Joseph André Ball. A thesis submitted in partial fulfillment of the requirements for the degree of. Master of Science in Civil Engineering

Transcription

1 Impacts of Climate Change on the Proposed Lake Tapps - White River Water Supply Joseph André Ball A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science in Civil Engineering University of Washington 2004 Program Authorized to Offer Degree: Department of Civil and Environmental Engineering

2 University of Washington Graduate School This is to certify that I have examined this copy of a master s thesis by Joseph André Ball and have found that it is complete and satisfactory in all respects, and that any and all revisions required by the final examining committee have been made. Committee Members: Richard N. Palmer Stephen J. Burges Date:

3 In presenting this thesis in partial fulfillment of the requirements for a Master s degree at the University of Washington, I agree that the Library shall make its copies freely available for inspection. I further agree that extensive copying of this thesis is allowable only for scholarly purposes, consistent with fair use as prescribed in the U.S. Copyright Law. Any other reproduction for any purposes or by any means shall not be allowed without my written permission. Signature Date

4 University of Washington Abstract Impacts of Climate Change on the Proposed Lake Tapps - White River Water Supply Joseph André Ball Chair of Supervisory Committee: Professor Richard N. Palmer Municipal water supplies in the Central Puget Sound Region are projected to be insufficient to meet demands in the near future. Plans are in place to develop Lake Tapps and the White River as a water resource which will provide additional municipal supply for the region. This research investigate the impacts that climate change is projected to have on the water resource and its stakeholders. Climate projections from multiple general circulation models were downscaled, applied to a hydrology model to yield streamflows which were input to a systems model of the water resource. Climate change is projected to alter the hydrology of the White River Basin, resulting in higher winter streamflows, earlier spring run off and lower summer streamflows. Changes to streamflow will reduce the reliability with which White River minimum instream flows (MIFs) for fish can be met. Additional supply will need to be released from Lake Tapps to meet Puyallup MIFs. The potential safe yield of the system is expected to decrease but will not fall below the water right until near Summertime recreational lake level targets will likely be met with reduced reliability as water is released for MIFs and withdrawn for municipal uses.

5 Table of Contents List of Figures... ii List of Tables... iii Section 1: Introduction...1 Section 2: Problem Statement Study Area Description Regional Stakeholders...6 Section 3: Experimental Design GCM Climate signals GCM Model Attributes GCM Ensemble Model Descriptions Emission Scenario Description Downscaling Climate Data DHSVM Hydrology Model System Simulation Model...19 Section 4: Results and Discussion Impacts to Meteorology Impacts to Temperature Impacts to Precipitation Impacts to Streamflow Historical Streamflow Characteristics Impacts to Annual White River Streamflows Distribution of Summer Flows Impacts to Water Resource System Lake Tapps Safe Yield Bypass Minimum Instream Flow Reliability Municipal Demand Reliability Recreational Lake Level...40 Section 5: Conclusions...44 Bibliography...46 i

6 List of Figures Figure Number Page 2-1 Map of White River Basin White & Puyallup Rivers and Lake Tapps Reservoir Schematic Plan. (Reproduced from ROE Lake Tapps Reservoir Water Supply Application) Graphic representation of primary downscaling steps Sixty-one year ( ) average annual hydrograph with quartile distributions comparing USGS flows to simulated DHSVM flows, three week average applied Maximum and minimum operation rule curves for Lake Tapps Historic Monthly Temperature and Precipitation for Buckley Meteorological Station, 61-year average data Projected average monthly temperature at Buckley met station Projected average monthly temperature change for Buckley met station Projected average monthly precipitation at Buckley met station Projected change in average monthly precipitation at Buckley met station Average annual hydrograph for White River near Buckley, 3-week running average applied to , USGS data Comparison of 1990 s average annual hydrograph to full record for White River near Buckley. 3-week running average applied to USGS data Predicted Average Weekly White River Streamflows above Buckley diversion for Change in percent annual flow, White River above Buckley Diversion, streamflows compared to modeled historic streamflows Non-exceedance probability of average daily summer (June, July, August) inflows to Lake Tapps Projected trend of climate impacted safe yield for the Lake Tapps water supply, showing ensemble average decline of 1.8 MGD per decade. (adapted from Wiley 2004) Proposed minimum instream flows for White River bypass reach White River bypass reach minimum instream flow annual reliability using historic and climate projected flows Municipal demand reliability Projected average annual, peak season, and peak monthly demand from Lake Tapps compared to systems capacity and water right constraints Minimum recreational lake level reliability...42 ii

7 Table Number List of Tables Page 4-1 System model options for safe yield analysis System model options for municipal reliability analysis System model options for recreation lake level reliability analysis Average lake elevations, average recreation level failure durations and average number of recreation failure years with no municipal demands Average lake elevations, average recreation level failure durations and average number of recreation failure years with 2054 municipal demands...43 iii

