different highly-valued ecosystem services. This  synthesis is
38    based on each case study's scientific publications, final reports, an expert meeting of case study
39    team members held partway through their assessment process, and a series of interviews held
40    with each case study team at the conclusion of their projects. The questions to which they
41    responded to in the interviews are the following:

      August_2006_...                                                   _ _.External Review Report:. Watershed. Case.S.fypfes
 1       1.  What are the major methodological advances developed in your case study?
 2       2.  What is the applicability of the methodologies that you employed to other watersheds? Is
 3          applicability tied to scale, assessment endpoints, regions, or some other factor(s)?
 4       3.  What do you regard as the most important/ interesting results?
 5       4.  To what extent do the case study findings apply to other watersheds? Is applicability tied
 6          to scale, assessment endpoints, regions, or some other factor(s)?
 7       5.  To what extent could the outputs of your project support decisions and what types of
 8          decisions are they? More specifically, would your results affect watershed management
 9          practices, and if so how? If you suspect they won't, what are the obstacles, if any?
10       6.  To what extent did you isolate land use change and climate change as driving factors and
11          what were the results?
12       7.  If you were to propose additional work, what do you think the next phase of the project
13          should entail?  What steps would be natural extensions of the work that has already been
14          done?
15       8.  What do you consider to be the most important lessons learned or recommendations for
16          future watershed assessments?
17    The watershed case study assessments and this report support EPA's strategic goal 4 (Healthy
18    Communities and Ecosystems) and the EPA Office of Water's responsibilities under the Clean
19    Water Act to "restore and maintain the chemical, physical and biological integrity of the
20    Nation's waters." Under the EPA GCRP 2008 annual performance goal #333, this synthesis
21    report fulfills the 2006 annual performance measure #529: "ExternalReview Draft report to
22    develop and evaluate information and tools on global change impacts in key watersheds. "

23    1.2  The Case Studies
24    1.2.1  Motivation for the Watershed Case Studies
25    The Global  Change Research Act of 1990 established the U.S. Global Change Research Program
26    (GCRP) to coordinate a comprehensive, multi-agency research program on global change.1
27    Within EPA, the Office of Research and Development (ORD) has the lead for conducting
28    research and assessments that examine the effect of climate, land use, and other factors on
29    aquatic ecosystems and providing decision support resources and adaptation options to
30    stakeholders.
31    GCRP initiated watershed case studies to gain a better understanding of the effects of global
32    change on aquatic ecosystems and water quality, and to build capacity to respond to these effects
33    at appropriate levels of decision making. That led to the choice of case study sites that differed
34    hydrologically and bioclimatically from each other,  but were representative of a broader array of
35    conditions within larger regions of the United States. These regions were the Western U.S., the
36    arid/semiarid Southwest,  and the Eastern U.S. We chose to focus at the watershed scale based on
37    the knowledge that the properties of aquatic systems are strongly influenced by  the surrounding
38    land and are often managed and analyzed as a component of a larger watershed. The case study
39    approach stems from a motivation to conduct assessments that fit into the existing watershed-
40    based  strategy used by U.S. water management programs to integrate water management

      August_2006_...                                                    _ _.External Review Report:. Watershed. Case.S.fypfes
 1    activities within hydrologically defined drainage basins or watersheds. Additionally, GCRP has
 2    historically had a program-wide emphasis on examining site- or region-specific impacts and
 3    adaptation measures.
 4    The assessment approach used for each case study integrates methods and concepts of ecological
 5    risk assessment, ecosystem services, scenario analysis, and stakeholder engagement processes.
 6    Climate change scenarios are used in conjunction with scenarios of other relevant global change
 7    stressors and quantitative and  conceptual models to examine the potential impacts of global
 8    change on aquatic ecosystems. Therefore, the following list of desired case study design
 9    elements were identified prior to selecting the case study sites:

10       •   Address the interaction of climate change with other stressors, especially land use
11           change. Over the past century, there has been a trend for a higher proportion of
12           precipitation to fall in intense events (e.g., more than 2 inches per event), and these
13           intense events contribute to non-point source pollution (Karl and Knight  1998). Climate
14           change is anticipated to amplify this effect. Land use change (especially urbanization)
15           modifies stream hydrology by affecting the proportion of precipitation that immediately
16           enters the stream as runoff, and, thus, can also result in a "flashier" flow pattern  (or
17           hydrograph) (Karl and Knight 1998). The case studies are designed to examine these (and
18           other) interactions.

19       •   Emphasize ecosystem  services. The concept of ecosystem services enables individuals
20           from a cross section  of society to express the values they hold for ecological processes or
21           functions using a common language that helps frame assessment questions relevant to
22           decision making. Most of the watershed management decisions address a subset of
23           ecosystem services that aquatic systems provide. These services-which include water
24           supply, hydropower, recreational amenities, habitat for species, and transportation-are
25           the amenities that motivate stakeholders. Thus, the case studies attempt to identify
26           assessment endpoints that relate to these services.

27       •   Involve stakeholders. The goals of an assessment are to communicate insights about the
28           possible consequences of global change and the potential for adaptive responses.
29           Stakeholder involvement is crucial throughout this process to ensure that the assessment
30           is timely and relevant,  and that results are communicated effectively.

31       •   Use a risk assessment approach. Consistent with the human health and ecological risk
32           assessment programs within ORD, we apply a risk assessment paradigm to our global
33           change assessments. The case studies were thus designed to clearly articulate the problem
34           and develop an analysis plan (problem formulation), conduct an exposure assessment,
35           effects assessment, and risk characterization, and to use best practices to produce high-
36           quality scientific results. Watershed assessments employ a modification of the strict
37           exposure-effects approach because multiple stressors are being examined. Climate and
38           land use scenarios are intended to serve as exposure scenarios in order to project a range
39           of potential effects.
40    With these design elements  serving as the genesis for the effort, EPA formulated the problem
41    that the case studies would address and selected a portfolio of three case studies.

      August_2006_...                                                    _ _.External Review Report:. Watershed. Case.S.fypfes
 1    1.2.2   Criteria for Selecting Case Studies
 2    The goal of the case studies was to build capacity at appropriate levels of decision making to
 3    assess and respond to potential global change impacts on aquatic ecosystems within watersheds.
 4    The scientifically complex environmental problems associated with global change are beginning
 5    to be addressed under circumstances of increasingly complicated decision making processes.
 6    Watershed management has become a process of balancing multiple objectives, such as drought
 7    and flood protection, habitat and species protection, and provision of adequate supplies of water
 8    for withdrawals for municipal,  industrial, and agricultural uses. Waters and watersheds
 9    increasingly are seen as complex systems comprising both ecological and human processes
10    (Webler and Tuler 1999). Undertaking a set of watershed case studies enabled us to do an
11    integrated examination of the processes of interest at scales that are amenable to decision making
12    and scientific analysis.
13    Potential case studies were evaluated using the following criteria:

14       •   The sites chosen should represent:
15           >  different geographic scales of a watershed system with respect to ecosystem services
16              and stakeholders;
17           >  different climate stressors;
18           >  different land use pressures; and
19           >  different vulnerabilities and intensities of use in the context of a variety of
20              current/existing stressors.

21       •   Each site chosen should have services that are highly valued by the local community (and
22           beyond the local community, if possible)

23       •   Because of limited resources, gathering original data was beyond the capability of the
24           GCRP. Therefore, sites chosen needed to have fairly detailed and  comprehensive data
25           sets. Supporting research conducted in the selected location(s) was considered an
26           additional benefit.
27    These criteria were used to select three watershed case studies that address different scales,
28    assessment endpoints, and hydrologic regimes.
29    1.2.3   The Portfolio of Case Studies
30    Three case study locations were chosen based on the above criteria. The selected case study sites
31    represented diverse geographic regions and aquatic ecosystem types, different land use pressures
32    (e.g., agricultural pressures, urban growth pressures), and different future climate-induced
33    changes (e.g., increased versus decreased runoff). Each site provided highly-valued ecosystem
34    services and had substantial amounts of data and existing research on which the study teams
35    were able to build. Table 1 provides a comparison of some of the key aspects  of each of the case
36    studies and Figure  1 shows the location of the case studies across the United States.
37    The Maryland case study focused on riverine systems and their associated riparian zones in four
38    selected watersheds of the greater Washington, DC, metropolitan area. Ecosystem services of
39    interest involved the maintenance of water quality, fish and invertebrate species, and primary
40    production and the availability  of detritus. Primary stressors of concern included climate change

    August_2006_...                                                   _ _.External Review Report:. Watershed. Case.S.fypfes
1    and land use change, specifically disturbances resulting from urbanization, increasing
2    imperviousness in watersheds, and destruction of streamside vegetation.
3                   Table 1: Comparison of the Three Watershed Case Studies
                                San Pedro
    (other than
    change and
    land use

              Sub-watershed scale (13-
              28 mi2).

              Variance in daily
              streamflow has changed
              dramatically over the
              past 50 years;  enhanced
              peak flows and reduced
              baseflows are  attributed
              to increased

              Habitat suitability for
              fish (temperature,
              siltation, flashiness,
              riparian zone condition,
              riffle vs. pool  habitat).
              Changes in water
              temperature, siltation
              rates, streamflow,
              riparian zone condition,
              and stress on aquatic
              habitats due to

              Linked climate,
              hydrology, ecosystem,
              land-use economics, and
              geomorphology models.
Basin scale (42,000 mi2
SF Bay watershed).

Flow maxima typically
occur during the late
winter through spring
period and flow minima
(dramatically reduced
relative to peak flows) in
the late summer and early

Services related to water
use (flow and
seasonality), ecological
functioning (flow and
seasonality), and water
quality improvement by
wetlands (flow,
seasonality, sediment
rate,  toxic and nutrient
loads, and salinity

In-stream water
withdrawals for urban
populations, agriculture,
and industry.
Linked climate,
hydrology, ecosystem,
and wetland models.
Watershed scale (-2,500

The majority of the flow
in the San Pedro River
comes from the
groundwater aquifer, but
there is some seasonal
Maintenance of avian
biodiversity, sustained
urban water supplies, and
recreational uses
(groundwater and surface
water flow, structure and
composition of
vegetation, avian habitat
suitability, and indicator
withdrawals for
agricultural and
municipal uses;
increasing water demand
due to population growth.
Linked climate,
hydrology, ecosystem,
groundwater flow, and
geomorphology models.
1  Source: Moglen G. E., K. C. Nelson, M. A. Palmer et al. 2004. Hydro-ecologic responses to land use in small
urbanizing watersheds within the Chesapeake Bay watershed. Pages 41-61 in R. DeFries, G. Asner, and R.
Houghton, editors. Ecosystems and Land Use Change. American Geophysical Union, Washington, DC.

      August_2006_...                                                   _ _.External Review Report:. Watershed. Case.S.fypfes
      The San Pedro case study was located in the Upper San Pedro River riparian ecosystem in
      southeastern Arizona and northern Sonora, Mexico. This area supports a riparian ecosystem that
      maintains biodiversity at the ecotones between the Sonoran and Chihuahuan deserts and the
      plains grassland. The area contains one of the richest assemblages of species and supports one of
      the most important migratory bird habitats in western North America. The ecosystem services of
      interest thus included the maintenance of avian biodiversity, sustained urban water supplies, and
      recreational uses. Primary stressors of concern included groundwater pumping, climate change,
      and population growth.
      The Sacramento case study was located in the Central Valley of California from the headwaters
      of the San Joaquin River in the south to the headwaters of the Sacramento River in the north. The
      area's ecosystem services included maintaining water quality and freshwater habitat for aquatic
      life; the watershed also provides water to the regional agricultural, municipal and industrial
      sectors. Primary stressors of concern included land use change, population growth, and climate
                       Study Area Watersheds
     Figure 1: Geographic locations of case studies across United States.
     The studies' fundamental approaches were similar; all three case studies linked climate,
     hydrology, and ecosystem models. At their core, the analytic frameworks of all three studies
     were driven by integrated modeling systems that start by simulating the effect of climate change
     on hydrologic characteristics, and subsequently address how changes in these characteristics
     affect ecosystem functioning. All three case studies also used results of large-scale climate
     models - known as General Circulation Models, or GCMs - to provide the bounds for, or to
     drive the climate scenarios. In the Sacramento and Maryland case studies, GCM outputs
     provided the basis for creating downscaled scenarios of temperature, precipitation, and derivative
     climate parameters. The San Pedro case study team developed climate scenarios that represent a
     reasonable set of potential climate trajectories, given the range of projections for the region

      August_2006_...                                                   _ _.External Review Report:. Watershed. Case.S.fypfes
 1    derived from climate models (SRAG, 2000). They used a 52-year daily time series of historic
 2    weather data (1951-2002) to create transient climate scenarios for the period 2003-2102. All case
 3    studies used multiple climate scenarios rather than limiting their investigation to one particular
 4    future projection. This attempt to bound the range of plausible futures was used in recognition of
 5    the documented uncertainties inherent in simulating future climate.
 6    The three case studies all examined climate change along with population and other land use-
 7    related stressors, but the choices of specific stressors were different. For example, the
 8    Sacramento River Watershed study included in-stream water withdrawals; the San Pedro study
 9    carefully examined groundwater withdrawals; and the Maryland study focused on sediment load
10    due to land use change. Because of the differences in focus, there were also differences in model
11    components that supplemented the core climate-hydrology-ecosystems models. In the
12    Sacramento River Watershed, model components were added to simulate ground water flow and
13    geomorphology. The San Pedro study  team developed a model to simulate the effects of flow
14    changes on riparian vegetation. The Maryland team developed geomorphological models to
15    simulate changes in sediment load and bed sediment composition.
16    The relative effects of climate change, land use change, and other stressors demonstrated by  each
17    of the groups showed a mixed response, with each of the systems exhibiting different
18    sensitivities based on the region, the current stressors, management goals, and anticipated
19    changes.

