Modeling the effects of climate change forecasts on streamflow in the Nooksack River basin

Transcription

1 Modeling the effects of climate change forecasts on streamflow in the Nooksack River basin Thesis Proposal for the Master of Science Degree, Department of Geology, Western Washington University, Bellingham, Washington Susan E. Dickerson January 2009 Approved by Advisory Committee Members: Dr. Robert Mitchell, Thesis Committee Chair Dr. Doug Clark, Thesis Committee Advisor Dr. Andrew Bunn, Thesis Committee Advisor

2 Table of Contents 1.0 Problem Statement Introduction Background Nooksack River Basin Physical characteristics of the Nooksack basin Use and allocation of the Nooksack Regional Climate Streamflow in the Nooksack River Climate Change Climate System Global Climate Change Climate change and water resources Climate Change Technical Committee Hydrologic Modeling Proposed Research Methods Scope of Work DHSVM DHSVM Setup DHSVM Algorithms DHSVM Calibration Climate Change Forecasts General Circulation Models GCM Emissions scenario couples GCM downscaling Modeling Timeline Expected Outcomes Significance of Research References Tables Figures

3 1.0 Problem Statement The goal of my research is to predict the timing and magnitude of streamflow in the Nooksack River under changing climate conditions. Streamflow in a high relief drainage basin has a complex relationship to changes in temperature and precipitation; an understanding of the timing and magnitude of future streamflow and extreme events under a range of possible climate conditions is critically important for effective water resources planning in Whatcom County, Washington. General Circulation Models (GCMs) provide predictions of the large-scale climate trends of the future, but forecasts of regional climate, including local-scale weather patterns and topographic effects, are required to characterize future streamflow. I propose to use statistical methods that utilize the record of historical weather variability in order to downscale GCM forecasts from a global scale to a local scale. I will use the downscaled data set as meteorological input and present-day basin characteristics as spatial inputs into the Distributed Hydrology-Soil-Vegetation model (DHSVM; Wigmosta et al., 1994) in order to predict the effects of climate change on the timing and magnitude of streamflow in the Nooksack River. 2.0 Introduction The Nooksack River has its headwaters in the North Cascade Mountains and a watershed that incorporates approximately 2000 square kilometers (Figure 1). Municipalities and industries in Whatcom County, WA, and the Nooksack Tribe and Lummi Nation depend on the Nooksack River for water use and for fish habitat. An understanding of the probable response of the Nooksack River to climate change is of vital importance to these agencies for water resource planning purposes. Global climate change may include changes to both temperature and precipitation patterns, which influence the timing and magnitude of streamflow in a snowmeltdominated basin such as the Nooksack River basin. Prediction of future streamflow depends on prediction of future climate, rather than on historical observations of the variability of streamflow. GCMs have been developed by researchers at institutions worldwide to predict changes in global climate under a range of future emissions scenarios. In order to use data produced by selected GCMs as the input for a regional- 3

4 scale hydrologic model I will use a statistical downscaling process described by Salathe, et al. (2007) to convert the GCM data from coarse resolution to fine resolution. I will then use the downscaled data as meteorological inputs to the Distributed Hydrology-Soil-Vegetation model (DHSVM; Wigmosta et al., 1994) to predict the impacts of climate change forecasts on streamflow in the Nooksack River basin. This project is a regional case study application of an important technique for incorporating climate change forecasts into hydrologic modeling. I will use the methodology of the Climate Change Technical Committee (CCTC), associated with the Climate Impacts Group at the University of Washington. The CCTC investigated hydrologic impacts of climate change on five river basins in Pierce, King, and Snohomish Counties; the research was funded by a variety of groups interested in the long-term management of water resources (Palmer, 2007b). 3.0 Background 3.1 Nooksack River Basin Physical characteristics of the Nooksack basin The Nooksack River basin is an approximately 2000 km 2 watershed, located primarily in Whatcom County, Washington, that provides freshwater for domestic and commercial use, agriculture, salmon and shellfish habitat, and a variety of recreational opportunities (Figure 1). The headwaters are in the North Cascade Mountains, and the North, Middle, and South forks flow approximately west to their convergence near Deming, WA; the main stem of the river meanders through the lowland until it discharges into Bellingham Bay. There are two distinct provinces of the basin, delineated by topography, geology, and land use: the upland headwaters in the Cascades, and the lowlands west of the confluence. The majority of runoff into the Nooksack River comes from the upland province, whereas the majority of water usage is in the lowland (Bach, 2002). Topography in the upland province is rugged; elevation varies from 300 m to over 3000 m. The upland province includes Paleozoic and Mesozoic metamorphic rocks of the North Cascade System crisscrossed by thrust faults and strike-slip faults, late Cretaceous through Eocene sandstones, shales, and conglomerates of the Chuckanut Formation, and Quaternary volcanic rocks. Landslide and lahar deposits overprint 4