8 1 Section 1: Introduction If water from the White River and Lake Tapps is used for municipal supply in 2025 it will be the first significant new surface water source in the central Puget Sound region since the South Fork Tolt was brought online in The central Puget Sound region includes Snohomish, King and Pierce counties which is home to 3 million people, nearly half the population of Washington State. Regional water planners have determined that the current water supply will not be sufficient to meet future demand in the region (Forum, 2001). This has prompted the exploration of different supply possibilities; such as conservation, water reuse, source interties, and new surface or groundwater sources. Initial steps have been taken to promote Lake Tapps as part of the solution to meeting future demands in the Puget Sound region. Water resource planning must consider long planning horizons from 25 to 50 years. It has become necessary to include the impacts of potential climate change and climate variability on water system reliability. Global climate change is forecasted to influence temperature and precipitation, in turn impacting water supply and demand. Regional responses will vary but mountainous watersheds similar to the White River are predicted to experience diminished snowfall and earlier melting of snowpack, with larger spring runoff occurring earlier in the year (Frederick and Gleick, 1999; Palmer and Hahn 2003; Wiley 2004). Planning for the use of Lake Tapps as a water resource has addressed the needs of humans and fish but failed to incorporate the quantifiable impacts that climate change may also have on the region. This report investigates the potential impacts that climate change will have on the White River basin and the proposed Lake Tapps water supply. Section 2 of this

9 2 report provides the thesis background, describing the White River basin and then discussing the stakeholders who have prominent interests in the future of the water resource provided by the White River. Section 3 outlines the modeling methods used to investigate climate change and its impact on the White River water resources and stakeholders interested in the water. Section 4 presents and discusses the results of the methods described in section three. The fifth and final section offers conclusions to the research.

10 3 Section 2: Problem Statement 2.1 Study Area Description The White River watershed is approximately 495 mi 2, making it one of the major rivers discharging into the southern Puget Sound. The White River extends from the summit of Mt Rainier (14,410 ft) to the Puget Sound. The White joins with the Puyallup River for roughly 10 miles before discharging into Commencement Bay. Within the basin there are two surface water impoundments, Lake Tapps and Mud Mountain Dam (Figure 2-1). Water is diverted from the White River to Lake Tapps. Lake Tapps is a manmade lake formed by dikes and embankments which were constructed around four small natural lakes to create one body of water. A diversion dam on river mile (RM) 24.3 of the White diverts water into a canal and then a pipeline before entering the lake. Water exits Lake Tapps through a hydropower facility and then returns to the White River at RM 3.0. Some water flows past the diversion dam and remains in the original river bed. This 21-mile long segment of river between the diversion dam and the return is referred to as the bypass reach (Figure 2-2). Lake Tapps was built to create storage for the White River Hydroelectric Project which came online in 1912 and suspended operation as of January 2004 Mud Mountain Dam is located upstream of the diversion to Lake Tapps on RM The US Army Corps of Engineers (Corps) operates Mud Mountain as a single-use, flood protection project. Large flows are impounded by the dam to prevent flood damage downstream and are released at a controlled rate. During non-flood operation, lower flows are bypassed through the dam.

11 4 Figure 2-1 Map of White River Basin. 4

12 5 Figure 2-2 White & Puyallup Rivers and Lake Tapps Reservoir Schematic Plan. (Reproduced from ROE Lake Tapps Reservoir Water Supply Application) 5

13 The White River is a transient river basin, dependent on rain, snow, and glacial melt for streamflow. Snowpack stores precipitation during the winter and releases it as runoff in early spring and summer. It is during this spring runoff period that reservoirs refill, storing water that is draw down during summer and fall to meet water demands. Water resources in the Puget Sound are dependent on annual refilling of reservoirs. Lake Tapps has a usable capacity of 46,600 acre-feet (AF), (57 million cubic meters or 15,300 million gallons). Above the diversion to Lake Tapps the average annual flow volume is 1,040,000 AF. The active storage is only able to store 4.5% of the mean annual flow. Thus the dam has little impact on monthly average flows below the dam during the high precipitation months of fall, winter and spring, and only significantly impacts streamflow during the summer. For comparison, Seattle Public Utilities operates two reservoirs with a combined storage of 89,000 AF, which represents 34% of their mean annual inflow. The U.S. Army Corps of Engineers Howard Hansen project is capable of storing 106,000 AF, 15% of the mean annual inflow 2.2 Regional Stakeholders The potential use of Lake Tapps as a municipal water supply has interested many stakeholders in the region. Cascade Water Alliance (Cascade) represents eight cities and municipalities which have historically obtained wholesale water from Seattle Public Utilities (SPU) and groundwater sources. Cascade was formed in 1999 to provide water for their customers independent of wholesale contracts. Cascade s 2002/2003 three stage plan of action is summarized below (Cascade, 2002): Enter a 50 year Declining Block Contract with SPU to reduce wholesale demand by 5 million gallons per day (MGD) every 5 years Purchase water from the Tacoma Second Supply Project Partners to meet interim demands 6