      August_2006_...                                                   _ _.External Review Report:. Watershed. Case.S.fypfes
                                     2.  Case Study Results
 2    The following section discusses the three case studies, including background on each of the
 3    regions, goals of the project, major stressors, assessment methods, results, adaptation options (if
 4    analyzed), and how the case studies are applicable to other regions.

 5    2.1  Maryland
 7    The research team for this study assessed the potential
 9    effects of the interaction of climate variability and
11    change with land use change, in terms of ecological
13    structure and functioning of stream ecosystems in four
15    selected watersheds of the greater Washington, DC,
17    metropolitan area.  The watersheds lie primarily within
19    the Piedmont physiographic province, and range in
21    size from 13 to 28  square miles—much smaller than
23    the watersheds addressed by the other case studies.
25    These sites were selected because they are all
27    experiencing major changes in land use but with
29    differing patterns.
31    As shown in the accompanying map, three of these
33    watersheds (marked 3, 8, and 9 in the map) are in a
35    single county (Montgomery County) but one of them
37    (Cattail, number 2  in the map) lies in adjacent Howard
39    County, which has different growth and planning
41    policies. All four watersheds have similar amounts of
42    land left in forest; however, Northwest Branch and Paint Branch (numbers 8 and 9 in the map,
43    respectively) have  more residential development, whereas Hawlings (number 3 in the map) and
44    Cattail have more agricultural land. Most of the urban development in these watersheds occurred
45    since World War II, with additional development episodes in the late 1960s and early 1970s. The
46    study area includes first-order through third-order streams in the four watersheds.
47    2.1.1   Goals of the Case Study Assessment
48    The proj ecf s goal  was to better understand how the effects of climate variability and change on
49    stream ecosystems depend on land use choices in surrounding areas. This understanding was
50    intended to provide decision makers with information about the ecological consequences of
51    alternative land use configurations that will assist them in developing potential strategies for
52    adapting to climate change and variability.
53    2.1.2   Major Stressors
54    Climate change  and land use and land cover change, specifically urbanization, increases in
55    impervious surface, and destruction of streamside vegetation are all causing stream degradation
56    at the Maryland case study sites (Nelson et al. 2006). Streams, which occupy topographic lows,
57    collect runoff and sediment discharge, making them highly vulnerable to land use and climate
58    change. Urbanization, in particular, is  a major stressor on habitats in Maryland, causing changes
59    in aquatic temperature, siltation rates,  streamflow, riparian zone condition, and the availability of
60    riffle versus pool-type habitats for fish (Nelson and Palmer 2006).

     August_2006_...                                                  _ _.External Review Report:. Watershed. Case.S.fypfes
Climate change is projected to cause a 3 to 5 degree Celsius wanning nationally by 2100 (Nelson
and Palmer 2006), but the consequences of this warming depend on the seasonality of
temperature shifts. Fewer but more intense storms could produce drought- and storm-related
heating in much the same way that urbanization does (Nelson and Palmer 2006).
The specific stressors resulting from climate change and land use change addressed in the case
study include water temperature increases, changes in sediment transport regimes, increases in
runoff-associated pollution, streambed disturbance, and several other closely related factors
(Moglen et al. 2006.).
2.1.3  Assessment Methods
As the endpoint of interest for their watershed study, the research team chose the suitability of a
stream environment for selected fish species. Fish, with their relatively long lifespans,
dependence on numerous resources, and sensitivity of reproductive success to environmental
conditions, are effective integrators of systemic stressors (Nelson et al. 2006).
The team developed and applied the Forecasted Index Simulation System (FISSh) (Nelson et al.
2006) to model the combined effects of land use change and climate change on stream fish
assemblages over the next century. FISSh integrates five submodels, including downscaled
climate projections, stream hydrology, geomorphology, water temperature, and biotic responses,
to forecast future habitat availability on a day-to-day basis. The continuous streamflow model
includes three different forms of runoff production: surface runoff, subsurface runoff, and
groundwater runoff (Pizzuto et al. 2006). FISSh also models the food resource base as a daily
function of the input submodels and then combines and forecasts conditions for fish growth and
reproduction in Piedmont headwater streams
by modeling the entire fish assemblage using
readily available information and seasonal
variability (Nelson et al. 2006).
The case study team chose two scenarios to
evaluate the impacts of land use change and
climate change. The first was termed managed
growth/no climate change (MGNCC) and the
second was termed urban sprawl/climate
change scenario (USCC). MGNCC used
downscaled HadCM3 General  Circulation
Model (GCM) data from 1995-2004 for daily
precipitation and temperature, with a year-
round average temperature of 13.1 degrees
Celsius. This scenario assumed 10 percent
   Table 2: Comparison of Maryland
Climate and Land Use Change Scenarios
Mean annual air

Total precipitation

Rainfall events > 4

Max single day
                     USCC    MGNCC
 16.7 °C
13.1 °C
1308cm    1188cm
26.7 cm    17.4 cm
impervious surfaces, 20 percent forested lands with intact forested buffers, and no construction
in the watershed; this scenario served as the baseline. The USCC model used downscaled
HadCM3 GCM data from 2085-2094 with an average year-round temperature of 16.7 degrees
Celsius. This model assumed an imperviousness of 30 percent, 1 percent forested lands with no
forested buffers, and 2 percent construction in the watershed (Pizzuto et al. 2006; Nelson et al.
2006). Table 2 summarizes these scenarios.
The hydrological submodel used projected daily streamflow over the course of the scenario and
required a daily precipitation and temperature time series as well as characteristics of the land

      August_2006_...                                                  _ _.External Review Report:. Watershed. Case.S.fypfes
 1    use and geology of the system. The model output is a daily time series of discharge (Nelson et al.
 2    2006). The geomorphological submodel, a sediment transport model, was used to compute
 3    changes to the stream bed. Output was on a daily time-step and included particle size
 4    distribution, bedload and suspended material discharge, turbidity, areal fraction of the bed in
 5    motion, and interstitial clogging (Pizzuto et al. 2006). The instream temperature submodel
 6    predicted minimum and maximum instream temperature using a daily air temperature series
 7    derived from the downscaled GCM. The model used air temperatures from the same day and two
 8    earlier days (Nelson and Palmer 2006).
 9    2.1.4  Impacts & Findings
10    The two climate change and land use change scenarios differed in that the mean annual air
11    temperature increased by 3.2 degrees Celsius in the climate stressed scenario (USCC) relative to
12    the baseline, and the maximum air temperature rose by 10.5 degrees Celsius (Nelson et al. 2006).
13    While total precipitation was similar in the USCC and MGNCC scenarios, the number of large
14    storms per decade increased from 4 in MGNCC to 10 in USCC. In particular, there was a more
15    than two-fold increase in discharge associated with the very large storms under USCC, which
16    was due to both the increased precipitation per storm event and decreased infiltration capacity
17    (Nelson et al. 2006). The projected precipitation trends were more significant than future
18    temperature trends in their influence on hydrological and ecological processes (Moglen et al.
19    2006).
20    In addition, results of the geomorphologic submodel indicated that sediment delivery increased
21    in the USCC scenario due to projected construction in the study area and the frequency of runoff
22    events.  The increased sediment delivery resulted in a higher frequency of high turbidity
23    conditions. Annual mean water temperature rose by 0.9 degrees Celsius and annual maximum
24    water temperature rose 5.4 degrees Celsius between the MGNCC and USCC models. Due to the
25    presence of riparian buffers, there were no storm-related temperature surges in the MGNCC
26    model; however, the USCC model showed surges on about 11  percent of summer days. In
27    addition, fish food sources showed significant seasonal fluctuations under the USCC scenario,
28    and the availability of detritus dropped to only 2 percent of the MGNCC level. Because of the
29    lower levels of detritus, invertebrates had food deficits from April to July in the USCC scenario
30    and small fish showed deficits for most of the year (Nelson et al. 2006).
31    Results from FISSh included results for each of the following indices for each species: spawning
32    day availability; spawning substrate; juvenile growth; washout on eggs and young-of-year; adult
33    growth; feeding efficiency, and thermal maximum (Nelson et al.  2006). The two indices offish
34    ecology most affected by land use change and climate change were the index of adult growth and
35    index of washout on eggs;  they were both influenced by hydrological alterations and siltation.
36    When all seven of the indices were considered jointly, modeling  showed that up to three-quarters
37    of the fish species would be highly stressed. This finding suggests that under the combined
38    influence of climate change and projected land use change, there could be a considerable change
39    in community composition and a loss of diversity. Almost three-fourths of the present fish
40    species (30 out of 39) in the Maryland Piedmont watersheds were projected to be negatively
41    affected by urban sprawl /  climate change scenario conditions  such as: a shift in spawning times
42    due to shift in temperature (14 species); reductions in growth due to either reduced food or
43    increased temperature (14  species); reduced feeding efficiency due to turbidity (22 species); loss
44    of coarse woody debris needed for spawning structure (3 species); and surges of lethally hot
45    water running off of pavement during summer thunderstorms (2 species). (Nelson et al. 2006).

      August_2006_...                                                   _ _.External Review Report:. Watershed. Case.S.fypfes
 1    Not all ecological processes are negatively influenced by the projected climate and land use
 2    changes, but when they are combined, predominantly negative effects emerge. In addition, non-
 3    native warm water fish species may be relatively advantaged by projected land use change and
 4    climate change (Nelson et al. 2006).
 5    2.1.5  Methods and Results Applicable to Other Watersheds
 6    The models used in the Maryland case study are transferable to regions where similar processes
 7    are dominant; however, all of the models would require re-parameterization to be used in other
 8    locations. The ecological models were designed specifically for the purpose of being portable;
 9    they use readily available information and could be applied in other regions. The fish assemblage
10    is specific to the Maryland Piedmont and would be easier to apply to a similar region such as the
11    North Carolina Piedmont rather than the Maryland coastal plain, even though the latter is
12    geographically closer to the current site.

13    2.2  San Pedro

14    The Upper San Pedro River riparian ecosystem in
15    southeastern Arizona and northern  Sonora, Mexico, is of
16    critical importance in maintaining regional biodiversity at
17    the ecotones between the Sonoran and Chihuahuan deserts
18    and the plains grassland.  It contains one of the richest
19    assemblages of species and supports one of the most
20    important migratory bird habitats in western North
21    America. The biodiversity found along the Upper San
22    Pedro River matches or exceeds that found almost
23    anywhere else in the United States. More than 20 different
24    biotic communities occur in the basin, and the river
25    sustains 3 biotic types that are considered threatened:
26    Fremont cottonwood/Goodding willow forests, cienegas,
27    and big sacaton grasslands (Price et al. 2005).
28    Riparian ecosystems are fragile, especially those found in
29    arid climates. Water is crucial to these communities, and
30    the abundance, diversity, and health of these ecosystems
31    are strongly influenced by the hydrologic regime, including
32    the amount, timing, and pattern of surface and groundwater
33    flow. Surface water flow varies considerably both between
34    and within years. During periods of storm flows, the
35    shallow alluvial aquifer is recharged by the stream. This, in
36    turn, supports the riparian vegetation and provides groundwater flow back to the stream from the
37    shallow aquifer following storms. During periods of low precipitation, the flow in the river
38    comes primarily from groundwater inflow. Many portions of the Upper San Pedro have had
39    perennial water flow, and this is reflected in the plant communities (Price et al. 2005). However,
40    the number and extent of areas with intermittent flows has increased in recent times (Price et al.
41    2005). Because the flow of water is essential for the survival of the riparian biota, it is of critical
42    importance in maintaining the vegetation and, in turn, the wildlife of the region.

      August_2006_...                                                   _ _.External Review Report:. Watershed. Case.S.fypfes
 1    2.2.1  Goals of the Case Study Assessment
 2    The primary goal of this case study was to model the likely effects of climate change, coupled
 3    with existing stressors (urbanization and groundwater withdrawals), on ecological resources and
 4    biodiversity in the San Pedro Riparian National Conservation Area (SPRNCA). Specifically, the
 5    case study evaluated the future ability of the SPRNCA to remain a high-quality, self-sustaining
 6    riparian ecosystem capable of maintaining the plant and wildlife habitat that results in its unique
 7    biodiversity. Researchers sought to model the effect of existing stressors and changes in the
 8    climate on the hydrologic conditions of the region, which would affect the structure,
 9    composition, and representation of vegetation communities in the SPRNCA. Changes in the
10    plant communities would, in turn, affect their ability to continue to support important ecological
11    resources and alter water quality and water supply (Price et al. 2005).
12    2.2.2  Major Stressors
13    The main climatic and geophysical drivers of the condition and quality of the San Pedro riparian
14    ecosystem include aquifer depletion, climate change, and population growth. The primary
15    stressor in the San Pedro River Ecosystem is water extraction through pumping, and much of this
16    water goes to agricultural use. Although some of the water pumped returns to the aquifer via
17    percolation, 70 percent of the water used for agriculture is lost. Other uses of water include
18    municipal water supply  wells, domestic water wells, stock wells, and some military and
19    industrial use. From 1940-1997, anthropogenic water use has caused a reduction in San Pedro
20    River streamflow, a reduction in evapotranspiration by riparian vegetation along the floodplain
21    of the river, and the formation of significant cones of depression near many communities which
22    accompany large losses of groundwater storage. Over the next few decades, agricultural water
23    use in the area is expected to decrease, while the amount of urbanized land and concomitant
24    water use is expected to increase (Price et al. 2005).
25    Climate change is another significant stress on the already threatened ecosystem.  Climate change
26    projections suggest an increase in mean seasonal temperatures of 2-7 degrees Celsius over the
27    next 100 years. While all GCMs broadly agree with this temperature increase, they  differ in their
28    projections for precipitation (Price et al. 2005). Case study findings indicate that a warmer wetter
29    future climate does not pose as significant a threat for vegetation as a warmer drier  climate
30    projection.
31    The case study did not separately quantify the effects of aquifer depletion and climate change.
32    Climate change will cause changes in the ecosystem and aquifer even  without water extraction
33    and other forms of human interaction. The impact of aquifer depletion, however,  is  expected to
34    be more dramatic in terms of scale than the effect of climate change alone  on the  San Pedro
35    River ecosystem. Together, these two stressors will have a major synergistic effect (Price et al.
36    2005).
37    2.2.3  Assessment Methods
38    The ecological endpoints modeled included changes in the main vegetation communities
39    (including riparian, mesic, and xeric), river baseflow, soil water content, channel  migration, and
40    the incidence and intensity of wildfires.
41    The team created five climate projections for 2003-2102 using a 52-year daily time series of
42    historic data from 1951-2002 from the National Weather Service station in Tombstone, Arizona.
43    The five scenarios included one with no climate change where the 1951-2002 daily temperature

