Nov 17 Temp Modeling R10

COLUMBIA RIVERTEMPERATURE MODEL: 

EPA REGION 10 COLUMBIA RIVER TEMPERATURE MODEL

Water Quality Modeling in Region 10: 

Water Quality Modeling in Region 10 3 staff in Office of Environmental Assessment Modeling to support decisions under the Clean Water Act (e.g., TMDLs, NPDES permits) Support work for R10 Offices of Water, Ecosystems and Communities, and Tribal Office

Columbia River Temperature Issue: 

Columbia River Temperature Issue Water temperatures exceed water quality standards to protect fish – Clean Water Act requires a restoration plan called a TMDL States have asked EPA to take lead on TMDL Columbia and Snake Rivers span 3 state jurisdictions EPA has experience in this area (RBM10 model) High profile, heavy investment project R10’s best example of CREM guidance measures (e.g., model framework development, peer review, detailed documentation)

Slide5: 

Dworshak

HOW TO ESTIMATE RIVER TEMPERATURE?: 

HOW TO ESTIMATE RIVER TEMPERATURE?

Two Ways to Estimate Temperatures : 

Two Ways to Estimate Temperatures River Temperature Measurements (Measurement Model) Long term scroll case readings at dams No historic data for freely flowing river Energy Budget (Process Model)

Concept for Measurement Model: 

Concept for Measurement Model Cross-sectionally averaged river temperatures can be estimated based upon: Temperature Measurements at Dams (Scroll Case, Forebay, and/or Tailrace)

Slide11: 

Evaluating Different Measurement Types

Concept for Process Model: 

Concept for Process Model Cross-sectionally averaged river temperatures can be estimated based upon: Temperature Measurements at System Boundaries mainstem boundaries and tributary inflows Estimates of Surface Heat Exchange

State Estimation Approach: 

State Estimation Approach Treat both measurements and model estimates as random variables Alternate approach to comparing model outputs to measurements that are assumed to be correct Use Kalman filter algorithm to obtain “best” parameter estimates, accounting for uncertainty in both process model and measurements Differences between observed and predicted are unbiased and uncorrelated in time

Need for Process Model: 

Need for Process Model Oregon and Washington Standards for Temperature in Columbia/Snake Rivers Allow small increases to temperature above natural conditions (e.g., 0.14 deg C in summer) Virtually no measurements of un-impounded conditions available Need to estimate temperatures in both impounded and un-impounded conditions

MODEL FRAMEWORK DEVELOPMENT: 

MODEL FRAMEWORK DEVELOPMENT

Goals of Model Development: 

Goals of Model Development Develop a temperature model that: accurately simulates river temperatures supports a TMDL analysis Keep it non-proprietary, computationally simple and flexible Conduct External Peer Review 5 Reviewers (2 funded by EPA, others by industry and agencies)

System Features: 

System Features Run-of-River Reservoirs Vertical temperature stratification relatively low Water surface elevation is relatively constant points to potential utility of 1-D model with constant impoundment elevation previous 1-D studies of Columbia River

MODEL SELECTION: 

1-Dimensional, Time Dependent For Impounded Condition, Estimates of Water Temperature from Process and Measurement Models Treated as Random Variables Mixed Lagrangian-Eulerian solution technique “Reverse Particle Tracking” reduces error due to numerical dispersion reduces numerical instability reduces computational burden of uncertainty evaluation MODEL SELECTION

Model Name: 

Model Name River Basin Model developed in EPA Region 10 RBM10 is written in Fortran code and can be adapted to simulate any large scale river

Slide20: 

ONE-DIMENSIONAL ENERGY BUDGET MODEL SURFACE HEAT TRANSFER MAIN STEM OUTFLOW MAIN STEM INFLOW

Slide21: 

Standing Still Watching Water Flow thru Parcel (Euler)

Slide22: 

T2 = T1 + Net Surface Heat + Incremental Tributary Heat + Error where, T1 = initial temperature T2 = temperature after one time step Riding the Parcel of Water Downstream - Lagrangian

Slide23: 

