MODELING THE SPOKANE RIVER-LAKE ROOSEVELT SYSTEM

Transcription

1 MODELING THE SPOKANE RIVER-LAKE ROOSEVELT SYSTEM Chris J. Berger 1, Robert L. Annear, Jr. 2, Michael L. McKillip 3, Vanessa Wells 4 and Scott A. Wells, ASCE 5 1 Senior Research Associate, Department of Civil and Environmental Engineering, Portland State University, P. O. Box 751, Portland, Oregon, , Voice: , FAX: , bergerc@ce.pdx.edu 2 Senior Engineer and Office Manager, DHI, Inc., 319 SW Washington, Suite 614, Portland, Oregon, 97204, Voice: , FAX: , robert.annear@dhi.us 3 Civil Engineer, Murray, Smith and Associates, 121 S.W. Salmon, Portland, Oregon, , Voice: , FAX: , mlm@msa-ep.com 4 Research Assistant, Department of Civil and Environmental Engineering, Portland State University, P. O. Box 751, Portland, Oregon, , Voice: , FAX: , vaniwells@hotmail.com 5 Professor and Chair, Department of Civil and Environmental Engineering, Portland State University, P. O. Box 751, Portland, Oregon, , Voice: , FAX: , scott@cecs.pdx.edu ABSTRACT Three hydrodynamic and water quality models of the Spokane River and Lake Roosevelt system were developed and linked together by Portland State University to develop Total Maximum Daily Loads (TMDL) for phosphorus. Model development was funded by the Washington Department of Ecology, the United States Environmental Protection Agency (EPA), and the Spokane Tribe. The Spokane River-Lake Roosevelt system consists of multiple river and reservoir sections and is located in the Northeastern part of Washington State and Idaho. The modeled section stretches from the Lake Coeur d Alene, Idaho to Lake Roosevelt - an impoundment behind Grand Coulee dam on the Columbia River. The goals of the modeling effort were to gather data to construct a computer simulation model of the Spokane River system including Lake Roosevelt, Long Lake Reservoir and the pools behind Nine Mile dam, Upper Falls dam, Upriver dam and Post Falls dam; and to ensure that the combined model accurately represents the system hydrodynamics and water quality. The hydrodynamic and water quality model CE-QUAL-W2 used to model the Spokane River-Lake Roosevelt system. CE-QUAL-W2 is a two dimensional (longitudinal-vertical), laterally averaged, hydrodynamic and water quality model that has been under development by the Corps of Engineers Waterways Experiments Station. Included in the model are all major dischargers and tributaries. The ability to simulate multiple CBOD compartments in CE-QUAL-W2 permitted the simulation of discharger specific CBOD compartments and decay rates. In general, the model reproduces the river and reservoir responses to the known boundary conditions. The model is well suited for evaluating the impacts of management strategies to improve water quality in the Spokane River and Lake Roosevelt. 33rd IAHR Congress: Water Engineering for a Sustainable Environment Copyright c 2009 by International Association of Hydraulic Engineering & Research (IAHR) ISBN:

2 INTRODUCTION The Spokane River-Lake Roosevelt system is located in the in Northern Idaho and the Northeastern part of Washington State and stretches from the Lake Coeur d'alene Idaho to Lake Roosevelt on the Columbia River. The modeled area includes Lake Roosevelt, Long Lake Reservoir and the pools behind Nine Mile dam, Upper Falls dam, Upriver dam and Post Falls dam Figure 1 shows the location of the modeled region of interest. Low dissolved oxygen (DO) concentrations and warm water temperatures have been measured in the Spokane River. Excessive water temperatures and low dissolved oxygen concentrations can affect salmonid rearing. There is significant stratification occurring in the Spokane arm of Lake Roosevelt and Long Lake, with water surface temperatures exceeding 25 C and hypolimnetic dissolved oxygen concentrations below 1 mg/l. Nutrient rich water originating from upstream in the Spokane River affects algae growth, and organic matter settling from the water column exerts a water column oxygen demand and a sediment oxygen demand after it settles. The Washington Department of Ecology is issuing a DO total maximum daily load (TMDL) for the Spokane River from the Idaho border to Long Lake Dam, which is just upstream of the Spokane Arm of Lake Roosevelt. The TMDL, along with a 401 certification for the FERC relicensing of Spokane River dams, will reduce phosphorus loadings and affect minimum in-stream flows in the Spokane River. The impact of the TMDL on water quality in the Spokane Arm is uncertain. CE-QUAL-W2 water quality of models of Lake Roosevelt and the Spokane River will be used to help determine the impact of the TMDL and the FERC relicensing on the Spokane Arm water quality. The models have been updated using recent data. The Lake Roosevelt model was originally developed by Portland State University for the Spokane Tribe of Indians and simulates temperature, dissolved oxygen, nutrients, algae, zooplankton, and organic matter (McKillip and Wells, 2006). The goals of this modeling effort were to: Construct a computer simulation model of the Spokane River system including Lake Roosevelt, Long Lake Reservoir and the pools behind Nine Mile dam, Upper Falls dam, Upriver dam and Post Falls dam Ensure that the model accurately represents the system hydrodynamics and water quality (flow, temperature, dissolved oxygen and nutrient dynamics) ensure that the combined model accurately represents the system hydrodynamics and water quality (flow, temperature, dissolved oxygen, phytoplankton, periphyton and nutrient dynamics) Develop and Run Modeling Scenarios 6224