      August_2006_...                                                  _ _.External Review Report:. Watershed. Case.S.fypfes
 1    and precipitation data was repeated. A second scenario (warm) had increasing temperature over
 2    100 years with a 6 degree Celsius increase in the minimum daily temperature and a 4 degree
 3    Celsius increase in the maximum daily temperature by 2102. The third scenario (warm dry) had
 4    the same temperature assumptions as the second scenario, but included a progressive decline in
 5    winter daily precipitation. The fourth (warm wet) and fifth (warm very wet) scenarios both had
 6    the same warming as in the second scenario but increased the precipitation by 50 percent and 100
 7    percent respectively by 2102 (Price et al. 2005).
 8    The team used a mechanistic bottom-up approach that combined multiple interdisciplinary
 9    methods to link hydrological, vegetation, habitat, and climate change models. The team first
10    analyzed species, vegetation, and habitat suitability and then used linked models to analyze the
11    impacts of existing stressors and their interaction with projected climate change. The team linked
12    models to incorporate vegetation data into the groundwater and surface water models and to tie
13    the fluctuations of groundwater level to evapotranspiration processes.  It also linked the above
14    models to a geomorphic model (MEANDER) to drive vegetation dynamics (Price  et al. 2005).
15    The vegetation model was developed using the STELLA II Dynamic Simulation software
16    (Peterson and Richmond 1996), which can be used to simulate the habitat response to changes in
17    environmental drivers such as groundwater, streamflow, precipitation, temperature, or fire. The
18    specific vegetative groups measured included ten species or functional groups of southwestern
19    riparian plants including Fremont cottonwood, Goodding's willow, salt cedar, velvet mesquite,
20    velvet ash, a hydromesic shrub group, a xeric riparian shrub group, herbaceous annuals, wetland
21    perennials, and mesic perennial grasses. Climate inputs to the vegetation model included solar
22    radiation, average minimum and maximum temperature, daily precipitation, relative humidity,
23    and maximum daily wind speed, all averaged over  the 5-day time step of the model. The team
24    ran the vegetation model under four of the five climate scenarios: no change, warm, warm wet,
25    and warm dry. Initial conditions for vegetation were represented by stem density and mean
26    effective diameter in study plots for five of the six most abundant tree species in the riparian
27    corridor. Additionally, all plots began with a 20 percent cover of both annuals and wetland
28    plants. The model also included simulation of fire occurrence and its effects on vegetation and
29    structure. (Price  et al. 2005)
30    The case study also used the Soil Water Assessment Tool (SWAT), a  physically-based
31    hydrologic model designed to project the effects of land management  practices on water and
32    sediment yield in complex watersheds over long time periods (Srinivasan et al. 1998). Input data
33    for SWAT included daily weather, soils, topography, vegetation, and land management practices,
34    and includes input parameters representing streamflow-groundwater interactions. The team used
35    SWAT outputs of daily streamflow data to run MEANDER (Odgaard 1989), a model used to
36    project changes in channel location under different climatic  scenarios  (Price et al. 2005).
37    The avian biodiversity methodology focused  on 87 abundant bird species in the SPRNCA. Four
38    variables were used to predict future population changes, including the species' dependence on
39    (1) dominance by riparian species in the vegetative community; (2) extensive and non-
40    fragmented stands of riparian forest; (3) wetland habitat; and (4) running or standing water. In
41    addition, five  species were evaluated using Habitat Suitability Index (HSI) models developed for
42    this case study. These models evaluated the likely impacts of changes in habitat quality on a
43    species' carrying capacity (Price et al. 2005).

      August_2006_...                                                    _ _.External Review Report:. Watershed. Case.S.fypfes
 1    The team found that a major constraint with the model linkages is that the various models
 2    produce output with different resolutions and scales. For example, the surface water modeling
 3    operated at a finer scale than the groundwater and geomorphology models (Galbraith et al. pers.
 4    comm.)2. Linking the various models was one of the major methodological challenges in this
 5    study.
 6    2.2.4   Impacts & Findings
 7    Vegetation modeling indicated that changing hydrology and climate change may fundamentally
 8    impact vegetation in the SPRNCA by fragmenting existing riparian and wetland communities
 9    and leading to their replacement by more mesic or xeric communities (i.e., vegetation more
10    typical of the  desert matrix). Model results suggested a decreasing trend in coverage by pioneer
11    woody vegetation across the floodplains of the upper San Pedro over the next 100 years. The
12    influence of climate change on pioneer riparian communities will depend on the magnitude and
13    direction of precipitation changes. A decrease in winter precipitation will likely result in fewer
14    winter floods, lower rates of channel migration, and much lower cottonwood  and willow
15    recruitment rates. An increase in winter precipitation is expected to result in larger and more
16    frequent winter floods, higher channel migration rates, and higher cottonwood and willow
17    recruitment rates. Models indicated that coverage by later success!onal communities such as
18    mesquite, ash patch types, and sacaton grassland should increase over the next 100 years.
19    Finally, the dominant factor influencing vegetation patterns along the San Pedro today may be its
20    geomorphic history; Price et al. (2005) noted that  "the channel incision and widening events of
21    the late 19th through early 20th century set the template that influenced subsequent development
22    of the new floodplain and riparian ecosystem."
23    The avian biodiversity modeling projected that 26 percent of the most abundant bird species
24    would likely to be vulnerable to and adversely affected by changes in the vegetative community
25    due to climate change. An additional 25 percent could be relatively unaffected, and 43 percent
26    could benefit. Results of the HSI models indicated that the species most dependent on the
27    cottonwood/willow gallery forest would show the greatest projected decreases. Even without
28    factoring climate change into future conditions, marked changes in habitat quality are projected
29    for two of the five species. This change is caused by a maturation and contraction of the
30    cottonwood/willow forest in the middle of this century; the change will  result in decreased
31    habitat for the yellow-billed cuckoo and an increase in habitat for the Botteri's sparrow as the
32    forest is replaced with grassland and shrublands (Price et al. 2005).
33    The no change, warmer, and warmer drier climate modeling scenarios all resulted in a loss of
34    riparian forest and wetlands and their replacement by mesic or xeric vegetation communities.
35    High avian biodiversity in the SPRNCA is supported by the proximity of riparian gallery forest
36    and wetland habitats within a matrix of desert scrub and grassland (Price et al. 2005). Loss of
37    either of the habitats would reduce the biodiversity of the SPRNCA, since the birds that currently
38    inhabit these areas are expected to be replaced by  current occupiers of the desert scrub matrix.
39    These findings suggest that the ecosystem services provided by the SPRNCA could be greatly
40    affected by climate change (Price et al. 2005).
       Subsequent citations of personal communication with the San Pedro team refer to the interview at the conclusion
      of the case studies. Team members interviewed included H. Galbraith, M. Dixon, and T. Maddock.

      August_2006_...                                                   _ _.External Review Report:. Watershed. Case.S.fypfes
 1    A decline in ecosystem services that sustain ecotourism could adversely impact demand for those
 2    activities within the region. The SPRNCA is a major attraction to wildlife viewers and
 3    ecotourists. If the gallery forest were to be fragmented or entirely lost as is projected under the
 4    three driest climate change scenarios, the area would be less attractive to the public. If future
 5    climate changes more closely resemble the warm and very wet scenario, adequate water supply
 6    to the ecosystem might help maintain ecosystem services (Price et al. 2005).
 7    2.2.5  Methods and Results Applicable to Other Watersheds
 8    The vegetation, hydrology, and wildlife data inputs used in the models make them specific to the
 9    Southwestern U.S. and other arid environments with groundwater-dependent riparian systems.
10    The evapotranspiration model (RIPET) is a riparian modeling system, but is applicable to any
11    system where vegetation is affected by changes in the water table. Currently, the model is being
12    applied in the Rio Grande and Kern watersheds in California. The general approach of linking
13    vegetation to channel migration is useful in other regions, as long as the specific vegetation data
14    are adapted. However, the approach is not applicable to systems where vegetation uses water
15    from the unsaturated zone (Price et al. 2005).
16    The results from this case study and, in particular, the challenges of aquifer depletion are
17    applicable to other areas. The San Pedro ecosystem could be a model for future changes of
18    similar riparian systems in the Southwestern U.S. due to climate change and other anthropogenic
19    stressors.

20    2.3  Sacramento
21    The Central Valley of California extends approximately 450 miles, from the headwaters of the
22    San Joaquin River in the south to the headwaters of the Sacramento River in the north. This area
23    of approximately 42,000 square miles is referred to as the Sacramento River Watershed. The
24    main water inputs are derived from the two rivers and their tributaries, which drain into the delta
25    and, eventually, into Sacramento and the Pacific Ocean. The hydrology of the Sacramento River
27    Watershed is dominated by winter snow fall with
29    subsequent spring melt and runoff, giving rise to spring
31    peak flow maxima and later summer flow minima in
33    most of the streams. Winter (November-April) is the
35    season of maximum rainfall, so the variability of
37    wintertime precipitation dominates the hydrology.
39    Although this dominance is still a characteristic of the
41    natural hydrology of the region, human intervention in
43    the hydrologic cycle has greatly disrupted this natural
45    hydrology.
47    Extreme variability in precipitation throughout the
49    Sacramento River Watershed makes California
51    particularly vulnerable to climate variability and
53    change. Greater variability  means that many  of the
55    area's ecosystem values are already coping with years
57    of adverse conditions.  Climate change, however, could
58    make water supplies more vulnerable due to reduced snow packs and, consequently, lower
59    summer streamflows, which would threaten aquatic ecosystem services such as the provision of
60    water for the agricultural sector and for valued aquatic wildlife habitat.
 «- /,

      August_2006_...                                                   _ _.External Review Report:. Watershed. Case.S.fypfes
 1    The existence of suitable spawning and rearing habitat for Chinook salmon in the upper
 2    Sacramento River is currently dependent on releases of cool water from reservoir hypolimnia
 3    between May and September. Without these releases, the water temperatures would exceed the
 4    physiological tolerances of the eggs and juveniles of the winter and spring runs by three or more
 5    degrees Centigrade. The Chinook salmon are able to reduce their vulnerability to the temperature
 6    increase by spawning earlier and later in the year, and, therefore, are less dependent on changes
 7    in water management practices. The midsummer high water temperatures in the lower river are
 8    controlled by high ambient air temperatures and slow and low flows. Releases from dams
 9    currently keep the river water temperatures below the physiological thresholds for migrating fish
10    like Chinook salmon (Yates et al. 2006). If releases of cool waters from the dams were not
11    continued under the climate  change scenarios, the mid-summer monthly mean water
12    temperatures in the lower Sacramento River would be about 6 degrees Celsius higher, a
13    potentially lethal temperature for adult and juvenile Chinook salmon (Yates et al. 2006).
14    The main land uses in the Sacramento River Watershed, particularly at lower elevations,  are
15    agricultural, municipal, and  industrial. Eight of California's 15 most agriculturally productive
16    counties are in the Central Valley, which makes it one of the most productive areas in the world
17    (California Research Bureau 1997). The area of agricultural land is currently stable or decreasing
18    slightly, at about 1.5 million hectares (these lands are irrigated using water from the main rivers).
19    The main crops are generally water-demanding (e.g.,  cotton, grapes, tomatoes, fruits,  hay, and
20    rice). The annual crop value is typically in excess of $14 billion, and more than 30 percent of the
21    total economy is attributed to agriculture (California Research Bureau 1997).
22    2.3.1  Goals of the Case Study Assessment
23    The broad goal of this case study was to assess how global change-related alterations  in water
24    supply and water demand may affect important freshwater ecosystem services in the Sacramento
25    Basin. The objectives of the case study were five-fold: (1) to understand the relationships among
26    stressors and ecological processes and the  aquatic ecosystem services they provide; (2) to use
27    this information, along with  water resource models, climate change scenarios, and assumptions
28    about the future intensities of existing stressors to project effects on the future functioning of
29    these services; (3) to provide stakeholders  with information on how valued ecosystem services
30    are likely to be affected, so that they can make informed decisions; (4) to develop appropriate
31    methodologies for assessing effects on ecosystem services that will be transferable to other large
32    watersheds in different locations and settings; and (5) to provide integrated decision support for
33    issues of reservoir location, Federal Energy Regulatory Commission (FERC) dam re-licensing,
34    and system operations to preserve the ecosystem services of interest or of regulatory necessity.
35    This case study found that the challenge was to balance the complex tradeoffs and interactions of
36    the multiple uses of water (e.g., water for food and water for environment), some of which are in
37    conflict and others which are not. Climate  change presents the possibility that the existing
38    balance will be upset and that adaptation will be required (Yates et al. 2006).
39    2.3.2  Major Stressors
40    The most crucial stressors on the Sacramento River Watersheds are population growth, land use
41    change, and climate change. Steady growth in population, particularly around existing urban
42    areas and transportation corridors, directly affects the demand for water in the Sacramento River
43    Basin. In addition, changing land use, in particular the extension of urban area into other land use