5 the bedrock geology and are an important influence in the geomorphology of the basin (Dragovich & others, 1997). Upland land use primarily includes state and federally managed land and conservation lands. The landscape is heavily forested and includes second growth stands of coniferous trees (Douglas Fir (Pseudotsuga menziesii), Western Hemlock (Tsuga heterophylla), Western Red Cedar (Thuja plicata), Red Alder (Alnus rubra), Maple (Acer), and dense undergrowth (Salal (Gaultheria shallon), Devil s Club (Echinopanax horridum), Huckleberry (Vaccinium), Oregon Grape (Berberis), ferns). Soils are formed from loess, volcanic ash, colluvium, and slope alluvium derived from weathered bedrock and Quaternary volcanic and glacial deposits, and range across the basin from shallow to very deep, and from moderately well drained to well drained (Golden, 1992). The lowland is characterized by low elevation (0-300 m) and low relief. The river meanders through Quaternary glacial sediments including recessional outwashes of the Vashon glaciation, Vashon till, and recent alluvial deposits. Lowland land use is dominated by agricultural, commercial, and residential use; significant agricultural operations include fruit and dairy farms. The lowland province of the basin will be excluded from this study because the strong influence of agricultural, industrial, and municipal water usage create a challenge to accurately modeling the hydrology of this portion of the basin. The high elevations and abundant winter precipitation of the upland Nooksack basin result in a significant snowmelt and/or glacial melt component to the streamflow of each fork. The headwaters include Mount Shuksan, Mount Baker, and the Twin Sisters, with approximately 16 to 40 percent of streamflow derived from snowmelt (Bach, 2002). The North Fork originates from the East Nooksack glacier on Mount Shuksan and the headwaters of the Middle Fork include the Deming glacier; the South Fork currently contains no glaciers. During the Pleistocene glacial maximum of the Fraser glaciation the Cordilleran Ice Sheet covered the Nooksack basin, except for Mt. Baker, Mt. Shuksan, and a few other peaks Use and allocation of the Nooksack The Nooksack River flows past several municipalities and tribal reservation lands in Whatcom County, including Deming, Everson, Lynden, Ferndale, the Nooksack reservation, and the Lummi reservation. Streamflow is consumed for drinking water, irrigation, and industrial processes. Streamflow is also used for recreation and for providing habitat to salmon and 5

6 shellfish. A 1500 kilowatt hydroelectric plant at Nooksack Falls on the North Fork has been operated by Puget Sound Hydro LLC since 2003 (FERC, 2004). Two salmon hatcheries are operated in the Nooksack basin: one operated on Skookum Creek (South Fork) by the Lummi Tribe, and the other operated on Kendall Creek (North Fork) by Washington Department of Fish and Wildlife. The Washington Watershed Management Act of 1998 provided the framework for local control of watershed planning. As a result, Water Resource Inventory Area No. 1 (WRIA 1), which encompasses the surface and ground water in the Nooksack River basin, was established; stakeholders include the Nooksack Tribe and Lummi Nation and Whatcom County municipalities, public utilities, industries, individuals, and farms that depend on the Nooksack River for freshwater fish habitat and domestic, commercial, municipal, industrial, and irrigation uses. The WRIA 1 Watershed Management Project includes assessment, planning, and action related to water quantity, water quality, fish habitat, and instream flows in the Nooksack River (WRIA 1, 2008). The City of Bellingham operates a diversion pipeline from the Middle Fork of the Nooksack River to Lake Whatcom in order to increase water quantity and quality. Approximately half of the population of Whatcom County, about 65,000 people, rely on Lake Whatcom as a drinking water source. Minimum instream flows for the MF Nooksack are regulated by the Washington Department of Ecology; diversion to Lake Whatcom may occur only when the minimum instream flow requirement is met (DOE, 1988) Regional Climate The climate of the Nooksack watershed is characterized by mild temperatures typical of a maritime climate; fall and winter are characterized by frequent, low intensity precipitation, whereas late spring and summer are relatively dry. Average annual precipitation ( ) ranges from 40 inches in the lowland to 140 inches at Mount Baker (PRISM, 2008). There is a steep topographic gradient from west to east which creates a negative lapse rate for temperature and a positive lapse rate for precipitation across the basin. As elevation generally increases from west to east, the temperature decreases, allowing for snow to fall in the mountains while rain falls in Bellingham. Conversely, the increase in elevation over the mountains causes an increase in precipitation due to the orographic effect. As moisture-laden air is lifted over the mountains it 6

7 experiences adiabatic cooling due to the decrease in pressure, causing more precipitation to fall as the air is lifted to higher elevations (Figure 2) Streamflow in the Nooksack River Streamflow in the Nooksack River is characteristic of a snow-melt dominated basin that lies within a mild, rainy climate at lower elevations (Figure 3). Streamflow increases and remains high during the fall and winter as precipitation increases, and increasing soil moisture content results in more and more water running off directly into the stream rather than infiltrating into the soil and flowing slowly in the subsurface. Streamflow decreases in mid- to late spring as precipitation decreases, followed by a peak in the hydrograph in late spring or early summer as the snow melts. Streamflow decreases throughout the summer as snowmelt is depleted and precipitation is low. Currently, low summer flows are buffered by snowmelt in glaciated basins of the Nooksack River, but changes to the timing of melting and to the mass of glacial ice may affect future streamflow in the drier summer months. Streamflow in the Nooksack River is monitored by the USGS at five real-time stations: Cascade Creek (North Fork), Wickersham (South Fork), Deming (Middle Fork), North Cedarville (Nooksack River), and Ferndale (Nooksack River). The USDA National Resources Conservation Center operates a SNOTEL sites on the Middle Fork of the Nooksack, at Wells Creek, and at Elbow Lake that monitor precipitation, snow water equivalent (SWE), and temperature in the watershed. Bach (2002) quantified the amount of streamflow derived from a glaciated basin (the North Fork Nooksack) as compared to a similar unglaciated basin (the South Fork Nooksack), by comparing streamflow measured by the USGS at the Glacier station (North Fork) and the Wickersham station (South Fork) to the total flow after the confluence of the forks. Bach estimated that 26.9 percent of summer streamflow in the Nooksack River is attributable to high elevation snow and glacier melt. Donnell (2007) used DHSVM to quantify the glacial melt water component of streamflow in the Middle Fork Nooksack. Estimated late summer glacial melt water contribution based on 2002 glacier conditions and 2006 meteorological data was %. Donnell also modeled the effect of glacier recession on streamflow using a linear rate of recession and modern 7

