93085 BEYOND DOWNSCALING A Bottom-up Approach to Climate Adaptation for Water Resources Management COPYRIGHT © 2014 The World Bank 1818 H Street NW Washington DC 20433 Telephone: 202-473-1000 Internet: www.worldbank.org Disclaimer The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Rights and Permissions The material in this work is subject to copyright. Photos are copyright John Matthews. Because The World Bank encour- ages dissemination of its knowledge, the text in this work may be reproduced, in whole or in part, for noncommercial purposes as long as full attribution to this work is given. Any queries on rights and licenses, including subsidiary rights, should be ad- dressed to the Office of the Publisher, The World Bank, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2422; e- mail: pubrights@worldbank.org Attribution Please cite this work as follows: García, L.E., J.H. Matthews, D.J. Rodriguez, M. Wijnen, K.N. DiFrancesco, P. Ray. 2014. Be- yond Downscaling: A Bottom-Up Approach to Climate Adaptation for Water Resources Management. AGWA Report 01. Washington, DC: World Bank Group. i Moving beyond downscaling ii “In grappling with long-term climate change, it is natural to turn to climate modeling for guidance … The models, which are essential for elucidating the global climate system, have been informative in some applications related to agriculture or water development over large regions. But for many planning and design applications, especially when applied to smaller areas, to precipitation, and to extreme events, models often give too wide a dispersion of readings to provide useful guidance. A review of the application of these models…found that they are often used as a backdrop for urging the adoption of ‘no-regret’ actions, and rarely for quantitative decision-making on options.” Adapting to Climate Change: Assessing the World Bank Group Experience. Independent Evaluation Group (IEG)- World Bank/IFC/MIGA. Washington, DC, 2012. Overview, pp xxii-xxiii iii ACRONYMS AGWA - Alliance for Global Water Adaptation BCA - Benefit-cost analysis CF - Change factor CIDA - Climate Informed Decision Analysis CRiSTAL - Community-based Risk Screening Tool - Adaptation and Livelihoods DSS - Decision Support System GCM - Global Climate Model or General Circulation Model GHG - Greenhouse gases GWP - Global Water Partnership IEG - World Bank’s Independent Evaluation Group IGDT - Information-gap decision theory IISD - International Institute for Sustainable Development Intergovernmental Panel on Climate Change IWRM - Integrated Water Resources Management ODI - Overseas Development Institute OR - Operations Research RCM - Regional Climate Model RDM - Robust decision-making TTL - World Bank’s Task Team Leader UNDP - United Nations Development Program USACE - United States Army Corps of Engineers WPP - World Bank’s Water Partnership Program iv FOREWORD Most developing countries around the world will need Complex global circulation models (GCMs) have been to invest heavily in infrastructure in order to meet the developed to project future climate conditions under vari- needs of their people and of their economy more broadly. ous greenhouse gas (GHG) emission scenarios. The results Estimates of the global financing gap to meet sector de- are only applicable at the continental scale, as their resolu- mands for water supply and sanitation vary, but are on the tion is very coarse, and have to be further processed to ob- magnitude of $100 billion a year (McMahon, Rodriguez, tain information useful for decision-making. Researchers and Berg 2012). The challenge facing developing countries have developed methodologies that follow a similar proc- is therefore how to get the most benefit from their limited ess: create an ensemble of climate outputs from several investment budget—which requires careful planning, and GCMs, downscale the GCM outputs, bias-correct the out- infrastructure that is designed for the long term. puts based on climate observations in the area of interest, and use them as input for a calibrated hydrologic model to A water resources plan will include infrastructure, but assess climate change impacts on a given endpoint for guid- will also have many components of a more institutional ance in practical project and program planning and analy- and dynamic nature. Infrastructure will last a long time, sis. These methods, although useful for setting a global and and there is only a limited amount of flexibility once it has regional long-term context, have proven of little practical been built. The task of the water resources planner is to opti- use for site-specific water resources management and water mize the risk vs. cost trade-off under basic hydrologic, socio- infrastructure project design decisions at a local level, as re- economic, and environmental assumptions. Decisions on ported in 2012 by the World Bank’s Independent Evalua- water-related activities and projects have always faced tion Group (IEG). Moreover, sometimes time and resources many uncertainties, to which those caused by climate are disproportionately assigned to application of these so- change are now added. Climate change adaptation is seen called top-down methods in detriment of analyses of other as adjusting to evolving conditions, potentially far in the fu- non-climate related, uncertainties that might prove more im- ture. Some tools are needed and may be useful for non- portant in the short and medium terms. climate pressures; some may be better for adjusting to cur- rent climate conditions; and others for adjusting to climate While searching for solutions to this dilemma, it was change. found that similar concerns also existed in other water re- sources organizations—including universities, government iv agencies, and private entities, such as those represented in suggestions that perhaps the GCM outputs should supple- the Alliance for Global Water Adaptation (AGWA), of ment and inform the predictions from hydrologic models which the World Bank is a founding and active member. rather than drive the hydrologic models, seemed to garner Thus, benefiting from the knowledge base of AGWA’s mem- support from many participants.” ber organizations and with the participation of world re- As a result of these discussions and consultations, a set nowned experts, an internal-external workshop was organ- of guidelines and bottom-up approaches for including cli- ized by the World Bank Water Partnership Program mate uncertainty in water resources planning and project (WPP)/AGWA in November 2011. As a result, an AGWA design were identified by the Bank for its own purposes technical working group was formed to explore a number and by AGWA. The basic principles supporting these guide- of alternative methods for risk-based decision-making and lines have evolved into the framework presented in the last adaptation of vulnerable water systems, considering the ef- chapter of this book. They have also formed the basis for fect of uncertain information. The focus was on the so- the World Bank/WPP initiative to develop a practical, risk- called bottom-up approaches. The development of practical based bottom-up decision-making-aide instrument (called guidelines for practitioners, project coordinators and, in the the “decision tree”). This work is in progress and we expect case of the Bank, Task Team Leaders (TTLs), was deemed its results to provide an alternative approach contributing necessary. to improvement in the quality and effectiveness of water re- As a member of this working group, the World Bank/ sources management planning and project design under cli- WPP continued on this path and with the participation of mate variability and change uncertainty, which can be op- renowned external and Bank experts organized a session erationally used by practitioners and TTLs in World Bank (Special Session 3) at the HydroPredict2012 international projects at site-specific locations. conference held in Vienna, Austria, in September 2012, and William Rex a Learning Session at the Bank SDN Forum 2013 in March of this year. As discussed in the HydroPredict2012 confer- Acting Practice Manager, Global Programs, Water The World Bank ence, practitioners involved in managing flood risk, or de- veloping infrastructure for water supply or hydropower, seemed particularly uncomfortable with the uncertainty of climate predictions. As one HydroPredict2012 participant pointed out, “in the discussion following Special Session 3, v INTRODUCTION An AGWA-supported approach to sustainable water management This section introduces an adaptation approach supported Media by AGWA and addresses the following questions: Video i.1 - Hydrologic engineering perspective on adaptation approaches • Who should read this book? Video i.2 - An introduction to AGWA and the prac- • Is there a genuine need for an alternative approach to resilient tice of climate adaptation water resources management? • What’s in the rest of the book? vi Who should read this book? future hydrological conditions (Video i.1)? This book focuses on how we achieve water sustainabil- In many cases, these individuals work in technical disci- ity over long timescales—decades, even centuries from plines or are engaged in active resource management—engi- now. These timescales are important and relevant to our de- neers, planners, economists, conservationists, finance spe- cisions about planning, infrastructure, and institutions to- cialists, academics, and scientific and analytical staff. How- day. Many of the methods we use to manage water, directly ever, the challenges of climate change for water resources or indirectly, commit us to future decision pathways and re- are important at many organizational levels. The technical strict us from making other, alternative decisions. implications will be meaningless without a broad under- This book is designed for individuals who are explor- standing of climate issues by high-level decision-makers, ing the best means of incorporating climate adaptation per- policymakers, and professional communicators, as these are spectives into their water resources management work, par- the individuals who consume the reports produced by tech- ticularly if they are interested in mainstreaming that work nical staff, as well as interact with, set the strategy for, and in the context of other drivers that affect water supply and manage this technical staff. demand such as urbanization, Is there a genuine need for an alternative approach to demographic change, ecologi- Video i.1 Hydrologic engineer- resilient water resources management? cal shifts, and economic cycles. ing perspective on adaptation approaches The mainstream approach to water resources manage- It is designed for individuals ment used in most countries for well over a century as- who are asking questions such sumes that the statistical properties of past water history re- as: How do we estimate climate mained unchanged over time and did not follow any change impacts? Do the prob- trends. This is an assumption widely referred to in the scien- lems of climate adaptation pose tific and engineering literature as “stationarity,” which is of- new challenges for water man- ten interpreted to mean that the past is a good predictor for agers that require new the future. This assumption is understood to be a simplify- decision-making methods, or ing, practical approach to working with water data and—at will existing, traditional tech- Dr. Luis García (The World Bank) refers to hydrologic en- best—an approximation of the real world. niques be sufficient? Can we make sound operating rules for gineering in the context of The water cycle—that is, the cycling of water molecules adaptation approaches sup- addressing uncertainty about from ocean evaporation to precipitation, surface and ported by AGWA. vii groundwater flows, and return to the ocean—has proven to ment applications requiring high-confidence quantitative be both extremely sensitive to climate shifts and very diffi- estimates of future states. cult to predict. Climate prediction is particularly difficult What’s in the rest of the book? and uncertain over the long timescales of a decade or more, to which many water management decisions commit us. Across the first four chapters, this book describes the The manifestation of climate change in the water cycle is challenges of including climate change in water manage- complex. For instance, precipitation may shift in form (rain, ment decision-making and provides an overview of current snow, fog, sleet, and so on), intensity (hard, light), seasonal- practices in the adaptation field. After considering the pros ity (e.g., the timing of monsoons), frequency and magni- and cons of these practices, the book concludes with a tude of extreme events (floods, droughts, tropical cyclones), framework for an adaptation approach supported by and the degree of inter-annual variability. AGWA. The first chapter provides an overview of climate change, highlighting important concepts for water re- Because of this complexity, most recent approaches to sources managers. A selection of mechanisms for sustain- climate-adaptive water management have been based on able water development and ways to mainstream adapta- qualitative assessments (e.g., “this region will become dry- tion into these mechanisms are described in Chapter 2. er”)—characterized as “no-regrets” or “low-regret” actions Chapters 3 and 4 describe tools, methodologies, and frame- (i.e., actions that do not conflict with a wide range of future works available to water managers for climate risk assess- scenarios)—or based on quantitative outputs from climate ment and climate risk management, respectively. Chapter 3 model projections (e.g., derived from GCMs, applied in a hones in on the bottom-up risk assessment approaches sup- top-down framework to define climate impacts based on a ported by AGWA, i.e., those approaches that begin by as- narrow range of climate variables). Approaches based on sessing system vulnerabilities to variations in climate and climate models attempt to provide quantitative projections only consider GCMs in later assessment stages. Chapter 4 of future climate states; however, climate models were not focuses on tools for identifying robust management actions developed for climate adaptation purposes and are associ- once climate risks have been assessed in relation to other ated with levels of uncertainty that—used in isolation or challenges water managers face. Lastly, the information pro- without careful qualification—are unacceptable as stand- vided in this book is synthesized in Chapter 5 by framing a alone sources of information for most water resources man- theoretical approach supported by AGWA to include adap- agement applications. This is particularly true of manage- tation in water resources management, planning, and invest- ment (Video i.2). Ongoing and future projects seek to opera- viii tionalize this framework Video i.2 Introduction to through case study applications AGWA and the practice of cli- of the adaptation approach sup- mate adaptation ported by AGWA. Our hope is that this book will spark a sophisticated dia- logue about how to make sys- tematic, credible, and quantita- tive decisions about sustainable water management that can be For more information, please used by a wide variety of audi- visit the AGWA site ences. We also hope that indi- (http://alliance4water.org) viduals engaged in the plan- ning, management, design, finance, economics, evaluation, and operations of water resources—directly or indirect- ly—will find this publication helpful. John H. Matthews, Coordinator, AGWA Diego Rodriguez, The World Bank ix CHAPTER 1 Understanding climate change Chapter 1 provides an overview of climate change and its Media impacts on water resources management by addressing the Video 1.1 - History of climatic changes on earth following questions: Video 1.2 - Climate variability and extremes • What is climate and how is it changing? • What does hydrologic non-stationarity mean for technical water resources management? • What does sustainability mean in the context of a non-stationary climate? 