      August_2006_...                                                   _ _.External Review Report:. Watershed. Case.S.fypfes
 1    types, stresses water supply and demand in the basin. Climate change is expected to exacerbate
 2    these demands on water (Yates et al. 2006).
 3    The primary problem caused by land use change and population growth in the Sacramento Basin
 4    is the movement of water out of the irrigated agricultural systems and into the urban environment
 5    (Yates et al. 2006). The scale of water development in California is among the most substantial
 6    in the world, with water often being shifted from one basin to another over distances of hundreds
 7    of kilometers to satisfy water demands (Yates et al. 2006). Much of the water in the basin is
 8    exported through pumps in the Delta in order to satisfy municipal and industrial demands along
 9    the Southern California Coastal Plain and agricultural water demands in other basins (Yates et al.
10    2006). Land use change and water development - particularly the construction of major
11    reservoirs on all of the major rivers - has altered surface water hydrology in the basin and
12    created peak flow conditions earlier in the winter and reduced spring flows. In addition, summer
13    flows are higher than under natural conditions since operators attempt to meet summer irrigation
14    demands by releasing water downstream (Yates et al. 2006).
15    Climate change - particularly projected increases in summer temperatures - is expected to cause
16    an increase in water supply requirements for all land uses (Yates et al. 2006). Modeling for this
17    case study took cropping and irrigation management patterns and climate change into  account
18    using two simulations. Cropping and irrigation management patterns remained fixed over the
19    course of a 100-year simulation in the  first formulation, while cropping and irrigation
20    management patterns evolved along with climate in the second. Climate and land use  stresses
21    were not isolated to determine their individual effects on the system (Yates et al. 2006).
22    2.3.3  Assessment Methods
23    In California's diverse and highly regulated environment, the ecosystem  services deemed most
24    necessary for examination by this study included a reliable water supply for agriculture, urban
25    consumption, and industry; habitat and food resources for salmonids and other fish species;
26    hydropower; recreation; aesthetic beauty; transportation; soil fertility; flood and drought
27    mitigation; water purification; and erosion control. The selection of services was based on the
28    relative importance of the service to the region, the availability of data to relate the service to
29    climate, the tractability of the modeling, and stakeholder priorities.
30    The case study team applied the Water Evaluation and Planning (WEAP21) modeling system to
31    link natural and anthropogenic considerations to allow for an analysis of tradeoffs, such as water
32    supplies for agricultural, natural, domestic, industrial, or recreational uses (Yates et al. 2005).
33    Biological requirements in the model such as fish mortality or reproduction were related to
34    projected climatic characteristics and hydrological and water quality characteristics. WEAP21
35    recognizes that water supply is defined by the amount of precipitation that falls on a watershed
36    and is depleted through natural watershed processes with evapotranspiration being the first
37    significant point of depletion (Yates et al. 2006). The residual supply is available to the water
38    management system. WEAP21 is thus able to link climate, land use/land cover conditions, and
39    water management (Yates et al. 2006).
40    The WEAP21 model extends the original WEAP model by including demand and supply
41    priorities that solve the water allocation problem in a linear programming heuristic. WEAP21
42    includes a transparent set of model objects and procedures that can be used to analyze a full
43    range of issues faced by water planners through a scenario-based approach. The list of issues
44    includes climate variability and change, watershed condition, anticipated demands, ecosystem

      August_2006_...                                                  _ _.External Review Report:. Watershed. Case.S.fypfes
 1    needs, the regulatory environment, operational objectives, and available infrastructure (Yates et
 2    al. 2006).
 3    In the modeling process, the Sacramento River Basin was divided into more than 100
 4    subcatchments, groundwater basins, irrigated areas, and urban demand centers in an attempt to
 5    completely characterize the forces that act on water in the basin. A monthly climate time series
 6    from 1962 to 1998 was used to drive a distributed hydrologic model that simulates runoff,
 7    groundwater-surf ace water interactions, and consumptive water demands. Water management
 8    infrastructure, including reservoirs, canals, and diversions, was superimposed over the physical
 9    watershed. Results show that the model was capable of reproducing both local and regional
10    water balances for the 37-year period, including managed and unmanaged streamflow, reservoir
11    storage, agriculture and urban water demands, and the allocation of ground water and surface
12    water supplies (Yates et al. 2006)
13    This study evaluated the impact of four climate scenarios on water management in the region and
14    whether water management adaptation could reduce the potential impacts of climate change. The
15    four climate scenarios were derived by downscaling the output from two GCMs (Parallel
16    Climate Model and Geophysical Dynamics Laboratory) and two emission scenarios (A2 and Bl)
17    to a 1/8 degree grid over California. The four climate scenarios were: (1) Parallel Climate Model
18    with an A2 scenario; (2) Parallel Climate Model with a Bl scenario; (3) Geophysical Dynamics
19    Laboratory with an A2 scenario; and (4) Geophysical Dynamics Laboratory with a Bl scenario.
20    2.3.4  Impacts & Findings
21    Two of the four scenarios examined predicted a decreasing trend in precipitation over the next
22    century, with the other two scenarios showing less pronounced changes - one scenario predicted
23    slightly wetter conditions at the end of the century and the other showed a decrease in
24    precipitation in normal-dry years and an increase in precipitation in normal-wet years. All four
25    scenarios predicted increases in average winter and summer temperatures over the next century
26    ranging from a lower bound increase of 1.5°C in winter and 1.4°C in summer to the higher
27    bound of 3.0°C in winter and 5.0°C in  summer.
28    Key hydrologic factors were examined by the case study team to determine whether existing
29    water management arrangements in the Sacramento were capable of responding to the potential
30    climate and land use changes. For the first hydrologic factor, annual inflows to reservoirs, two
31    scenarios projected increased annual inflows to the major reservoirs and two projected lower
32    annual inflows. The second hydrologic factor was changes in the timing of stream flows. All
33    scenarios showed an earlier timing in stream flows as compared to historic conditions, which
34    would have the greatest effect on those basins dependent on snow melt runoff (e.g., Sacramento
35    watershed above Lake Shasta). Persistence of drought conditions, the final hydrologic factor,
36    was projected to be less severe than the historical record under two scenarios.  A third scenario
37    projected that droughts comparable in magnitude to the early '90s drought would occur with
38    regularity. The fourth scenario projected a very severe drought during the last 15 years of the
39    century. (Yates et al. 2006).
40    All four scenarios showed an increasing trend in water requirements with time. These increasing
41    supply requirements were due primarily to increasing summer temperatures for each of the four
42    scenarios. The cause was likely the increasing crop water demands as summer temperatures
43    increased. Groundwater pumping was projected to be relatively stable for all scenarios for the
44    periods covering 1960 to 2064. In the last period, 2070-2099, pumping increased significantly in

      August_2006_...                                                   _ _.External Review Report:. Watershed. Case.S.fypfes
 1    dry years for one scenario when surface water deliveries were less reliable. Aquifers in the
 2    region showed relatively stable fluctuations around a mean for most of the period between 1960
 3    and 2070. During this period the surface water deliveries were increasing with growing crop
 4    water requirements so that groundwater pumping levels were only marginally increased. During
 5    the final period of analysis (2070-2099), however, an extended ten-year drought in one scenario
 6    shifted agricultural water supplies to groundwater. As a result, groundwater levels decreased
 7    sharply.  (Yates et al. 2006).
 8    Future climate changes, particularly shifts in temperature and precipitation patterns, were
 9    projected to alter the river's temperature regime. This change could lead to further reductions of
10    the Chinook salmon's fragmented habitat. Specifically, increased water temperatures were
11    projected to result in exceedences of critical spawning and rearing temperatures, thus
12    jeopardizing the productivity of Chinook salmon (Yates et al. 2006).
13    2.3.5  Adaptation Options
14    The Sacramento team considered two forms of adaptation options in this case study. The first
15    was incorporated into the WEAP21 model and consisted of strategies to adapt cropping
16    practices. The cropping options analyzed included improved irrigation efficiency and changes in
17    cropping patterns in response to water supply conditions. The results showed a decline in water
18    supply requirements as improvements in irrigation efficiency were implemented (Yates et al.
19    2006).
20    The second adaptation option the team considered was managed releases of cold water stored in
21    reservoirs. This case study demonstrated that the management structures and practices that had
22    historically affected the fish adversely may provide an opportunity to alleviate some of the future
23    impacts  of climate change. More 'natural' and unmanaged systems may provide fewer
24    opportunities for adaptation to climate change effects (Yates et al. 2006).  Since Chinook salmon
25    are coldwater fish, they may be vulnerable to climate change, particularly increasing water
26    temperatures. Rising water temperatures in their natal rivers could adversely affect the  salmon's
27    ability to find suitable breeding habitats, especially since that habitat has already been reduced
28    by dam construction. However, dams allow scheduled releases of cold water stored in reservoirs,
29    such that the frequency  and timing of these releases could be used to aid salmon survival during
30    spawning (Yates et al. 2006).
31    2.3.6  Methods and Results Applicable to Other Watersheds
32    The WEAP21 model framework has proven resilient, with wide applicability to  multiple
33    locations and systems. The intrinsic logic behind WEAP21 is universal and could be easily
34    adapted  for other locations using site specific data. Further, the results from this  case study
35    would apply to other watersheds that are similar in character and nature. For example, if
36    agricultural sector water demand is scaled back due to improved irrigation efficiency and
37    changes in cropping practices, there will be more water for other sectors. Given that all water
38    resource systems and hydrologic systems are unique, however, the specific results would not be

     August_2006_...                                                          _ _.External Review Report:. Watershed. Case.S.fypfes
1    transferable. In other words, the results would be applicable qualitatively, but not quantitatively
2    (Purkey and Yates pers. comm.)3.
     3 Subsequent citations of personal communication with the Sacramento team refer to the interview at the conclusion
     of the case studies. Team members interviewed included David Purkey and David Yates.

      August_2006_...                                                   _ _.External Review Report:. Watershed. Case.S.fypfes
 i                                 3. Cross-cutting Findings

 2    The GCRP established broad case study goals from the outset to gain an understanding of the
 3    potential consequences of global change by: (1) improving the scientific capabilities and basis
 4    for projecting and evaluating effects and vulnerabilities of global change in the context of other
 5    stressors and human dimensions (as people catalyze and respond to global change); (2)
 6    conducting assessments of the ecological, human health, and socioeconomic risks and
 7    opportunities presented by global change; and (3) assessing adaptation options to improve
 8    society's ability to effectively respond to the risks and opportunities presented by global change
 9    as they emerge (US EPA 2003). All of the case studies clearly addressed and achieved the aims
10    set out in (1) and the relevant ecological aim in (2). The Sacramento team was also able to assess
11    adaptation options (3) due largely to a second round of funding and a decision-driven approach
12    throughout the assessment process.
13    The case studies have demonstrated that certain factors help ensure that there is sufficient
14    "capacity" for doing assessments. Factors such as public awareness of issues and stressors, good
15    data, good models, clear goals, and clear decisions facilitate the study teams' ability to establish
16    and maintain legitimacy, credibility, and relevance. This section discusses these factors in further
17    detail through a discussion of the lessons learned, or cross-cutting findings, across  the case
18    studies. It is broken into three subsections that address the assessment process, the  stakeholder
19    process, and the relevance of the case studies to decision making and climate change and land
20    use change.

21    3.1  Assessment Processes
22    Given the multidisciplinary nature of the assessments, and the complexity of the models used to
23    produce the results described in Section 2, the processes used by the teams were designed to
24    accommodate multiple investigators and disciplines (addressed below in Section 3.1.1, Team
25    Composition and Management). The teams also addressed several of the key issues we identified
26    in the initial scope of work for the assessments: research approaches, watershed scales, and
27    uncertainty.
28    3.1.1   Team Composition  and Management
29    All three teams involved a large set of researchers with expertise in different disciplines. The San
30    Pedro and Maryland teams included experts in hydrology, geomorphology, and aquatic and avian
31    ecosystems. The principal challenges in both of those case studies involved defining suitable
32    assessment endpoints and linking a series of separate models to provide an integrated analytical
33    capability. Both San Pedro and Maryland invested considerable effort in developing new
34    research approaches. These innovative approaches required varying levels of modification to
35    existing models and linking the models to provide a complete capability starting with physical
36    effects of climate change and continuing along the assessment chain to culminate in measures of
37    ecosystem services.
38    The Sacramento team started with a preconceived notion of the key analytic endpoints (i.e.,
39    flows and quantities of water). There was consequently less of an emphasis on developing new
40    models and linking existing models, and more of an emphasis on improving and demonstrating
41    an existing analytic framework (i.e., the WEAP model).