14 7 Develop Lake Tapps Reservoir as a municipal water supply source completing the first stage in 2024 and the second stage in 2034 A contract formalizing the declining block contract between Cascade and SPU was completed in December 2003 (SPU, 2003). Tacoma s Second Supply Project is almost complete and negotiations are under way for Cascade to purchase a portion of this water as interim supply. Prior to the cessation of the White River Hydroelectric project in 2004, Puget Sound Energy (PSE) applied for three interrelated water rights from the Department of Ecology (Ecology) to divert water from Lake Tapps for municipal consumption. As part of the application, a Report of Examination (ROE) was submitted which outlined the impacts of the intended water use. The three water rights were approved by the Department of Ecology in June These water rights were obtained for the purpose of being sold to Cascade at a later time. Several parties, including local tribes, have contested these water rights and an appeal was filed with Washington State s Pollution Control Hearings Board (PCHB) which has authority to review water rights issued by Ecology. In August 2004 PCHB issued an order determining that the original ROE approved by Ecology was insufficient to grant the water rights because the benefits and harms of a non-hydropower use were not evaluated. An updated ROE must address the nonhydropower use scenario and then must be approved by Ecology (PCHB, 2004). Lake Tapps is a popular recreation site for boating and swimming. Parks and boat launches offer public access to the lake while hundreds of homeowners have personal docks. The Save Lake Tapps Coalition (Coalition) represents the longstanding homeowners associations in the Lake Tapps vicinity. The Coalition was formed in 1999 when the lake s future was uncertain. At that time it appeared that PSE would soon discontinue operation of the hydropower project. This might have caused the lake to be emptied, ending recreation uses and decreasing property values. The Coalition has worked with PSE to explore possible options to preserve

15 8 Lake Tapps. In April of 2004 a legal agreement was signed between PSE and the Coalition which sought to protect the lake s future existence. If neither PSE or Cascade operates Lake Tapps, the Coalition will be provided with an opportunity to purchase the infrastructure and water rights so that the lake could be maintained. The contract also stated that the target elevation would be full pool level from April 15 th through October 31 st. Fish species, especially salmon, are an important natural resource in the Puget Sound Region. The White River and the Puyallup River have salmon habitats which are impacted by the operation of Lake Tapps. The proposed Lake Tapps water supply plan addresses the impacts to fish on both rivers, though some stakeholders are not satisfied with the proposal. Minimum instream flows have been established for the Puyallup River below the confluence of the White/Puyallup. Minimum instream flows for the bypass reach of the White River are likely to be increased in light of biological review by fishery agencies. Advocates for salmon include federal agencies and local tribes including the Muckleshoot and the Puyallup Indian Tribes. Operating Lake Tapps to meet the needs of multiple stakeholders requires extensive planning to maximize the regional benefits. For water supply purposes it is desirable to keep the lake full at all times and to divert as much water as possible through the lake to maintain high water quality. Keeping the lake full assures that the maximum amount of storage is available during any drawdown periods. Drawdown periods occur when the withdrawal exceeds the inflow. This occurs typically in summer and early fall. Recreational and aesthetic interests also benefit the most when the lake is full of clean, fresh water. The conflict with municipal operation is that the peak summer recreational season coincides with the drawdown period. For fish in the White River, it is desirable to have a natural flow regime. This conflicts with diversions to Lake Tapps. Minimum instream flows measured at Puyallup could benefit from additional releases from Lake Tapps during low flow

16 9 conditions since the point is downstream of Lake Tapps. However, these releases would diminish the amount of water in the lake for recreation and municipal supply. Appropriate planning is needed to assure that enough water is set aside for fish while the remaining water is managed in such a way to meet the needs of human users.

17 10 Section 3: Experimental Design This report evaluates the potential impacts of climate change on Lake Tapps. The assessment follows the basic format of other climate change impact assessments (Frederick and Gleick, 1999; Palmer and Hahn, 2003; Wiley, 2004). The application of this approach can be summarized by the following four steps: Obtain GCM climate signal output, Rescale climate signals to a regional scale with finer temporal resolution, Apply downscaled climate signals to a hydrology model to develop streamflows, and Apply streamflows to a systems simulation model. Evaluation of climate impacts on water demand was the only omission from the original impact assessment method. Instead impacts to water demand were explored through sensitivity analysis. 3.1 GCM Climate signals General circulation models (GCMs) offer an insight into the potential impacts of climate change. These complex models are designed to simulate Earth s natural processes and the impacts of anthropogenic climate change. GCMs are driven by assumptions of emissions and concentrations of atmospheric greenhouse gasses. The models maintain energy and mass balances and result in global temperature and precipitation projections. GCMs require significant computational resources. The models are maintained by a small number of large governmental research institutes, but the derived results are readily available to the scientific community. GCMs vary in their projections due to model characteristics such as resolution and components as well as driving assumptions (Covey, 2003). To produce an ensemble of climate change projections for this research, four GCMs were chosen