8 climate data and predicted up to 8.6% decrease in streamflow in the Middle Fork during the next fifty years s a result of glacier shrinkage. 3.2 Climate Change Climate System The Earth s climate is a complex system composed of the interaction of subsystems that include the atmosphere, oceans, biosphere, land surface, and cryosphere. Global temperature responds to a net change in the energy balance; an energy surplus leads to warming, whereas an energy deficit leads to cooling. Energy inputs include short-wave radiation from the sun, longwave radiation reflected off the Earth and long-wave radiation re-emitted from the Earth. Energy outputs include radiation that is reflected and re-emitted. The climate system is an inertial system due to the high specific heat of water and the large percentage of the Earth covered by oceans. Therefore it may take years for the climate system to reach an equilibrium temperature after an energy imbalance occurs. Climate is the statistical average of weather, commonly averaged on a 30 year cycle. Within a climate, the daily or hourly weather can be extremely variable due to local weather patterns and topography; additionally, the climate system, and average global temperature, varies due to internal and external factors. Pseudo-periodic internal variations such as the El Nino Southern Oscillation and the Pacific Decadal Oscillation result from global circulation of ocean currents and air currents within the climate system. Global temperature also changes due to variations in radiative forcings, external energy inputs or outputs, due to solar variations caused the periodic Milankovitch cycles, as well as random changes such as volcanic events. A large volcanic event that emits reflective sulfate aerosols into the stratosphere can create a negative forcing on the climate system for up to tens of years (IPCC, 2007). Feedbacks are complex mechanisms that add to climate variability. A positive feedback amplifies the effect of a radiative forcing, whereas a negative feedback dampens the effect. Some phenomena can act as both a positive and a negative feedback on the same or different timescales. For example, the input of CO 2 to the atmosphere increases the growth rate of vegetation which causes an increase in surface albedo and a negative radiative forcing, while the 8

9 increased level of CO 2 in the atmosphere results in a positive radiative forcing due the increased absorption and storage of energy Global Climate Change Evidence indicates that the Earth s climate is warming. Average global temperature has increased 0.74 C in the last 100 years ( ), sea level has risen 1.8 mm/year from thermal expansion and melting of continental ice sheets, and global ice cover has decreased (IPCC, 2007). Proxy data that are sensitive to changes in climate such as tree rings, pollen, ice cores, and corals, have been used to reconstruct global temperature into the last thousand or more years (National Research Council, 2006). Studies of these different proxy data have produced climate reconstructions that show that average global temperature has fluctuated over the last one thousand years. The modern warming trend, however, departs sharply from the known upper bounds of recent natural variability (National Research Council, 2006). The warming of Earth s climate is attributed to the post-industrial increase of both naturally occurring greenhouse gases (e.g., CO 2, CH 4 ) and purely anthropogenic greenhouse gases (e.g., chlorofluorocarbons (CFCs)) to the atmosphere. The gases absorb and retain energy, which increases the net storage of energy in the climate and increases global temperature; the concentration of each greenhouse gas can be related directly to a radiative forcing that changes the energy balance of the atmosphere. According to the 2007 Fourth Assessment Report (AR4) from the Intergovernmental Panel on Climate Change (IPCC) there is a greater than 90% probability that most of the observed warming is due to increases in anthropogenic greenhouse gases, and predicted regional impacts including decreased snowpack, increased flooding and decreased summer streamflow in western North America (IPCC, 2007) Climate change and water resources Three decades of studies on the effects of climate change on water resources indicate that both temperature and precipitation have direct effects on the timing and magnitude of streamflow, and that the response of a given stream is highly variable and non-linear based on basin characteristics and local and regional climate (Alexander et al., 2007). Climate-driven effects on streamflow include changes to the ratio of precipitation in the form of rain to snow, amount of total precipitation, timing of snowmelt, and timing in seasonal changes in soil moisture content. An average warming rate of 0.3 F/decade is projected for the Pacific Northwest (Mote et al., 9

10 2005; Alexander et al., 2007). The temperature change is expected to result in higher winter and spring streamflow and an earlier melt season in snowmelt-dominated basins, resulting in decreasing summer flows at the height of water usage demand. Projected changes to rainfall in the Pacific Northwest are variable and modest; most models predict an increase in winter precipitation and a decrease in summer precipitation (Mote et al., 2007). Investigations on Western river basins, which are commonly snowmelt-dominated, have focused on the impact of temperature on the type and amount of precipitation and on the timing of snowmelt. A number of studies have documented the shift of the timing of spring snowmelt to earlier in the year under warming climate conditions (e.g., Lettenmaier and Gan, 1990; Cayan et al., 2001; Regonda et al., 2005; Stewart et al., 2004). Gleick and Chalecki (1999) pointed out that despite a variety of different methods, assumptions, and models that had been used to evaluate effects of increasing temperatures on the Sacramento River, that every study showed changes to the timing and magnitude of runoff as a result of temperature-driven changes to the snow dynamics in the river basin. Spatially distributed hydrology models have been used to predict the response and to evaluate sensitivity of streamflow to different climate conditions, and to understand the causes of the changes to observed streamflow. Leung and Wigmosta (1999) used DHSVM and regionally downscaled GCM forecasts to predict the sensitivity of two Pacific Northwest basins, the American River (coastal) and the Middle Fork Flathead River (continental), to changes in climate under a doubling in atmospheric CO 2 concentration. The American River responded with a 60% decrease in basin SWE and an early spring melt, whereas the SWE in the MF Flathead was reduced only by 18% and the timing of spring streamflow remained the same. Their study demonstrated the regional variability of the magnitude of effects of climate change on snowmelt-dominated basins. Mote and others (2005) documented a decreasing trend in SWE in the Western U.S. from based on observed data, and argued for a predominantly climatic cause for the trend; the observed data agreed with simulations using the Variable Infiltration Capacity (VIC) hydrologic model, and with the spatial pattern of climatic data from the same time frame. Largest relative decreases were in the Pacific Northwest, which pointed to the increased sensitivity of SWE to elevation, and mean winter temperature. Mote and others (2008) more recently investigated the relationship of spring snowpack in the Cascades to 10