      August_2006_...                                                   _ _.External Review Report:. Watershed. Case.S.fypfes
 1    The Team Managers/ Principal Investigators of all three teams were faced with a challenging
 2    series of tasks, including

 3       •  Guiding their research teams through an extensive scoping and method development
 4          phase;

 5       •  Deciding how to depict climate change and variability in a way that related to the
 6          ecosystem services of concern;

 7       •  Developing and implementing a plan to engage stakeholders;

 8       •  Implementing a model development and integration phase; and

 9       •  Reporting results in a variety of forums and formats.
10    The managers all guided the projects based on the particular expertise they brought to the table.
11    Ecologists led the San Pedro and Maryland teams, and many but certainly not all of the technical
12    advances were related to methods for simulating important ecological processes. A
13    hydroclimatologist led the Sacramento study, and much of the effort on that study - particularly
14    in the initial stages of the project - was devoted to developing an innovative technique to
15    downscale regional climate from larger-scale GCM results.
16    One of the strategic issues we face is whether future watershed assessments and other similar
17    projects could benefit from some degree of standardization in terms of tools or expertise. One
18    point that was made by all three teams is that there was a common and fundamental need for
19    several "keystone" skills, especially expertise in interpreting climate change scenarios. Dr. David
20    Yates, the hydroclimatologist who led the Sacramento team, acted as an informal consultant to
21    the other two assessment teams in developing their climate scenarios. Similarly, Dr. Hector
22    Galbraith, an ecologist who specializes in evaluating the suitability of habitats for key species of
23    concern, was a member of the San Pedro team and also provided assistance to the Sacramento
24    team.
25    Despite the emphasis on stakeholder engagement,  only one team (Sacramento) included a
26    member whose specific focus was to facilitate stakeholder interactions or to communicate with
27    stakeholders and decision  makers. The resources devoted to eliciting stakeholder input were still
28    fairly modest (about $10,000 out of a project budget of almost $500k) even in that study. This
29    gap was filled to varying degrees by existing connections between the scientists on the team and
30    decision makers. The success of the Sacramento study in orienting its research to decision-
31    relevant issues was probably less because they included a team member to facilitate interactions
32    with stakeholders and more tied to their enduring and intimate engagement with decision making
33    processes and their understanding of what types of information decision makers in that area
34    need. Moreover, as discussed in more detail in Section 3.2 on stakeholder processes,  there was
35    an evolution in the views of both the study teams and the project managers in terms of the types
36    of stakeholder interactions that were productive. In retrospect, it probably improved the cost-
37    effectiveness of the studies that they did not invest in unfocussed stakeholder interactions.
38    3.1.2  Research Approach
39    All three case study teams developed approaches that relied on integration of a chain of
40    submodels to simulate  physical, hydrological, geomorphologic, and ecological components that
41    ultimately related to ecosystem services. All three of the study teams also reported challenges in
42    coupling the model elements.

      August_2006_...                                                   _ _.External Review Report:. Watershed. Case.S.fypfes
 1    The Maryland team invested considerable effort in integration, and even developed a paper on
 2    the topic (Nelson et al. 2006). The San Pedro team pushed the state of the art in several of the
 3    individual submodels, but those submodels remained largely discrete, and required considerable
 4    integration. The Sacramento team started with a pre-existing model framework that linked
 5    hydrology and water management. The team invested in improvements in the model's ability to
 6    incorporate climate scenarios, but generally appeared to have fewer difficulties in integrating
 7    component elements than the other two projects.
 8    Geomorphology took on a central role in two of the three studies (San Pedro and Maryland). One
 9    of the key aspects of climate change - more intense precipitation as well as longer dry periods -
10    translates to higher high flows and lower low flows in the hydrographs  of streams. This change
11    in the hydrologic regime, in turn, affects the processes that create transitional ecosystems vital
12    for certain avian species, and govern sediment transport and stability that affects  spawning
13    success for fish. The resulting changes in avian and aquatic habitat were key drivers in the San
14    Pedro and Maryland studies.
15    All three teams appear to have benefited from the mid-project workshop held in January 2005.
16    That workshop offered an opportunity to compare notes on methods, assessment  endpoints,
17    relevance to decision making, model integration, stakeholder interactions, uncertainty analysis,
18    and other cross-cutting topics.  To the extent that we plan to conduct similar large, multi-year
19    projects like the watershed case studies, such mid-project workshops are a worthwhile
20    investment for the research teams.
21    3.1.3  Scale
22    Each group worked at a variety of spatial scales, a result both of the phenomena they were
23    investigating and the scale at which decisions are being addressed.
24    The group investigating small watersheds in Maryland worked at a sub-watershed scale (13 - 28
25    mi2), in part because urban growth is regulated at the county-level. Individual parcels of land
26    rather than pixilated representations were represented in this analysis, and surrounding land uses
27    fed back into subsequent land use change dynamics for each parcel.
28    The San Pedro group examined the upper portions of the San Pedro basin (2,500  mi2), where
29    most of the remaining perennial or near-perennial river reaches exist, making this stretch of
30    greatest importance for the study's primary focuses: the maintenance of avian biodiversity,
31    sustained urban water supplies, and recreational uses. Model representations were limited to plot-
32    scale information, however, meaning that the simulations were not run  simultaneously for the
33    entire landscape, but only for certain representative patches within it.
34    The Sacramento River Watershed study worked at the basin scale (42,000 mi2), but designed the
35    study modularly, so that smaller sub-basins that performed ecosystem services of particular value
36    (such as Chinook salmon spawning) were nested separately within the design, and stand-alone
37    results could be produced for those areas. The primary decisions being  addressed here, including
38    water allocation and the balance of competing legislative and regulatory authority, occur at the
39    state-level, and consequently, it was necessary to consider the watershed as a whole.
40    One of the challenges addressed by all three teams was that available data may not be suited to
41    the questions under consideration. For example, the geomorphologic processes that shape
42    channel migration act at a localized level within a stream reach, but information on the
43    hydrologic and geologic factors that control these processes may be available only at a much

      August_2006_...                                                    _ _.External Review Report:. Watershed. Case.S.fypfes
 1    broader scale. The case study teams dealt with this issue by developing scenarios and scaling up
 2    their results for sample situations to the larger watershed.
 3    Another important scale issue that the teams addressed is that ecosystem services are delivered at
 4    varying scales and with different levels of "connectedness" to other resources outside the study
 5    area. For example, in San Pedro, with its focus on migratory neotropical birds, the birds are
 6    dependent on the availability of suitable habitat at other locations and other times. Similarly, in
 7    Sacramento, one of the key endpoints - instream flows to support salmon - is necessary but not
 8    sufficient for sustaining threatened and endangered salmon populations. Climate change could
 9    affect the other critical resources needed to support these populations, but to keep the scope
10    manageable the study teams assumed that conditions outside their study's boundaries were
11    essentially static.
12    3.1.4   Uncertainty
13    Given the variety of academic disciplines and datasets involved in creating decision support for
14    the effects of climate and land use change on water resources, it is interesting to note that the
15    teams identified the same sources of uncertainty when asked which sources were most important.
16    The first section below discusses these sources, and the second briefly reviews the approaches
17    that were considered to communicate uncertainty to stakeholders.
18    Types and the Extent of Uncertainty
19    None of the case studies conducted a comprehensive quantitative uncertainty analysis to
20    determine the relative importance of various inputs. However, in response to a set of queries
21    posed by EPA regarding uncertainty, they all provided a qualitative listing of likely uncertainty
22    sources, and all expressed interest in conducting more detailed uncertainty analysis if additional
23    resources could be made available to support it. Because the findings on types and extent of
24    uncertainty are based primarily on judgment and qualitative analysis, they should be viewed  as
25    tentative.
26    The magnitude and distribution of the hydrologic effects of regional climate change (i.e.,  how
27    changes in temperature and precipitation affect streamflow and the "downstream" analytic
28    components) were considered by all  case study teams to be the primary sources of uncertainty in
29    their study results. The applicability  of GCM projections to the regional scale is not generally
30    considered particularly reliable (Wilby et al. 2004). The input requirements of the modeling
31    systems was different for each research team, and the algorithms used by researchers to
32    downscale the datasets to fit with their models varied accordingly These input requirements and
33    downscaling algorithms were another source of uncertainty.
34    Study teams applied different approaches to constraining this uncertainty, including generating
35    scenarios based on multiple GCM realizations (assuming that more common results indicate
36    more probable outcomes), and using Monte Carlo-type analysis. The San Pedro research team
37    used a set of scenarios, for example,  to bracket a large range of uncertainty regarding projected
38    future precipitation patterns for the area.
39    Land use and land use change are highly uncertain aspects of the watersheds in question, which
40    also contribute significantly to uncertainty in the study results.  Small Maryland watersheds,
41    given their proximity to rapidly urbanizing  landscapes, were particularly susceptible to this
42    influence. Land use changes were found to  be quite sensitive to regulatory changes, population
43    growth, and income changes. Once the potential land open for development  has been developed,

      August_2006_...                                                    _ _.External Review Report:. Watershed. Case.S.fypfes
 1    the possible responses by landowners is unknown, but possible outcomes are intensification of
 2    developed areas or reclassification of previously undevelopable land and further landscape
 3    alteration. Additionally, while land use may be predicted on a large scale with some degree of
 4    certainty, the idiosyncrasies of individual land owners and managers may never be anticipated
 5    with complete predictability.
 6    Hydrological responses to changing land use were identified by both the Maryland and
 7    Sacramento groups as primary contributors to overall uncertainty. Geomorphological responses
 8    were another large contributor to uncertainty, particularly in the San Pedro watershed, for which
 9    management goals are predicated in part on the occurrence of transitional ecological states.
10    These transitional states are highly dependent upon sporadic hydrological events such as
11    flooding, which translate to the ecosystem through their geomorphological effects. The creation
12    of unvegetated areas on islands within the river or on the river's banks by a flood event allows
13    for colonization by plants which would otherwise be unable to compete with established plant
14    communities, and thereby increases the overall habitat diversity of the area. With time, these
15    colonizing plants are replaced by more stable assemblages, and other unvegetated areas are
16    created by intense hydrological events. Any one plot may show little change, but over a larger
17    area, the patchiness of habitat types allows high avian diversity to be maintained. In general, it is
18    more challenging to predict and monitor processes associated with unusual or extreme events
19    and transitional conditions, and analyses of this type tend to be more uncertain than those  dealing
20    with processes that are driven by average conditions.
21    Other sources of uncertainty listed by the researchers from Maryland include a lack of
22    knowledge regarding the effects of the interactions of multiple stressors in streams, the biology
23    of understudied fish species, and predictions of habitat suitability.
24    Communicating the Importance of Uncertainty to Stakeholders
25    In communicating the uncertainty to stakeholders associated with climate and land use change
26    effects on ecological and water resources, a concern of the researchers was that doing so may
27    undermine the contribution that results may have to the decision making processes. At issue was
28    whether managers would be willing to place confidence in results that are initially presented as
29    uncertain, particularly when the uncertainties may be used to justify setting aside the results. This
30    is particularly true under resource-limited circumstances where more pressing and more certain
31    matters may justify immediate attention and action.
32    Another issue involved determining which form of uncertainty analysis would be most useful to
33    stakeholders. The Sacramento team found that their stakeholders were interested in "stylized"
34    scenarios. For example, the El Dorado Irrigation district currently uses the worst three-year
35    drought of record as the basis for developing drought plans. To be responsive to and consistent
36    with the existing decision framework, the Sacramento team developed alternative scenarios such
37    as a 4-year drought of similar magnitude to the 3 year drought, but one that is also 2°C warmer.
38    Some decision makers - particularly those who are statistically sophisticated - will only be
39    willing to place confidence in results if researchers are able to provide them with quantitative
40    estimates of uncertainty. Rather than circumventing a quantitative uncertainty analysis,
41    researchers should ideally embrace it; conversations about uncertainty are a very useful  frame for
42    engagement between the two communities. As a practical matter, however, it appeared that by
43    the time the teams were prepared to conduct an uncertainty analysis, their resources were
44    dwindling and other priorities took precedence.

      August_2006_...                                                   _ _.External Review Report:. Watershed. Case.S.fypfes
 1    The researchers expressed interest in the possibility of framing climate change uncertainty (an
 2    unfamiliar type of uncertainty) using comparisons with more familiar sources of uncertainty. For
 3    example, long-range water resource plans generally make assumptions on population growth,
 4    changes in demand for key uses (e.g., agriculture), and changes in per capita demand. While
 5    these assumptions are sometimes heroic, they are nevertheless familiar (unlike climate change-
 6    related factors). The teams agreed that comparing sources of uncertainty would help stakeholders
 7    realize that all long-term decisions are made in an uncertain context, perhaps reducing reluctance
 8    to incorporate climate change in decision making.
 9    Uncertainty issues represented a serious challenge to the research teams, both in terms of
10    developing techniques to address uncertainty within the modeling frameworks, and for
11    communicating their results in a way that is understandable and useful to stakeholders.