18 based on their ability to replicate historic streamflows when used in a similar research project involving the City of Seattle water supply (Wiley, 2004): ECHAM4 (Max Plank Institute) GFDL_R30 (Geophysical Fluid Dynamics Laboratory) HadCM3 (Hadley Centre for Climate Prediction and Research) DOE PCM (Department of Energy) All of the models were driven with same emission scenario, A2, which specifies assumptions of population, economy and technology resulting quantified values of greenhouse gases GCM Model Attributes GCM results vary due to model attributes, including their components, resolution, flux-adjustment, and emission scenario forcing. Components refer to the individual processes modeled by smaller models with in a given GCM. Current GCMs are referred to as coupled models because they are comprised of linked components which model physical processes such as the atmosphere, oceans, land surfaces and sea ice. Atmospheric and ocean components are represented as grid cells in all GCMs while the representation of land surfaces and sea ice vary more. Land surface hydrology may be modeled in one of four ways, each increasing in complexity. The most basic approach is bucket hydrology where each grid area has a given water capacity which accumulates water from precipitation and loses it through evaporation. Once the water capacity is reached runoff results. The complex hydrology scheme uses multiple grid cell layers for moisture and temperature. Sea ice plays an important role in climate dynamics and has been included in all of the models used in this study. Spatial resolution defines the unit grid cells which represent the Earth s surface. It is expressed as degrees latitude x longitude. It is assumed that greater spatial resolution results in improved accuracy but computational efficiency is reduced. Different components may have different resolutions within a GCM. The four 11

19 models used in this study have a range of 2.25 to 3.75 resolution for the atmosphere and 0.67 to 2.28 for the ocean. Flux adjustment is a remnant of early GCMs where ad hoc correction factors were included to reproduced observed climate. Flux adjustment has been discouraged since it does not represent any physical process and violates conservation of mass and energy. Some current GCMs do not use flux adjustment though others still do. Forcing scenarios define atmospheric gas concentrations which influence input of solar radiation and outgoing long-wave radiation. Forcing scenarios are representative of assumptions about the political economic, and technological state of the world and thus the human production of greenhouse gases GCM Ensemble Model Descriptions The four GCMs used in this research are among the most respected and widely used. The ECHAM4 model was developed by the Max Planck Institute in Hamburg, Germany. It is a modification of an existing weather forecast model created by the European Center for Medium Range Weather Forecast, Hamburg. ECHAM4 s grid resolution is 2.8 x 2.8 with 19 atmospheric layers and 11 ocean layers. Land surface hydrology is modeled using a complex scheme. ECHAM4 includes heat and water flux adjustments (Roeckner et al, 1996). The GFDL_R30 model was developed by the Geophysical Fluid Dynamics Lab located in Princeton, NJ. This model was developed to study climate variability and change. The model has high spatial resolution but retains simple descriptive physics so that it its computationally efficient. The atmospheric component has 14 layers with 3.75 x 2.25 resolution and the ocean component has 18 layers with x 2.25 resolution. This model uses the basic hydrology method. Flux adjustment is applied at the ocean-atmosphere interface to minimize drift in the model (Delworth et al, 2002). The HadCM3 models is the third in a series of models developed by the Hadley Center in the United Kingdom. HadCM3 is for both long climate projections and 12

20 weather forecasting. The atmospheric component resolution is 2.5 x 3.75 with 19 layers and the ocean component resolution is 1.25 x 1.25 with 20 layers. This model uses a complex hydrology modeling scheme. No flux adjustment is use in the HadCM3 model (Gordon et al, 2000). The Parallel Climate Model (PCM) model was developed by the US Department of Energy. It couples atmospheric and land surface models developed by National Center For Atmospheric Research (NCAR) with the DOE Parallel Ocean Program (POP) and a sea-ice model created by the Naval Postgraduate School. The atmospheric resolution is 2.8 x 2.8 with 18 layers and the ocean resolution is 0.67 x 0.67 with 32 layers. Land surface hydrology is represented by the complex scheme. The sea ice component is a thermodynamic model with ice rheology included. No flux adjustment is used in this GCM (Washington et al, 2000) Emission Scenario Description The Special Report on Emission Scenarios (SRES) included in the third IPCC report recommends six specific emission scenarios to be used to drive GCM simulations (SRES, 2000). These scenarios offer unique atmospheric gas concentrations based on a variety of assumptions about global political, economical and technological states. The scenarios do not have probabilities associated with them. For a regional study using one emission scenario can be done so that a particular political/economical state is emphasized. In this case the A2 emission scenario was chosen to drive all of the GCMs. Since the input was identical for all of the GCMs, internal model assumptions were what produced variation in the projections. The A2 emission scenario depicts a heterogeneous world which has distinct regional identities. Fertility, technological advancement and economy change at different rates across the globe. Different global regions rely on differing sources of energy thus some areas have high carbon emission depending upon energy source. 13