RBM10 Calculation Method Mixed Lagrangian-Eulerian technique “Reverse Particle Tracking” reduces error due to numerical dispersion reduces numerical instability reduces computational burden of uncertainty evaluation allows incorporation of diffusion processes

Slide24: 

Reverse Particle Tracking 1. Compute travel time/velocities through each element 2. Track a parcel of water to its position at beginning of time step 3. Position will probably be between element boundaries 4. Estimate previous temp based on interpolation 5. Run parcel downstream, adding surface heat at each element 6. If element has a tributary, calculate flow-weighted avg temp 7. Continue adding heat and incorporating tribs 8. When time step has elapsed, stop

Elements of RBM10 Framework: 

Elements of RBM10 Framework Boundaries of Simulated System System Topology Geometry/Hydrodynamics Boundary Inputs (Flow and Temperature) Heat Budget Inputs based on Meteorology

Slide26: 

Lewiston Airport Snake River Clearwater above N.Fork Dworshak (North Fork) Snake @ Brownlee Dam Tucannon segments elements Boundaries and Topology Advective Inputs (flow and temperature) Weather Station Simulated Reach

Geometry/Hydrodynamics: 

Geometry/Hydrodynamics Mainstem Geometry for Free-Flowing Reaches Need cross-sectional profiles of the river bottom Open channel hydraulics relationships HEC-RAS model gradually varied flow is assumed provides cross-sectional area and top width over the range of observed flows area used to estimate velocity, width used to estimate surface area for heat exchange

Geometry/Hydrodynamics: 

Geometry/Hydrodynamics Freely-Flowing Reaches – cont. Leopold relationships determine width and cross sectional area for un-impounded reaches (e.g., Clearwater River) Y = a*X^b where, Y = cross sectional area or width X = flow a, b = coefficients based on regression

Geometry/Hydrodynamics: 

Geometry/Hydrodynamics Mainstem Geometry for Impounded Reaches Simple : constant geometry of model elements (e.g., cross-sectional area, width, volume) Exception: Grand Coulee Dam Storage reservoir with variable elevation Volume-elevation relationships are used to vary geometry of model elements

Geometry/Hydrodynamics: 

Geometry/Hydrodynamics Velocity Continuity V = Flow / X-Area For reservoir segments, x-area is constant For river segments, x-area varies based on flow

Slide31: 

Surface Heat Exchange Surface Heat Exchange = Net Absorbed Radiation - Water Dependent Exchange (J1 + J2) (J3 + J4 + J5) where, J1 = Net Solar Short Wave Radiation J2 = Net Atmospheric Long Wave Radiation J3 = Longwave Back Radiation From Water J4 = Conduction J5 = Evaporation

Slide32: 

Meteorological Data Needed to Compute Heat Budget Air Temperature Dew Point Wind Speed Atmospheric Pressure Cloud Cover

AVAILABLE DATA : 

AVAILABLE DATA

Available Data: 

Available Data On the one hand… Long term records are available for meteorology, tributary flow, and water temperature, enabling: long term simulations evaluation of system variability, and comparison of simulations to monitored temps

Slide35: 

On the other hand… Mainstem Temperature Monitoring Monitoring at Dams Not Designed for Assessment of River Temperature Limited Quality Control/Quality Assurance Tributary Temperature Monitoring Discontinuous Record Unknown Quality Control/Quality Assurance Meteorology Limited Geographical Coverage Data Limitations

IMPORTANT ASSUMPTIONS (and Sources of Uncertainty): 

IMPORTANT ASSUMPTIONS (and Sources of Uncertainty)

Important Assumptions: 

Important Assumptions Meteorology Described by five regional weather stations Mainstem Flow Constant elevation for impounded reaches except Grand Coulee Leopold relations developed from gradually-varied flow methods for un-impounded reaches Tributary Temperatures Mohseni relations developed from local air temperature and weekly/monthly river monitoring

Important Assumptions: 

Important Assumptions Groundwater Hyporheic flow does not significantly change the cross-sectionally averaged temperature in un-impounded conditions Measurement Model Tailrace monitoring represents best available measure of cross-sectionally averaged temperatures

