Predicting Effects of Climate Changes: A Study of the Skagit River Hydro Project

A warming climate could have significant effects on the hydrologic balance of the watersheds on which hydropower depends. Model studies performed by Seattle City Light indicate that over the next 40 years, projected climate changes could require the utility to reconsider its current operating procedures.

By Wing Cheng and Amy L. Sansone

The potential effects of climate change on regional resources are a growing concern, not only to scientists and environmentalists, but also to regional governments and energy producers. In Washington State, the Office of the Governor recently released a report discussing the effects of climate change on the Puget Sound region. The projected changes include warmer air and water temperatures, alteration of streamflows, and increased flooding.1 The potential effects of climate change on streamflow and snowpack are of particular interest to utilities and local governments in western Washington State, where hydropower makes up approximately 80 percent of the electricity generated and where spring snowmelt is critical for reservoir recharge.2 Warmer temperatures could affect electricity demand, but they also could affect hydropower generation in mountainous watersheds by increasing freezing levels, decreasing snowpack, and increasing winter streamflows. The Climate Impacts Group at the University of Washington has published numerous studies that describe the effect of warmer temperatures on the hydrology of the Columbia River.

The effects of climate change on smaller watersheds in western Washington are also a concern from a water supply and power generation standpoint. One such watershed supports Seattle City Light ’s Skagit River hydroelectric project. In 2005, Seattle City Light teamed with 3Tier Environmental Forecast Group of Seattle to examine how expected levels of climate change in the 2020s and 2040s could affect inflow to the project.

The Skagit River resource

Seattle City Light, one of the largest municipal utilities in Washington, provides power to a service area population of more than 700,000. Approximately 89 percent of that power is generated from hydro resources.2 Currently, Seattle City Light is a surplus utility and is able to use resources within its energy portfolio to meet the electricity demand over 85 percent of the time. In very dry years, however, the utility may need to purchase power from outside sources to meet demand. Warmer temperatures have a direct effect on the seasonal availability of water and on power and revenue generation.

The Skagit River hydroelectric project is located in the North Cascades Region of Washington. It consists of three reservoirs and three powerhouses: 460-MW Ross, 168-MW Diablo, and 177-MW Gorge. The most upstream reservoir, Ross, has a storage capacity of 1,298 million cubic meters, compared to 34 million cubic meters at Diablo and 6.6 million cubic meters at Gorge. Since Ross Reservoir provides the majority of the storage capacity and releases water directly to the Diablo and Gorge powerhouses, 3Tier ’s analysis focused on the inflow into the Ross Reservoir. The Ross powerhouse provides 24 percent of Seattle City Light ’s total generating capacity of 1,889 MW.

Modeling the watershed response

The Ross Reservoir is located on the west side of the North Cascade mountain range with a drainage area of approximately 2,590 square kilometers. Basin elevations range from approximately 1,100 feet to 7,900 feet. Eighty-five percent of the basin is above the average winter freezing level of 2,500 feet, which means that snowpack is a major component of the water available for reservoir recharge in the spring. Therefore, a model that captures the spatial variability in precipitation and air temperature is critical for accurately depicting the effects of climate change on the hydrology in the watershed.

We modeled the Ross Reservoir watershed with the Distributed Hydrology Soil Vegetation Model (DHSVM). DHSVM is a physically-based model originally developed at the University of Washington for use in mountainous watersheds.3 The model represents the spatial distribution of soil moisture, snow cover, evapotranspiration, and surface and subsurface flow. For this analysis, the watershed was modeled at a horizontal resolution of 1,500 feet.

Figure 1: The Distributed Hydrology Soil Vegetation Model predicts snow cover throughout a water year for a given precipitation and temperature sequence. The model indicates that snow water equivalent over the basin will drop sharply as a result of projected climate changes in the 2020s and the 2040s.
Click here to enlarge image

We evaluated the effects of climate change on Ross Reservoir hydrology by creating a benchmark runoff simulation based on historical climatology, then creating additional simulations using the expected 2020s and 2040s climatology. The benchmark simulation of snowpack and reservoir inflow was based on the historic air temperature and precipitation record at the Diablo meteorological station. To create a benchmark representing recent conditions and still capture the yearly variability of the climate, we used the historic air temperature record from 1936 to 2004, but adjusted it so that the average temperature over the entire period was equal to the average from 1990 to 1999. This adjustment was equivalent to a warming of 0.6 degree Fahrenheit.

