Climate-Change Scenarios for Water Planning Studies Pilot Applications in the Pacific Northwest

Transcription

1 Climate-Change Scenarios for Water Planning Studies Pilot Applications in the Pacific Northwest BY AMY K. SNOVER, ALAN F. HAMLET, AND DENNIS P. LETTENMAIER AFFILIATIONS: SNOVER Climate Impacts Group (CIG), Center for Science in the Earth System, Joint Institute for the Study of the Atmosphere and Ocean, University of Washington; HAMLET AND LETTENMAIER CIG/Department of Civil and Environmental Engineering, University of Washington, Seattle, Washington CORRESPONDING AUTHOR: Amy Snover, Ph.D., Climate Impacts Group, Center for Science in the Earth System, Joint Institute for the Study of the Atmosphere and Ocean, University of Washington, Box , Seattle, WA American Meteorological Society Despite uncertainties in predictions of the magnitude and timing of anthropogenic climate change, many Pacific Northwest (PNW) water managers now recognize climate change as a significant issue that should be addressed in water resources planning. This represents a significant change over the last several years. These managers are requesting detailed information about potential climate impacts on their systems, in a form suitable for inclusion in planning. Academic climate-change assessments, which have been available since the mid-1980s, often produce information that is inconsistent with the specific periods of the historic streamflow record and/or the internally developed water management models used by individual water management agencies in formal water planning studies. While water managers may find an academic assessment of a system s sensitivity to climate change (produced by coupling academically developed climate, hydrologic, and water management models) to be perfectly believable in the abstract, they are unlikely to consider such an assessment directly relevant to the decisions they make about the system they manage. This is one reason that water-planning studies have rarely addressed the implications of climate change in any more than a very cursory manner. We therefore believe that water management agencies would be more apt to incorporate climate change in their planning in the short term if they had access to scenarios of future streamflow that could easily and inexpensively be used within the confines of their existing planning frameworks. A few water management agencies in the PNW have taken the lead in incorporating climate change in their planning efforts. The Portland (Oregon) Water Bureau, for example, recently completed a study in partnership with the University of Washington s Climate Impacts Group (CIG) that evaluated several long-range planning alternatives under four different scenarios of future climate. Seattle Public Utilities (which manages Seattle s water supply) has begun a similar planning process. This paper describes how we are working to expand the availability of climate-change streamflow scenarios that would allow more planners to assess the impacts of climate change on PNW water systems. In the pilot project described here, streamflow scenarios are being generated to support two large-scale PNW water-planning studies. EXPECTED CLIMATE CHANGE IN THE PNW. Regional climate-change projections are uncertain. However, the magnitude of projected warming combined with a strong regional reliance on mountain snowpack creates some consistency in the implications of climate change for PNW water resources. Global climate models project mid-twenty-firstcentury increases in PNW temperature that are well outside the range of observed twentieth-century climate. Most models suggest modest changes in annual average precipitation, but wetter winters and drier summers (Fig. 1). Even with wetter winters, warmer temperatures would elevate the typical winter snowline, decreasing the snow-covered area in the mountains and total winter snowpack. For the 2040s, for example, we expect to see strong reductions in snow water equivalent in the southern part of the Columbia River basin AMERICAN METEOROLOGICAL SOCIETY NOVEMBER