PARAMETER ESTIMATION: 

PARAMETER ESTIMATION

PARAMETER ESTIMATION: 

Identify parameters that govern rates of energy transfer in the system Some are well known (e.g., solar declination) Some are less well known (e.g., evaporation rates) Two parameters that are less known are to be varied evaporation rates assignment of area covered by 5 meteorological stations PARAMETER ESTIMATION

ACCEPTANCE CRITERIA: 

Estimates for evaporation rates and meteorological station assignment are varied to satisfy criteria for model acceptance Acceptance criteria (Kalman Filter characteristics): Differences between simulated and measured are unbiased and uncorrelated in time Observed variance of differences equals theoretical variance of differences ACCEPTANCE CRITERIA

MODEL EVALUATION ANDTESTING (“CORROBORATION”): 

MODEL EVALUATION AND TESTING (“CORROBORATION”)

Model Evaluation Process: 

Model Evaluation Process A variety of evaluations have been conducted throughout the model development process, including: Measurement variation evaluation (forebay, tailrace, scrollcase) Comparison of simulated to measured (tailrace) temperatures Comparison of un-impounded simulations with measurements from nearly un-impounded river (one dam) Comparison: simulations vs transect measurements (special study) Review of other temperature modeling studies in the region Sensitivity to tributary inputs Code tests: Eulerian vs Reverse Particle Tracking schemes

Slide49: 

RBM10 Results for 1990-1994

Slide50: 

RISLEY (1997) - Tualatin River Max Mean Difference = 3 Deg C Mostly < 1 Deg C BATTELLE-MASS1 (2001) - Columbia River RMS Error = 0.59 - 1.52 Deg C HDR/PORTLAND STATE/IPC (1999) - Snake River AME = 0.6-2.3 Deg C (1992 data) AME = 0.5-2.0 Deg C (1995 data) CHEN (1996) - Grande Ronde River Error = -2.20 - 8.28 Deg C (Summer Max) Error = -1.21 - 7.69 Deg C (Avg 7-day Max) Error Estimates from Other Studies

MODEL APPLICATION: 

MODEL APPLICATION

Impact of Dams on Natural Condition: 

Impact of Dams on Natural Condition Two scenarios are run using identical boundary inputs (weather, tributary flows/temperatures, etc.) 1. Existing Condition 2. Un-impounded Condition Dams are “mathematically removed” – altered geometry Corroboration not feasible – no observations Un-impounded Condition is not the “natural condition” – model domain does not reach to headwaters

Slide55: 

Impact of Individual Dams on Daily Cross Sectional Average Temperature in the Columbia River

Slide56: 

Impact of Tributaries on Mainstem Temperatures

Slide57: 

Impact of Point Sources on Mainstem Temperatures Simulated Increases in Temperature at River Mile 42 in the Columbia River due to the Existing Point Sources

Conclusions: 

Conclusions Water temperatures in the Columbia and Snake rivers are elevated in the late summer and fall in the impounded river system compared to the un-impounded river system. Development of hydroelectric projects on the Columbia and Snake rivers have resulted in major changes in the temperature regimes of these rivers. Tributaries and point sources have a small effect on mainstem temperatures except at the upper end of the Snake River (e.g., Salmon R.) To meet water quality standards, actions should focus on measures the hydro system can make to reduce the temporal shift in water temperatures toward un-impounded conditions.

Conclusions - continued: 

Conclusions - continued 1-D model can be used to simulate temperatures in the Columbia and Snake mainstems Inter-annual variability in river temperatures is significant and complicates TMDL allocation process Both point measurements and process model simulations are valuable sources of information from which to estimate cross-sectional average river temperature Reverse Particle Tracking is a promising solution technique for minimizing numerical dispersion and model run time

Documentation: 

Documentation Yearsley, J., Karna D., Peene S., and Watson, B. 2001. Application of a 1-D Heat Budget Model to the Columbia River System. EPA 910-R-01-004, EPA Region 10, Seattle, WA.