1 What is climate and how is it changing? stronger link between observed warming over the last 50 The earth’s climate has changed in the past and will years and human-source GHG concentrations. Recent IPCC continue to change into the future. “Climate” refers to how communication (Stocker, Dahe, and Plattner 2013) states the atmosphere behaves over relatively long time periods that human influence on the climate system is clear and (decades to thousands of years), as opposed to “weather,” that it has been the dominant driver of shifts in the global which describes atmospheric conditions over short periods climate system since 1950. Over the past 150 years, GHG of time (hourly to annually). In the last 700,000 years, gla- levels have increased by 40 percent, mainly from the burn- cial periods have occurred about every 100,000 years, and ing of fossil fuels (US Department of Commerce 2013). the earth has experienced both colder and warmer periods Climate models and recent observations indicate both than today (Video 1.1). The historical record of changes re- changes in mean climate as well as increases in climate vari- veals that our climate is highly sensitive to relatively small ability (e.g.,Dai, Trenberth, and Karl 1998; Hulme, Osborn, changes in heat retention and atmospheric circulation, such and Johns 1998; Lettenmaier et al. 1999; Lins and Slack as those caused by shifts in human- and natural-source 2005; Jones et al. 1998). Mean climate refers to broad gener- greenhouse gas (GHG) concen- alizations about regional climate, such as the total annual trations, e.g., carbon dioxide Video 1.1 History of climatic changes on earth precipitation and the mean annual temperature. Variability and methane. includes intra-annual variability—seasonal patterns and If climate change is in it- shifts—and inter-annual variability, the degree to which self not new, then what is novel one year can be characterized as climatologically similar to about the current changes the other years. earth is undergoing? In recent Water resources will likely be the principal medium by years, we’ve increasingly come which these climate change impacts are felt and mitigated to realize that the pace of mod- (UN Water, 2010). Indications of hydrologic change renew ern climatic changes has been Dr. Eugene Stakhiv (US attention for the assumption of “non-stationarity” and call accelerated by human actions. Army Corps of Engineers) into question whether the statistics of the historical record Each subsequent report by the discusses climatic changes are an accurate and useful descriptor for the future. For United Nation’s Intergovern- that occurred historically on earth, using a case study most water managers, the main concern lies with changes mental Panel on Climate from the Great Lakes Region, in variability and extremes (Video 1.2). A warmer climate Change (IPCC) recognizes a US-Canada border. 2 could increase the risk of floods Video 1.2 Climate variability changes by the end of the 21st century, either model uncer- and droughts of greater magni- and extremes tainty or uncertainties associated with emissions scenarios tude, duration, and frequency used becomes dominant, depending on the extreme. with respect to recent observa- Given the uncertainties in projected climate, it is important tions (IPCC 2007). In most re- to consider climate in the context of other trends affecting a gions, changes in extremes are water system’s performance, particularly in the shorter occurring more rapidly than term, such as shifts in demography, land use, economic pat- changes in long-term, average terns, and urbanization. patterns of drying or wetting. Observations since the 1950s Dr. Juan Valdéz (University What does hydrologic non-stationarity mean for techni- show changes in the hydrologic of Arizona) and Dr. Kenneth cal water resources management? extremes in all critical variable Strzepek (University of Colo- The implications of non-stationarity have received rado) discuss the importance dimensions: intensity, fre- much attention since the publication of a critique of climate of climate variability and ex- quency, spatial extent, duration, change–neutral approaches to water management by Milly tremes for water resources timing, and probability distribu- management and how these and colleagues (2008). The probability distributions of sta- tion functions (IPCC 2012). may change in the future. tionary hydrologic processes do not change with time, i.e., the mean and the variance are constant in the long term. Unfortunately, projections regarding climate extremes However, even stationary variables may show regular natu- remain highly uncertain, vary by region, and may be over- ral variability and periodic oscillations (Kundzewicz 2011). shadowed by natural variability, at least in the short term. In contrast, the statistical properties of non-stationary proc- According to IPCC (2012), esses vary over time. For instance, droughts may become Projected changes in climate extremes under different emis- more (or less) severe or frequent, or the mean timing of sea- sions scenarios generally do not strongly diverge in the com- sonal monsoons may advance (or retreat). ing two to three decades, but these signals are relatively Stationarity has generally been interpreted as the rule small compared to natural climate variability over this time that the past is a guide to the future, which has shaped frame. Even the sign of projected changes (i.e., trends to- most water resources planning, operations, and manage- wards greater or lesser precipitation) in some climate ex- ment in the modern era. While it has long been known that tremes over this time frame is uncertain. For projected the assumption of stationarity is not correct, it was believed 3 to be a reasonable, simplifying approximation until re- Over the past three decades, hydrologists and water re- cently. The assumption that streamflow was a stationary sources specialists have been concerned with the issue of process facilitated the generation of plausible “future” se- non-stationarity arising from several factors. First is the ef- quences of stochastic inputs. If non-stationarity is viewed fect of human intervention on the landscape that may cause as a deterministic component of a time series, then no con- changes in the precipitation–runoff relationships at various ceptual difficulty is introduced in dealing with non- temporal and spatial scales, such as deforestation and urbani- stationary inputs (Matalas 2012). Furthermore, several ap- zation. Second is the occurrence of natural events such as proaches have been proposed in the literature to address volcanic explosions or forest fires that may cause changes in non-stationarity, as reported by Salas et al. (2012), including the composition of the air, thenonst soil surface, and geomor- stochastic approaches to simulate, for example, monthly phology. Third is the low-frequency component of oceanic–at- and yearly hydrologic processes such as streamflows (e.g., mospheric phenomena that may have significant effects on for drought studies and designing reservoirs). the variability of hydrological processes such as annual run- off, peak flows, and droughts. Fourth is global warming, Matalas (2012) also stressed the need to qualify the “as- which may cause changes to oceanic and atmospheric proc- sertion that the past was stationary, that the present is not esses, thereby affecting the hydrological cycle at various tem- stationary, and that the future will never be stationary.” Ac- poral and spatial scales. cording to Matalas (2012), since at least 1938 it has been known that time series are composed of deterministic and Although it is essential that water managers recognize the stochastic components, evolving to the present view that hydrologic non-stationarity associated with climate change, these components include a trend, fluctuations about the this must be considered in the context of other sources of trend, seasonal movement, and irregular or random move- non-stationarity and the specific decision context. ment. Time series are not simply stationary or non- What does sustainability mean in the context of a non- stationary; they may be stationary in some components and stationary climate? non-stationary in others. The implications of a non-stationary climate for While there has been much discussion and even attri- decision-makers are both significant and subtle. If we have bution of non-stationarity solely to the effects of climate overestimated our ability to reliably and predictably plan change, climate change is only one of the possible causes of for the future, then we face a serious crisis in how we make non-stationarity. According to Salas et al. (2012): water management decisions for energy, water supply and sanitation, natural resource management, navigation and 4 transportation, and the myriad other uses to which we put too soon or too late. In effect, sustainable water manage- water. We are in effect making long-term decisions based on ment is a pathway of decisions, some of which may reverse short-term information and potentially limiting our options or contradict previous decisions. and economies for the future. In some cases, we may need However, water managers and decision-makers are not to make such potentially regrettable decisions. However, simply investors: they are also resource managers of this text is meant to show that there are pathways that can aquatic ecosystems and the broader eco-hydrological land- avoid or reduce those regrets. scape. “Sustainability” presents significant difficulties here Perhaps the most difficult implication of non- as well, since all species and even areas such as groundwa- stationarity is what sustainability itself means. In most ter recharge zones respond dynamically and in complex, of- cases, water managers and decision-makers have viewed ten unpredictable ways, to shifts in climate (Matthews and “sustainable” water resources management as a fixed tar- Wickel 2009). get, as if there were a single optimal balance point for alloca- Ultimately, a sustainable vision of water resources man- tion, water infrastructure design, governance, and opera- agement must encompass both ecological and engineering tions. By recognizing that the water cycle can undergo sig- perspectives of non-stationary change (Matthews et al. nificant shifts over relatively short timescales, we have to 2011). While this topic has not been widely addressed by transition to a view of “sustainabilities,” with multiple and the conservation or water management communities, cli- evolving (“unfixed”) targets. For instance, the Murray- mate change may provide an opening to conjoin these per- Darling basin in Australia and the lower Colorado River ba- spectives into a more coherent whole—which will be cov- sin in North America may be seeing long-term declines in ered more extensively in another publication. precipitation patterns. Are these decadal droughts or part of a relatively permanent shift in climate? In response, should resilience mean “bouncing back” following extreme events, or maintaining systemic integrity as environmental and economic conditions undergo major transformations? As investors, our ability to see the future is limited, so our challenge now may be closer to the latter model: avoid- ing decisions that commit us to too much or too little—or 5 CHAPTER 2 Mainstreaming adaptation into water resources management Chapter 2 discusses how adaptation can be mainstreamed Media into water resources management through existing sustain- Video 2.1 - Uncertainty, confidence, and able management mechanisms, including those that ad- consequences in water resources management dress: Video 2.2 - IWRM under climate change • Uncertainty and confidence in planning information, Video 2.3 - Environmental flows under climate • Investment under uncertainty, change • Integrated water resources management (IWRM), and • Natural and environmental flow regimes. 6 Uncertainty and confidence in planning information current state of climate modeling. While some standard hy- drological practices, based on assumptions of a stationary Water resources engineers and decision-makers have climate and variability, can be extended to accommodate as- long dealt with uncertainty and variability. Standard engi- pects of climate uncertainty, new issues and approaches neering practices account for uncertainties through risk must be considered (Stakhiv 2010). Some of these are dis- management techniques; design redundancy; and adding cussed below. safety factors to deal with the unknowns (e.g., by adding a levee “freeboard” onto a “standard project flood” to accom- Investment under uncertainty modate the uncertainties associated with historic climate Climate change poses significant challenges for invest- variability). However, water managers face greater magni- ments in the water sector, particularly due to the long life- tudes of future uncertainty than historically experienced. span and large upfront costs associated with many water Climate change projections are highly uncertain—in fact, projects. While water practitioners have long contended the unknowns about climate change dynamics go beyond with variability and uncertainty in hydrology, the circum- our understanding of classical Video 2.1 Uncertainty, confi- stances surrounding climate change and the inability to de- risk and uncertainty analysis. dence, and consequences in wa- rive probabilities of future scenarios require a major shift in This requires new perspectives ter resources management thinking, planning, and designing water investments of the on risk and uncertainty analy- future (Qaddumi et al. 2009). At the same time, adaptation sis (Video 2.1). As stated by to climate change must continue to build on conventional Kundzewicz (2011), “We know interventions while also addressing immediate challenges increasingly well that we do and needs, such as disaster management, ecological restora- not know enough.” tion, and poverty alleviation. Investment decisions must There is currently no con- place the climate change dimension in the context of other sensus on how best to ap- factors, such as population growth, land management, and proach the planning and hydro- Dr. Guillermo Mendoza (US economic markets, which in some cases may be far more sig- logic design of water resources Army Corps of Engineers) nificant and critical than that of climate change in some discusses the relationship be- projects under climate uncer- parts of the world (Qaddumi et al. 2009). tween confidence and conse- tainty. Unlike uncertainties of quence and how these may The discount rate used in economic models to assess ad- the past, climate uncertainties shape decisions regarding ad- aptation costs and benefits plays a key role in determining cannot be estimated with the aptation approaches. 7 the trade-offs between the benefits and costs of actions tion for low-impact adaptation is the adjustment of water taken now versus actions taken later. Put simply, discount system operating rules developed for existing and planned rates help determine how much people are willing to pay water infrastructure, especially relevant to reservoirs (e.g. today for benefits accrued in the future. Due to the long life- Ghosh et al. 2010, Raje and Mujumdar 2010). span of water infrastructure (typically 30-50 years, although Integrated Water Resources Management (IWRM) under in some cases it may reach 100-200 years), applying the 12 climate change percent discount rate generally used by multi-lateral devel- opment banks makes it difficult to justify spending much IWRM provides a useful framework for integrating cli- money on climate adaptation today. Thus, some econo- mate adaptation into water resources management. The mists, notably Nicolas Stern in the Stern Review (2007), advo- goal of IWRM is the sustainable management and develop- cate the use of significantly lower discount rates when as- ment of water resources, taking into account social, eco- sessing climate policy, or discount rates that decline nomic, and environmental inter- throughout project lifespans (Groom et al. 2005; Gollier, ests. It recognizes the interde- Video 2.2 Integrated water re- Koundouri, and Pantelidis 2008). The choice of the discount pendence of many different sources management (IWRM) rate can strongly influence the perceived economic value of competing interest groups and under climate change a project. This choice remains controversial and no consen- considers the effects of each use sus exists on what discount rate to use when assessing cli- on the others when making wa- mate policies or adaptation projects. ter allocations and manage- ment decisions (GWP 2009). As The dominant effect of the discount rate on future wa- a framework, IWRM can di- ter system benefits and costs can be partially mitigated by rectly assist communities in reducing the economic lifetime of current projects. One way coping with climate change of this is by adaptive management, which adds the flexibil- and climate variability, espe- Dr. Torkil Jønck Clausen ity to upgrade current, less expensive investments in the fu- cially since good management (DHI Water Policy), one of ture. Upgrades occur only if new information confirms the of systems allows the right in- the most influential people in need for the extra investment and reduces the factors of centives to be passed on to wa- the Global Water Partner- safety required. “Real options theory” is an example of a ter users (Cap Net 2009, Video ship's embracement of the systematic procedure for incorporating flexibility into pro- IWRM concept, discusses 2.2). Although IWRM is not a ject designs, further discussed in Chapter 4. A second op- IWRM in relation to climate new idea, many questions re- adaptation. 