2 FIG. 1. Decadal average changes in PNW temperature (top) and precipitation (bottom) projected by four global climate model simulations for the decades of the 2020s (left) and the 2040s (right). Temperature changes are shown as absolute increases relative to each model s control climate; precipitation changes are shown as fractional changes. The red line indicates the monthly mean of the four GCM perturbations. and at moderate elevations in the Cascade Mountains, areas that have significant spring snowpack in the current climate. Some of the most sensitive areas (like the Cascades in southern Oregon) could be essentially snow free in the future (Fig. 2, upper panels). Warmer winters would cause more precipitation to fall as rain, increasing winter streamflow. Both spring peak flows and summer and fall streamflow would decrease. A greater fraction of the snow would melt early in the season, moving spring peak flows earlier in the year and increasing the time between snowmelt and fall rains (Fig. 2, lower panels). These changes would result in less summer streamflow and increased frequency of significant low-flow events, even assuming increases in winter precipitation. In the transient snow zone, where midwinter temperatures are currently close to freezing, streamflow timing would shift even more for the same warming (Fig. 2, lower right). Because PNW water systems have relatively little reservoir storage (typically 10% 30% of annual flow), they are quite sensitive to changes in streamflow timing. In broad terms, these scenarios suggest that for watersheds at moderate elevations in the Cascade Mountains and in the southern interior of the Columbia River basin (e.g., the Snake River basin), significant impacts on water resources are likely to appear within the next few decades. In large snow melt dominated systems, like the northern headwaters of the Columbia River, impacts may not become apparent for several more decades. Given the long lead time required to change water resources infrastructure and policy, however, we argue that PNW policy makers and water management agencies need to begin to assess these potential impacts in their planning as soon as possible. PLANNING FOR CLIMATE CHANGE IN THE PNW: WORKING WITH REGIONAL STAKEHOLDERS. The first comprehensive examination of the potential consequences of climate change for the PNW, conducted by CIG in 1997, demonstrated that PNW managers and policymakers were largely unaware of the potential impacts and unprepared to respond to information about them. It was clear that continued outreach and a much more focused description of potential impacts and adaptation strategies were needed. As a result, CIG continued research into regionally specific climate impacts and provided this infor NOVEMBER 2003

3 FIG. 2. Upper panels: Simulated PNW Apr 1 average snow water equivalent (above a 10-mm threshold) for the current climate ( , left) and for the 2020s (middle) and 2040s (right) under the mean climate-change scenario shown in Fig. 1. Black cells indicate a snow water equivalent greater than 4000 mm. Note that under the warmer climate, the snowpack at moderate elevations and in the southern part of the Columbia River basin is reduced for the warming scenarios. Lower panels: (left) Simulated average natural streamflow in the Columbia River at The Dalles, OR, and (right) inflows to Chester Morse Lake, WA, for the current climate (black), the 2020s (green), and the 2040s (red) under the mean climate-change scenario shown in Fig. 1. Streamflow at The Dalles integrates the runoff from the entire Columbia basin, whereas Chester Morse Lake is fed by a small watershed in the Cascades at an elevation of about 1000 m. (Figure of Chester Morse Lake inflow courtesy of Richard Palmer, CIG and Civil and Environmental Engineering, University of Washington.) mation to local, state, and regional natural resource agencies, media groups, and elected officials. In July 2001, CIG convened a climate-change and waterpolicy workshop for high-level PNW resource managers, policy makers, and water users. Despite the persisting scientific uncertainties, stakeholders in 2001 recognized climate change as a potentially significant threat to regional water resources. Many stated that climate-change information would be critical to future planning. Few agencies, however, possessed the technical and financial resources to develop their own hydrologic scenarios for planning. Workshop participants requested that CIG provide suggestions for integrating climate-change information into existing planning initiatives. They wanted information about what kinds of climate-change information could be used in existing planning/operating models, about climate impacts in finer spatial and temporal detail, and about the water quality and temperature implications of future minimum flows. They also asked for sample case studies showing how climate-change projections could be incorporated into basin planning. Several participants commented that more detailed information at the catchment, wa- AMERICAN METEOROLOGICAL SOCIETY NOVEMBER