3 Spokane River City of Spokane, WA Figure 1. Upper Spokane River in Washington MODEL SELECTION The model to be used for Spokane River-Lake Roosevelt system is the public domain model, CE-QUAL-W2 (Cole and Wells, 2008). This model is a 2-dimensional (longitudinal-vertical) hydrodynamic and water quality model capable of predicting water surface, velocity, temperature, nutrients, multiple algae, zooplankton, periphyton, and macrophyte species, dissolved oxygen, ph, alkalinity, multiple CBOD groups, multiple suspended solids groups, multiple generic constituents (such as tracer, bacteria, toxics), and multiple organic matter groups, both dissolved and particulate. The model is set up to predict these state variables at longitudinal segments and vertical layers. Typical model longitudinal resolution is between m; vertical resolution is usually between 0.5 m and 2 m. The model can also be used in quasi-3-d mode, where embayments are treated as separate model branches off the main stem of the reservoir. The user manual and documentation can be found at the PSU website for the model: Since 2000, this model has been used extensively throughout the world in 116 different countries. MODEL DATA NEEDS In order to model the system, the following data were required: Spokane River flow, Columbia River flow, water level and water quality data; tributary inflows and water quality; Meteorological conditions; bathymetry of the Spokane River, the dam pools along the river, Long Lake Reservoir, Lake Roosevelt; point source (wastewater treatment plants, WWTPs) inflows and water quality characteristics. Water quality data were provided primarily by the Washington State Department of Ecology (DOE) and the Spokane Tribe. Additional flow, temperature and water quality data were provided by the USGS in WA and ID, dischargers along the Spokane River and operators of some of the dam facilities. The data collected at sampling sites consisted of periodic grab samples, which were used to generate 6225

4 longitudinal profiles of the water quality parameters, and vertical profile data used for comparing vertical profiles in various parts of the river on the same day. MODEL FORCING DATA The model forcing data consist of the system bathymetry developed into the model grid; the boundary condition flow, temperature and water quality; the tributary and discharger flow, temperature and water quality; the groundwater flow, temperature, and water quality; and the meteorological forcing data. The river/reservoir system was broken into model branches and water-bodies based on (1) how groundwater inflow/recharge was computed for the Spokane River, (2) how the vertical slope changed from reach to reach, and (3) where there were pools or dams. The upstream boundary condition for the model is Lake Coeur d'alene and the downstream boundary condition is Grand Coulee dam on Lake Roosevelt. The upstream boundary condition was characterized by flow, water temperature, and water quality. The downstream boundary condition was characterized by flow rate. The model used internal interpolation to fill in the boundary conditions between the data. The tributary inflows of the Little Spokane River, Hangman Creek, Skalan Creek and Coulee Creek were also characterized using flow, temperature and water quality data. Groundwater inflows were modeled as distributed tributaries with flow rates estimated by analyzing river flow gains/losses between gaging stations. The water quality and temperature of the groundwater was estimated from well data. Water Quality data were available for the following constituents: alkalinity, chloride, conductivity, total dissolved solids, ammonia nitrogen, nitrite-nitrate nitrogen, phosphorus (soluble reactive phosphorus), fecal coliform, CBOD ultimate, and dissolved oxygen. Organic matter in the upstream boundary condition, tributaries, and point sources was simulated using CBOD ultimate data and multiple CBOD compartments in CE-QUAL-W2. Each point source had a separate CBOD compartment, and a decay rate, and the upstream boundary condition was had separate CBOD compartments and tributary CBOD were grouped into a single CBOD compartment. The decay rates used for each compartment were obtained from the Washington Department of Ecology. There are seven significant point sources modeled along the Spokane River including the Coeur D'Alene WWTP, Hayden Area POTW, Post Falls WWTP, Liberty Lake WWTP, Kaiser Aluminum, Inland Empire Paper Company and Spokane River WWTP. The data were obtained from the National Pollutant Discharge Elimination System (NPDES) through the WA Department of Ecology and additional data were obtained either directly from the dischargers or from WA Department of Ecology, which acquired the data from the dischargers. Each point source is characterized by flow, temperature, and additional water quality constituent concentrations. 6226