8 main regarding IWRM implementation and success in prac- following chapter. In Video 2.3, Video 2.3 Environmental flows tice. Dr. LeRoy Poff discusses natu- under climate change ral and environmental flows in Natural and environmental flow regimes: Towards a relation to climate and climate management model of eco-engineering change. For freshwater ecosystems, climate change is expected to have its most profound effects through changes in the his- One of the biggest chal- torical natural flow regime (Le Quesne et al. 2010). The flow lenges for natural resource man- regime is a primary determinant of freshwater ecosystem agers of freshwater ecosystems structure and processes—the so-called “master variable” is how we define success and Dr. LeRoy Poff (Colorado (Poff et al. 1997). Changes in the volume and timing of fresh- set management targets. Histori- State University) discusses water flows are already a leading driver of global declines cally, these targets have been how climate shapes flow re- in freshwater biodiversity from abstractions and infrastruc- set based on a past reference gimes; how both may change state, which was presumably a in the future; and what we ture; the impacts of climate change are accelerating this healthier or more intact system. can do to maintain freshwa- pressure. Ongoing changes in precipitation and evapotran- ter ecological integrity. spiration regimes are altering many aspects of water quality Even the definition of environ- and quantity, while some of the most important long-term mental allocations globally has ecological impacts now come from impacts on water timing been based on historical conditions in most cases. However, (Le Quesne et al. 2010). while the human drivers of the current period of climate change are “new” and unprecedented, climate change itself The World Bank report Flowing Forward points out the is not new for ecosystems and species. Paleoecological stud- existence of opportunities to undertake assessments of vul- ies show dramatic shifts, transformations, and reassembling nerability to climate change in a range of planning activities of ecosystems and ecological processes. Thus, a manage- and operations. Flowing Forward acknowledges the consider- ment system based solely on a past reference state may be able uncertainty about ecosystem impacts of climate ineffective, even counterproductive, during periods of change. In view of this, it recommends risk-based sensitiv- rapid climate shifts. Can we manage ecosystems in a way ity and vulnerability assessments for freshwater ecosys- that allows them to maintain integrity and auto-adapt as tems, rather than impact assessments (Le Quesne et al. much as possible? 2010), similar to the bottom-up approaches described in the 9 Traditionally, freshwater ecosystem management deci- sions have been after and in response to water infrastructure and management decisions. As suggested in Chapter 1, however, we may be able to use shifts in risk assessment as an opportunity to better integrate these perspectives through “eco-engineering” systems that include more flexi- ble environmental allocations—which can better balance op- erational, user, and environmental allocations—and that link dynamic ecological and engineering performance mark- ers. 10 CHAPTER 3 Key tools supporting climate risk assessment Chapter 3 provides an overview of the available tools, frame- Media works, and methodologies to support climate change risk Table 3.1 - Classification of adaptation tools assessment in the water sector, by addressing the questions: Video 3.1 - What are climate adaptation tools? • What are climate risk assessment tools? Video 3.2 - Intended use of models • Screening tools: how relevant is climate change to my project? Video 3.3 - Climate uncertainty • Data tools: what data is available and how could it be used? Video 3.4 - Top-down versus bottom-up • Impact assessment tools: how do top-down and bottom-up climate Figure 3.1 - The cascade of uncertainty impact assessments differ? Figure 3.2 - Top-down versus bottom-up 11 What are climate risk assessment tools? considerations in tool selection include: The Overseas Development Institute (ODI) has identi- • What problem is the tool intended to address? fied over 100 climate change adaptation tools relevant to the water sector, defining these tools as “documents, com- • Is a tool needed at all? puter programs or websites that clearly and thoroughly op- • What relevance does it hold for the particular context, lan- erationalize a set of principles or practices that could build guage, users, region? the resilience of water services to current climate variability • What type of tool(s) is(are) needed? or future climate impacts, preferably in an engaging and • How complex is the tool? (What is the local capacity to use user-friendly manner” (Doczi 2013) (Video 3.1). These tools the tool? Will training be needed?) can be classified by function into three major types: process guidance tools, data and information tools, and knowledge • What is the price? sharing tools. Also, tools can be classified by target users • Is user support available? (Is the tool kept up-to-date?) and sector, such as tools for general climate adaptation, best This chapter focuses on tools that support climate risk management practices, or specifically for water sector cli- assessment, namely, climate screening tools, data tools, and mate adaptation (Table 3.1). risk assessment frameworks. There is a natural progression Considering the wide Video 3.1 What are climate ad- to questions regarding how to assess climate risk and start range and variety of available aptation tools? the adaptation process. Whereas traditional approaches be- climate adaptation tools, choos- gan by asking how the climate conditions in the future ing the most appropriate tool(s) would differ from the past, we propose a more strategic requires identification of the us- starting point for climate change adaptation: an exploration ers’ specific adaptation needs of the vulnerabilities of the water system to changes in his- and concerns at the outset. Us- toric climate conditions. Once the conditions to which the ing the wrong tools may lead to water system is vulnerable have been identified, questions inappropriate adaptation ac- on the likelihood of those conditions arising can be ad- tions, which could actually in- Dr. Julian Doczi (Overseas dressed in a more efficient, targeted manner. Methods for crease system vulnerability to Development Institute, progressing through these questions are referred to as cli- climate change, termed mal- ODI) discusses climate risk mate risk assessment tools. Beyond simply identifying management tools for the wa- climate-related risks, climate risk assessment tools place adaptations. Some important ter sector. 12 Table 3.1 Classification of climate adaptation tools based on function climate-related vulnerabilities in the context of all other vul- (Doczi 2013). nerabilities facing a project. The two broad methods for assessing the effects of cli- mate change on water resources use data tools differently and begin the impact and vulnerability assessment from dif- ferent directions. Traditional methods to assess climate risk and vulnerability take a top-down approach, by downscal- ing a necessarily limited selection of individual projections from GCMs to identify snapshots of potential climate im- pacts. The water system’s vulnerability to those particular scenarios is then assessed by forcing hydrological and wa- ter systems models with each scenario’s climate informa- tion. Bottom-up approaches reverse this assessment process by first identifying system vulnerabilities to a very wide range of future climates (beyond that projected by GCMs) and then determining the plausibility of particular climate impacts using the best available and most credible climate information. This chapter begins by building the conceptual and sci- entific basis for bottom-up approaches, and then describes methods for bottom-up climate change risk assessment, us- ing a method called decision scaling as an example. Once climate-related vulnerabilities have been quantified, many other questions related to adaptation can be addressed, such as: “What to do?” “When to start action?” “How fast to proceed?” “How to incorporate updating mechanisms?” and “What are the costs of action, compared to the costs of 13 inaction?” Kundzewicz (2011). Tools for answering these ample, tools such as The Nature Conservancy’s Climate questions can be categorized as climate risk management Wizard are helpful for quickly identifying possible broad tools, discussed in Chapter 4. trends in temperature and precipitation relevant to the loca- tion of the planned water project. Similarly, The World Screening tools: how relevant is climate change to my Bank’s Climate Change Knowledge Portal, and Climate and project? Disaster Risk Screening Tools (available January 2015) can A subset of climate risk assessment tools is climate identify potential changes in climate conditions that may screening tools. Comprehensive climate risk assessment is po- affect projects and communities, while also allowing consid- tentially expensive and laborious. Screening tools classify eration of climate change relative to other types of risk. The water systems and water projects according to broad catego- United Nation’s Development Programme’s (UNDP’s) ries of climate sensitivity. For example, in the short term Adaptation Learning Mechanism offers geographically- (<20 years out), natural, internal climate variability is likely targeted resources for climate change adaptation, including to dominate uncertainties in the climate parameters of rele- overviews of current adaptation practices and needs. The vance to water resources system planning (Deser et al. 2012, International Institute for Sustainable Development’s (IISD) Lownsbery 2014). Thus, water projects with economic life- Community-based Risk Screening Tool (CRiSTAL), among times of less than 20 years are not likely to be sensitive to others, involves community members, planners, and man- climatic changes in that timeframe. In order to efficiently agers in the process of determining the relevance of climate use financial, computational, and human resources, it is im- change to a specific project. portant that climate risk assessment tools allocate effort in a way that is consistent with the potential sensitivity to cli- In cases in which the screening process indicates that mate risk. climate sensitivity poses a significant threat to project per- formance, further assessment of climate risks is warranted. Screening tools are important for project managers in The following sections describe the data and information the early stages of project development and climate risk as- tools available for such an assessment. sessment. Climate change is not relevant to all water re- source management work, and other factors may play a Data tools: what data is available and how could it be larger role in prescribing and designing management ac- used? tions. Screening tools allow for a quick assessment of the Data tools provide the information necessary for climate sensitivity of a system or project to climate change. For ex- risk assessment. These include historical water system per- 14 formance, hydrological statistics based on the historical re- focus and believed to capture some dynamics better, cord, paleodata, stakeholder experience, and projections of though this has recently been called into question (Kerr future changes in hydrology, demographics, economics, 2013). Further, the computational burden and resource re- technology, and land use. Traditionally, too much emphasis quirements associated with RCM may overshadow the has been placed on the application of climate models, the value they added (Kerr 2013). Both GCMs and RCMs at- generation of emissions scenarios, and the translation of tempt to represent climate dynamics and how the global cli- those scenarios to long-range projections of climate pat- mate system may respond to changes in external forcings, terns. While these models and scenarios provide insight particularly elevated GHG concentrations. Therefore, they into the scale and character of epistemic climate uncer- may be useful in national- and regional-scale vulnerability tainty, climate projections continue to underperform in assessments (IEG 2012). However, there is growing consen- terms of the information needed by water resources engi- sus that climate models are ill-equipped to support robust neers, operators, and managers for effective adaptation water resource management decisions (e.g., see the AGWA (Brown and Wilby 2012; Mendoza and Gilroy 2012). white paper Caveat Adaptor). Global Circulation Models Video 3.2 Intended use of mod- The problems in using (GCMs) els GCMs to model the likelihood Video 3.3 Climate uncertainty GCMs have emerged as the of future hydrologic events standard tool for projecting fu- stem from the cascade of uncer- ture climate conditions. How- tainty propagated through cli- ever, the use of GCMs for as- mate projections and downscal- sessing the implications of cli- ing processes (Wilby and Des- mate change for water re- sai 2010) (Video 3.3 and Figure sources needs to be carefully 3.1). Foremost, future GHG considered in the context of the emissions are uncertain since Dr. Eugene Stakhiv (US Dr. Robert Wilby (Loughbor- decision at hand since climate Army Corps of Engineers) they depend on hard-to-predict ough University) discusses models were not designed for provides a perspective on the future human behavior. Sec- the cascade of uncertainty adaptation planning (Video 3. use of “fancy models” and ond, GCMs contain uncertain- that occurs during downscal- 2). Regional Circulation Models Dr. Robert Wilby (Loughbor- ties in model parameters and ing along with distinguish- (RCMs) are narrower in spatial ough University) discuss the structure (Stainforth et al. ing between the different value and limits of GCMs. sources of uncertainty. 