11 temperature and precipitation, and determined that temperature was the dominant long-term influence. The study quantified the decline of April 1 SWE at 15-35% since the 1950s. The understanding of the first-order effects of changing climate of water resources is necessary precursor to the understudied area of second-, third- and fourth- order effects, such as hydropower, electricity prices, and national security, respectively (Chalecki and Gleick, 1999). Many studies are focused on the effects of climate change on streamflow in order to plan for future water resource availability and needs. Milly et al. (2008) asserted that water resource planners can no longer rely on historical records to characterize an unchanging range of weather variability; rather, planners need to consider future water resources based on changing probabilities of future weather frequency Climate Change Technical Committee The Climate Change Technical Committee (CCTC) was formed as part of the Regional Water Supply Planning Process, a collaborative effort between the Climate Impacts Group (CIG) at the University of Washington, the WA Department of Ecology, public utilities, tribes, and other community groups. The goal of the group was to gather data and tools to assess the impact of climate change on local water resources, and thus to assist in water resource management and planning (Alexander et al., 2007). The CCTC used DHSVM and downscaled GCM predictions to model the effects of climate change forecasts on five river basins in King, Pierce, and Snohomish Counties: the Cedar, Green, White, Sultan, and Tolt Rivers. The CCTC outlined background, methods, and results of their research in a series of eight technical memos and made their results available in a variety of formats to interested groups and individuals (e.g., Polebitski, 2007a; Results of the simulations included an overall shift in the hydrographs of each river to an earlier spring melt off (Figure 4). Each river showed different changes in the magnitude of streamflow under a range of future climate conditions; the average change in all five rivers was a positive net increase in annual flow, with increasingly negative changes to summer flow and increasingly positive changes to winter flow in the next 75 years (Table 1). 3.3 Hydrologic Modeling Hydrologic modeling to predict streamflow began in the 1950s and 1960s with spatially lumped models that used a water balance approach and meteorological data averaged over an 11

12 entire watershed to forecast streamflow (Storck et al., 1998). However, the availability of digital spatial data, such as digital elevation models and soil maps, combined with advancements in computing power, led to the development of spatially distributed hydrology models that simulate rainfall and runoff in individual pixels of a spatially heterogeneous watershed (Storck et al., 1998). The Distributed Hydrology-Soil-Vegetation Model (DHSVM) is a physically based, spatially distributed hydrology model that was developed at the University of Washington; the model was originally tested and validated in the Middle Fork Flathead River basin in Montana (Wigmosta et al., 1994). The model s explicit spatial representation of watershed characteristics has allowed applications to understanding hydrologic impacts of land use changes. Storck et al. (1995) investigated the effect of forestry impacts on peak flows in the Snoqualmie River basin and modified the model to simulate flood events in maritime mountainous watersheds. An accurate representation of the Pacific Northwest basin required a variable time step, with an hourly time step to model flood events and a daily time step during periods of consistent base flow, a precipitation lapse rate, and a two-layer snowpack component. DHSVM has been used to model the effects of timber harvesting, including change in vegetation cover and addition of roads, on the magnitude of flood events (e.g., Storck et al., 1998; Wigmosta and Perkins, 2001). DHSVM has recently been applied to partially urbanized watersheds, with representations of impervious surfaces and retention ponds in the model (Cuo et al., 2008). The DHSVM combines spatially variable watershed characteristics, including elevation, soil type, soil thickness, and vegetation with temporally variable meteorological information such as temperature and precipitation to predict the magnitude and timing of streamflow in the watershed. The DHSVM utilizes the physical relationships in the hydrologic cycle, such as the relationship between temperature and evaporation, to calculate the flux of water and energy in and out of each grid box, or pixel, in a digital elevation model (DEM). Water and energy can be stored in a pixel or can move between adjacent pixels, with direction and rate dependent on topography, soil type, and other factors; water flows across the surface and through the subsurface, and collects in stream valleys, which translates to the simulated discharge of the stream. Thus, the DHSVM provides a tool for understanding the surface water hydrology in a mountainous watershed given information about past, present, or future spatial characteristics and climate. 12

13 The DHSVM is calibrated and validated against historical records of meteorology and measured streamflow in order to use the model as a predictive tool. The small size of each pixel (30 m m on a side), allows the spatial heterogeneity of the watershed to be represented in the model; small variations in elevation and topography have an important effect on regional hydrology, as illustrated in the temperature and precipitation lapse rates across the Nooksack basin. 4.0 Proposed Research I propose to apply the methods of the CCTC and currently available climate change forecasts to model the effects of climate change on streamflow in the Nooksack River. I will use the methodology of the CCTC, outlined in eight technical memos (e.g., Polebitski, 2007a), for three reasons: 1) The CCTC uses downscaling methods that have been validated as a way to use coarse-scale climate predictions with the modeling of regional-scale hydrologic processes, and used DHSVM as the runoff model, which is available for use at WWU; 2) support in learning the methods is available, and has been offered, from contacts at the CCTC; and 3) outcomes for the Nooksack will be valid for comparison to the other Western Washington river basins that the CCTC studied; a direct regional comparison will be useful to this project and to the CCTC in understanding regional hydrology and monitoring the forecasts into the future. 5.0 Methods 5.1 Scope of Work This project will involve six main steps: 1) Setup DHSVM to the South Fork, Middle Fork, and North Fork basins of the Nooksack River drainage. 2) Collect and process meteorological input data. The DHSVM requires a meteorological input file that includes daily maximum temperature, precipitation, long wave radiation, short wave radiation, wind speed, and relative humidity. Additionally, a 30 to 50 year historical time series is required for the downscaling process. 13