4 tershed, or sub-basin level was a condition for using climate-change information; others stated that they would only use climate information if it s easy to apply to the problem at hand. A SHORT-TERM STRATEGY FOR PROVID- ING CLIMATE-CHANGE INFORMATION. In response to these requests, CIG is working to produce climate-change streamflow scenarios that will be useful for existing water planning efforts. Long-range water planning in the United States is frequently based on critical period analysis (i.e., an evaluation of system performance under the most adverse streamflow conditions in the historic record). Within this planning framework, water managers tend to base their assessment of planning options on an evaluation of how the system would respond to specific historic events [e.g., How would the system respond to the drought of the 1930s with the (current; alternate) reservoir operating policy? ]. These analyses typically use a specific period of historical streamflow (naturalized to remove effects of reservoirs and other management) and some combination of: (a) a water management model representing the system of dams, reservoirs, and other infrastructure; (b) present and projected future water demand; and (c) the water management policies and/or proposed system changes under evaluation. Whatever the overall merits or deficiencies of critical period planning techniques, in the context of anticipated climate change the use of an historic streamflow record to represent future hydrologic variability is an obvious weakness. Climate related changes in PNW streamflow timing and reductions in summer water availability, for example, are likely to create specific challenges to water management that would be overlooked if the observed streamflow record is used for planning. To address this issue, we have focused on producing alternate (climate change-perturbed) streamflow scenarios that can be used in place of the historic record in existing planning processes. SELECTION OF CLIMATE SCENARIOS AND LINKAGE TO HYDROLOGIC MODELS. In the pilot project described here, we used projected changes in PNW climate simulated by four global climate system models (CMs): HadCM2, HadCM3, ECHAM4, and PCM3 (see Fig. 1). These models were selected primarily because of their higher spatial resolution for the atmospheric component and their use of relatively sophisticated land surface schemes. All simulations incorporated increasing climate forcing by carbon dioxide (IS92a or ~1% per year) and anthropogenic sulfate aerosols (IS92a). To construct quantitative hydrologic scenarios, we use a simple downscaling technique frequently referred to as the delta method. This technique retains the fundamental temporal and spatial variability of observed regional climate but adjusts the observed daily climate records in each month by the projected changes in monthly mean precipitation and temperature from the CM scenarios. This perturbed climate record is then used as the input to a hydrologic model. Although scenarios constructed using the delta method intentionally sidestep some important research questions, we use this simple method, rather than a more sophisticated downscaling technique, for several reasons. First, in the short term, streamflow scenarios that retain the chronological sequencing of observed flows are a good fit with planning studies that rely on an evaluation of how a system would respond to specific historic events. Streamflow scenarios constructed in this manner allow planners to compare the performance of each planning alternative for the historic record and various climate-change scenarios. In other words, planners can explore the question: What would water year X look like for our water system if the average climate changed as these climate model simulations suggest? This process is useful in that it places potential changes in streamflow in the context of the historic record. Second, despite advancements in the performance of CMs, the fundamental uncertainty in large-scale climate-change simulations (in particular, differences in regional precipitation between different CM scenarios) has a much larger effect on hydrologic assessments than does choice of downscaling technique. These relatively simple climate-change assessments are intended to provide a starting point for incorporating information about climate change and climate uncertainty into existing water planning processes. As planners gain more experience with incorporating climate uncertainty in their planning, it is our hope that they will begin to utilize more sophisticated assessment techniques. CLIMATE-CHANGE STREAMFLOW SCE- NARIOS. Forty-year time series of monthly streamflows for the historic climate and for the projected changes in monthly average temperature and precipitation were simulated using the Variable Infil NOVEMBER 2003