15 Figure 3.1 The cascade of climate change uncertainty (adopted from runoff estimates (Fekete et al. 2004). Water resource manag- Wilby and Dessai 2010). ers are primarily concerned with planning and design at the local and regional scale, yet precipitation (and to a lesser extent temperature) output from GCMs is only con- sidered spatially credible at coarse resolution grid cells (100s of km) and temporally credible at a monthly time step. Perhaps the most critical weakness of climate projec- tions is that they are less reliable in regard to the variables that are most important for water resources projects, such as hydrologic extremes (e.g., flood and drought). Those ex- treme events are located at the tails of distributions of cli- mate variables and percentage-wise will change more rap- idly than the mean in a changing climate (Dai et al. 1998). As a final complication, imperfect hydrologic models may take GCM climate parameters as input, translating climate 2007). Third, natural climate variability adds uncertainty to variables into water resources variables, and adding an- projections, since it continues to be unpredictable at lead other level of uncertainty to the cascade. times longer than a few seasons. Finally, there is uncer- tainty in the underlying science, since all the complex inter- Scientists employ several techniques to try to overcome actions of the Earth system are not well understood. Even a some of the limitations inherent in GCMs. The Coupled consensus between a broad set of models and scenarios is Model Inter-comparison Project (CMIP) establishes stan- no guarantee that the future will in fact mirror projected dard experimental protocols for studying GCM output, ena- outcomes. bling scientists to analyze GCMs in a systematic fashion. Further, CMIP supports model diagnosis, validation, inter- More specifically, climate projections provide limited comparison, documentation, and data access. The IPCC’s and often biased explorations of the effects of natural cli- most recent (Fifth) Assessment Report uses the CMIP Phase mate variability, especially precipitation variability (Rocheta et al. 2014), with amplified carry-over effects for 16 5 (CMIP5) framework for coordinated climate change ex- to provide probabilistic representations of the uncertainty periments. itself (Hall 2007). Because risk is a function of both probabil- ity and impact (Dessai and Hulme 2004), the inability of cli- To overcome the limitations of GCM resolution, scien- mate projections to probabilistically represent uncertainty is tists apply downscaling methods to construct climate infor- a substantial obstruction to assessing and mitigating mation at the higher resolutions needed for water resources climate-related risks to proposed water projects if used in a management. For example, a new archive of downscaled conventional “predict-then-act” framework. Impact model CMIP5 climate projections is being developed at a spatial structures and parameters also contribute significant uncer- resolution of approximately 800 meters for the coterminous tainty to the overall cascade (e.g., Dobler et al. 2012; Wilby United States (Thrasher et al. 2013). While applying down- and Harris 2006). In practice, therefore, there are insuffi- scaling techniques can produce higher-resolution regional cient resources to explore exhaustively all components in and local projections, they will not correct for large-scale er- the uncertainty cascade so the inferred uncertainty range is rors in GCMs (Barsugli et al. 2009, Olsen and Gilroy 2012). almost certainly an underestimate of the true range. New generations of GCMs, RCMs, and downscaling tech- niques all have the potential to better characterize uncer- Historical record and weather generators tainty; however, these new models and techniques will by If GCM output is unreliable, how can projections of fu- no means eliminate uncertainty, and instead may even in- ture climate be generated? One way is to perturb the histori- crease uncertainty in future climate projections (Roe and cal climate record in a manner that is consistent with the Baker 2007; Knutti and Sedláček 2013). best current understanding of climate change effects on the statistical properties of the historical climate signal (e.g., While GCM-based climate change projections may indi- mean, low-frequency variability, duration, autocorrelation, cate a range of possible challenges for water systems, they etc.). do not typically reduce the uncertainty of future climate relevant for water systems planning; climate projections are Weather generators are computer algorithms capable of in fact unlikely to describe the limits of the range of possi- producing long series of synthetic daily weather data. The ble climatic changes. As a result, climate model-based pro- parameters of the model are conditioned on existing mete- jections may have difficulty providing managers or orological records to ensure that the characteristics of his- decision-makers with the climate-related information they toric weather emerge in the daily stochastic process. require (Kundzewicz and Stakhiv 2010). Nor are they able Weather generators are a common tool for extending mete- 17 orological records (Richardson 1985), supplementing sets (multi-decadal to multi-centennial), to understand the weather data in a region of data scarcity (Hutchinson 1995), risks to which the water system could be exposed. These disaggregating seasonal hydroclimatic forecasts (Wilks are based on observations from the history of a given loca- 2002), and downscaling coarse, long-term climate projec- tion and not on contentious projections of what “unprece- tions to fine-resolution, daily weather for impact studies dented” conditions might arise in the future. This is impor- (Kilsby et al. 2007, Wilks 1992). A major benefit afforded by tant, particularly if the period of modern engineering prac- most weather generators is their utility in performing cli- tice has coincided with a relatively benign epoch (as in the mate sensitivity analyses (Wilks and Wilby 1999). Several western United States). Adapting to past low-frequency studies have used weather generators to systematically test variations in water resources presents significant challenges the climate sensitivity of impact models, particularly in the even before considering the additional risks posed by an- agricultural sector (e.g., Confalonieri 2012, Dubrovsky et al. thropogenic forcing. 2000). These sensitivity studies systematically change pa- Risks associated with a non-stationary climate have rameters in the model to produce new sequences of been presented as deviations from observations of the past weather variables (e.g., precipitation) that exhibit a wide 100 years or so of record; however, natural climate cycles re- range of change in their characteristics (e.g., average sulting in extreme flood and drought that repeat on periods amount, frequency, intensity, duration). The permutations greater than a single century are likely to provide much bet- created by the weather generator are not dependent on any ter information regarding the risks faced in the economic climate projections, allowing for a wide range of possible lifetime of long-lived water infrastructure such as dams. future climates to be generated while avoiding biases propa- gated from the projections. However, the particular permu- Unfortunately, paleodata are often only available for tations generated can be informed by available projections specific variables and at coarse temporal resolution (annual to ensure that they more than encompass the range of GCM or decadal). Improvements in the development and process- projections. ing of paleodata hold potential to greatly improve our un- Paleodata derstanding of natural climate variability, and the longer- term risks facing our water systems. While the use of paleodata has traditionally received little attention in risk estimation, paleodata are becoming Local expertise more important to inform expectations for future climate Along with recorded data from hydrometeorological scenarios. Paleodata allow for the use of very extended data stations, much information can be garnered from the experi- 18 ences of local people regarding the occurrence and impacts Figure 3.2 Top-down versus bottom-up risk assessment of historical weather patterns and extreme events. Farmers, Traditional Approach Decision Scaling for example, have strong institutional memory regarding 1. Downscale a few floods, droughts, precipitation patterns, seasonal transi- climate model projections 3. Determine climate risks to project performance tions, and planting and harvesting times. Those living near water bodies tend to retain stories regarding floods or surges and to observe gradual changes in water levels. Lo- 2. Generate a few water cal newspapers are often good sources of information on supply series http://www.cccsn.ec.gc.ca/?page=downscaling noteworthy historical climate events. These observations 3. Determine whether system performance is acceptable for these series. 2. Map climate domain onto vulnerability domain and anecdotal information are a very valuable supplement to the hydrometeorological record. Impact assessment tools: How do top-down and bottom-up climate impact assessments differ? 1. Determine the vulnerability domain Top-down climate assessments Figure 3.2 and Video 3.4 compare the traditional top- A top-down framework can help quantify the relative down approach for climate change risk assessment with de- contribution of different components to overall uncertainty cision scaling, which is a particular example of a bottom-up for extremes such as low flows (e.g., Wilby and Harris, approach. Top-down approaches begin by downscaling a 2006). Moreover, very high resolution RCMs are now being few climate model predictions (from low-resolution GCM) used to investigate the sensitivity of extreme precipitation and run the downscaled climate projections through vari- to temperature forcing (e.g., Kendon et al., 2014). In other ous models to develop expectations for changes in hydrol- words, climate models and downscaling methods can be ogy, vegetation, social systems, etc. Those few selected sce- usefully deployed to enhance understanding of the physi- narios (shown as GCM-informed point estimates on the 3rd cal processes or critical thresholds that drive hydrological level of the traditional analysis in Figure 3.2) are then evalu- extremes. ated for their effect on the expected net benefits of the pro- Limitations of top-down assessments ject under evaluation. Top-down climate assessments rely heavily on GCM outputs for describing local and regional climate impacts. 19 Most top-down approaches be- Video 3.4 Top-down versus bot- tial role of likelihood concepts in risk assessment (where gin with a small selection of fu- tom up climate assessment risk is a function of impact and probability of that impact), ture scenarios from GCM out- top-down methods tend not to provide the insights needed put, which themselves, even if for water resources system planning. taking all GCM output, repre- Bottom-up climate assessments: Decision scaling and sent only a subset of all possible other methods climate futures. As a result, top- In contrast to top-down approaches, bottom-up climate down methods do not sample assessments begin in the vulnerability domain. They take from the full range of climate important system characteristics and local capacities into ac- futures. And, as described ear- Dr. Casey Brown (Univer- sity of Massachusetts- count before the sensitivity and robustness of possible adap- lier, the process of downscaling tation options are tested against climate projections, such as GCMs results in a cascade of un- Amherst) compares tradi- tional top-down approaches GCM outputs. Bottom-up approaches account for particular certainty. Further, all models with the bottom-up Decision intrinsic system characteristics such as exposure, sensitivity, have similar resolution and Scaling approach. and adaptive capacity as important elements for describing must parameterize the same risk (Bouwer 2013). This is in contrast to top-down ap- processes (Tebaldi and Knutti 2007). Uncertainties that are proaches that use GCM downscaling to “predict, then act” related to the underlying science will be the same in differ- in response to a narrow range of climate variables (Weaver ent models. et al. 2013). While top-down climate change analyses present a Decision scaling (also referred to as Climate Informed wide range of possible mean future climate conditions, the Decision Analysis or CIDA) is a bottom-up approach to inte- models do not adequately describe the range of potential grate the best current methods for climate risk assessment future conditions more generally (Stainforth et al. 2007b). In and robust decision analysis with simple procedures for addition, top-down analyses provide limited insight into risk management. It is also a robustness-based approach to the changes in climate drivers (such as monsoon patterns water system planning making use of a stress test for the and atmospheric rivers), and climate extremes (Olsen and identification of system vulnerabilities, and simple, direct Gilroy 2012). As a result, deriving probability distributions techniques for the iterative reduction of system vulnerabili- from an ensemble of GCMs is problematic, making it impos- ties through targeted design modifications. The decision sible to predict which future is most likely. Given the essen- 20 scaling methodology has been presented in a number of benefits, the relative likelihood of maintaining a state of no publications (e.g., Brown 2010a, Brown et al. 2011, Brown et regret for each design, and violations of performance thresh- al. 2012). olds. An example method for conducting the stress test is provided in Steinschneider and Brown (2013). The decision scaling stress test consists of three major steps, as shown in Figure 3.2. First, the vulnerabilities of the In the second step, as various sources of climate infor- system to changes in climate are evaluated throughout a mation can be applied without rerunning the modeling large climate space using a “weather generator.” Weather analysis, decision scaling can make use of all sources of cli- generators are developed for the region of interest to pro- mate information, such as a frequency analysis of GCM out- duce numerous stochastic time series that preserve the vari- put, historical data, stochastically-generated climate simula- ability, seasonal, and spatial correlations of the historical re- tions, paleodata, and the expert judgment of scientists and cord. This may be done either by resampling directly from stakeholders. the historical record, or by generating new time series In the third step, collectively, all of these sources of cli- based on the perturbations of the statistical characteristics mate information can then inform the likelihoods of differ- of the historical record. The parameters are systematically ent types of climate change. When climate information is changed to produce new sequences of weather variables deemed fairly reliable and projections are consistent, this al- such as precipitation, which exhibit a wide range of change lows for model-based probabilistic estimates of risk and in their characteristics (e.g., annual average, frequency, in- risk-weighted decision-making. If, on the other hand, projec- tensity, duration). Trends can be added to the precipitation tions based on the various sources are contradictory, not and temperature of the numerous stochastic time series to relevant, or not credible, the process enables the identifica- simulate climate change on a range informed by the avail- tion of climate sensitivities and provides a framework for able downscaled GCMs. Using the stochastic time series, addressing potential hazards through robustness ap- the hydrologic and water resources system model is then run repeatedly over the entire period for many future cli- proaches. mates for each of the water system plans considered. The Decision scaling supports the use of bottom-up ap- performance of each proposed plan is evaluated over a proaches for defining decision-making pathways. The first range of future climate states and the results are presented step here is a stakeholder consultation for identification and on a climate response map. Examples of system perform- characterization of historical system performance and vul- ance evaluations could include cost-benefit ratio, total net nerabilities to change. While standard decision analysis re- 21 quires well-characterized uncertainties, decision scaling Prudhomme et al. (2010) favor a procedure very similar was developed to handle poorly characterized uncertainties to the climate stress test used in decision scaling. The and make the best use of available information. A further authors use a “change factor” (CF) to apply an absolute per- advantage of the bottom-up approaches is that non-climatic centage change to temperature and precipitation in line stressors of the system are readily accommodated. This en- with that suggested by the GCMs, and then use a harmonic ables a more holistic approach to risk screening, thereby function to model the seasonal pattern of precipitation and avoiding what some have termed “climate exceptionalism.” temperature. By performing repeated simulations using a hydrologic model to observe flood peaks across scenarios, Crucially, decision scaling determines whether the the procedure generates valuable information (risk analy- time- and effort-intensive process of downscaling is likely sis) on the critical climate conditions at which a water sys- to be beneficial. The resulting climate response function pro- tem fails. vides