21 Anthropogenic CO 2 emissions remain high through 2100 due to the continued use of coal as an energy source in the A2 scenario. World population increases at a constant rate and is estimated at 15 billion by 2100, the greatest of the six emission scenarios but lower than the 17 billion projected by the United Nations. Overall the global economy is also one of the slowest growing with developed countries growing disproportionately faster than undeveloped countries (SRES, 2000). 3.2 Downscaling Climate Data Downscaling interprets large scale regional climate projections and disaggregates their values so they may be applied to watersheds at appropriate spatial and temporal resolution. GCM output variables are available as monthly temperature and precipitation values for a given grid cell. These values are too coarse both spatially and temporally to be directly applied to a regional hydrology model. Statistical downscaling is used to resolve the discrepancy in scales between GCMs and regional hydrology. The downscaling process utilized in this research consists of three primary steps (Wood and Mauer, 2002). First, GCM projections are modified so that they become scenarios which represent the entire range of historic regional climate variability. Second, the future regional climate scenarios are bias corrected and spatially downscaled into monthly temperature and precipitation records for multiple meteorological stations. Lastly, the monthly station climate scenarios are downscaled to daily scenarios. The spatial and temporal downscaling steps are graphically represented in Figure

22 Figure 3-1 Graphic representation of primary downscaling steps. 15

23 16 The first step of the downscaling process for this study utilizes a 75-year record of observed monthly meteorological observations for the region. Twenty-one year segments of GCM projections were used to create a steady-state representation of decadal climate projections with the full range of observed variability. Regional historic climate data was used from the International Research Institute for Climate Prediction (IRI) regional climate data set. The IRI data are observation based and are available as monthly averages and anomalies for temperature (Jones et al. 1999, Jones and Moberg 2003) and precipitation (Hulme 1992, 1994). The distribution properties of the regional historic climate are then obtained by plotting cumulative probability distribution functions (CDFs) of both monthly temperature and precipitation averages using the Cunnane plotting position, assuming an unknown distribution. i 0.40 q i =, Equation 3-1 n where i is the i th largest value of n values. Distributional properties for the GCM future climate projections are calculated by first selecting 21 years of monthly records centered on the period of investigation and then plotting a CDF using the Cunnane plotting position. Quantile mapping is then used to produce a future climate scenario by modifying the historic regional record so that it has the distribution properties of a given GCM time series. The second step involves spatial downscaling the regional GCM projections to coincide with different meteorological observation stations (met stations). Historic station records and historic GCM runs are correlated, resulting in transfer functions to convert future GCM scenarios into multiple station scenarios. The transfer functions are applied to the future climate scenarios output from the first step. This results in 16 different monthly time step scenarios, four GCMs each with four periods of investigation, and each scenario represented by 10 different met stations.

24 The station scenarios are compared to the historic monthly station records and then changes to temperature and precipitation are calculated. The third step downscales the monthly time scenarios to a daily record. Step two resulted in monthly station temperature and precipitation changes for each of the 16 scenarios. The changes are applied to the original monthly historic values creating a modified monthly data series. Daily met station records, maximum temperature, minimum temperature and total precipitation are then modified so that the monthly averages are equal the modified monthly time series. The resulting daily met records are translations of the GCM climate projections and used to drive the hydrology model. 3.3 DHSVM Hydrology Model The Distributed Hydrology Soil Vegetation Model (DHSVM) was used to translate the climate scenarios into streamflows. DHSVM is a physically based, mesoscale, rainfall - runoff model developed by the University of Washington and Princeton University (Wigmosta et al, 1994). Unique models of river basins are built using grid layers based upon geographic information system (GIS) data which represent elevation, soil type, soil depth, vegetation type, stream network, solar exposure, and precipitation distribution. The resolution of the model can be defined between 30m and 450m grid cells. Meteorological forcing data are applied to a compiled DHSVM model producing streamflow output while maintaining energy and mass balances. DHSVM was used to build a model of the White Basin from the Mt. Rainier headwaters to the Puyallup confluence using a 450 meter spatial resolution and a three hour time step. Streamflow calibration was performed using two USGS gauging stations: , White River near Buckley, WA and , White River at Greenwater, WA (Figure 2-1). The period of recorded daily streamflow is 1928-Present (excluding ) and respectively. An eleven year calibration period was used including water years A comparison of the 17

25 18 61-year average annual hydrograph for observed and simulated flows above the diversion dam is shown in Figure 3-2. The annual water balance ratio between simulated and observed is Calibration was performed by altering the distribution of precipitation through the basin. Figure 3-2 Sixty-one year ( ) average annual hydrograph with quartile distributions comparing USGS flows to simulated DHSVM flows, three week average applied. The upper portion of the White River basin includes glaciers and permanent snowfields on Mt. Rainier. Basins with glaciers and permanent snowfields present a challenge to streamflow calibration since precipitation is stored and released over time periods greater than a year. Other basins in the Puget Sound Region that have been modeled with DHSVM are subject only to perennial snow packs which make calibration easier since the water balance between water years are less related. Lake Tapps and the diversion structures are not modeled in DHSVM. The streamflow network represents where water would flow unobstructed. The White River flows unrestricted up to the diversion dam to Lake Tapps. The only exception is Mud Mountain dam which is located at RM Mud Mountain dam is operated