14 3) Collect and process historical streamflow data for use in model calibration and validation. 4) Calibrate DHSVM to the South Fork, Middle Fork, and North Fork basins of the Nooksack drainage using data from 2003 to ) Use statistical downscaling techniques to create a local climate change forecast dataset. These data will be processed to obtain the remaining meteorological input required for DHSVM (e.g., long wave radiation). 6) Perform hydrologic simulations using the downscaled climate change forecast data as the meteorological input, and assess changes in snowpack, SWE, evapotranspiration, and streamflow timing and magnitude. 5.2 DHSVM DHSVM Setup The DHSVM requires inputs of meteorological data and spatial data in order to simulate the hydrology of the basin (Wigmosta et al., 2002). Meteorological inputs include temperature, precipitation, wind speed, relative humidity, incoming shortwave radiation, and incoming long wave radiation. Meteorological data are required for the time step at which the model is run, ranging from hourly to daily. The model requires six maps as spatial inputs, including elevation, watershed boundary, land cover, soil type, soil depth, and stream network; the spatial inputs are managed in ArcGIS. Elevation data are available as 10 m Digital Elevation Models (DEMs), and provide the base layer of spatial information. Landcover data, including vegetation and glacial coverage, is based on the 2001 National Oceanic and Atmospheric Administration (NOAA) land cover grid, and soil type comes from the United States Department of Agriculture State Soil Geographic (STATSGO) database. Watershed boundaries, soil depth, and stream networks are created from elevation using ArcGIS tools. Flow routing, including the stream network and road network, is also determined using ArcGIS tools (Wigmosta et al., 2002). For this study I will define the lower extent of the basin just below the confluence at Nugent; west of Deming the streamflow is impacted by removal of water for municipal, agricultural, and commercial use. I will collect spatial data for the Nooksack River basin from USGS 7.5 minute, 10-meter DEM files, the STATSGO database, and NOAA; I will use ArcGIS to set up and manage the basin information as a series of layers. 14

15 5.2.3 DHSVM Algorithms At each time step (hourly to daily) and for each pixel, the model provides simultaneous solutions to water and energy balance equations for seven hydrologic processes: evapotranspiration, snowpack accumulation and melt, canopy snow interception and release, unsaturated moisture movement, saturated subsurface flow, surface overland flow, and channel flow (Wigmosta et al., 2002). Evapotranspiration is represented through a two layer canopy, with each layer partitioned into a wet and dry percentage; the rate of evaporation and transpiration is calculated based on meteorological factors, vegetation type, and soil type. Snowpack is represented by a surface layer and a pack layer with energy and mass exchanged between the layers. Snow that is intercepted by the canopy is represented by a single layer that exchanges mass and energy with the air and ground below through interception, sublimation, and melt. Vertical movement of water through the unsaturated zone is represented through three soil layers in the model. Water that accumulates on the surface at a rate higher than the user-defined infiltration rate is routed as excess overland flow; water that infiltrates moves into the layer below at a rate described by Darcy s Law, or is removed through transpiration based on the type of vegetation in the rooting zone. Water that reaches the water table is routed laterally as subsurface flow. The direction and rate of saturated subsurface flow is determined via hydraulic gradients. Overland flow in the model includes infiltration excess runoff, saturation excess runoff when precipitation falls on a saturated soil surface, and return flow when the water table rises to the ground surface. Channel flow includes flow in stream channels and in road drainage ditches, and is routed through a linear storage routing algorithm in which outflow from the channel is linearly related to storage in the channel (Wigmosta et al., 2002). Version 2.4 of DHSVM includes a glacier movement component that may be useful to incorporate in this study. In version 2.0 and 3.0 of DHSVM, glaciers are represented as permanent snowpack, with all accumulation of snow remaining where it falls until it melts, which leads to an unrealistic accumulation of ice in high altitude basins, such as the Nooksack River basin. The new component characterizes the flow of accumulated ice into the ablation zone based on basal shear stress and glacier velocity (University of Washington, 2008). 15

16 5.2.4 DHSVM Calibration The DHSVM model is calibrated to a watershed by comparing simulated streamflow and SWE data to measured values for years in which there is recorded streamflow, SWE, and meteorological data available. Historic streamflow data are available from USGS gauging stations on the Nooksack and SWE data are available from SNOTEL monitoring stations. Sensitive parameters in the model that were estimated as inputs such as soil thickness, lateral hydraulic conductivity, and temperature and precipitation lapse rates are adjusted to fit the simulated data to the historic data. Once the calibration is complete the model is then validated by running simulations for a different time period for which there is historic data available. Initial conditions for the simulations are created by running one year of data and using the resulting soil moisture conditions as the initial conditions for subsequent simulations. 5.3 Climate Change Forecasts General Circulation Models General Circulation Models (GCMs) are coupled ocean-atmosphere 3-D models that divide the surface into grid boxes that extend vertically into the atmosphere. At each time step, within each box the model calculates the transfer of energy, mass and momentum based on known physical relationships. GCMs have been developed utilizing the first relationships described by atmospheric physics. Computational efficiency of the models requires a coarse spatial resolution (on the order of 100s of km per side) and thus requiring parameterization of relationships that are below the resolution of the model (e.g., cloud formation). The models are developed and run by large research institutions such as the Goddard Space Science Institute that make their forecasts publicly available. In order to understand the impact of anthropogenic changes on the global climate system, the emissions of greenhouse gases are related directly to a positive radiative forcing due to the net energy imbalance created by the increased storage of energy. The GCM is run using the change in net energy in the climate system and the model characterizes changes in air temperature, water temperature, precipitation, and other climatic factors (IPCC, 1997). 16