12    3.2  Stakeholder Processes
13    Given the formidable technical hurdles and advanced scientific nature of the topics under
14    consideration by these investigators, it is surprising that stakeholder interactions were deemed
15    among the most challenging aspects of these projects. Each case study research team approached
16    stakeholder inclusion somewhat differently, though a number of features and impressions were
17    common among them.
18    In general, the teams relied on existing stakeholder relationships and processes. Interactions with
19    these stakeholder groups were moderate in the frequency, scope, and intensity. The information
20    flows were primarily unidirectional  - from the research teams to stakeholders, although
21    stakeholders were given opportunities to provide input on each study's endpoints. This may have
22    been due to the types of stakeholders who were engaged in these particular case studies. A better
23    understanding of the appropriate stakeholders for whom this information is relevant might result
24    in a closer working relationship with better information flows. A kickoff meeting was a common
25    feature among all of the case studies. After that point, the Sacramento study continued with
26    structured elicitation of stakeholder  input. Stakeholder involvement for the Maryland and San
27    Pedro studies was generally less structured and more opportunistic after the initial kickoff.
28    Teams found that stakeholder involvement can be a resource-intensive exercise, which can divert
29    the attentions and drain the ability of researchers to focus on the technical and decision making
30    issues that were of greatest interest to EPA and the study teams. This and other challenges the
31    teams faced in engaging stakeholders are discussed in more detail below.
32    3.2.1  Defining and Identifying Stakeholders
33    The concept of stakeholder engagement, required by EPA in its specifications for the case study
34    cooperative agreements, was not clearly defined from the outset. Stakeholders included a wide
35    array of individuals from decision makers to non-governmental organizations to interested
36    citizens. The teams found that determining who to engage and then exchanging meaningful
37    dialogue with those key players could be paralyzing to the assessment process if it was not
38    focused  along the way.
39    The Maryland team initiated a broad-based introductory meeting for stakeholders. Professional,
40    academic, and EPA personnel were  the initial contact points for assembling this group. The goal
41    of the meeting was primarily information dissemination from the researchers to interested
42    parties. Subsequent interactions and input from stakeholders was minimal. Researchers did
43    maintain a close collaborative interaction with the Montgomery County Department of

      August_2006_...                                                   _ _.External Review Report:. Watershed. Case.S.fypfes
 1    Environmental Protection - a primary stakeholder - which resulted in the sharing of data and
 2    incorporation of researchers' findings into a wider planning context.
 3    Researchers in the Upper San Pedro Basin began work in a setting that already had an active
 4    stakeholder group—the Upper San Pedro Partnership (USPP). The USPP is a consortium of
 5    local, state, and federal government organizations, Fort Huachuca Army Base, businesses,
 6    citizens, and conservation groups. The USPP had been working to develop decision support tools
 7    for the analysis of alternative water management regimes, and members of the case study group
 8    became participants in this  ongoing process. Additionally, their affiliation with the Sustainability
 9    of Semi-Arid Hydrology and Riparian Areas (SAHRA)—a National Science Foundation-
10    supported research center at the University of Arizona—provided them with indirect access to
11    stakeholders.
12    The group working in the Sacramento River Watershed took a more active role in
13    comprehensively identifying stakeholders, particularly with the intention of influencing ongoing
14    decisions regarding water allocations. The team assembled a Technical  Advisory Panel (TAP)
15    formed of regionally-based academics from the physical, ecological, and economic sciences.
16    This panel made recommendations regarding other experts to consult and the development and
17    application of the modeling framework. Professional connections and trust in the competence of
18    the study team members were very important to this process because of the access these
19    connections provided to stakeholder groups. Additionally, these researchers hired a consultant
20    (CONCUR) to interview high-level water and ecosystem management organizations and make
21    strategic recommendations  regarding potential  applications of this research.
22    3.2.2  Balancing Stakeholder and Researcher Interests
23    Stakeholders engaged in each case study had diverse views on which services should have
24    highest priority for study resources. Teams had to balance those competing stakeholder interests
25    with their own research interests. The team members recognized the dilemma that the primary
26    professional and career focus of many scientists—credibility—requires academic achievements
27    that can come at the expense of spending time with  stakeholders to prove the relevancy and
28    legitimacy of their work to  decisions. Stakeholder interactions increase  the likelihood of a
29    project's usefulness. However, efforts to maintain those interactions can be onerous and
30    resource-intensive, and thus come at the expense of other priorities. The inverse  of this challenge
31    is also true: managers do not necessarily recognize the  relevance of research (especially research
32    on long-term problems), often deeming it merely an academic exercise with no application to
33    their pressing concerns.
34    Climate change is a particularly acute example of this situation. All research teams reported that
35    climate change is simply not recognized by managers and decision makers as a concern that must
36    be addressed within a relevant timeframe. Decision makers generally focus on quantifying water
37    use and supply allocation, for which they do not yet recognize climate change as a significant
38    stressor. They have not yet  concluded that investing time in climate change-related work will
39    assist them with what they consider to be the more immediate concerns of their charge. For
40    example, the explicit identification of ecosystem services that depend highly on  ongoing climate
41    change and variability may require an interactive stakeholder process that neither side
42    (researchers and stakeholders seem prepared to participate in.
43    The Maryland and San Pedro studies reflected researcher interests. The study findings were later
44    shared with decision makers and supported decision making indirectly.  The Sacramento case

      August_2006_...                                                   _ _.External Review Report:. Watershed. Case.S.fypfes
 1    study reflected interests of some of the stakeholders. Stakeholders in the Sacramento River
 2    Watershed guided the selection and prioritization of analytic activities, helped establish project
 3    goals, shared expertise, and provided information on a variety of areas including public values,
 4    equity considerations, and relevant decision processes (Yates et al. 2006). Researchers in
 5    Sacramento solicited input from stakeholders to develop a list of current water management
 6    decisions that may be sensitive to climate. They then applied several criteria to the list to narrow
 7    it down to a reasonable number of decision making processes amenable to climate change
 8    assessment (Yates et al. 2006). One of the criteria ensured that some segment of the stakeholder
 9    community had expressed a concern about the potential impact of climate change  on the project
10    (Yates et al. 2006).
11    3.2.3  Engaging Stakeholder Communities
12    All teams presented their proposals to stakeholder groups at the onset of these projects. These
13    meetings were well attended; however, researchers from at least two of the case studies reported
14    that attendees were not very responsive or specific when asked what information stakeholders
15    themselves could provide to the study or what information  or tools would be valuable for the
16    research effort to produce. For the most part, the general public and laypeople who attended the
17    initial meetings perceived their role as passive rather than participatory; their interests were
18    primarily in hearing what researchers themselves were doing, not in playing an active role in
19    study development.
20    Teams found it challenging to determine when to involve stakeholders in the assessment process
21    after the kick-off meeting. The teams recognized that it would be ideal to have regularly
22    scheduled meetings/consultations, but technical and modeling challenges prevented regular
23    production of results. Researchers were disinclined to hold meetings without results. On the
24    other side, many of the stakeholders seemed to lack both an understanding of whether and how
25    study findings would be useful to them and a commitment to use the findings.
26    The Maryland case study team was able to develop an interactive relationship with the
27    Montgomery County Department of Environmental Protection (MCDEP) that led  to an exchange
28    of data and some collaboration regarding desirable  study products. Collaborative interactions
29    with MCDEP were mostly limited to the land use components of the study, since they were less
30    interested in the climate change components (Palmer et al. pers.  comm.)4.
31    The San Pedro team's association with USPP and SAHRA allowed them access to an existing
32    stakeholder process to  which they could contribute  research, planning, and management
33    information. These stakeholders were  not actively engaged in the research planning or
34    implementation, but study results will be made available to them.
35    The Sacramento study  team convened a technical advisory panel (TAP) and met with them at the
36    beginning, middle, and end of the investigation to receive technical as well as stakeholder input
37    to the analysis. The team engaged other stakeholders via ad hoc  meetings with several high level
38    decision making organizations. They found that the key to establishing working relationships
      4 Subsequent citations of personal communication with the Maryland team refer to the interview at the conclusion of
      the case studies. Team members present were Margaret Palmer, Jim Pizzuto, Karen Nelson, Nancy Bockstael, and
      Glenn Moglen.

      August_2006_...                                                  _ _.External Review Report:. Watershed. Case.S.fypfes
 1    with the targeted stakeholders was to demonstrate that their work was relevant to the decisions at
 2    hand, scientifically credible, and legitimate as an approach to climate change analysis—three
 3    concepts that guided their stakeholder meetings. The Sacramento team also stressed the
 4    importance of finding an advocate among the participating stakeholders. One person who acts as
 5    a champion for collaborating with researchers and participating in the assessment process can
 6    help sustain and facilitate the relationship (Purkey and Yates pers. comm.).
 7    3.2.4  Establishing Credibility
 8    The Sacramento team noted that climate change may be viewed as a less important or non-vital
 9    concern to stakeholders. Climate change is not considered in most water resources planning
10    efforts in California because there is a general perception that significant changes will not occur
11    within the typical 20 to 30 year planning horizon used in most NEPA and CEQA studies (Yates
12    et al. 2006). This view is widespread and presents a particular challenge for teams seeking to
13    establish relevance.
14    The Sacramento team found that continuous input from stakeholders enhanced the relevance and
15    credibility  of results (Yates et al. 2006). They also concluded that it is important to spend time
16    interfacing with stakeholders throughout the process to make them more comfortable with the
17    assessment process.
18    3.2.5  Communicating Results
19    All of the case study teams have used scientific publications (e.g., peer-reviewed journals,
20    books) as a primary means  of dissemination, but many stakeholders do not find these types of
21    publications to be particularly accessible, user-friendly, or succinct sources of information.
22    Further, results presented in scientific publications may not be directly applicable to decision
23    making.
24    Case study teams have found that one-on-one interaction with decision makers or stakeholder
25    meetings to share decision-relevant results is an effective means of communicating with
26    stakeholders. The Maryland team worked closely with the Montgomery County officials
27    throughout the process and shared findings as they became available. The San Pedro team shared
28    results with water resource  managers and the Bureau of Land Management. The Sacramento case
29    study group is participating in a multi-stakeholder state-wide planning process to project water
30    resource supplies and needs for the State over the next 50 years.
31    The Sacramento and San Pedro case studies both noted the importance of using the concept of
32    "ecosystem services" to enable individuals of different backgrounds to speak in a common
33    language about the values they hold for ecological processes and functions (Yates et al. 2006).
34    All three of the teams noted that, in principal, communicating decision-relevant results to
35    decision makers in terms that avoid excessive technical jargon facilitates a more effective use of
36    scientific findings.

37    3.3  Relevance of Impacts and Adaptation to Decision Making
38    EPA's third major goal in conducting the watershed case  studies was to assess adaptation options
39    to improve decision-makers' ability to respond to global change. EPA recognizes, however, that
40    it is often a challenge to move beyond impact assessments and address adaptation options. Only
41    the Sacramento case study assessed adaptation options, due largely to a second round of funding.
42    The Sacramento team relied on a directly decision-driven approach that resulted in findings

      August_2006_...                                                  _ _.External Review Report:. Watershed. Case.S.fypfes
 1    relevant to planning processes that were soon to be initiated or already underway. The different
 2    approaches taken by the study teams - based on the goals communicated by EPA, available
 3    funding, and their own needs and resources - offer valuable lessons for future watershed
 4    assessments. The comparison of the teams' approaches that follows informs the later conclusions
 5    and recommendations for future GCRP efforts.
 6    The Maryland and San Pedro teams focused primarily on conducting impact assessments—
 7    determining the effects of global change (including climate change, land use change) on water
 8    quantity and quality and the consequences for aquatic ecosystems. These assessments made
 9    significant contributions by linking multiple models to better understand stressor interactions and
10    responses. The assessments were not conducted from an "information demand" or decision-
11    driven perspective, and thus were not immediately and directly applicable to specific decisions.
12    However, both the Maryland and San Pedro assessments ultimately supported subsequent
13    decisions or provided corroborating evidence for proposed decisions.
14    The Maryland team found that decision makers at the county level (Montgomery County) were
15    more interested in the land use change component of the case study than the climate change
16    component. Montgomery County officials seemed to be less interested in possible future stresses
17    and focused on problems they are facing in the immediate term. However, the Maryland team's
18    findings indicate that decision makers should be considering climate change. For example,
19    stormwater management facilities are currently being retrofitted or built without accounting for
20    potential changes in the intensity of future rainfall events that could affect the performance of
21    these facilities. On the other hand, County decision makers took an interest in the findings
22    regarding nutrient concentrations in various streams and were pleased to see evidence that
23    riparian buffers have a definite impact on those concentrations. They have used the study
24    findings to validate recommendations they had already made and research they already have in
25    progress (Palmer et al. pers. comm.).
26    The San Pedro team's findings were used by the Bureau of Land Management (BLM) and water
27    resource managers. The latter have used some of the team's models to assess flow rules for
28    reservoir releases from a dam in the middle San Pedro area. The team's results regarding loss of
29    water in the system led to a BLM decision to reintroduce beaver in an attempt to impound and
30    detain water rather than letting it flow downstream. In this case, decision makers used the study
31    findings, but did not consult the study team on the proposed management solution (which
32    researchers anticipate could cause changes in habitat, topography, ecology, and impacts to the
33    gallery forest). The San Pedro team noted that the politics in this area are an obstacle for the
34    incorporation of study findings into the decision making process.  Resource managers also seem
35    reluctant to make decisions based on climate change options due to the level of uncertainty and
36    other factors that take precedence in management decisions (Galbraith et al. pers. comm).
37    The Sacramento team took a different approach than the other two case studies — in the second
38    phase of this project, they started by identifying key water-related decisions and then tailored
39    their analytical work to meet the needs of decision makers. They consulted stakeholders to
40    develop a list of ongoing decision making processes in the California water system that might be
41    sensitive to climate change. The list was then narrowed to include only decision-making
42    processes that met three criteria:

43       •    The success of a project to be implemented would be strongly influenced by hydrologic
44           variability;

     August_2006_...                                                  _ _.External Review Report:. Watershed. Case.S.fypfes
 1       •  The investment in a project would be substantial enough to merit the consideration of
 2          climate change impacts; and

 3       •  Some segment of the stakeholder community was concerned about the potential impact of
 4          climate change on the project (Yates et al. 2006).
 5   The shorthand for these criteria are Sensitivity, Significance, and Stakeholder support,
 6   abbreviated as the "3S standard" (Purkey et al. 2006).
 7   The Sacramento team targeted  assessment endpoints that were intrinsic to the decision making
 8   processes (e.g., water flow). The WEAP model itself is a management tool that was modified for
 9   this assessment to include climate change. The Sacramento team included in their scenarios
10   adaptive measures to assess the ability to mitigate climate change impacts.
11   The Sacramento team was able to apply findings to decision making processes to a greater
12   degree than the other teams because they received a second phase of funding.  These decision
13   making processes included: (1) the Integrated Regional Water Management Plan (IRWMP) for
14   the Consumes, American, Bear, and Yuba (CABY) watersheds; (2) the California Department of
15   Water Resources' 5-year water planning process (Bulletin 160); (3) an assessment by several
16   water utilities generating hydropower in the American river basin of how vulnerable they are to
17   climate change and their ability to meet more stringent inflow water demands; and (4) the 2006
18   Climate Action Team Final Report to Governor Schwarzenegger and the California Legislature.