26 for flood control, storing peak flood flows and releasing them as soon as possible. This alteration to flow volume and timing is not critical to this research since the binding constraints on the water resource system occur during low flow periods. 3.4 System Simulation Model The streamflows produced by the hydrology model served as input to a simulation model of the water resource infrastructure and operations in the White Basin. Different metrics were developed to explore the sensitivity of particular stakeholders to impacts of climate change The simulation model was built with a weekly time step to represent the physical infrastructure forming Lake Tapps and the multiple scenarios of streamflows, consumptive demands, and minimum instream flows (MIFs). For this study an alternate water year, April 1 st March 31 st was defined in place of the traditional October 1 st September 30 th water year. Water year 1929 is defined as April 1 st 1929 through March 31 st The reason for this is that low flow typically occurs across the traditional October- September water year. Periods of low flow are important for describing annual reliability. The systems model uses two streamflow outputs produced by DHSVM as flow inputs and synthetic Puyallup River flows. The primary input is just above the diversion dam and the second represents local inflows to the bypass reach before the return from Lake Tapps. The Puyallup flows represent the Puyallup Basin above the White/Puyallup confluence. For each streamflow scenario there are 73 years of data from April 1929 to March Cascade Water Alliance originally sought approval to develop a water resource systems on Lake Tapps in addition to the hydroelectric project. However, the primary operational objective changed dramatically in January of 2004 when a FERC license to continue to produce hydropower was rejected. The present plan is for Lake Tapps to be operated solely as a municipal supply. The first phase of construction is due to be completed in In the interim the U.S. Army Corps of 19

27 20 Engineers will operate the diversion and release structures to maintain the lake, but not produce hydropower or municipal supply. Model assumptions were made to represent the anticipated status quo once the lake becomes used for municipal supply. It is unlikely that the dam will be used for hydroelectric generation, either as a primary or secondary use since a FERC license is needed for any hydropower operations. Cascade has stated projected demands and infrastructure project timelines for the years , these were used as 30 years of investigation in the systems model (Ecology 2003). Even though the primary operation of the lake will be for municipal supply, obligations to fish must be met first. Minimum instream flows, as negotiated by tribes and federal regulators, must be left in the bypass reach before diversions to the lake are made. Bypass minimum instream flows are currently under adjudication and multiple proposals exist. The water supply plan also included addition avoidance and augmentation releases from the reservoir to help meet established fish flows at Puyallup during times of low flow. First avoidance water is released up to the amount withdrawn for consumption. If Puyallup MIFs are still not being met and augmentation operation will prevent Lake Tapps from refilling, thus releases are equal to inflow to the lake. After water is allocated to fish, water may be divert to and stored in Lake Tapps to be withdrawn for municipal demands. Once the water needs of fish and municipal demands have been met operations will be made to maintain a high lake level for the purpose of recreation. The maximum and minimum rule curves bounding the operating range of lake levels are depicted below in Figure 3-3. During the spring and summer the target lake elevation is full pool but the lake is drawn down during fall and winter to help combat milfoil and maintain water quality.

28 Lake Elevation Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Figure 3-3 Maximum and minimum operation rule curves for Lake Tapps The systems created for this research is different than the system model used in the evaluation of the proposed water supply project. The systems model used in the initial application ROE for the water right included local inflows to the White River and Lake Tapps. Certain flows were adjusted due to discrepancies between gauged inflows and releases from the reservoir relative to calculated reservoir volumes. Many of the gauges documenting smaller local inflows had short records thus the ROE model only simulates the 14-year historic period from water year 1988 through water year The ROE model was also run on a daily time step where as the model in this research used a weekly time step. Together, the downscaled climate projections, modeled hydrology, and systems simulation were used to make an assessment of climate change impacts to the White River Lake Tapps. The results of this assessment are presented in the next section.

29 22 Section 4: Results and Discussion Analysis was performed after each step in the modeling process, demonstrating the impacts of future climate change through meteorology, hydrology and human systems. Meteorological impacts are represented in terms of changes to average temperature and precipitation. Hydrologic impacts are shown in changes to streamflow characteristics such as timing, volume and frequency. Impacts to human and natural systems are represented by changes in safe yield and metric reliability. All 16 GCM scenarios were individually run through the three steps of the modeling process and later the results from the GCM scenarios were averaged into ensembles for each of the four periods of investigation (2000, 2020, 2040, 2060). 4.1 Impacts to Meteorology Downscaled climate change projections result in temperature and precipitation changes that are added and multiplied, respectively, to historical records to result in climate scenarios. On a global scale, climate is expected to shift towards warmer, wetter weather but regional scale impacts will vary (IPCC, 2001). Other studies in the region have shown increased temperature in addition to more winter precipitation and less summer precipitation (Palmer and Hahn, 2003; Wiley, 2004). The Buckley meteorological station, which is located near the USGS streamflow gauge used for the DHSVM model calibration, is used to demonstrate the projected changes to meteorology. The historic monthly averages for the period of 1930 through 1990 is presented below in Figure 4-1. Historically December has received the most precipitation and been the coldest month while July has been the driest and warmest month.