The next step was to examine how runoff patterns might change as a result of a warming climate. We adjusted the benchmark record again to incorporate expected changes in the Pacific Northwest climate, using average annual expected changes for the Pacific Northwest region derived from a consensus of eight global climate simulations. Two future time horizons of climate change were used: the 2020s and 2040s. For the decade of the 2020s, the net effect was a warming of 2.7 degrees Fahrenheit, compared to the average annual air temperature of the 1990s. For the decade of the 2040s, the net warming was 4.1 degrees Fahrenheit relative to the 1990s.4 Although the majority of climate models suggest a slight increase in winter precipitation and decrease in summer precipitation, the expected change in precipitation associated with global warming for the region is difficult to distinguish from natural variability in precipitation.1 Therefore, for this analysis only a change in the temperature was evaluated.

Climate change and basin hydrology

The effects of expected levels of climate change on the hydrology of the Ross Reservoir are significant. Our model indicated that the timing of the seasonal inflow hydrograph would shift, with an increase in average monthly flow in the winter months and a decrease in average monthly flow in the spring months. This shift could be critical for the Skagit River project, since the inflow in the spring months is used to refill the Ross Reservoir for power generation during the following high-demand winter period. Under the current reservoir operation guidelines, there is a high probability that spring snowmelt and summer low flows will be insufficient to meet minimum instream flow requirements during the summer and simultaneously refill the reservoir for the upcoming winter.

Figure 2: Watershed modeling suggests that, under future climate conditions, annual maximum floods would increase over all exceedance probabilities as a result of warmer winter temperatures.
Click here to enlarge image



Figure 1 shows the seasonal snow accumulation in the Ross Reservoir watershed for the benchmark climate and expected climates of the 2020s and 2040s. The simulation results show a marked decrease in average accumulated snow water equivalent (SWE) over the basin. The accumulated SWE drops from a spring peak of approximately 24 inches SWE under benchmark conditions to 16 inches under 2020 conditions and 12 inches under 2040 conditions. Since the winter snowpack acts as a second reservoir of water that becomes available during spring snowmelt, a loss in accumulated SWE translates directly into a loss of stored water that would be available for reservoir refill. The expected loss in average accumulated SWE for the 2020s is equivalent to 33 percent of the active storage capacity of Ross Reservoir. In the 2040s, the change is even greater, with the expected loss in SWE equivalent to 49 percent of the active storage capacity.

Figure 3: This chart shows the number of times, by month, that daily flows would exceed the present median flow over a 69-year analysis period. For benchmark conditions (in black), the highest flows normally occur in May and June, but under future climate conditions (2020s in green and 2040s in red) the highest flows shift to midwinter.
Click here to enlarge image


Flood Frequency

Figure 2 on page 61 shows the exceedance probabilities for the annual maximum daily flows. The maximum daily flows increase for all exceedance probabilities under 2020 and 2040 conditions. Warmer winter temperatures result in more precipitation falling as rain instead of snow, which increases the magnitude and frequency of winter flooding. Figure 3 shows the number of times for each month of the year that the daily average streamflow exceeds 23,000 cubic feet per second (cfs) for benchmark, 2020, and 2040 conditions. The 23,000-cfs streamflow corresponds to the two-year recurrence interval flow under benchmark conditions. In comparison to benchmark conditions, the number of times the present two-year streamflow is exceeded increases by 14 percent and 60 percent under 2020 and 2040 conditions, respectively.

Not only is there an increase in the number of times the present two-year flow is exceeded, there is also a shift in when the flow is exceeded. Under benchmark conditions, 45 percent of exceedances occur during November through February, with a secondary spike during May and June. Under 2020 and 2040 conditions, 84 percent and 93 percent of exceedances occur from November to February, respectively, with almost no exceedances during the summer. The number of exceedances expected to occur at the start of the rainy season (October) actually decreases under the warmer scenarios. This decrease is due to an earlier snowmelt and increased evapotranspiration during the summer, resulting in lower soil moistures at the start of the rainy season.