insight into the expected performance of the system in an uncertain future. The procedure does not include an IGDT characterizes the uncertainty of system perform- explicit framework for risk management, as will be dis- ance as a group of nested sets. The method requires the cussed in Chapter 4, but the methodology does contribute user to identify a best estimate of each unknown parameter many of the informational elements required for a decision from which to start the uncertainty analysis. Next, each of tool to be effective. the input parameters is bounded in an interval, the range of which is meant to encompass most of the uncertainty par- Bottom-up approaches similar to decision scaling ticular to that parameter. Whereas in the stress tests devel- Other examples of bottom-up approaches to climate oped within decision scaling and scenario-neutral modeling risk assessment are the scenario-neutral approach (Prud- a single increment of uncertainty is explored—the total homme et al. 2010), the information-gap decision theory range of average annual temperature and precipitation over (IGDT, Ben-Haim 2006), and risk-informed decision- which the performance of a water project is evaluat- making (Olsen and Gilroy 2012). These approaches focus on ed—IGDT explores the range of performance within sub- the decision at hand and then scale climate information sets of the total uncertainty space, which are referred to as based on what is needed to best inform that decision. This “horizons.” Careful attention must be given to the selection allows water managers or planners to ask specific questions of the best estimate of the uncertain parameter, and the hori- about the relevance of climate change to a project or deci- zon of uncertainty explored should be chosen large enough sion. to encompass all reasonable parameter realizations. In this 22 way, decision scaling represents an improvement on the maintenance or stress testing of water infrastructure have IGDT approach by starting from a logical point (climate nor- different governance and decision-making needs compared mal) and then using projections to inform the probabilities to national or global priority-setting exercises to allocate of the space that can be derived. limited capacity or funds (Wilby and Dessai 2010, Brown 2010b). Limitations of bottom-up assessments The bottom-up approach relies on top-down informa- tion to inform the likelihoods of future climate conditions; this is essential. The scientific understanding of physical cli- mate mechanisms (and specifically, response to changes in forcing) informs the experiments performed using bottom- up techniques. Without these inputs from the physical cli- mate modeling community, the bottom-up approach would lack a basis for selecting the range over which to test the vulnerability of the system. The vulnerability exploration would be imprecise and unbounded, and of limited decision-making value. Concluding remarks on bottom-up approaches For most risk-assessment applications in water re- sources management, bottom-up approaches are more rele- vant than top-down approaches since climate impacts are difficult to untangle or correlate with hydrologic changes (Matthews and Wickel 2009, Parmesan et al. 2011). How- ever, both top-down and bottom-up approaches can poten- tially provide complementary information (Le Quesne et al. 2010). The selection of an approach, alone or in combina- tion, should be guided by the level of specificity and confi- dence necessary: local scales, operations decisions, and the 23 CHAPTER 4 Identifying robust adaptation strategies Chapter 4 reviews some of • No-regret / low-regret, Media the most prominent ap- • Precautionary principle/ Video 4.1 - Low-regret climate adaptation proaches to identify and safety margins, Video 4.2 - Adaptive institutions evaluate robust adaptation • Sensitivity analysis, strategies for water pro- • Benefit-cost analysis, jects, including: • Stochastic optimization, • Adaptive management, • Real options, and • Robust decision-making. 24 An overview of approaches to evaluate and include ad- ronment for adaptation and the implementation of activi- aptation in water projects ties to manage future flood risk. Developing and applying a How do we move from the diagnosis and assessment robust adaptation strategy requires an enabling environ- of potential climate impacts to planning, design, and ac- ment, supported by activities such as routine monitoring, tion? Given uncertainties in the magnitude and direction of flood forecasting, data exchange, institutional reform, bridg- climate change, project planners are ill-equipped to assess ing organizations, contingency planning for disasters, and the trade-offs of adaptation options to reduce the effects of insurance and legal incentives to reduce vulnerability. climate change on water resources systems relative to alter- These enabling activities are “low-regret” in that they yield native actions intended to address changes in other vari- benefits regardless of the climate scenario. On the other ables such as population, technology, and demand (the mag- hand, reducing vulnerability to plausible future climates nitudes and directions of which are also uncertain). Project may require implementing activities that go beyond low- planners are consequently unable to incorporate climate in- regret enabling activities, including climate safety factors formation into a broader assessment of a project’s probabil- for new build, upgrading the resistance and resilience of ex- ity of success, and to make intelligent modifications to the isting infrastructure, modifying operating rules, develop- project design to reduce its vulnerabilities to failure. Project ment control, flood forecasting, temporary and permanent planners faced with these challenges should not expect cli- retreat from hazardous areas, and periodic review and adap- mate science to develop a single, clearly defined, “most tive management. While implementing activities have high likely” future. potential for vulnerability reductions, they are generally more expensive, less flexible, and less reversible than ena- Under these conditions, robust adaptation is the most bling activities, opening the window to regrets in the event effective approach. Robust adaptation strategies prioritize that the future climate differs from that for which the adap- the ability of projects to perform well over a wide range of tation was developed (Wilby and Keenan 2012). climate and non-climate uncertainties rather than attempt- ing to define a single set of targets. Robust adaptation strate- As previously implied, robustness typically increases gies can take many forms and be classified as “no-regret,” project cost, and it’s economically and physically impossi- reversible and flexible, incorporating safety margins, em- bly to design a project that can perform under the full range ploying “soft” solutions, or reducing decision timeframes of uncertainties. In view of this, vulnerability thresholds are (Hallegatte 2009). Wilby and Keenan (2012) further distin- commonly established for robustness to many, but not all, guish between activities related to creating an enabling envi- possible climate futures. There is further concern that inter- 25 ventions intended to increase adaptation in one sector When the complexities of the application demand it, the might inadvertently increase total system vulnerability by, risk-assessment aspects of decision scaling work in concert for example, increasing carbon emissions or transferring with a combination of risk management tools described risks from one group to another (Barnett and O’Neill 2010), here to create a holistic climate risk assessment and manage- reinforcing the need to develop holistic adaptation strate- ment approach. gies. No-regret / low-regret Approaches developed to identify a most efficient path In the absence of accurate climate prediction models, through a subset of the available adaptation actions/ the “no-regret” or (perhaps more aptly named “low- activities have to this point mostly been founded on modifi- regret”) approach gives priority to actions that are prudent cations to traditional decision-making models. Risk- and ro- regardless of future climate conditions (Video 4.1). For ex- bustness–based approaches to decision-making under un- ample, it is always good to save water (hence prospect for certainty trade off initial investment costs with benefits re- water demand management) and improve water use effi- turned and potentially future costs avoided over the life- ciency in agriculture (“more crop from a drop”). Low-regret time of the project. Recently, a number of suggestions have adaptation decisions perform reasonably well compared to been made for developing and implementing robust de- the alternatives over a wide range of future climate states signs and policies that accommodate uncertain, non- and typically have positive net benefits over the entire stationary information (Salas et al. 2012). In this chapter we range of anticipated future climate states (Field et al. 2012). provide snapshots of the most prominent approaches for In contrast to the low-regret approach, a decision based on the identification of robust adaptation strategies. The useful- a small number of possible climate futures may lead to mal- ness of each approach depends on the individual situation, adaptation if the actual future doesn’t match the limited and in many cases a combination of approaches may prove number of scenarios considered. advantageous. The “soft path’’ to climate adaptation often features The essential approach of decision scaling uses itera- prominently in low-regret decision-making (e.g. Gleick tive application of the climate stress test to systematic, tar- 2003, Pearce 2004). The soft path may include non- geted modifications of the preliminary design or existing structural measures such as water conservation, demand system in order to identify configurations that are more ro- management (e.g., water pricing), floodplain zoning, disas- bust than others to the potential future climate domain. ter relief and emergency preparedness (e.g., flood forecast- 26 ing, warning, and evacuation Video 4.1 Adaptation chal- analysis (BCA) under uncertainty, stochastic optimization, plans), flood and drought insur- lenges and low-regrets adaptive management, real options analysis, and robust ance, optimization of existing decision-making (RDM). systems (e.g., reservoir opera- Precautionary principle / safety margins tion rules), water-efficient crop- ping patterns and indigenous A simple and effective strategy for decision-making un- agriculture, watershed manage- der uncertainty is to be conservative. Uncertainty associ- ment and protection of water ated with estimation errors and acknowledged faults in the quality, adjustments in river stationarity assumption were historically addressed using transportation standards, en- Dr. Zbigniew Kundzewicz the “precautionary principle” and safety margins. For exam- hancement of water storage (Potsdam Institute for Cli- ple, planners oversized dams and added extra height or and other aquifer augmenta- mate Impact Research (PIK)) freeboard to levees above the size analytically deemed nec- describes adaptation chal- tion, and low-impact utiliza- essary (Stakhiv 2010). Of course, the magnitude of the lenges and a low-regret ap- tion of run-of-the-river hydro- safety margin is affected by many factors, including the cost proach to climate adaptation. power. Reservoir reoperation, of additional capacity, the consequences of system failure, in particular, has been shown to be a cost-effective adapta- the economic lifetime of the project, the flexibility of the de- tion strategy (e.g., Watts et al. 2011; Vonk et al. 2014), with sign, and the likelihood that better forecasts of future condi- the understanding that the opportunities, constraints, and tions will become available in time to add additional capac- goals for dam reoperation are region- and site-specific, and ity at a later stage. strongly influenced by the main operating purpose(s) of the dam (e.g., flood mitigation, production of hydropower, wa- In many cases, the projected future hydrologic and so- ter supply) (Richter and Thomas 2007). cioeconomic conditions challenge the theory that design conservatism and safety margins can adequately address However, a soft path by itself would not be sufficient future uncertainties. The magnitude of future uncertainties for the needs of most of the developing world. When combi- affecting water resources management is far greater than nations of hard infrastructure, soft-path practices, and insti- the uncertainty assumed in the past (Hall and Murphy tutional adjustment are required for robust adaptation, 2012; Wilby and Dessai 2010). Also, due to budget con- more advanced tools for trading off benefits and costs may straints and growing demands for water, energy, and envi- be needed. Examples highlighted here include benefit-cost ronmental protection, many water and economic budgets 27 no longer have room to allow for operational and economic Benefit-cost analysis (BCA) under uncertainty inefficiencies associated with the historical conservative ap- Traditionally, BCA has often been used in water re- proach to designing water resources systems (Frederick et sources development to choose among alternative projects. al. 1997). BCA under uncertainty generally requires estimates of pos- Sensitivity analysis sible future states as well as the probability of those states occurring. This information can then be used to calculate Sensitivity analysis is a method for assessing the effect the expected net present value of future benefits and costs of uncertainty on system performance, which considers the of competing projects. Subsequently, an optimal solution possible costs of making alternative choices to some “opti- can be found that maximizes economic benefit or some mal” decision. According to Loucks and van Beek (2005), other performance criterion (Olsen and Gilroy 2012). Opera- “A sensitivity analysis attempts to determine the change in tions Research (OR), developed during World War II, has model output values that results from modest changes in provided the tools for modern decision analysis of this type model input values. A sensitivity analysis thus measures (Hillier and Lieberman 2005). the change in the model output in a localized region of the space of inputs.” A sensitivity analysis, however, is not the Historically, probability distributions for future hydro- same as a thorough analysis of the uncertainties potentially logic states have been estimated statistically based on the affecting system performance (together with their probabil- observed record and the assumption that the statistical ity of occurrence), and it does not address the question of properties of hydrologic variables in the future will be statis- what decision should be made when the future is unknown tically similar to the observed record. However, as dis- or unknowable (Loucks et al. 1981). Furthermore, as argued cussed in Chapter 1, this stationarity assumption is no by Lempert et al. (2006), the attachment of sensitivity analy- longer appropriate (Milly et al. 2008). Further, as mentioned sis to traditional decision analysis techniques is an ade- in Chapter 2, no consensus exists regarding the appropriate quate measure for risk exploration only when the optimum discount rate used to assess future costs and benefits under strategy is relatively insensitive to key assumptions. When climate change. As a result, depending on the project, BCA it is not, sensitivity analysis techniques can lead to strate- may be extremely dependent on parameters for which there gies vulnerable to surprises that might have been countered is either no scientific agreement (probabilities of future hy- had available information been used differently (Lempert et drologic states) or no consensus (discount rate). al. 2002). 