17 5.3.2 GCM Emissions scenario couples Since emissions of greenhouse gases are directly related to a positive radiative forcing, it is necessary to describe future emissions in order to model future global climate. Forty emissions scenarios were developed by modeling teams associated with the IPCC in order to model a range of future climate possibilities. Each scenario is a narrative that describes the population, technology, and economy of the world into the future, and relates levels of emissions to the fictitious future world. The scenarios are considered equally likely, and are a way of forecasting future climate based on a range of possible future emissions. The A2 scenario describes a future that includes continued population increase and an economy based on the intensive use of fossil fuels. The B1 scenario characterizes a future in which world population peaks and then declines, with a focus on alternative energies and economies (IPCC, 2000). GCMs are run using a specific emission scenario and the associated radiative forcing related to the emissions levels described in that scenario. Each model represents the climate system differently, based on spatial resolution, and levels of parameterization of processes. Thus, even utilizing the same emissions scenario, each GCM will provide a different forecast. For average temperature change in the Pacific Northwest, ten GCMs predict warming of C by the 2040s, with a range of possible warming provided by both the different GCMs and the two emissions scenarios (Figure 5; Mote et al., 2005). The same group of GCMs disagree as to whether precipitation in the PNW will increase or decrease into the 2040s. With future climate being based in part on different conditions, and with the varying levels of spatial resolution and complexity represented by the different GCMs, it is necessary to use a suite of models and scenarios to predict a range of possible future climate conditions. I will use the same three GCM-Emission scenario couples used by the CCTC, each representing a group of GCM predictions for temperature and precipitation in the PNW by the 2040s (Figure 6). These include the: IPSL_CM4_A2 (GCM from the Institut Pierre Simon Laplace, with A2 emissions scenario) which represents a group of couples that predict increase in temperature of 2-5 C and 8-9% increase in precipitation by Echam5_A2 (GCM from the Max Planck Institute for Meteorology, with A2 emissions scenario), which represents a middle of the road scenario with 2% precipitation increase and 1.7 ºC increase. 17

18 GISS_ER_B1 (GCM from the Goddard Institute for Space Studies, with B1 emissions scenario), which represents the group of couples that predicts a % decrease in precipitation along with a 2-5 C increase in temperature (Mote et al., 2005; Polebitski, 2007b). GCM data are publicly available from the research institutions developing the models GCM downscaling GCM outputs are provided on a very coarse scale, on the order of 100 km per grid cell side, and are inappropriate for use in a regional hydrologic study because the prediction does not reflect the regional spatial variability of weather. The meteorological data predicted by a GCM may be for just four to six locations for Washington State; therefore, the data must be converted to a much finer scale in order to provide useful information about a watershed that is subject to both local and regional meteorological influences such as topography and local weather patterns (Polebitski, et al., 2007). Through statistical downscaling, I will use the GCM output combined with historic weather data for the Nooksack basin to develop a local-scale climate prediction that I can use as the meteorological forcings in the DHSVM. Statistical downscaling relates the statistical properties of the predicted data to those of the historic data in order to better represent local variability in the data. Quantile mapping is a type of statistical downscaling that utilizes a long historic record of weather, and maps the predicted future climate onto the historic record of climate. The historic record is used because it reflects the variability between consecutive years, including extreme events. Cumulative distribution functions are calculated for monthly temperature and precipitation data from the historic record and from GCM simulations of the same period; transform functions are created from the relationship between the two data sets and used to correct the GCM predictions to the local scale. The goal of quantile mapping is to use the GCM forecast to provide the underlying future climate trend while preserving the full range of spatial variability of weather seen on a local level (Polebitski, et al., 2007; Wiley, et al., 2006). I will collect historical meteorological data from weather stations and SNOTEL stations, including Clearbrook weather station, Bellingham International Airport, and the Middle Fork Nooksack SNOTEL station. Point measurements can be distributed through observed lapse rates, interpolation methods, or through gridded climate estimate maps, such as the PRISM maps 18

19 developed by NRCS National Water and Climate Center (NWCC) and Oregon State University (OSU). 5.4 Modeling Modeling will take place after the DHSVM is calibrated and validated to the Nooksack basin and climate forecasts have been downscaled. DHSVM is written in ANSI-C and can be run on a variety of platforms. I will use a Linux operating system on a Dell Precision with dual 2GHz processors in Dr. Robert Mitchell s hydrology modeling lab. I will run the model using three meteorological data sets, each downscaled from a different GCM, for three periods in the future, 2025, 2050, Each year of simulation will represent a thirty-one year average of climate change forecasts, centered on the simulated year, and the watershed variability including in the downscaling process by mapping the predictions to fifty years of historical weather data. 5.5 Timeline Step Planned Completion Setup of DHSVM Winter 2009 Collect and process meteorological input data Winter 2009 Collect and process historical streamflow data Spring 2009 Calibrate DHSVM Summer 2009 Downscale climate change forecasts Winter/Spring 2009 Modeling Summer/Fall Expected Outcomes The final product of my project will be a simulation of the watershed hydrology of the Nooksack River basin during several points in the next century (e.g., 2025, 2050, 2075). Each year of simulation will represent a thirty-one year average of climate change forecasts, centered on the simulated year, and the watershed variability represented in fifty years of historical 19