Monthly inflows

Figure 4 shows the changes to the yearly Ross Reservoir inflow hydrograph due to expected temperature changes. Although there is a slight increase in annual inflow, the most significant changes are seen in the timing of the runoff and the total inflow during the April to July refill period. Reservoir inflows increase during the winter and decrease during the spring due to higher freezing elevations associated with warmer temperatures.

Figure 4: The average annual inflow hydrograph to Ross Reservoir is expected to shift toward higher flows in midwinter under future climate conditions. The shift poses problems if water relied upon for future generation arrives during the season when flood control requirements limit the useable storage.
Click here to enlarge image

Under 2020 and 2040 conditions, the probability of exceeding the median April through July inflow of the benchmark condition is significantly reduced. Figure 5 on page 64 shows the exceedance probabilities for the April through July average inflow for benchmark, 2020, and 2040 conditions. Under benchmark conditions, the probability of exceeding an inflow of 5,100 cfs is 50 percent (by definition, the median flow). Under 2020 and 2040 conditions, the probability of exceeding 5,100 cfs drops to 18 percent and 7 percent, respectively.

Refilling the reservoir

The average refill from April to July under benchmark climate conditions is approximately 1.29 million acre-feet. The simulated average refill volume drops to 0.96 million acre-feet in the 2020s and to 0.77 million acre-feet by 2040. The reduction in inflow volume in the 2020s is roughly equivalent to 31 percent of the active storage capacity of the reservoir. The reduction in inflow in the 2040s is approximately 50 percent of the active storage capacity.

Figure 5: This chart shows the projected change in exceedance probability for monthly inflow in the refill months of April through July. The flow that is currently the median flow would be exceeded only 18 percent of the time under 2020 conditions and only 7 percent of the time under 2040 conditions.
Click here to enlarge image

To maintain the same amount of inflow volume during the refill period as under benchmark conditions, the refill period has to begin in late February under 2020 conditions and late January under 2040 conditions. Beginning refill during the winter months has significant effects on reservoir operations, since the refill period under 2020 and 2040 conditions overlaps with the traditional flood control season for the watershed. Operators are required to have 11 percent of Ross Reservoir ’s active storage capacity available for flood control from December 1 through March 15.5 A longer refill period that overlaps with the flood control season will result in one of two scenarios under the current reservoir operation guidelines:

— The amount of water used for power generation in the spring will be reduced in order to meet power obligations in the following winter high demand period; or

— Maintaining current power generation capabilities in the spring will result in reduced generation during the following winter.

We used the Seattle City Light system control center ’s hourly electricity demand model to evaluate the effects of expected air temperature increases on demand. We found that the expected increase in air temperatures reduces the winter load in the 2040s by only 2 percent. Thus, the expected change in system load is not expected to significantly affect Ross Reservoir operations.

Operating in a warmer climate

Seattle City Light uses a monthly reservoir storage model to plan for meeting electricity demand while observing other regulatory constraints. We used the system control center ’s monthly Ross Reservoir operations model to evaluate measures that could mitigate the effects of the expected air temperature increases in the 2020s and 2040s. This analysis used median inflows to produce power closely matching the current level of monthly generation.

The analysis indicated that the utility could use the storage capability of the Ross Reservoir to mitigate the effects of the 2020 climate conditions. At the start of the runoff period, the increased winter inflows will raise the lake level approximately 42 feet higher than the current normal. The current operational target for the Skagit hydroelectric project is to refill Ross Reservoir by the end of July while still meeting recreation, flood control, and fishery flow requirements. Because of the significantly lower July inflow, Seattle City Light will have to raise the lake level to its allowable maximum by the end of June. In wetter years, water will have to be spilled during June. In drier years, releases will have to be curtailed during May and June. Discharge in July will have to be reduced to maintain the target elevation in Ross Reservoir. With these modifications, Seattle City Light can still meet current instream flow requirements.