28 BCA can be useful in water management decision- bust optimization extends stochastic optimization to explic- making, particularly in situations in which uncertainty is itly make it more robust to challenging scenarios (Ray et al. quantifiable or limited. However, conducting a BCA under 2014). the deep uncertainty of climate change and other drivers Adaptive management poses considerable challenges. When representing the un- certainty associated with climate change indices (e.g., tem- Adaptive management “promotes flexible decision- perature and precipitation) with Gaussian or other asymp- making that can be adjusted in the face of uncertainties as totically diminishing probability distribution functions, the outcomes from management actions and other events be- BCA under uncertainty method is extremely sensitive to come better understood” (NRC 2004). It is a structured, it- tails of the distribution functions (Weitzman 2009). In situa- erative process that requires adaptive system components, tions of deep uncertainty, therefore, BCA is best used as a including institutions (Video 4.2), infrastructure, policy and screening tool (Hallegatte et al. 2012). regulations, etc. In the context of climate change, documen- tation and monitoring of each step and all outcomes ad- Stochastic optimization vances the scientific understanding of climate change and While approaches aimed at producing a narrow concep- informs adjustments in policy or operations as part of an it- tion of optimality (“one future”) have traditionally been at erative learning process. Adaptive management is a continu- odds with approaches aiming at robustness (“many fu- ous process of adjustment that tures”), stochastic optimization is a technique in which mul- attempts to deal with the in- Video 4.2 Adaptive institutions tiple future scenarios are weighted probabilistically. The creasingly rapid changes in our “best” design performs reasonably well across the range of climate, societies, economies, considered futures. In all likelihood, a stochastically opti- and technologies. It increases mized solution is not the best-performing design for any the ability of decision-makers single future. Stochastic optimization offers a straightfor- to formulate timely responses ward, first-order approximation of hedging against unfeasi- to new information. Adaptive bility, and is thus a step toward robustness. For summaries institutions are essential to of stochastic optimization techniques that apply probabilis- adaptive management. Dr. Margot Hill Clarvis tic uncertainty paradigms to water systems decision- As noted by Stakhiv (2011), (University of Geneva) dis- making, see Revelle and colleagues (2004), Loucks and van cusses what it means to be the water resources manage- Beek (2005), and Sen and Higle (1999). Multi-objective ro- an adaptive institution. 29 ment sector has developed a variety of strategies to deal Real options with periods of high demand and low water availability. Real options analysis is an established probabilistic de- They consist of longer-term infrastructure ‘‘adaptation’’ to cision process by which adaptability can be explicitly incor- stationary climate signals and shorter-term ‘‘adaptive man- porated into project designs in an effort to avoid potential agement’’ measures that center mostly on flexible opera- regrets associated with either over-investment or under- tions, forecasting, and innovative uses of existing delivery investment in adaptation measures. Real options encour- and supply infrastructure to meet unexpected demands ages staged decision-making through which more expen- and match changing extremes. There are five ways that wa- sive and more highly-irreversible decisions are reserved un- ter managers have of adapting to climate variability and til more information is available on which to base those de- change, and different water management strategies employ cisions. The philosophical underpinning of real options has various combinations of all the categories listed below: roots in the work of Dewey (1927), who promoted policies • Planning new investments or capacity expansion (reser- with continual learning and adaptation in response to expe- voirs, irrigation systems, levees, water supply, wastewater rience over time, as well as Rosenhead (1989), who defined treatment). flexibility and keeping options open as an indicator for evaluating the robustness of strategies under uncertainty. • Operation, monitoring, and regulation of existing systems The mechanism for real options is founded on the analysis to accommodate new uses or conditions (ecology, climate of financial decision-making (Arrow and Fisher 1974, change, population growth). Henry 1974, Myers 1984, Copeland and Antikarov 2001). A • Maintenance and major rehabilitation of existing systems real options analysis can be integrated into a stochastic opti- (dams, barrages, irrigation systems, canals, pumps). mization strategy. • Modifications in processes and demands (water conserva- A strong water system management plan combines ele- tion, pricing, regulation, legislation) for existing systems ments of adaptability, flexibility, diversification, and robust- and water users. ness. Real options analysis is applicable when uncertainty • Introduction of new, more efficient technologies (desalina- is more “dynamic” than “deep” (i.e., the quality of our tion, drip irrigation, wastewater reuse, recycling) (Stakhiv knowledge should improve over time) and the project in- 2011). volves potentially irreversible decisions, such as major infra- structure investments. Some adaptation strategies will be more flexible than others in the future. The expected value 30 of each option—its degree of flexibility—can be calculated noot et al. 2013, HMT DEFRA 2009, Ingham et al. 2007, and compared. The objective in this formulation is still to Merz et al. 2010, Woodward et al. 2011, Woodward et al. maximize net present value, but the adaptability of design 2011). options is explicitly considered. The government of the Robust decision-making (RDM) United Kingdom, for example, requires that climate change adaptation analyses account for “the value of flexibility in Robust decision-making (RDM) attempts to more strate- the structure of the activity” (HMT DEFRA 2009). gically use deeply uncertain climate information to answer adaptation questions. RDM uses an iterative decision frame- Examples of real options for water supply include in- work to identify strategies that perform reasonably well vestments in pumps to draw upon dead storage, pipelines over a wide range of plausible future scenarios (Lempert et to connect to storage at another impoundment, or infrastruc- al. 2003; Lempert et al. 2006). RDM inverts traditional sensi- ture to tap groundwater resources. Demand-oriented real tivity analysis, seeking strategies whose good performance options for water supply are also possible, such as invest- is insensitive to the most significant uncertainties. The proc- ments in household metering and a strong public outreach ess begins with scenario generation, based on the principles campaign that could be implemented at some cost to help of scenario planning (Schwartz 1996) and informed by enforce future conservation efforts (Steinschneider and downscaled GCMs and stakeholder-derived information on Brown 2012). Real option water transfers provide a mecha- expected local conditions. The scenario-generation process nism by which water supply can be augmented without the is designed to encompass a very wide range of possible fu- need for large-scale infrastructure expansion. A number of tures. Typically, RDM uses a priori internally consistent sce- studies have demonstrated how financial instruments such narios and bases climate forecasts directly on time series of as leases, option contracts, and water banks can facilitate downscaled GCMs. the trade of water between low- and high-priority uses dur- ing a localized water shortage (see, for example, Brown and Once the climate scenarios have been developed, the Carriquiry 2007, Characklis et al. 2006, Kirsch et al. 2009, next step in RDM is the identification of a proposed robust Lund and Israel 1995, Palmer and Characklis 2009, strategy, through an initial ranking or screening, along with Steinschneider and Brown 2012). Applications to water re- the identification and characterization of one or more clus- sources problems with a focus on the mitigation of flood ters of future states in which each of the strategies perform damages have also become common (e.g., Gersonius et al. poorly. These clusters, unweighted by probabilities, are de- 2010, Gersonius et al. 2013, Hall and Harvey 2009, Haas- signed to be considered even if decision-makers find them 31 unlikely or inconvenient. Faced with potential futures (or gies designed to evolve over time in response to new infor- clusters of futures) in which a proposed strategy (or sys- mation. tem) performs poorly, the RDM procedure makes modifica- tions to the strategy (or system) that hedge against vulner- abilities. Finally, the trade-offs involved in the choice among the hedging options are explored. Because RDM samples from all combinations of uncer- tain system parameters, it explores futures both more be- nign and more dire than the present. In a full RDM analy- sis, various aspects of candidate strategies would be succes- sively altered and resubmitted to the RDM process until a suitably robust strategy was identified. A particular strength of RDM is its ability to model complex systems; its framework enables it to analyze very large numbers of sce- narios in which any or all system and design parameters are altered in any number of configurations (RAND 2013). Using an approach similar to decision scaling, RDM characterizes uncertainty in the context of a particular deci- sion. However, RDM applies equal probability to all consid- ered climate futures and identifies adaptation strategies that perform well across as wide a range of those potential futures as possible. This type of approach thus makes it very difficult to weight extreme scenarios to which the ad- aptation strategy is vulnerable in proportion to the many less extreme scenarios to which it is robust. Importantly, RDM also includes iterative and adaptive decision strate- 32 CHAPTER 5 Moving beyond downscaling Chapter 5 makes the case for moving beyond just downscal- Media ing GCMs for climate adaptation in the water sector. In- Video 5.1 - A bottom-up adaptation approach stead, AGWA supports a bottom-up approach to adaptation, supported by AGWA described subsequently through discussion of: • The impetus for a bottom-up approach to adaptation, • A framework for an adaptation approach supported by AGWA, and • Towards the AGWA Decision Support System (DSS) 33 The impetus for a bottom-up approach to adaptation approaches for managing the uncertainties of future climate There is a tremendous need for practical guidance that conditions have proven unsatisfactory and ineffective for supports water resources management under climate quantitative engineering, long-term sustainable resource change. For applications relating to water management, the management, and many investment decisions. Under these elusive “gold standard” for climate adaptation has been ac- low-confidence and high-consequence conditions, AGWA curate, confident, and quantitative estimates of future cli- suggests shifting to a bottom-up, adaptive management mate states—ideally, decades from now, over the full opera- strategy aimed at creating a more robust system given the tional lifetime of water infrastructure and long-term plan- relatively high uncertainty. ning horizons. However, the data produced by downscal- As a first product of the collective effort, AGWA is de- ing GCM output are far from this standard and not appro- veloping a Decision Support System (DSS) to guide water priate for use as a starting point for water resources risk as- management planners, investment officers, and practitio- sessment. Consequently, improvements in data develop- ners in combining existing tools, research, and data prod- ment and acquisition, particularly in developing countries, ucts into an evidence-based system to inform water manage- must be prioritized. Also, stakeholders must define perform- ment decision-making processes. The DSS is meant to pro- ance metrics and performance thresholds. From its incep- vide a generalized methodology for (1) analyzing risk using tion, AGWA has focused on this data problem, which is ar- “bottom-up” methodologies, (2) integrating ecological and guably the single most technically challenging issue sur- engineering approaches to achieve resilient and robust wa- rounding climate adaptation. ter management, (3) using economic tools to enable and pro- Building upon previous recommendations to move be- mote flexible decision pathways, as well as (4) governance yond downscaling (Fowler and Wilby 2006), AGWA has mechanisms that represent broad allocation needs and en- adopted an approach to assist in the selection of appropri- able consensus-based approaches. Beyond Downscaling par- ate strategies for robust water resources design and plan- ticularly targets the first component, but attempts to touch ning under uncertainty. This approach requires an evalua- on all of the topics. tion of both the confidence in the available data and the po- A framework for an adaptation approach supported by tential consequences of climatic changes. Under high confi- AGWA dence and/or low-consequence situations, AGWA supports The adaptation approach supported by AGWA recog- the use of traditional planning and design methods based nizes that robust, quantitative approaches and insights into on stationary, probabilistic concepts. However, traditional 34 climate adaptation have been accruing over the past dec- ble as we once believed they were. As a result, we can lock ade, and that these insights span a wide range of disci- in ineffective investments for very long periods into the fu- plines: engineering, economics, hydrology and ecology, ture if we are not robust to a wide range of potential shifts governance and law, climate science, and finance, among we may experience. As a result, decision-making processes others. Each discipline’s accrued knowledge however, has around the adaptation approach supported by AGWA em- largely been developed in isolation and without clear refer- body several strategies and assumptions, namely: ence to complementary or conflicting perspectives from other disciplines. This book represents great progress on • Climate change is not relevant to all water resources man- agement work, nor is climate change equally important to the integration of engineering, economics, climate science, all problems when climate impacts will be relevant. The and hydrology. However, there is a great need for further approach supported by AGWA recognizes the need to inte- integration into water resources management of ecology, grate climate adaptation into existing decision-making governance and law, finance, and many other human fac- processes around water management rather than invent- tors, including urban/rural issues, manufacturing/ ing completely new methodologies. agricultural water allocations and trade, transboundary water sharing (hydrohegemony), water-related aspects of • Climate vulnerability assessments are widely understood poverty reduction, and social/religious/cultural links to to be a critical component to determine risks for water re- and valuation of natural water resources. Further, climate sources management under climate change, relative to analysts need to work more directly with decision-makers other threats and opportunities. AGWA advocates bottom- to co-explore and co-produce knowledge about climate up approaches to vulnerability assessment, which reflect risks and adaptation options. inherent system limits and serve as an effective means of framing uncertainties about future climate projections AGWA believes that the convergence of disciplines, rather than top-down methodologies, which rely heavily tools, and expertise represent the ascendance of a new on climate models to frame vulnerability (Video 5.1). paradigm for water management that integrates climate Stakeholders are a key gap—engaging and educating resilience with non-stationary water perspectives. Critical stakeholders can both help define systemic vulnerabilities to this paradigm is the insight that current decision- and opportunities and serve as a platform for dialogue making processes represent the weakest and most climate- with a decision-scaling coach, fostering consensus and vulnerable element in how we approach water manage- problem solving. This book attempts to begin enabling ment. Our decisions are not as credible, effective, or dura- these methodologies for water management decisions 35 (visit the AGWA website for Video 5.1 A