5 MODEL CALIBRATION The calibration effort focused on model predictions of hydrodynamics (flow and water level), temperature, and eutrophication model parameters (such as nutrients, algae, dissolved oxygen, organic matter, coliform). The model calibration period was from January 1, 2001 to October 31, 2001 and the January 1, to December (Lake Roosevelt model only). The hydrodynamic calibration of the Spokane River-Lake Roosevelt system included comparing model-data comparisons of flow rate and water level. Figure 2 compares water level data and model results in Lake Roosevelt for /31/05 2/9/06 3/21/06 4/30/06 6/9/06 7/19/06 8/28/06 10/7/06 11/16/06 12/26/ Water Surface Elevation, m Data, USACOE GCL Model Water Surface Elevation, ft Grand Coulee Dam Forebay (USACOE GCL) Model Segment Julian Day Figure 2. Water Level predictions in Grand Coulee Dam forebay in Lake Roosevelt compared with data for Temperature Calibration Parameters affecting temperature calibration included wind sheltering coefficients, groundwater inflow temperature, and the accurate representation of reservoir outflows. Temperature predictions in Long Lake and Nine Mile Reservoir were particularly sensitive to the wind-sheltering coefficient. In these reservoirs, wind sheltering was increased during the summer in order to simulate the reservoir s vertical temperature profile. For other sections of the river, wind-sheltering coefficients between 0.5 and 1.40 were applied for the entire year. Groundwater temperatures were estimated from well data. Figure 3 shows the model-data comparison in Long Lake for Average mean error (AME) and Root mean square error (RMS) were less than 1 degree Celsius

6 Elev, m NGVD Julian Day :47 8/8/ Temperature, C Julian Day :30 8/29/ Temperature, C Figure 3. Temperature predictions and data for station LL3 on Long Lake. The red triangles represent data and the blue line shows model predictions. Water Quality Calibration The general approach toward water quality calibration was to keep coefficient values close to commonly accepted literature values If during the process of calibration a particular combination of coefficient values did not produce good results, values would then be changed back to their default values, and a new avenue would be investigated. Vertical profile and time series water quality data were collected at multiple sites throughout in the Spokane basin. CBOD was modeled using separate CBOD groups for each discharger. This facilitated accurate simulation of the oxygen demand exerted by effluent originating from each discharger since each CBOD group had its own decay rate. The first-order decay rates of the CBOD compartments were developed from laboratory data supplied by the Washington Department of Ecology. Figure 4 shows the comparisons between model predictions and year 2006 vertical profile DO data in the Spokane Arm of Lake Roosevelt. Total Phosphorus predictions are compared with Long Lake 2001 vertical profile data in Figure

7 Figure 4. Comparisons between model predicted dissolved oxygen concentrations with data for 8/21/2006 in Lake Roosevelt. The Spokane Arm site are stations SA1 through SA6. The red triangles represent data and the blue line shows model predictions Elev, m NGVD Julian Day :00 8/9/ Total P, mg/l Julian Day :00 8/30/ Total P, mg/l Figure 5. Total phosphorus predictions and data for station LL3 on Long Lake. The red triangles represent data and the blue line shows model predictions. 6229

8 CONCLUSION This article summarizes the development and calibration of the CE-QUAL-W2 Version 3.6 model of the Spokane River from Lake Coeur d'alene to Lake Roosevelt. Since the CE-QUAL-W2 model allows the user to separate the river basin into separate branches (collections of model longitudinal segments or computational cells) and waterbodies (collections of branches with similar kinetic coefficients, turbulence closure, and meteorological forcing). The W2 model was composed of both riverine and reservoir sections, such as the Spokane River, Nine Mile Dam pool, Upriver Dam pool, Upper Falls Dam pool, Long Lake, and Lake Roosevelt. The system model required that boundary conditions and the topography of the river and reservoir sections be determined. This includes data such as dynamic inflow/discharge rates, dynamic inflow/discharge temperatures, dynamic inflow/discharge water quality constituents dynamic meteorological data (air temperature, dew point temperature, wind speed, wind direction and cloud cover or short wave solar radiation), and the bathymetry of each model segment. Model predictions were compared to field data for the following parameters: water level, flow rate, temperature, dissolved oxygen, nitrate-nitrite nitrogen, periphyton biomass, chlorophyll a, ammonia-nitrogen, total phosphorus, soluble reactive phosphorus and ph. Field data used in the model-data comparisons included near-surface grab sample data, continuous water quality data (temperature, dissolved oxygen and ph), and vertical profiles data. Grab sample data were compared to field measurements at over 15 river-reservoir locations along the Spokane River. Vertical profiles comparisons were made at over 20 locations. REFERENCES Cole, T. and Wells, S. (2008). CE-QUAL-W2: A Two-Dimensional, Laterally Averaged, Hydrodynamic and Water Quality Model, Version 3.6 Department of Civil and Environmental Engineering, Portland State University, Portland, OR. McKillip, M. and Wells, S. (2006). Lake Roosevelt Water Quality and Hydrodynamic Model Calibration with Fish Bioenergetics, Technical Report EWR-03-06, Department of Civil and Environmental Engineering, Portland State University, Portland, Oregon. 6230