20 weather data. Additionally, each simulation year will include three simulations based on the three different GCM-Emission scenario couples, thus representing a range of possibilities. Both the average prediction and the range of predictions are potentially useful in assessing the impacts of climate change on future water resources in Whatcom County. This study will closely follow the methods of the CCTC, and thus my final predictions will be comparable to their predictions for the five rivers in Pierce, King, and Snohomish Counties. This direct comparison will be useful for monitoring the accuracy of the predictions in comparison to other regional predictions. The Nooksack River basin will be set-up and calibrated in DHSVM and available for future use in understanding the hydrology of the basin and in predicting the effects on streamflow from changes in climate, land-use, and vegetation. Additionally, a downscaled climate forecast data set will be available for use in future simulations of local hydrologic processes such as streamflow in the Lake Whatcom watershed. The largest source of uncertainty in my study arises from the GCM forecasts, and GCMs will become more sophisticated as the computing power to run the models efficiently at finer and finer resolutions becomes possible, reducing the need to parameterize fine-scale processes. As new GCM forecasts become available it will be possible to re-run streamflow simulations for the Nooksack River since the basin will be set-up and calibrated in DHSVM, and historical weather data will be available for the downscaling process. Simulations of future streamflow can be updated as more detailed climate change predictions become available. 6.0 Significance of Research Under changing climate conditions the range of weather variability will shift as the underlying climate characteristics shift. A complete understanding of the sensitivity of the Nooksack River to climatic changes is critical to future water resources in Whatcom County. In order to plan for adaptation to future water availability we must anticipate the range of possible change to the timing and magnitude of streamflow. Additionally, this project will lead to a greater understanding of the complex hydrology of the Nooksack River basin, and a calibrated and validated hydrology model available for future simulations. 20

21 7.0 References Alexander, D., R.N. Palmer, and A. Polebitski (2007), Technical Memorandum #1: Literature review of research incorporating climate change into water resources planning, A report prepared by the Climate Change Technical Subcommittee of the Regional Water Supply Planning Process, Seattle, WA. Bach, A. (2002), Snowshed contributions to the Nooksack River Watershed, North Cascades Range, Washington, Geographical Review, 92(2), Cayan, D.R., S.A. Kammerdiener, M.D. Dettinger, J.M. Caprio, and D.H. Peterson (2001), Changes in the onset of spring in the western United States, Bulletin of the American Meteorological Society, 82, Chalecki, E.L., and P.H. Gleick (1999), A framework of ordered climate effects on water resources: A comprehensive bibliography, Journal of the American Water Resources Association, 35(6), Cuo, L., D.P. Lettenmaier, B.V. Mattheussen, P. Storck, and M. Wiley (2008), Hydrologic prediction for urban watersheds with the Distributed Hydrology-Soil-Vegetation Model, Hydrological Processes, 22, Donnell, C.A. (2007), Quantifying the glacial meltwater component of streamflow in the Middle Fork Nooksack River, Whatcom County, WA, using a distributed hydrology model, M.S. Thesis, Western Washington University. Dragovich, J.D., D.K. Norman, R.A. Haugerud, and P.T. Pringle (1997), Geologic map and interpreted geologic history of the Kendall and Deming 7.5 minute quadrangles, Western Whatcom County, Washington, Open file report 97-2, U.S. Department of Natural Resources. Federal Energy Regulatory Commission (FERC) (2004), Order on rehearing and dismissing petition as moot, Docket No. JR , Gleick, P.H., and E.L. Chalecki (1999), The impacts of climatic changes for water resources of the Colorado and Sacramento-San Joaquin River basins, Journal of the American Water Resources Association, 35(6), Golden, A. (1992), Soil survey of Whatcom County area, Washington, United States Department of Agriculture. Intergovernmental Panel on Climate Change (IPCC) (1997), An introduction to simple climate models used in the IPCC second assessment report, IPCC Technical Paper II, IPCC Working Group I, Cambridge University Press. IPCC (2000), IPCC Special Report: Emissions scenarios, Summary for policymakers, Cambridge University Press. 21