5 FIG. 3. A pilot Web-based decision support system for water resources management is currently available for 16 locations in the Columbia River basin (left) at CR_cc.htm. Climate-change streamflow time series (i.e., adjusted realizations of the historic streamflow record) are generated by bias-correcting simulations from a physically based hydrologic model driven by scenarios of projected climate change (see text). The water resources manager can retrieve the adjusted streamflow data both graphically (center; for the time series mean) and in tabular form (right; for the entire 40-yr time series). tration Capacity (VIC) hydrologic model implemented over the Columbia River basin and coastal areas in the PNW at 1/8-degree resolution (see link to technical documentation below). Hydrologic simulations invariably show some bias compared to observations. In planning studies, where model results of system performance derived from the historic record of streamflows are directly compared to results derived from alternate hydrologic simulations, errors in the mean for particular months or significant errors in particular parts of the time series can make this comparison difficult. In order to minimize these problems, we remove the bias from the simulated ( raw ) streamflow time series, using a probability-mapping technique (see Lettenmaier/permanent_archive/hamleaf/bams_paper/ technical_documentation.pdf). This technique removes systematic bias in the monthly time-step hydrologic simulations by mapping between a simulated and observed probability distribution for each calendar month. The procedure involves careful treatment of values outside the range of historical simulations and insures that mass balance in the channel system (and across seasons) is preserved. The resulting climate-change streamflow scenarios are numerically consistent with, and for purposes of evaluating system performance directly comparable to, the observed streamflow time series. FROM ACADEMIA TO AGENCY. Climatechange streamflow scenarios created using these methods are currently available on the Web for 16 streamflow locations in the Columbia River basin for the 2020s and 2040s (Fig. 3). The scenarios are based on a consensus climate-change scenario (the average of the scenarios from the four climate models, Fig. 1) and cover the period , when the hydrologic simulations and the historic record overlap (observed streamflows are currently available for ). The number of regional climate-change scenarios, streamflow locations, and years for which data are provided will all be expanded in ongoing work. In order to (1) ensure that the climate-change streamflow scenarios are useful to regional water planners, and (2) support the development of case studies AMERICAN METEOROLOGICAL SOCIETY NOVEMBER

6 illustrating the use of climate-change information in basin planning, the scenarios are being tailored for use in regional-scale planning studies at the Northwest Power Planning Council (focused primarily on regional hydropower resources) and Idaho s Department of Water Resources (focused primarily on irrigation in the Snake River basin). The streamflow scenarios are also freely available on the Web. We anticipate that others in the region will be able to use these scenarios, since many water management agencies use the same period of historical streamflow data at the same locations in their planning. The methods used for producing the streamflow scenarios in this project are quite portable. For example, CIG is working closely with the Washington State Department of Ecology s statewide Watershed Planning Program to help planners and stakeholders understand and quantify the implications of climate change for their watersheds. Although these watersheds are generally much smaller than the Columbia basin (which implies larger uncertainties in projected streamflow changes), the same techniques for scenario development can be applied. Although we use the delta method here, these data processing techniques can be used with any CM downscaling approach. CONCLUSIONS. Climate change is projected to have important impacts on PNW water resources within the next several decades, including decreased snowpack, increased winter runoff, decreased spring and summer runoff, and increasing competition among water users. Initial efforts to bring this message to water resources managers generally involved (1) describing the scientific (i.e., academic) understanding of the water resources management system and its perceived weakness to projected climate changes and (2) exhorting managers to decrease their potential vulnerability by including climate-change information in their planning. This type of outreach has had noticeable success. The potential impacts of expected changes in climate and streamflow are now reasonably well recognized in the PNW water management community, and the dialogue has recently shifted away from whether climate change will affect the PNW toward how best to include climatechange information in long-term planning. Without additional effort on the part of the academic community to translate information about regional climate change into a form compatible with the tools and objectives of the water resources management community, however, water planning studies are unlikely to address the implications of climate change in any more than a very cursory manner in the near term. The alternate climatechange streamflow scenarios described here can easily and inexpensively be incorporated into existing planning methods in place of the historic record typically used, enabling water resource managers to evaluate their system s vulnerability to climate change. This simple method is intended to inject climate-change information into existing planning processes as rapidly as possible, and to serve as the first step toward more sophisticated vulnerability assessments. ACKNOWLEDGMENTS. This publication is funded by the Joint Institute for the Study of the Atmosphere and Ocean (JISAO) under NOAA Cooperative Agreement No. NA67RJ0155. Contribution #965. REFERENCES. Selected references for this paper are permanently archived at: Lettenmaier/permanent_archive/hamleaf/bams_paper/ refs.pdf NOVEMBER 2003