     August_2006_...                                                 _ _.External Review Report:. Watershed. Case.S.fypfes
 i          4.  Conclusions and Recommendations for Future Watershed
 2                                             Assessments

 3    4.1  Contributions to Body of Knowledge on Climate Change Impacts
 4    As noted earlier, these case studies addressed many issues associated with climate change and
 5    land use change and their impacts on the environment. These studies contributed both
 6    methodological advances and key results to this body of knowledge. The place-based approach
 7    used to generate climate impacts information was a key factor in ensuring its usefulness to
 8    specific decision makers, and the lessons learned and results may benefit or be applied to other
 9    assessment teams and geographic regions. This section discusses the methods and key results
10    developed by the studies.
11    4.1.1  Methodology Development
12    The most significant methodological advance across the three case studies has been the
13    integration of interdisciplinary models. The Sacramento team integrated a climate and hydrologic
14    model within the water planning environment — other water planning models typically hold
15    hydrology as unchanging from the historic record — and conceptually linked it to an ecological
16    model. The San Pedro case study team integrated hydrologic, vegetation, habitat and climate
17    change models, and the Maryland case study team integrated a hydrologic, geomorphologic, and
18    ecological model using driver data from climate models. Each of the teams recognized the value
19    of linking these types of models and expressed an interest in integrating additional models to add
20    more dimensions to the studies.
21    This methodological advance was significant from both the scientific and decision support points
22    of view. Scientifically, modeling system behavior reduces uncertainties associated with ignoring
23    system interactions that cannot be modeled without an integrated approach. Similarly, integrated
24    modeling reduces uncertainty for decision makers in terms of forecasting effects and providing
25    new information about a whole suite of components quickly. In this way, integrated models
26    make it easier and quicker to address decision-relevant questions using scientific models of
27    system behavior. This approach also makes it  easier to analyze multiple scenarios.
28    The Sacramento case study team's assessment framework helps decision makers evaluate
29    possible adaptation strategies and identify tradeoffs among important ecosystem services. This
30    framework, known as WEAP21, is able to provide integrated water resource management
31    support to other regions as well. The decision-relevant approach taken by the team and the
32    subsequent direct applicability of the results to decision making processes provides an excellent
33    model for future GCRP watershed analyses (Yates et al. 2006).
34    The San Pedro team had several ideas for how to improve the methodological approach in the
35    future. First, the models should incorporate a greater degree of complexity with  respect to habitat
36    responses to ecosystem changes. This  would require observations to be done of  subtle changes in
37    the ecosystem to detect modifications  of habitat. Their sense is that small changes in the
38    ecological matrix can cause large swings in the wildlife community. Second, more knowledge is
39    needed about the life history of species in the system to avoid only looking at conditions in an
40    equilibrium situation. Third, the assessment should be expanded beyond the riparian area to look
41    at climate change impacts on the entire system, including vegetation in the uplands or the
42    recharge processes involved in mountainous areas  (Galbraith et al. pers. comm.).

      August_2006_...                                                   _ _.External Review Report:. Watershed. Case.S.fypfes
 1    The methods developed in the Maryland and San Pedro case studies are transferable to regions
 2    with similar conditions, although the input data would have to be modified accordingly. For
 3    example, the Maryland models could be applicable in the North Carolina Piedmont and other
 4    areas that have similar ecosystems. The San Pedro model, although it is specific to riparian
 5    systems in the Southwestern U.S., could also be applied to other similar ecosystems, and is
 6    currently being used in the Rio Grande and Kern watersheds in California. In addition, the
 7    RIPET submodel could also be transferable to specific types of wetlands, such as the Everglades.
 8    The most transferable modeling system may be the WEAP model. Since the intrinsic logic
 9    behind WEAP is universal, it could be produced for other locations with site specific data over a
10    relatively short time frame.
11    4.1.2   Key Results
12    The Maryland team separately quantified the effects of land use change and climate change, but
13    determined that it was unclear which stressor had a larger impact; they found that each
14    contributes to the same impacts but in slightly different proportions. Land use change provides
15    more sediment due to increased construction and increased impervious surface, and climate
16    change causes more increased storm flow,  disturbances to the streambed, and variability in
17    conditions than land use. Effects on ecological processes are thus generally negatively influenced
18    by the projected climate and land use changes, and when the stressors are combined,
19    predominantly negative effects emerge. (Moglen et al. 2006).
20    The Sacramento and San Pedro teams did not separate the effects of land use change and climate
21    change, but expressed interest in evaluating these stressors independently in the future.
22    Sacramento researchers noted that they would like to systematically separate ecosystem stressors
23    given their particular challenge of moving  water out of irrigation and into the urban environment
24    (Purkey and Yates pers. comm.). The San Pedro team noted that climate change is not as
25    significant a stressor as aquifer depletion; however, when both stressors are applied together,
26    there is a synergistic effect. The team also  acknowledged that isolating the effects of land use
27    change and climate change in the future could provide information about runoff and surface flow
28    (Galbraith et al. pers. comm.).
29    The Maryland team found that up to three-quarters of the fish species would be highly stressed
30    under the combined effects of land use change and climate change and that this outcome could
31    be mitigated by maintaining riparian buffers and decreasing urbanization. The team also
32    concluded that all ecological processes were not negatively influenced by projected climate
33    change and land use change; however, when they are combined, predominantly negative effects
34    emerge. In addition, low flow modeling indicates that future precipitation trends will influence
35    hydrologic and ecological processes more than future temperature trends, and the frequency of
36    low flow events of a given magnitude will  increase under future climate  and land use changes
37    (Moglen et al. 2006).
38    The San Pedro team found that among their five climate scenarios, the warmer drier scenario
39    could exacerbate current water use conflicts between the human and natural ecosystems of the
40    upper San Pedro basin and could accelerate the decline of cottonwood-willow gallery forests. A
41    wetter future could partially mitigate the impacts of human water use (Dixon et al. 2006).
42    In one analysis, the Sacramento team addressed the issue of adapting to climate change by
43    looking at three future alternatives including a simulation without adaptation, a simulation with
44    increases in irrigation efficiency, and a simulation with improved irrigation efficiency and shifts

      August_2006_...                                                    _ _.External Review Report:. Watershed. Case.S.fypfes
 1    in cropping patterns related to the simulated status of available water supplies. The results
 2    showed that improvements in irrigation efficiency led to a decline in supply requirements. When
 3    coupled, the effect of improved irrigation efficiency and a dynamic crop pattern was a decrease
 4    in water supply requirements. In addition, the study showed that the management structures and
 5    practices that adversely affected the fish populations historically may provide an opportunity to
 6    alleviate some  of the future impacts of climate change. More 'natural' and unmanaged systems
 7    may provide fewer opportunities for salmonid conservation (Yates et al. 2006).
 8    All three of these place-based assessments provided impacts information that will be useful to
 9    specific decision makers as they develop management responses. Each case study team was able
10    to examine the interaction of climate change with other stressors already present, particularly
11    land use change, and was able to conclude that climate change will  exacerbate those effects.
12    Where stressors were examined separately by the Maryland case study, results revealed that the
13    interactive effects were strongly negative and more apparent than when the stressors were
14    considered separately.

15    4.2  Assessment Process
16    As noted earlier, a "portfolio approach" was used to select case studies and commission
17    assessments in three distinctly different watersheds with differing ecosystem services, scales, and
18    decision-making processes. One benefit of this approach is that although the study sites differ
19    from each other hydrologically and bioclimatically, they are representative of a broader array of
20    conditions within larger regions of the United States. Another benefit of this approach is that it
21    demonstrates the diversity of methods, results, and processes over a range of settings. Even
22    though there are many distinctions and differences in the assessment processes used in the case
23    studies, associated with their unique objectives and endpoints, several cross-cutting findings
24    emerged that can be extrapolated to other watersheds and regions and that can shape GCRP's
25    and other research institutions' strategies for designing the process for similar  assessments.  The
26    three main ways that results are more broadly useful are:
27        1.  Extrapolation of the results themselves: individual results can be extrapolated for each
28           watershed to similar systems. For example, San Pedro can be extrapolated to other arid
29           systems relying on groundwater; Maryland results can be extrapolated to other Piedmont
30           rivers;  the interactive effects of climate and land use change can be generalized to other
31           watersheds, and the results that climate will exacerbate existing effects of stressors
32       2.  The methods used to link process models across disciplines may be used by other
33           assessment teams and in other geographic regions of the country.
34       3.  The insights gained about the assessment process, such as the standardization of methods
35           for climate scenarios, stakeholder processes, and other topics described below, will be
36           helpful to any research institution seeking to produce useful climate impacts information
37           for decision makers.
38    4.2.1   Provide Keystone Capabilities and Tools to Assessment Teams
39    As noted earlier, there were several areas where all three teams expended considerable effort on
40    similar tasks. It is reasonable to expect that other watershed-level assessments  would need to
41    undergo a similar process; when conducting similar assessments in the future,  several keystone
42    capabilities and tools might be useful:

      August_2006_...                                                   _ _.External Review Report:. Watershed. Case.S.fypfes
 1       •   Tools for converting GCM output to watershed modeling input. The teams all had an
 2           initial focus on reviewing and interpreting GCM runs to develop their climate scenarios.
 3           GCMs analyze temperature and precipitation on short time steps ranging from 15 minutes
 4           to half a day. These results are often stored as averages over longer time scales of (often
 5           monthly) averages, however, because of data storage constraints. Most hydrologic
 6           processes require daily (or even hourly) inputs over a finer geographic scale, and need to
 7           be downscaled on both a temporal and geographic basis. GCM runs are also available for
 8           many different combinations of emission scenarios and climate sensitivities, and it can be
 9           daunting to choose among the scenarios. The lesson  learned from these case studies is to
10           provide expertise to future assessment teams to aid in selecting, interpreting, and
11           downscaling GCM output. Future assessments could attempt to establish a tool or
12           methodology for handling climate information and explicitly implementing (and
13           evaluating) those practices.

14       •   Tools to develop or apply trend analysis of precipitation and hydrology to complement
15           GCM output. Downscaled GCM outputs have a high level of uncertainty associated with
16           them and therefore may attract scientific controversy. A simple trend analysis of climate
17           variables may be a complementary approach to create future scenarios (Denault et al.
18           2006). However, conventional precipitation and hydrology analyses of intensity,
19           duration, and frequency are based on the assumption that there is no underlying trend in
20           the record, i.e., that a record from 100 years ago has  equal relevance to predicting
21           tomorrow's conditions as a record from 100 days ago. Several powerful statistical
22           techniques have been developed to evaluate trends, and could be made available to
23           watershed researchers to complement GCM output (Denault et al. 2006).
24       •   Consulting services in keystone skills. Assessment teams could benefit from having
25           access to consulting help in key areas such as  climate scenario development and habitat
26           suitability analysis. To the extent that many of the assessments begin with the same
27           inputs and end with habitat-related outputs, it  may streamline the assessment process to
28           provide access to experts.
29       •   Techniques for expressing uncertainty.  As noted earlier, the assessment teams generally
30           found it  difficult to express uncertainty, and several noted that it would be helpful to have
31           guidance on how to characterize and communicate uncertainty results. In addition,
32           several noted that it would be useful to be able to compare climate change-related
33           uncertainty to uncertainty from other, more familiar  sources relevant to long-term water
34           resource decision making, (e.g., population, land use change, per capita water demand).
35    For all  of these keystone capabilities and tools, the benefit of providing them to the watershed
36    teams would have to be balanced against the objective of building broad-based technical capacity
37    and testing alternative approaches, which argues for less, rather than more, concentration and
3 8    standardization of expertise.

      August_2006_...                                                   _ _.External Review Report:. Watershed. Case.S.fypfes
 1    4.2.2  Emphasize Model Linkages
 2    Given the multi disciplinary nature of these projects and the need for assessment teams to develop
 3    new modeling capabilities to analyze climate change impacts or opportunities for decision
 4    support, one of the key challenges is to facilitate smooth links between submodels. This was one
 5    of the most difficult challenges for the teams to overcome.
 6    Although the disciplines that were well-represented among the teams (e.g., ecology, hydrology,
 7    geomorphology) tend to espouse "systems thinking," many of the details involved in assuring
 8    seamless integration of models are viewed most effectively from an information technology
 9    perspective. There are trade-offs between setting up  an IT-intensive interface for linked models
10    versus a "hand-crafted" solution, and the different teams dealt with the trade-offs in different
11    ways. In retrospect, however, it appears that in several cases the teams would have benefited
12    from more design work up front in clearly defining inputs, outputs, and interactions among
13    submodels.
14    4.2.3  Change Selection Criteria for Assessments
15    One of the recurring themes in this report is that the  case studies were quite thorough and
16    innovative in assessing climate change impacts, but did not necessarily attain the objective of
17    providing decision support. This may be due, in part, to the emphasis on selecting case studies
18    where there was a good foundation of existing  data and models so that impacts could be assessed
19    as efficiently as possible. For future watershed assessments, it may be useful to modify the
20    selection criteria to emphasize case studies where it is clear that (1) decisions are being made that
21    are sensitive to climate change, and (2) where there are  existing relationships with decision
22    makers that would enable the project team to provide relevant decision support products.
23    Another factor that should be re-evaluated in terms of selection criteria and study design relates
24    to scale. The research teams noted that there were  some mismatches between available data
25    versus the scale of data needed to support assessment and decision making. All three teams were
26    able to bridge the gaps.  Nevertheless, in developing a strategy for future work, it would be
27    useful to consider the scale at which GCM and watershed-level information is available, the
28    scale at which key endpoints are assessed, and  the uncertainty introduced by bridging the gap, to
29    assure that it will be feasible to produce good science and sound decision support.
30    4.2.4  Establish Forum for Researchers to Compare Notes
31    The mid-project review meeting was very productive in terms of providing an opportunity for
32    cross-fertilization among the project teams. It not only encouraged technology transfer on an
33    inter-team basis, but it also provided an impetus for intra-team coordination, which was a
34    continuing challenge within the teams.  Preparation for the review meeting encouraged attention
35    to several cross-cutting issues and envisioning  likely results, and thus facilitated several
36    adjustments and improvements to the methods. It also provided an opportunity for the GCRP to
37    reiterate and clarify its desired outputs. If similar case studies are  done simultaneously in the
38    future, a mid-project forum should be convened to enable investigators to share notes on
39    progress and to make mid-course corrections in project goals and methods.
40    4.2.5  Require an Uncertainty Analysis Plan
41    In addition to the earlier recommendation to develop a set of tools or methods for analyzing
42    uncertainty in assessments and decision support, future assessments should include a requirement
43    to address uncertainty more comprehensively, and, where possible and appropriate,

      August_2006_...                                                    _ _.External Review Report:. Watershed. Case.S.fypfes
 1    quantitatively. This would be helpful both from a perspective of scientific rigor and from a
 2    perspective of decision relevance. It should be made clear that a comprehensive uncertainty
 3    analysis is a key output of the study, and that resources need to be reserved to produce this
 4    output.
 5    It would be useful to set out an uncertainty analysis plan as one of the initial deliverables in the
 6    process, and to design the plan to reflect feedback from decision makers on what types of
 7    uncertainty analysis would be most relevant to decision making.  This feedback would help
 8    determine whether the plan should focus on  developing several alternative scenarios, or a risk
 9    analytic approach that explicitly assigns probabilities to various components, or some other
10    qualitative or quantitative characterization of uncertainty.
11    Many of the elements of uncertainty in watershed assessments like those described in this report
12    are unknown or extremely difficult to quantify or parameterize, making it difficult to express
13    uncertainty in terms of probabilities. Moreover, there is a growing literature that argues that it is
14    not necessary to fully characterize probabilities of climate change to develop climate adaptation
15    policy, but rather to focus on developing strategies of resilience and adaptive environmental
16    management that enhances coping capacity across a broad range of climate outcomes (Yates
17    pers. comm.)5. Nevertheless, regardless of the specific approach to analyze uncertainty, future
18    watershed assessments should make a plan to conduct this analysis and then execute the plan.