22 IPCC (2007) Climate Change 2007: The Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change [Solomon, S., D. Qin, M. Manning, Z. Chen, M. Marquis, K.B. Averyt, M. Tignor and H.L. Miller (eds.)], Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA, 996 pp. Kovanen, D.J., and D.J. Easterbrook (2001), Late Pleistocene, post-vashon alpine glaciations of the Nooksack drainage, North Cascades, Washington, GSA Bulletin, 113(2), Lettenmaier, D.P. and T.Y. Gan (1990), Hydrologic sensitivities of the Sacramento-San Joaquin River Basin, California, to global warming, Water Resources Research, 26(1), Lettenmaier, D.P., A.W. Wood, R.N. Palmer, E.F. Wood, and E.Z. Stakhiv (1999), Water resources implications of global warming: A U.S. regional perspective, Climatic Change, 43, Leung, L.R., and M.S. Wigmosta (1999), Potential climate change impacts on mountain watersheds in the Pacific Northwest, Journal of the American Water Resources Association, 35(6), Milly, P.C.D., J. Betancourt, M. Falkenmark, R.M. Hirsch, Z.W. Kundzewicz, D.P. Lettenmaier, R.J. Stouffer (2008), Climate change - Stationarity is dead: Whither water management? Science, 319, Mote, P.W., A.F. Hamlet, M.P. Clark, D.P. Lettenmaier (2005), Declining mountain snowpack in western North America, Bulletin of the Meteorological Society, 86(1), National Research Council (2006), Surface Temperature Reconstructions for the Last 2,000 Years, National Academy of Sciences, National Academy Press. Palmer, R.N. (2007a), Technical Memorandum #6: Framework for incorporating climate change into water resources planning, A reported prepared by the Climate Change Technical Subcommittee of the Regional Water Supply Planning Process, Seattle, WA. Palmer, R.N. (2007b), Final report of the Climate Change Technical Committee, A report prepared by the Climate Change Technical Subcommittee of the Regional Water Supply Planning Process, Seattle, WA. Palmer, R.N., M.W. Wiley, A. Polebitski, B. Enfield, K. King, C. O Neil, and L. Traynham (2006), Climate change building blocks, A report prepared by the Climate Change Technical Subcommittee of the Regional Water Supply Planning Process, Seattle, WA. Polebitski, A., M.W. Wiley, and R.N. Palmer (2007a), Technical Memorandum #2: Methodology for downscaling meteorological data for evaluating climate change, A report prepared by the Climate Change Technical Subcommittee of the Regional Water Supply Planning Process, Seattle, WA. 22

23 Polebitski, A., L. Traynham, and R.N. Palmer (2007b), Technical Memorandum #4: Approach for developing climate impacted meteorological data and its quality assurance/quality control, A report prepared by the Climate Change Technical Subcommittee of the Regional Water Supply Planning Process, Seattle, WA. Polebitski, A., L. Traynham, and R.N. Palmer (2007c), Technical Memorandum #5: Approach for developing climate impacted streamflow data and its quality assurance/quality control, A report prepared by the Climate Change Technical Subcommittee of the Regional Water Supply Planning Process, Seattle, WA. PRISM Group (2008), Oregon State University, created 23 June Regonda, S.K., B. Rajagopalan, M. Clark, and J. Pitlick (2005), Seasonal cycle shifts in hydroclimatology over the western United States, Journal of Climate, 18, Stewart, I.T., D.R. Cayan, and M.D. Dettinger (2004), Changes in snowmelt runoff timing in western North America under a Business as Usual climate change scenario, Climatic Change, 62, Storck, P., D.P. Lettenmaier, B.A. Connelly, T.W. Cundy (1995), Implications of forest practices on downstream flooding: Phase II Final Report, Washington Forest Protection Association, TFW-SH , 100. Storck, P., L. Bowling, P. Wetherbee, and D. Lettenmaier (1998), Application of a GIS-based distributed hydrology model for prediction of forest harvest effects on peak stream flow in the Pacific Northwest, Hydrologic Processes, 12, United States Geological Survey (USGS) (2008), USGS Water Science Center, Water Resources Inventory Area 1 Watershed Management, University of Washington, Water Resources Management and Drought Planning Group (2008), website description of new components included in Version 2.4 of DHSVM, Washington Department of Ecology (DOE) (1988), Instream Resources Protection Program Nooksack Water Resource Inventory Area (WRIA) 1, Chapter WAC, Washington State Department of Ecology (DOE) (2008), Watershed Planning, WRIA 1, Wigmosta, M.S., L.W. Vail, and D.P. Lettenmaier (1994), A distributed hydrology-vegetation model for complex terrain. Water Resources Research, 30, 6,

24 Wigmosta, M.S. and W.A. Perkins (2001), Simulating the effect of forest roads on watershed hydrology, in Land Use and Watersheds: Human Influence on Hydrology and Geomorphology in Urban and Forest Areas, M.S. Wigmosta and S.J. Burges, eds., AGU Water Science and Application, 2, Wigmosta, M.S., B. Nijssen, and P. Storck (2002), The distributed hydrology-soil-vegetation model, In Mathematical Models of Small Watershed Hydrology and Applications, V.P. Singh and D. Frevert, eds., Water Resources Publications, Wiley M.W., Palmer R.N., Salathe Jr., E.P. (2006), The development of GCM based climate scenarios for use in water resource impact evaluations, Submitted to Journal of Water Resources Planning and Management. Water Resources Inventory Area 1 (WRIA 1) (2008), Watershed Management Project, 24

25 8.0 Tables Table 1. Seasonal averages for predicted streamflow for the Sultan, Tolt, Cedar, Green, and White Rivers, Washington (Palmer, 2007b). 25

26 9.0 Figures Figure 1. Generalized map of the Nooksack River watershed (USGS, 2008). Figure 2. Mean annual precipitation ( ) in Water Resources Inventory Area (WRIA) 1, which includes the Nooksack watershed, Lake Whatcom watershed and several small adjacent watersheds; contours are in inches of precipitation (USGS, 2008). 26

27 Figure 3. Discharge (cfs) in the Middle Fork of the Nooksack River, measured near Deming, WA for WY2005 (USGS, 2008). Figure 4. Temperature anomaly ( C) indicated by the instrumental record and a variety of paleoclimate reconstructions; increasing uncertainty represented by darkening gray color (National Research Council, 2006). 27

28 Figure 5. Predicted 2050 hydrographs (colored lines) as compared to historic average (black) for the Sultan River, WA (CCTC, ). Figure s change in temperature and precipitation for the Pacific Northwest, as predicted by twenty GCM-emission scenario couples; three representative couples used by the CCTC, and to be used in this study are in bold (Mote and others, 2005). 28