bottom-up adapta- (a) estimate the costs of maintaining multiple options and bottom-up case study pres- tion approach supported by flexibility, (b) evaluate the trade-offs between waiting for entations). AGWA more certain information before implementation versus • The use of explicit, system- acting in the short term with less information (presumably atic decision trees based on requiring more robust and expensive solutions), and (c) existing water resources design multiple decision-making pathways. management approaches, • The challenge of sustainability itself contains philosophi- such as the approaches be- cal issues. Sustainable water resources management must ing developed by the US merge perspectives on resilience and robustness from both Army Corps of Engineers engineering and ecological perspectives. Previously, these Dr. Guillermo Mendoza (US (USACE) and the World visions of resilience have been in tension and opposition, Army Corps of Engineers) Bank (“Including Climate but bottom-up approaches can serve as a powerful frame- describes a bottom-up adapta- Uncertainty in Water Re- tion approach supported by work for integration by making dynamic ecosystem integ- sources Planning and Pro- AGWA. rity a performance marker for water sustainability. ject Design – Decision Tree • Finally, flexibility must be implemented and expressed Initiative”), will enable separate individuals to come to through real-world governance mechanisms and institu- similar conclusions about vulnerabilities and effective ad- tional processes. Integrating into water resources manage- aptation responses for the same project, assuming they ment the use of flexible governance mechanisms that as- have access to the same initial datasets. In addition, these sume allocations can be adjusted in response to or anticipa- decision trees should help water managers and planners tion of dynamic water conditions is essential to reducing “track” the emergence of alternative futures over time and the potential for conflict and crisis-induced decision- detect decision-making tipping points, which will enable making. long-term flexible management, operations, and imple- Towards the AGWA Decision Support System (DSS) mentation. The adaptation approach supported by AGWA is on • Closely connected is the process of creating explicitly flexi- its way to becoming a formal methodology called the ble decision pathways, so that the risk of making all-at- AGWA Decision Support System (DSS). As an organiza- once stationary decisions is minimized. Critical here is the tion, AGWA seeks to harvest expert knowledge and place development of economic analytical methodologies that it in a format that can be used for systematic, consistent, 36 repeatable applications for “bottom-up” risk assessment, sessments are now more commonly required as part of integration of ecological and engineering resilience into broader project evaluations, it is important that such water management, and economic analysis that promotes evaluations be accomplished in the most efficient and di- flexibility and robustness. The DSS will encompass the rect manner possible. The bottom-up processes developed aforementioned strategies to provide water managers with as part of the AGWA DSS are designed to be the most tar- a decision system that will help them select the appropri- geted and efficient tools available for climate change risk ate techniques and tools for resilient water resources de- assessment and risk management. As the AGWA DSS sign and planning. grows, project planners will have access to the methodo- logical frameworks, information, and community of prac- The content work streams for the AGWA DSS draw di- tice to empower targeted and comprehensive risk manage- rectly on the diversity of knowledge critical to making ment in proportion to the risks faced, resulting in robust, more resilient decisions. The work streams have been or- cost-effective project designs and management plans. ganized into four clusters, the first three of which were launched as a result of the World Bank/AGWA workshop (2011): • Hydrology and climate science • Economics and finance • Engineering and ecology • Governance The AGWA DSS is intended as a resource center offer- ing documents that describe how to implement the adapta- tion approach supported by AGWA and access to software tools for decision support, as well as a series of connected strategy and implementation guidance documents to sup- port resource managers and technical staff, infrastructure designers and operators, and policy and planning staff across a wide range of sectors. As climate change risk as- 37 REFERENCES Arrow, K. J. and A. Fisher. 1974. “Environmental preserva- Brown, C. and M. Carriquiry. 2007. “Managing hydroclima- tion, uncertainty, and irreversibility.” The Quarterly Jour- tological risk to water supply with option contracts and nal of Economics 88(2): 312–19. reservoir index insurance.” Water Resources Research 43 (11). doi: 10.1029/2007WR006093. Barnett, J and S. O’Neill. 2010. “Maladaptation.” Global Envi- ronmental Change 20: 211–13. Brown, C., Y. Ghile, M. Laverty, and K. Li. 2012. “Decision Scaling: Linking Bottom-up Vulnerability Analysis with Barsugli, J. J., C. J. Anderson, J. B. Smith, and J. M. Vogel. Climate Projections in the Water Sector.” Water Resources 2009. “What’s Needed from Climate Modeling to Ad- Research 48 (9). doi:10.1029/2011WR011212. vance Actionable Science for Water Utilities?” AGU Fall Meeting (December): 08. Brown, C., W. Werick, W. Leger, and D. Fay. 2011. “A Decision-Analytic Approach to Managing Climate Risks: Ben-Haim, Y. 2006. Info-Gap Decision Theory: Decisions Under Application to the Upper Great Lakes.” JAWRA Journal of Severe Uncertainty, 2nd ed. London, UK: Academic Press. the American Water Resources Association 47 (3): 524–34. Bouwer, L.M. 2013. “Projections of Future Extreme Weather doi:10.1111/j.1752-1688.2011.00552.x. Losses Under Changes in Climate and Exposure.” Risk Brown, C., and R.L. Wilby. 2012. “An Alternate Approach to Analysis 33 (5): 915–30. Assessing Climate Risks.” Eos, Transactions American Geo- doi:10.1111/j.1539-6924.2012.01880.x. physical Union 93 (41): 401–2. doi:10.1029/2012EO410001. Brown, C. 2010a. “Decision-scaling for Robust Planning Characklis, G. W., B. R. Kirsch, J. Ramsey, K. Dillard, and C. and Policy under Climate Uncertainty.” World Resources T. Kelley. 2006. “Developing portfolios of water supply Report, Washington, D.C. transfers.” Water Resources Research 42 (5). Brown, C. 2010b. “The End of Reliability.” Journal of Water DOI: 10.1029/2005WR004424. Resources Planning and Management 136 (2): 143–45. Confalonieri, R. 2012. “Combining a weather generator and doi:10.1061/(ASCE)WR.1943-5452.65. a standard sensitivity analysis method to quantify the relevance of weather variables on agrometeorological 38 models outputs.” Theoretical Applied Climatology. 108 (1- Dubrovsky, M., Z. Zalud, and M. Stastna. 2000. “Sensitivity 2): 19–30. of ceres-maize yields to statistical structure of daily weather variables.” Climatic Change. 46 (4): 447–72. Copeland, T. and V. Antikarov. 2001. Real Options: A Practi- tioner's Guide, 1st ed., 320 pgs. New York, NY: Texere. Fekete, B., C. Vorosmarty, J. Roads, and C. Willmott. 2004. “Uncertainties in precipitation and their impacts on run- Dai, A., K. Trenberth, and T. Karl. 1998. “Global variations off estimates.” Journal of Climate 17(2): 294–304. in droughts and wet spells: 1900-1995.” Geophysical Re- search Letters 25(17): 3367–70. Field, C. B., V. Barros, T. F. Stocker, and Q. Dahe. 2012. Man- aging the Risks of Extreme Events and Disasters to Advance Deser, C., R. Knutti, S. Solomon, and A. Phillips. 2012. Climate Change Adaptation: Special Report of the Intergovern- “Communication of the role of natural variability in fu- mental Panel on Climate Change. Cambridge University ture North American climate”. Nature Climate Change 2: Press. 775-9, DOI: 10.1038/NCLIMATE1562. Frederick, K. D, D. C Major, and E. Z Stakhiv. 1997. “Water Dessai, S., and M. Hulme. 2004. “Does Climate Adaptation Resources Planning Principles and Evaluation Criteria Policy Need Probabilities.” Climate Policy 4 (2): 107–28. for Climate Change: Summary and Conclusions.” Cli- Dewey, J. 1927. The Public and its Problems. New York, NY, matic Change 37 (1): 291–313. USA: Holt and Company. Fowler, H. J., and R. L. Wilby. 2007. “Beyond the Downscal- Dobler, C., S. Hagemann, R. L. Wilby, and J. Stötter. 2012. ing Comparison Study.” International Journal of Climatol- “Quantifying Different Sources of Uncertainty in Hydro- ogy 27 (12): 1543–45. logical Projections in an Alpine Watershed.” Hydrology Gersonius, B., R. Ashley, A. Pathirana, and  C. Zevenbergen. and Earth System Sciences 16 (11): 4343–60. 2010. “Managing the flooding system's resiliency to cli- Doczi, J. 2013. “Introduction to Tools for Climate Adapta- mate change.” Proceedings of the Institution of Civil tion & Risk Management for the Water Sector in a Trans- Engineers-Engineering Sustainability 163: 15–22. boundary Context.” Presentation in Geneva. Gersonius, B., R. Ashley, A. Pathirana, and C. Zevenbergen. 2013. “Climate Change Uncertainty: Building Flexibility 39 into Water and Flood Risk Infrastructure.” Climatic for Crafting Robust Decisions for a Deeply Uncertain Change 116 (2): 411–23. World.” Global Environmental Change 23 (2): 485–98. doi:10.1016/j.gloenvcha.2012.12.006. Ghosh, S., D. Raje, and P. P. Mujumdar. 2010. “Mahanadi Streamflow: Climate Change Impact Assessment and Hall, J. 2007. “Probabilistic climate scenarios may misrepre- Adaptive Strategies.” Current Science (Bangalore) 98 (8): sent uncertainty and lead to bad adaptation decisions.” 1084–91. Hydrological Processes 21 (8): 1127–29. Gleick, P.H. 2003. “Global Freshwater Resources: Soft-Path Hall, J., and H. Harvey. 2009. “Decision-Making under Se- Solutions for the 21st Century.” Science 302 (5650): 1524– vere Uncertainties for Flood Risk Management: A Case 28. doi:10.1126/science.1089967. Study of Info-Gap Robustness Analysis.” In 8th Interna- tional Conference on Hydroinformatics. Global Water Partnership (GWP). 2009. “A Handbook for Integrated Water Resources Management in Basins.” Hall, J., and C. Murphy. 2012. “Adapting Water Supply Sys- Elanders, Sweden. tems in a Changing Climate.” In Water Supply Systems, http://www.gwptoolbox.org/images/stories/Docs/gw Distribution and Environmental Effects. Hauppauge, NY: p_inbo%20handbook%20for%20iwrm%20in%20basins_e Nova Science Publishers, Inc. ng.pdf. Hallegatte, Stéphane. 2009. “Strategies to Adapt to an Un- Gollier, C., P. Koundouri, and T. Pantelidis. 2008. “Declin- certain Climate Change.” Global Environmental Change 19 ing Discount Rates: Economic Justifications and Implica- (2): 240–47. tions for Long-Run Policy.” Economic Policy 23 (56): 757– Hallegatte, S., A. Shah, C. Brown, R. Lempert, and S. Gill. 95. 2012. “Investment Decision-Making Under Deep Uncer- Groom, B., C. Hepburn, P. Koundouri, and D. Pearce. 2005. tainty—Application to Climate Change.” SSRN Scholarly “Declining Discount Rates: The Long and the Short of Paper ID 2143067. Rochester, NY: Social Science Research It.” Environmental and Resource Economics 32 (4): 445–93. Network. http://papers.ssrn.com/abstract=2143067. doi:10.1007/s10640-005-4681-y. Henry, C. 1974. “Investment decisions under uncertainty: Haasnoot, M., J.H. Kwakkel, W.E. Walker, and J. ter Maat. the irreversibility effect.” The American Economic Review. 2013. “Dynamic Adaptive Policy Pathways: A Method 64 (6): 1006–12. 40 Hillier, F. S. and G. J. Lieberman. 2005. Introduction to Opera- Intergovernmental Panel on Climate Change (IPCC). 2007. tions Research, 8th ed. Boston, MA: McGraw Hill. Climate Change 2007: Impacts, Adaptation and Vulner- ability: Contribution of Working Group II to the Fourth As- HM Treasury and Department for Environment, Food and sessment Report of the Intergovernmental Panel on Climate Rural Affairs (HMT-DEFRA). 2009. “Accounting for the Change. Edited by M Parry. Cambridge U.K.; New York: Effects of Climate Change: Supplementary Green Book Cambridge University Press. Guidance.” www.hm-treasury.gov.uk/data_greenbook_supguidance ———. 2012. Summary for Policymakers. In: Managing the .htm. Risks of Extreme Events and Disasters to Advance Climate Change Adaptation. A Special Report of Working Groups I Hulme, Mike, Timothy J. Osborn, and Timothy C. Johns. and II of the Intergovernmental Panel on Climate Change. C. 1998. “Precipitation Sensitivity to Global Warming: Com- Cambridge, U.K. and New York, NY, USA. parison of Observations with HadCM2 Simulations.” http://ipcc-wg2.gov/SREX/images/uploads/SREX-SP Geophysical Research Letters 25 (17): 3379–82. Mbrochure_FINAL.pdf. doi:10.1029/98GL02562. Jones, T. H., L. J. Thompson, J. H. Lawton, T. M. Bezemer, R. Hutchinson, M. F. 1995. “Interpolating Mean Rainfall Using D. Bardgett, T. M. Blackburn, K. D. Bruce, et al. 1998. “Im- Thin Plate Smoothing Splines.” International Journal of pacts of Rising Atmospheric Carbon Dioxide on Model Geographical Information Systems 9 (4): 385–403. Terrestrial Ecosystems.” Science 280 (5362): 441–43. Independent Evaluation Group (IEG). 2012. “Adapting to Kendon, Elizabeth J., Nigel M. Roberts, Hayley J. Fowler, Climate Change: Assessing World Bank Group Experi- Malcolm J. Roberts, Steven C. Chan, and Catherine A. ence.” Washington, D.C.: The World Bank. Senior. 2014. “Heavier Summer Downpours with Cli- http://ieg.worldbankgroup.org/evaluations/adapting-c mate Change Revealed by Weather Forecast Resolution limate-change-assessing-world-bank-group-experience. Model.” Nature Climate Change. Ingham, A., J. Ma, and  A. Ulph. 2007. “Climate change, Kerr, R.A. 2013. “Forecasting Regional Climate Change mitigation and adaptation with uncertainty and learn- Flunks Its First Test.” Science 339 (6120): 638. ing.” Energy Policy 35 (11): 5354–69. doi:10.1126/science.339.6120.638. 41 Kilsby, C.G., P.D. Jones, A. Burton, A.C. Ford, H.J. Fowler, Notes. Washington, D.C.: World Bank. C. Harpham, P. James, A. Smith, and R.L. Wilby. 2007. “A http://www.flowingforward.org/. Daily Weather Generator for Use in Climate Change Lempert, R.J., S.C. Bankes, and S.W. Popper. 2003. Shaping Studies.” Environmental Modelling & Software 22 (12): the Next One Hundred Years: New Methods for Quantitative, 1705–19. doi:10.1016/j.envsoft.2007.02.005. Long-Term Policy Analysis. Santa Monica, CA: RAND Cor- Kirsch, B.R., G.W. Characklis, K.E.M. Dillard, and C.T. Kel- poration. ley. 2009. “More Efficient Optimization of Long-Term Wa- http://www.rand.org/pubs/monograph_reports/MR16 ter Supply Portfolios.” Water Resources Research 45 (3): 26.html. W03414. doi:10.1029/2008WR007018. Lempert, R., S. Popper, and S. Bankes. 2002. “Confronting Knutti, Reto, and Jan Sedláček. 2013. “Robustness and Un- Surprise.” Social Science Computer Review 20 (4): 420–40. certainties in the New CMIP5 Climate Model Projec- doi:10.1177/089443902237320. tions.” Nature Climate Change 3 (4): 369–73. Lempert, R.J., D.G. Groves, S.W. Popper, and S.C. Bankes. doi:10.1038/nclimate1716. 2006. “A General, Analytic Method for Generating Ro- Kundzewicz, Z.W. 2011. “Comparative Assessment: Fact or bust Strategies and Narrative Scenarios.” Management Sci- Fiction?” Presented at the Workshop: Including long ence 52 (4): 514–28. doi:10.1287/mnsc.1050.0472. term climate change in hydrologic design. World Bank, Lettenmaier, Dennis P., Andrew W. Wood, Richard N. Pal- Washington, D.C., USA, on November 21, 2011. mer, Eric F. Wood, and Eugene Z. Stakhiv. 1999. “Water Kundzewicz, Z.W., and E.Z. Stakhiv. 2010. “Are Climate Resources Implications of Global Warming: A US Re- Models ‘ready for Prime Time’ in Water Resources Man- gional Perspective.” Climatic Change 43 (3): 537–79. agement Applications, or Is More Research Needed?” Hy- Lins, Harry F., and James R. Slack. 2005. “Seasonal and Re- drological Sciences Journal 55 (7): 1085–89. gional Characteristics of US Streamflow Trends in the doi:10.1080/02626667.2010.513211. United States from 1940 to 1999.” Physical Geography 26 Le Quesne, T., J.H. Matthews, and C. Von der Heyden. 2010. (6): 489–501. “Flowing Forward: Freshwater Ecosystem Adaptation to Climate Change in Water Resources Management and Biodiversity Conservation.” Note No. 28. Water Working 42 Loucks, D. P., J. R. Stedinger, and D. A. Haith. 