Based on the same analysis, the effects of the 2040 climate conditions cannot be fully mitigated using the storage capability of the Ross Reservoir. In years with median monthly inflows, Seattle City Light will not be able to meet both the minimum fishery flow requirements and the end-of-July target reservoir elevation. During November and December, releases will need to be increased, possibly to include spilling, to keep the reservoir levels below the flood control limit. At the start of the runoff season, the reservoir level will be approximately 57 feet higher than the current normal. During May and June, the lake level will most likely be at its maximum. In wetter years, water will have to be spilled in May and June to keep the lake level at or below the maximum allowable. Discharge in July will have to be reduced below the minimum downstream fishery flow in order to maintain the reservoir target elevation. If the July target elevation were reduced by 2 feet, minimum fishery flow requirements could be met but the minimum lake level requirement for recreation would be violated in late August and early September.

The outlook for reservoir management

The effects of expected levels of climate change on the hydrology of the Skagit River watershed are significant. Overall, the issues posed by climate change for the Skagit River project are about water storage, not water shortage. One specific complication of the expected changes is that the refill period overlaps with the flood season. The management of the reservoir for flood control and power generation becomes more complicated as both objectives may need to be addressed simultaneously.

Based on our analysis, the current reservoir operation guidelines could be modified to meet the current level of generation to serve load, instream flow requirements, and flood control objectives under 2020 conditions but not under 2040 conditions. In addition to the projected loss of ability to meet customer demand, Seattle City Light could also be affected by revenue losses stemming from forced changes in operation. Although our analysis did not address revenue, it is undoubtedly one of the many aspects of utility operation that will be affected by a warming climate.

The Skagit River analysis was the first of its type for Seattle City Light, and represents about 40 percent of the utility ’s generating capacity. For the remaining 60 percent, most of which is situated in the Columbia River Basin, Seattle City Light is awaiting the results of a more detailed regional climate change study now nearing completion at the University of Washington. The forthcoming study will provide information to be used in support of systemwide planning.

Wing Cheng may be contacted at Seattle City Light, 700 5th Avenue, Suite 330, Seattle, WA 98104; (1) 206-706-0163; E-mail: Amy Sansone may be contacted at 3Tier Environmental Forecast Group, 2001 Sixth Avenue, Suite 2100, Seattle, WA 98121; (1) 206-325-1573; E-mail:


  1. Snover, A.K., P.W. Mote, L. Whitely Binder, A.F. Hamlet, and N.J. Mantua, Uncertain Future: Climate Change and its Effects on the Puget Sound, produced for the Puget Sound Action Team by the Climate Impacts Group of the Joint Institute for the Study of Atmosphere and Oceans, University of Washington, Seattle, 2005.
  2. U.S. Department of Energy, Energy Information Administration, Official Energy Statistics from the U.S. Government: State Electricity Profiles, at electricity/st_profiles/e_profiles_sum.html, 2002.
  3. Wigmosta, M.S., B. Nijssen, P. Storck, and D.P. Lettenmaier, “The Distributed Hydrology Soil Vegetation Model, ” in Mathematical Models of Small Watershed Hydrology and Applications, V.P. Singh and D.K. Frevert, eds., Water Resources Publications, Littleton, Colo., 2002, pages 7-42.
  4. Mote, P.W., et al., “Preparing for Climatic Change: the Water, Salmon, and Forests of the Pacific Northwest, ” Climatic Change, 2003, Volume 61, No. 1, pages 45-88.
  5. U.S. Army Corps of Engineers, Columbia Basin Water Management Division, Water Management, at http://www. reports/bluebook/1997/Ch3.pdf.


The authors acknowledge Pascal Storck, PhD, president of 3Tier, who provided technical review of the study described in this article.

Wing Cheng, P.E., senior mechanical engineer at Seattle City Light, performed the electricity demand and reservoir operations analysis for the study described in this article. Amy Sansone, P.E., senior hydrologist at 3Tier Environmental Forecast Group, performed the hydrologic analyses for the study.


Previous articleSMUD urges 50-year term to relicense 648-MW Upper American River
Next articleAssessment of Waterpower Potential and Development Needs

No posts to display