19    4.3  Stakeholder Process
20    Streamlining and focusing stakeholder-related efforts is necessary and desirable. The lessons
21    learned from this portfolio of case studies have led to an evolution in our view of effective
22    stakeholder interaction. Our goal to produce case study outputs which directly impact the
23    policies that avert potentially negative changes in areas sensitive to land use change or climate
24    change may require more guidance to study  investigators in the selection of priority stakeholders
25    and managing the process to ensure useful outcomes. The case studies showed that stakeholder
26    relationships may not need to extend to all potentially interested members of the lay public;
27    instead, they should target specific decision  makers who have an identifiable stake in the study's
28    goals. The broad definition of stakeholders that GCRP initially embraced and the associated
29    challenges that arose may have additionally  hampered attainment of the objective to provide
30    decision support. The recommendations that follow stem from these lessons learned in an effort
31    to improve the stakeholder process.
32    4.3.1   Focus on Decision Makers
33    Decision makers are the most important stakeholders to engage during the assessment process.
34    Study teams should focus on engaging decision makers early and interacting with them
35    throughout the assessment. Collaborating with decision makers will  ensure that case study
36    findings are decision relevant and immediately applicable. Working closely with decision makers
37    to supply information based upon their needs and demands will also help facilitate later
38    transferability of case study results and processes. Decision-driven research tends to be more
39    transferable, since it seeks to answer questions that decision makers  elsewhere are  likely to face.
40    Also, where possible, focus on developing collaborative working partnerships with members of
      5 E-mail from David Yates to Randall Freed, July 31, 2006.

      August_2006_...                                                    _ _.External Review Report:. Watershed. Case.S.fypfes
 1    the decision making body to gain their interest and trust in the assessment results. A working
 2    partnership builds technical capacity within the decision making body that increases the
 3    likelihood of climate change impacts being considered beyond the particular assessment.
 4    4.3.2   Build on Existing Stakeholder Relationships
 5    Research teams already work with stakeholders in many areas and have developed long-standing
 6    relationships with decision makers. Research teams should build on these existing relationships
 7    when seeking input from stakeholders. Strengthening existing relationships will take less time
 8    and resources than trying to establish new relationships with numerous stakeholders. Further,
 9    existing relationships can open doors to meeting and collaborating with new stakeholders who
10    may be similarly interested in  study findings. In much the same way that the selection criteria for
11    this set of projects involved availability of existing data, future projects may also benefit from a
12    selection criterion to evaluate availability of existing relationships.
13    4.3.3   Facilitate Interaction of Researchers and Decision Makers
14    Researchers and decision makers often speak in different languages or use different terminology.
15    Focusing on ecosystem services helps to bridge the gaps (see Section 4.3.5, Communicate
16    Results using Common Language below), but interactions may need further facilitation.
17    Researchers may be preoccupied with the technical aspects of the work and professionally ill-
18    equipped to manage external interests. One possible solution suggested at the expert workshop
19    was to establish a separate office (within EPA or at the university where the study is being led) to
20    manage public inputs to the study itself, rather than relying on the assessment team alone. Other
21    means for facilitation should be considered to help foster open communication between
22    researchers and decision makers.
23    4.3.4   Establish Credibility with Decision Makers
24    Establishing credibility and providing relevant findings takes time and commitment on the part
25    of researchers. Study teams must set aside time to interface with decision makers to make them
26    more comfortable with the assessment process and findings and how they fit into ongoing
27    decision processes. Finding one or more champions amongst the participating stakeholders can
28    help keep the lines of communication open between researchers and decision makers.
29    4.3.5   Communicate Results using Common Language
30    The concept of "ecosystem services" provides  an excellent means for communicating  scientific
31    results to a wide variety of stakeholders. It allows researchers and stakeholders to speak in a
32    common language that is not overly reliant on technical or industry-specific jargon. Distilling
33    study results down to the findings that are decision relevant also goes a long way to bridging the
34    gap that so often forms between researchers  and decision makers.

      August_2006_...                                                   _ _.External Review Report:. Watershed. Case.S.fypfes

        Table 3: Summary of Recommendations for Future Watershed Assessments
Assessment Process
• Provide Keystone Capabilities and
Tools to Assessment Teams
• Emphasize Model Linkages
• Change Selection Criteria for
• Establish Forum for Researchers to
Compare Notes
• Require a Comprehensive,
Quantitative Uncertainty Analysis
Stakeholder Process
• Focus on Decision Makers
• Build on Existing Stakeholder
• Facilitate Interaction of Researchers
and Decision Makers
• Establish Credibility with Decision
• Communicate Results using Common

4.4   Final Thoughts and Future Directions for GCRP
The case study approach yields richness of detail in terms of methods and results, and propels the
research team well up the learning curve on climate change issues. It has proven extremely
effective in pushing forward the state of the art in impact assessment and characterizing the
potential effects of climate change and land use change on ecosystem services, which is an
essential foundation for adaptation.
It is important to ensure that capacity for doing assessments is in place before the project starts.
As discussed at the beginning of the section on cross-cutting findings (see Section 3), there are a
number of critical success factors for the case studies, including: good data, good models, clear
goals, public awareness of the issue, etc. Identifying that these critical success factors are in
place before launching an assessment will help determine if there is sufficient capacity to
undertake such an assessment.
It is not enough to test ecological models;  there is a clear need to also support decisions.
Decision-relevant approaches need to be developed and employed. A need for transferability of
assessment methodologies and model results arises out of the need to support decisions. If goals
are defined upfront and the assessment approach is decision-relevant, different regions and
watersheds with the same goals will be more likely to produce transferable methodologies and
results. Approaching assessments with a focus on ecosystem services works well, but it is
important that decisions drive the selection of which ecosystem  services to focus on.
Moving forward, it may be constructive for study teams to develop a framework for assuring that
research is decision-driven,  not necessarily by  early and  frequent exposure to a broad set of
stakeholders, but instead by a focus on a narrow set of decisions and stakeholders where the
information will be most useful. The GCRP is  developing a decision-driven approach through
some of our other activities. For example,  the GCRP is conducting a pilot study to inventory and
analyze climate-sensitive decisions,  and has developed a research design on which future
assessments may be modeled. This project may provide the foundation data and analysis to aid in

      August_2006_...                                                    _ _.External Review Report:. Watershed. Case.S.fypfes
 1    developing a long-term strategy for providing effective decision support to the relevant decision
 2    makers.
 3    An effective way to establish adaptation priorities is to classify decisions into three categories:
 4           (1)     decisions unlikely to be affected by climate change
 5           (2)     decisions probably affected by climate change that could benefit from adaptive
 6                  decision support in the short term
 7           (3)     decisions that will probably be affected but where adaptive actions can (and often
 8                  should) be addressed later.
 9    Decisions that fall into category (2) present opportunities for immediate decision support in the
10    form of scientific findings or other support products. Decisions in category (3) would benefit
11    from adaptive management approach. Ongoing research and assessments that are directly
12    decision relevant would thus be particularly useful for decisions that fall into this category.
13    Decision support products for decisions in categories (2) and (3) should be designed to be
14    compatible with any pre-existing decision frameworks.
15    This classification framework and the decision-driven approach will motivate future assessments
16    and other project activities that aim to build capacity to assess and respond to global change
17    impacts on water quality and aquatic ecosystems, but there is still much work to be done to
18    understand how best to implement such a decision-driven approach. It is our hope to learn more
19    about this approach through our future projects.

     August_2006_...                                                 _ _.External Review Report:. Watershed. Case.S.fypfes
 1    Literature Cited
 3    California Research Bureau. A Statistical Tour of California's Great Central Valley. CRB.  1997.
 4    Sacramento, CA.
 5    Denault, C., R.G. Millar, and BJ. Lence. 2006. Assessment of Possible Impacts of Climate
 6    Change in an Urban Catchment. Journal of the American Resources Association. 42:685-697.
 7    Dixon, M. D., J. C. Stromberg, J. T. Price et al. 2006. Potential Effects of Climate Change on the
 8    Upper San Pedro Riparian Ecosystem: Boon or Bane?
 9    Karl, T. R. and R. W. Knight. 1998. Secular trends of precipitation amount, frequency, and
10    intensity in the United States. Bulletin of the American Meteorological Society, 79:231-241.
11    Moglen, Glenn E., Palmer, Margaret A., Bockstael, Nancy E., Pizzuto, James E. Jointly
12    Changing Climate and Land Use in the Mid-Atlantic: Modeling Drivers and Consequences in
13    Economics, Hydrology, Geomorphology, and  Ecology. Final Report to U.S. EPA. 2006.
14    Nelson, K. C. and M. A. Palmer. 2006. Stream Temperature Surges under Urbanization and
15    Climate Change: Data, Models, and Responses. Journal of the American Water Resources
16    Association.
17    Nelson, Karen C., Palmer, Margaret A., Angermeier, Paul L., Pizzuto, James E., Moglen, Glenn
18    E., Dettinger, Michael, and Hayhoe, Katharine. 2006. Modeling the vulnerability of stream-fish
19    assemblages to urbanization and climate change. To submit to Ecological Modelling.
20    Odgaard, A. Jacob. 1989.  River-Meander Model: I. Development. Journal of Hydraulic
21    Engineering, 115(11): 1433-1450.
22    Peterson, S.  and Richmond, B., 1996. STELLA and STELLA Research Software: Technical
23    Documentation. High Performance Systems, Inc., Hanover.
24    Pizzuto, J. E., G. E. Moglen, M. A. Palmer, and K. C. Nelson. 2006. Two model scenarios
25    illustrating the effects of land use  and climate  change on gravel riverbeds of suburban Maryland,
26    U.S.A.
27    Price, Jeff T., Galbraith, Hector, Dixon, Mark  D., Stromberg, Juliet C., Root, Terry,
28    MacMykowski, Dena, Maddock, Tom, and Baird, Kate. Potential Impacts of Climate Change on
29    Ecological Resources and Biodiversity in the San Pedro Riparian National Conservation Area,
30    Arizona. Final Report to U.S. EPA. 2005.
31    Purkey, David R., Huber-Lee, Annette, Yates,  David, Hanemann, M.,  and Julius, Susan H.
32    Integrating a Climate Change Assessment Tool into Stakeholder-Driven Water Management
33    Decision-Making Processes in California. To appear in a special issue of Water Resources
34    Management, entitled WARM/ Advances in Global Change Research. 2006.
35    SRAG (Southwest Regional Assessment Group). 2000. Preparing for a changing climate: the
3 6   potential consequences of climate variability and change.
38    Srinivasan, R.S., Arnold, J.G., and Jones, C.A. 1998.Hydrologic modeling of the United States
39    with the soil and water assessment tool. Water Resources Development 14(3):315-325.

     August_2006_...                                                  _ _.External Review Report:. Watershed. Case.S.fypfes
 1   US EPA, GCRP. Problem Formulation Report for the Assessment of the Consequences of
 2   Global Change for Aquatic Ecosystems. 2003. Washington, DC, EPA\ORD\NCEA.
 3   Webler, T. and S. Tuler. 1999. Integrating Technical Analysis with Deliberation in Regional
 4   Watershed Management Planning: Applying the National Research Council Approach. Policy
 5   Studies Journal, 27:530-543.
 6   Wilby, R. L., Charlies, S. P., Zorita, E., Timbal, B., Whetton, P., and Mearns, L. O. Guidelines
 7   for use of climate scenarios developed from statistical downsacaling methods. 1-27. 2004.
 8   United Kingdom, Environment Agency of England and Wales.
 9   Yates, David, Sieber, Jack, Purkey, David, and Huber-Lee, Annette. 2005. WEAP21- A
10   Demand-, Priority-, and Preference-Driven Water Planning Model, Part I: Model Characteristics.
11   Water International 30(4):487-500.
12   Yates, David, Purkey, David, Galbraith, Hector, Huber-Lee, Annette, Sieber, Jack, and Joyce,
13   Brian. Global Change Impacts on Freshwater Ecosystem Services of the Sacramento Watershed,
14   California USA. 3-136. Final Report to U.S. EPA. 2006.