30 Precipitation (mm) Temperature (C ) Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Prec MaxT AveT MinT Figure 4-1 Historic Monthly Temperature and Precipitation for Buckley Meteorological Station, 61-year average data Impacts to Temperature Within the White River basin, a trend can be seen in rising temperatures. Figure 4-2 displays the average temperature projections at the Buckley station for each GCM scenario, while Figure 4-3 displays the temperature changes relative to the historic record. Increases on average 0.25 C per decade in the winter and 0.5 C per decade during the summer. These trends are subtle but consistent among the ensemble of GCMs. Increasing temperature can cause snowpack to runoff sooner and increasing the need to use reservoir storage during the summer. Temperature also impacts demand, particularly during the summer. Warmer temperatures promote more outdoor water use. 0.0

31 Temperature (C ) Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Historic Figure 4-2 Projected average monthly temperature at Buckley met station Change in Temperature (C ) Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Historic Figure 4-3 Projected average monthly temperature change for Buckley met station.

32 4.1.2 Impacts to Precipitation Precipitation projections have more uncertainty than temperature due to the challenges associated modeling the complex micro-physics describing rain. The trends in precipitation for the White Basin are less distinct than the temperature changes. Figure 4-4 displays the average historic precipitation in comparison to the projected changes. The monthly precipitation amounts in the 2000 projection exceeds the historic for the majority of months while the 2020 projection amounts fall slightly below the historic. The 2040 and 2060 projections have less precipitation than the historic. In almost every scenario the projections for winter precipitation exceed the historic and summer precipitation falls below. This suggests a trend towards a shift in the annual timing of precipitation with more precipitation occurring in the winter and less in the summer. Figure 4-5 presents the changes in precipitation relative to the historic record Precipitation (mm) Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Historic Figure 4-4 Projected average monthly precipitation at Buckley met station.

33 26 60% 40% Change in Precipitation 20% 0% -20% -40% -60% Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Historic Figure 4-5 Projected change in average monthly precipitation at Buckley met station. 4.2 Impacts to Streamflow Historical Streamflow Characteristics Historical streamflow records demonstrate that the White River has the typical two peak hydrograph characteristics of transient watersheds (Figure 4-6). The larger peak occurs in late spring and is associated with snow-melt run off. The winter peak occurs due to increased seasonal precipitation, the average volume is less than the spring but the instantaneous peaks are greater. Observing the annual hydrograph using a 3-week average centered on the week of interest smoothes instantaneous peak flows and emphasizes the seasonal timing of flow volumes.

34 % average Median 75% AF/wk Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Figure 4-6 Average annual hydrograph for White River near Buckley, 3-week running average applied to , USGS data. Historic records define the amount of natural variability that has been observed in streamflows. Streamflow has been recorded daily on the White River, USGS , since water year 1929, with the exception of water years , resulting in a 72 year record to date. By investigating more recent times within the historic record, deviation from the average can be seen. As Figure 4-7 shows, the average annual streamflows in the 1990 s varied from those of entire record. The average volume was greater and the timing differed with more flow during fall and winter and less in the spring. These changes in observed flow are similar to the impacts climate change is projected to have on streamflow.

35 AF/wk Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Figure 4-7 Comparison of 1990 s average annual hydrograph to full record for White River near Buckley. 3-week running average applied to USGS data Impacts to Annual White River Streamflows GCM ensemble streamflows for four periods of investigation: 2000, 2020, 2040, and 2060 were compared to modeled historic streamflows. Figure 4-8 displays the progressive change in the average annual hydrograph over the next six decades. The general trend is that more of the annual streamflow will occur during the winter and less during the summer. Annual flow volumes were all in the range of 5-9% less that the historic run. Both the shift in streamflow distribution and decrease in volume can be attributed to a combination of warmer temperatures and less annual precipitation. Increased streamflows during winter months indicates a shift from precipitation as snow to precipitation as rain, or an earlier runoff. During spring, streamflows decrease due to lack of snowpack and lower precipitation. Summer streamflows are lower than historic due to a reduction in summer rainfall. The timing of annual peaks is not shown to shift. Figure 4-8 shows the winter spring streamflow peaks occurring during the same months. Changes in the magnitude of annual peaks reveal a change in the seasonal distribution of annual

36 29 streamflow. Figure 4-8 demonstrates the transition to more runoff during the winter and less in spring. By 2060 the spring peak is significantly less pronounced than the historic peak Average Flow, AF/wk Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Historic Figure 4-8 Predicted Average Weekly White River Streamflows above Buckley diversion for Figure 4-9 further demonstrates the shift in the annual distribution of streamflow. What is shown is the change in fraction of annual flow for a given month, relative to the modeled historic record. For instance, if on average 8% of the annual flow volume occurred during February during the historic record and 10% occurred during the 2040 ensemble then the change in percent annual flow is a 25% increase. This representation is meant to highlight the change in timing separate of the change in volume.