1981. Water ter, Infrastructure, and Institutions.” PLoS Biol 9(9): Resource Systems Planning and Analysis. Englewood Cliffs, e1001159. doi:10.1371/journal.pbio.1001159. New Jersey: Prentice Hall. McMahon, Amanda, Diego J. Rodriguez, and Caroline van Loucks, D.P. and E. Van Beek. 2005. Water Resources Systems den Berg. 2012. Investing in Water Infrastructure : Capi- Planning and Management: An Introduction to Methods, tal, Operations, and Maintenance. The World Bank. Models and Applications. Paris, France: UNESCO Publish- http://documents.worldbank.org/curated/en/2012/11 ing, /17007405/investing-water-infrastructure-capital-operat ions-maintenance. Lownsbery, K. E. 2014. “Quantifying the impacts of future uncertainties on the Apalachicola-Chattahoochee-Flint Mendoza, Guillermo, and Kristin Gilroy. 2012. “A Guidance basin.” M.S. thesis, 1-80 pp., University of Massachu- Model for Resilient Water Resources Planning and De- setts, Amherst. Amherst, Massachusetts, USA, April sign.” In . 3rd International Interdisciplinary Conference 2014. on Predictions for Hydrology, Ecology, and Water Re- sources Management. Vienna. Lund, J.R. and M. Israel. 1995. “Optimization of transfers in urban water supply planning.” Journal of Water Resources Merz, B., J. Hall, M. Disse, and A. Schumann. 2010. “Fluvial Planning and Management 121 (1): 41–8. flood risk management in a changing world.” Natural Hazards and Earth System Sciences 10: 509–27. Matalas, N.C. 2012. “Comment on the Announced Death of Stationarity.” Journal of Water Resources Planning and Man- Milly, P.C.D., J. Betancourt, M. Falkenmark, R.M. Hirsch, agement 138 (4): 311–12. Z.W. Kundzewicz, D.P. Lettenmaier, and R.J. Stouffer. 2008. “Stationarity Is Dead: Whither Water Manage- Matthews, J.H., and A.J. Wickel. 2009. “Embracing Uncer- ment?” Science 319 (5863): 573–74. tainty in Freshwater Climate Change Adaptation: A Natu- doi:10.1126/science.1151915. ral History Approach.” Climate and Development 1(3): 269–79. doi:10.3763/cdev.2009.0018. Myers, S.C. 1984. “Finance Theory and Financial Strategy.” Interfaces 14 (1): 126–37. Matthews, J.H., A.J. Wickel, and S. Freeman. 2011. “Con- verging Currents in Climate-Relevant Conservation: Wa- National Research Council (NRC). 2004. Adaptive Manage- ment for Water Resources Project Planning. Washington, 43 DC: The National Academies. Qaddumi, H.M., E. Dickson, S.M. Diez, R.F. Hirji, G. Puz, http://www.nap.edu/openbook.php?isbn=0309091918. A.V. Danilenko, M. Jacobsen, V. Alavian, C. Pizarro, and B. Blankespoor. 2009. “Water and Climate Change: Un- Olsen, J. R., and K. Gilroy. 2012. “Risk Informed Decision- derstanding the Risks and Making Climate-Smart Invest- Making in a Changing Climate.” In 3rd International In- ment Decisions.” 52911. Washington, D.C.: The World terdisciplinary Conference on Predictions for Hydrology, Bank. Ecology, and Water Resources Management. Vienna. http://documents.worldbank.org/curated/en/2009/11 Palmer, R.N. and G.W. Characklis. 2009. “Reducing the /11717870/water-climate-change-understanding-risks-m costs of meeting regional water demand through risk- aking-climate-smart-investment-decisions. based transfer agreements.” Journal of Environmental Man- Raje, D., and P. P. Mujumdar. 2010. “Reservoir Performance agement 90 (5): 1703–14. under Uncertainty in Hydrologic Impacts of Climate Parmesan, C., C. Duarte, E. Poloczanska, A.J. Richardson, Change.” Advances in Water Resources 33 (3): 312–26. and M.C. Singer. 2011. “Overstretching Attribution.” Na- Ray, P.A., P.H. Kirshen, and D.W. Watkins Jr. 2012. “Staged ture Climate Change 1 (1): 2–4. doi:10.1038/nclimate1056. Climate Change Adaptation Planning for Water Supply Pearce, F. 2004. Keepers of the Spring: Reclaiming Our Water in in Amman, Jordan.” Journal of Water Resources Planning an Age of Globalization. Washington, D.C.: Island Press. and Management 138 (5):403–11. Poff, N.L., J.D. Allan, M.B. Bain, J.R. Karr, K.L. Prestegaard, Ray, P.A., D.W. Watkins, Jr., R.M. Vogel, and P.H. Kirshen. B.D. Richter, R.E. Sparks, and J.C. Stromberg. 1997. “The 2014. “A performance-based evaluation of an improved Natural Flow Regime.” BioScience 47 (11): 769–84. robust optimization formulation.” Journal of Water Re- doi:10.2307/1313099. sources Planning and Management (in press). Prudhomme, C., R.L. Wilby, S. Crooks, A.L. Kay, and N.S. RAND. 2013. “Making Good Decisions Without Predic- Reynard. 2010. “Scenario-Neutral Approach to Climate tions.” Change Impact Studies: Application to Flood Risk.” Jour- http://www.rand.org/pubs/research_briefs/RB9701/in nal of Hydrology 390 (3-4): 198–209. dex1.html. doi:10.1016/j.jhydrol.2010.06.043. 44 Revelle, C.S., E.E. Whitlatch, and J.R. Wright. 2004. Civil and Sen, S. and J.L. Higle. 1999. “An introductory tutorial on sto- Environmental Systems Engineering, Vol. 2. New Jersey: chastic linear programming models.” Interfaces 29 (2): Prentice Hall. 33–61. Richardson, C.W. 1985. “Weather simulation for crop man- Stainforth, D.A., M.R. Allen, E.R. Tredger, and L.A. Smith. agement models.” Transactions of the American Society of 2007a. “Confidence, Uncertainty and Decision-Support Agricultural Engineers 28:1602–06. Relevance in Climate Predictions.” Philosophical Transac- tions of the Royal Society A: Mathematical, Physical and Engi- Richter, B. D., and G. A. Thomas. 2007. “Restoring Environ- neering Sciences 365 (1857): 2145–61. mental Flows by Modifying Dam Operations.” Ecology and Society 12 (1): 12. Stainforth, D.A., T.E. Downing, R. Washington, A. Lopez, and M. New. 2007b. “Issues in the Interpretation of Cli- Roe, G. H., and M. B. Baker. 2007. “Why Is Climate Sensitiv- mate Model Ensembles to Inform Decisions.” Philosophi- ity So Unpredictable?” Science 318 (5850): 629–32. cal Transactions of the Royal Society A: Mathematical, Physi- doi:10.1126/science.1144735. cal and Engineering Sciences 365 (1857): 2163–77. Rosenhead, J. 1989. Robustness analysis: Keeping your options Stakhiv, E.Z. 2010. “Practical Approaches to Water Manage- open, in Rational Analysis for a Problematic World, edited ment under Climate Change Uncertainty.” In Workshop by Jonathan Rosenhead, pp. 181–207. Wiley. on Non-Stationarity, Hydrologic Frequency Analysis, and Wa- Salas, J., B. Rajagopalan, L. Saito, and C. Brown. 2012. “Spe- ter Management, edited by J. Rolf Olsen, Julie Kiang, and cial Section on Climate Change and Water Resources: Cli- Reagan Waskom. Information Series No. 109. Boulder, mate Non-Stationarity and Water Resources Manage- CO: Colorado Water Institute. ment.” Journal of Water Resources Planning and Manage- http://www.cwi.colostate.edu/Non-StationarityWorksh ment 138 (5): 385–88. op/index.shtml. doi:10.1061/(ASCE)WR.1943-5452.0000279. ———. 2011. “Pragmatic Approaches for Water Manage- Schwartz, P. 1996. The Art of the Long View. New York, NY: ment Under Climate Change Uncertainty.” JAWRA Jour- Double Day. nal of the American Water Resources Association 47 (6): 1183–96. doi:10.1111/j.1752-1688.2011.00589.x. 45 Steinschneider, S. and C. Brown. 2012. “Dynamic reservoir Suitable for Resource Management.” Eos, Transactions management with real-option risk hedging as a robust American Geophysical Union 94 (37): 321–23. adaptation to non-stationary climate.” Water Resources Re- doi:10.1002/2013EO370002. search 48 (5): W05524. UN Water. 2010. “Climate Change Adaptation: The Pivotal Steinschneider, S. and C. Brown. 2013. “A semiparametric Role of Water.” Policy Brief. multivariate, multi-site weather generator with low- US Department of Commerce, NOAA. 2013. “Trends in Car- frequency variability for use in climate risk assess- bon Dioxide.” Accessed July 19. ments.” Water Resources Research 49(11): 7205–20. http://www.esrl.noaa.gov/gmd/ccgg/trends/. Stern, N. Nicholas Herbert. 2007. The Economics of Climate Vonk, E., Y. P. Xu, M. J. Booij, X. Zhang, and D. C. M. Augus- Change: The Stern Review. Cambridge University Press. tijn. 2014. “Adapting Multireservoir Operation to Shift- http://books.google.com/books?hl=en&lr=&id=U-VmI ing Patterns of Water Supply and Demand.” Water Re- rGGZgAC&oi=fnd&pg=PA1&dq=stern+review&ots=9c sources Management 28 (3): 625–43. uY7wivqc&sig=dr3KDb5sQ5ayhYml9TXWmhG4i3M. Watts, R.J., B.D. Richter, J.J. Opperman, and K.H. Bowmer. Stocker, T.F., Q. Dahe, and G. Plattner. 2013. “Climate 2011. "Dam reoperation in an era of climate change." Ma- Change 2013: The Physical Science Basis.” Working rine and Freshwater Research 62: 321–27. Group I Contribution to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. Sum- Weaver, C.P., R.J. Lempert, C. Brown, J.A. Hall, D. Revell, mary for Policymakers (IPCC, 2013). and D. Sarewitz. 2013. “Improving the Contribution of http://www.climatechange2013.org/images/report/W Climate Model Information to Decision-Making: The G1AR5_Frontmatter_FINAL.pdf. Value and Demands of Robust Decision Frameworks.” Wiley Interdisciplinary Reviews: Climate Change 4 (1): 39– Tebaldi, C., and R. Knutti. 2007. “The Use of the Multi- 60. doi:10.1002/wcc.202. Model Ensemble in Probabilistic Climate Projections.” Philosophical Transactions of the Royal Society A: Mathemati- Weitzman, M. L. 2009. “On modeling and interpreting the cal, Physical and Engineering Sciences 365 (1857): 2053–75. economics of catastrophic climate change.” The Review of Economics and Statistics 91 (1): 1–19. Thrasher, B., J. Xiong, W. Wang, F. Melton, A. Michaelis, and R. Nemani. 2013. “Downscaled Climate Projections 46 Wilby, R.L, and S. Dessai. 2010. “Robust Adaptation to Cli- Woodward, M., Z. Kapelan, and B. Gouldby. 2014. “Adap- mate Change.” Weather 65 (7): 180–85. tive Flood Risk Management Under Climate Change Un- doi:10.1002/wea.543. certainty Using Real Options and Optimization.” Risk Analysis 34 (1): 75–92. Wilby, Robert L., and I. Harris. 2006. “A Framework for As- sessing Uncertainties in Climate Change Impacts: Low- Flow Scenarios for the River Thames, UK.” Water Re- sources Research 42 (2). Wilby, Robert L., and Rod Keenan. 2012. “Adapting to Flood Risk under Climate Change.” Progress in Physical Geography 36 (3): 348–78. doi:10.1177/0309133312438908. Wilks, D.S. 1992. “Adapting stochastic weather generation algorithms for climate change studies.” Climatic Change 22 (1): 67-84. Wilks, D.S. 2002. “Realizations of daily weather in forecast seasonal climate.” Journal of Hydrometeorology 3 (2): 195- 207. Wilks, D.S. and R.L. Wilby. 1999. “The weather generation game: a review of stochastic weather models.” Progress in Physical Geography 23 (3): 329–57. Woodward, M., B. Gouldby, Z. Kapelan, S. Khu, and I. Townend. 2011. “Real Options in flood risk management decision-making.” Journal of Flood Risk Management 4(4): 339–49. 47 ACKNOWLEDGMENTS This work is a product of the following members of the Alli- Coordinating and Production Work ance for Global Water Adaptation (AGWA) network: The Kara N. DiFrancesco (Oregon State University/WPP/ World Bank, Conservation International, the Institute of Wa- AGWA) ter Resources at the US Army Corps of Engineers, and the University of Massachusetts Amherst, with external contri- Primary Authors butions from many others. Kara N. DiFrancesco (Oregon State University/WPP/ It was made possible by financing contributions from AGWA) the Water Partnership Program (WPP) of the World Bank as well as the Institute of Water Resources at the US Army Patrick Ray (University of Massachusetts Amherst) Corps of Engineers and Conservation International. We ac- Additional Written, Audio and Video Contributions knowledge and appreciate their support. Casey Brown (University of Massachusetts Amherst) We are grateful to AGWA—as a network, source of vi- sion, and wellspring of insight—as well as to the many that Margot Hill Clarvis (University of Geneva) devoted their effort to managing, coordinating, writing, con- tributing, reviewing, editing, laying out, and supporting Torkil Jønck Clausen (DHI Water Policy) this effort. Julian Doczi (Overseas Development Institute) Managing and Editing Luis E. García (The World Bank) Luis E. García (The World Bank) Kristen Gilroy (US Army Corps of Engineers Institute John H. Matthews (AGWA Coordinator, Colorado State for Water Resources) University Water Center) Zbigniew Kundzewicz (Potsdam Institute for Climate Diego J. Rodriguez (The World Bank) Impact Research) Marcus Wijnen (The World Bank) 48 John H. Matthews (AGWA Coordinator, Colorado • Including Climate Change in Hydrologic Design State University Water Center) workshop, The World Bank, Washington, DC, USA, 21 November 2011, Guillermo Mendoza (US Army Corps of Engineers In- stitute for Water Resources) • HydroPredict 2012: Predictions for Hydrology, Ecol- ogy, and Water Resources Management, Session S3: Diego J. Rodriguez (The World Bank) Special Session on Choosing models for resilient wa- Rolf Olsen (US Army Corps of Engineers Institute for ter resources management, Vienna, Austria, 24-27 Water Resources) September 2012, LeRoy Poff (Colorado State University) • Sustainable Development Network Forum 2013: Wa- ter Learning Day 2, The World Bank, Washington, William Rex (The World Bank) DC, USA, 8 March 2013, and Eugene Stakhiv (US Army Corps of Engineers Institute • Climate informed water resources project design, for Water Resources) World Water Week, Stockholm, Sweden, 2 Septem- ber 2013. Kenneth Strzepeck (University of Colorado) Reviewers Juan Valdés (University of Arizona) Special thanks to our reviewers, who provided valu- Robert Wilby (Loughborough University) able guidance and suggestions for this report: These contributions were collected during a series of Víctor Vázquez Alvarez (The World Bank) World Bank and AGWA events as well as private interview sessions, including, but not limited to, the following: Jorge José Escurra (The World Bank) • Hydrologic Analysis to Inform Bank Policies and Ian Harrison (Conservation International) Projects: Bridging the Gap workshop, The World Bank, Washington, DC, USA, 24-25 November, Robert Wilby (Loughborough University) 2008. 49 Additional Support Brian Loo, Alex Mauroner, and Bunyod Holmatov (Conservation International) Gina Lizardi, Nansia Constantinou, and Linda Walker Adigwe (The World Bank) 50 ABOUT AGWA down and no-regret approaches to climate adaptation. We acknowledge the need for a new paradigm for sustainable water resources management, and recognize that the chal- lenge of climate adaptation requires the ability to bridge dis- ciplinary, institutional, political, and sectoral boundaries, to harvest the best practices and approaches, and to connect The Alliance for Global Water Adaptation (AGWA) was them into a coherent paradigm. As a network, AGWA has founded in August 2010 as a network of institutions fo- come together to fill the decision-making gap by making cused on how to develop effective, practical methods to in- contributions from multiple perspectives and disciplines, corporate the emerging best practices for climate adapta- strengthening collaborations, reducing duplications and tion. Our network spans a diverse array of multilateral insti- overlaps, and promoting coherence and effectiveness across tutions, governments, non-governmental bodies, and pri- institutions and sectors. Our fundamental goal is to provi- vate sector. The AGWA steering committee includes a wide sion tools, partnerships, and technical assistance to improve range of institutions and individuals: Casey Brown (Univer- operational decision-making, governance, and analytical sity of Massachusetts), Christine Chan (a consultant based processes in water resources management, with a focus on in Hong Kong), Joppe Cramwinckel (World Business Coun- the scales relevant to climate adaptation and climate cil for Sustainable Development), Paul Fleming (Seattle Pub- change. lic Utilities), Rebecca Tharme (The Nature Conservancy), Cees van de Guchte (Deltares), Karin Lexén (Stockholm In- AGWA is especially interested in supporting resilient ternational Water Institute), Robert Pietrowsky (U.S. Army water management in the data-poor regions of the develop- Corps of Engineers Institute for Water Resources), and Di- ing world. ego Rodriguez (World Bank). AGWA is co-chaired by the World Bank and SIWI, and the secretariat is funded by How can I join AGWA? SIWI and led by John H. Matthews. AGWA welcomes new members. We have a flexible charter and governance system. To discuss membership, Philosophically, AGWA has arisen as a result of dissatis- participation, and consultation, we ask that you contact us faction with the past decade of experimentation with top- via the AGWA site: http://alliance4water.org. 51