37 % 30.0% Change in Percent of Annual flow 20.0% 10.0% 0.0% -10.0% -20.0% -30.0% -40.0% Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Historic Figure 4-9 Change in percent annual flow, White River above Buckley Diversion, streamflows compared to modeled historic streamflows Distribution of Summer Flows Summer is a critical time for municipal water supplies and instream flows for certain species. Temperatures increase and precipitation decreases creating peak season outdoor municipal demands. While demand peaks in the summer, reservoir inflow wanes during the summer. Seasonal snowpack is usually spent by June leaving streamflow to decrease through July and August. Lower summer flows increase the likelihood of shortfalls for fish and humans. Figure 4-10 depicts the projected impacts that climate change will have on summer flow volumes. Total summer flows, June-August, for each of the 16 GCM scenarios were arranged in ascending order then averaged across each period of investigation. The ensemble averages were then plotted as a CDF using the Cunnane plotting position. Total summer volume is represented as an average daily flow in cubic feet per second (cfs). The 2000 ensemble is lower than the historic

38 31 record with the exception of high flow years. The subsequent streamflow ensembles continue to show a declining departure from the historic summer inflow cfs Historic Figure 4-10 Non-exceedance probability of average daily summer (June, July, August) inflows to Lake Tapps. 4.3 Impacts to Water Resource System The final step in modeling projected climate change determines the impacts on the Lake Tapps White River water resource system. It is here that the greatest impacts will be felt by humans and the systems they manage. The systems level also offers the greatest opportunity to mitigate negative impacts by incorporating adaptive management. Safe yield and climate impacted safe yield demonstrate any changes in the amount of water that can be reliably provided by the natural system. The sensitivity of the bypass reach minimum instream flows, municipal demand and recreational lake level to climate change were measured in terms of reliability.

39 The plan to use Lake Tapps for municipal demand involves existing infrastructure as well as new construction. The water will be withdrawn near the existing output point, the hydroelectric turbines. A filtration and treatment plant will be built to comply with water quality standards. Construction will be done in two phases both estimated to begin in Phase I should be complete in 2025 and have a capacity of 30 MGD. Phase II should be complete in 2035 adding an additional 35 MGD bringing the total capacity to 65 MGD. Phase II capacity corresponds with the 65 MGD water right for withdrawal from the lake that has been issued by the Department of Ecology Lake Tapps Safe Yield Safe yield is a common attribute used to describe a municipal water resource. At this time, no safe yield has been published for Lake Tapps. This is because the lake has historically been used for non-consumptive use storage. Safe yield analysis simulates which historic streamflows would have caused failures for given assumptions about the system and demands. Because this is a simulation, when a year is designated as a failure year it does not mean that historically the system failed during that year. In the Pacific Northwest safe yield is typically described as a annual average demand which can be met with 98% reliability. The reliability is determined by the historic streamflow record. With 73 years of streamflow, 98% reliability corresponds to one allowable failure year. Therefore the second failure year defines the safe yield. A water yield by water year analysis was performed in the ROE application, Technical Memorandum 15, but this analysis was limited to White River flows above the diversion and did not incorporate storage or stakeholder issues (Ecology, 2003). Operating assumptions were made which represent the most likely constraints to be placed on system when it is operated for consumptive purposes. This included avoidance and augmentation releases from the reservoir for meeting Puyallup MIFs 32

40 33 which severely limit the yield. The assumptions made when evaluating the safe yield and climate impacted safe yields are as follows presented below in Table 4-1. Table 4-1 System model options for safe yield analysis. Inflows Diversion Restrictions Withdrawal Restrictions Consumptive Demands Minimum Instream Flows Recreational Restrictions Historic or climate flows, April-March water year Up to 2000 cfs at Buckley No water right or system capacity constraints Base demand with seasonal peaking factors Agency 10j Bypass MIFs, Avoidance and Augmentation releases for Puyallup MIFs No limit to minimum lake level 98% uncertainty boundary 90% uncertainty boundary 75% uncertainty boundary Ensemble Average E4 G30 HAD3 MGD PCM 10 Historic Figure 4-11 Projected trend of climate impacted safe yield for the Lake Tapps water supply, showing ensemble average decline of 1.8 MGD per decade. (adapted from Wiley 2004) The trend in safe yield shows a loss of about 1.8 MGD per decade, uncertainty boundaries were calculated based on the regression parameters and using Student s t distribution (Figure 4-11). The system s yield is particularly sensitive to the 2020 ensemble of streamflow projections which make a linear representation of the trend