1. Hydrology Laboratory Research Modeling System (HL-RMS) Introduction: Office of Hydrologic Development National Weather Service National Oceanic and Atmospheric Administration Fekadu Moreda Presented to Mid-Atlantic River Forecasting Center February, 15, 2005

2. Over View 1)Historical Perspective 2)Motivation 3)Definition of a Distributed Hydrologic Model 4)Structure of HL-RMS and Components 5)Parameterization 6)Forcings (Precipitation, Temperature, Evaporation) 7)Case Study

3. (1) Historical Perspective Rational formula Unit Hydrograph Event based model Continuous simulation models Semi-distributed models Fully Distributed models

4. (2) Motivation for distributed Models Availability of high resolution data: basin properties and /forcings Better stream flow forecasting River and flash flood forecasting, Soil moisture products Snow cover Potential extension to environmental models –Non-point source pollution –Land-use change (can account for burn areas) –Erosion studies –Landslide/mudslide/soil strength applications Land-atmosphere interactions for meteorological and climate applications Groundwater recharge and contamination studies Others

5. (3) Definition of a Distributed Hydrologic model -(informal definition) a model which accounts for the spatial variability of factors affecting runoff generation : -precipitation -temperature -terrain -soils -vegetation -land use -channel shape

6. Discharge hydrograph at the outlet Generic Modeling Steps Lumped Model Distributed Model Discharge hydrograph at any model element Lumped runoff and soil moisture states Distributed runoff and Soil moisture states Apply distributed routing model Apply unit hydrograph Derive mean areal precipitation (MAP) Compute basin runoff Derive model element precipitation Compute model element runoff

7. 1.Rainfall, properties averaged over basin 2.One rainfall/runoff model 3.Prediction at only one point 1.Rainfall, properties in each grid 2.Rainfall/runoff model in each grid 3.Prediction at many points Lumped Distributed Hydrologic Modeling Approaches

8. Hydrology Lab Distributed Model ( HL-Research Modeling System HL-RMS) Modular, flexible modeling system Gridded (or small basin) structure Independent rain+melt calculations for each grid cell (SNOW-17) Independent rainfall-runoff calculations for each grid cell –Sacramento Soil Moisture Accounting (SAC-SMA) –Continuous Antecedent Precipitation Index (CONT-API) Grid to grid routing of runoff (kinematic) Channel routing (kinematic & Muskingum-Cunge)

9. The surface and base flow components for each grid is obtained from a SAC-SMA rainfall –runoff model HL-RMS Elements

10. 1 st Quadrant 4 th Quadrant 3 rd Quadrant 2 nd Quadrant SMI/SMIX=1.0 =0.9 F s =FRSX.CR AI F FRSX FsFs 0.0 0.5 1.0 F g =CG (AI f -AICR) FgFg AICR AI f AI API AIXW.CW API AIXD.CD API AIXD AIXW The API MODEL The surface and base flow components for each grid is obtained from a CONT-API rainfall –runoff model

11. SNOW-17 SNOW 17 model is used in each element

12. Distributed routing Translates distributed runoff into distributed stream flow With distributed routing, flow velocity in each element is dependent on flow level Different flows (states) are computed for each element in a stream network. Unit graph only produces flows at basin outlets. Commonly used approach: numerical solution to the 1-D equations for momentum and mass conservation 2. Lumped and distributed modeling

13. Surface Runoff SAC-SMA /CONT-API Base flow Hillslope routing Channel routing Components of HL-RMS SNOW Model SNOW-17 Stream Flow (P, T) rain+melt

14. (4) Parameterization a)Basic watershed properties b)SNOW-17 model parameters c)Cont-API parameters d)Routing parameters

15. (a) Basic watershed properties Digital Elevation Model (DEM) Available for each of RFC with 400m resolution. 4km resolution (HRAP) is used in HL-RMS Directly used in the SNOW-17 model Flow Direction and Accumulations are derived from DEM Location of outlets (lat long  HRAP) Connectivity file – ASCII file

16. Connectivity of Pixels Basins in MARFC Saxton

17. Connectivity file

18. (b) SNOW-17 Parameter Grids Ongoing work to develop distributed snow parameters Use of Elevation (DEM) at HRAP grid cell The traditional snow depletion curve may be replaced by two methods. –i) Assuming SI=0 => for a given time step in a pixel this snow or no snow –ii) Assuming a 45 degree depletion line for each grid. Since the 4km grid is much smaller than a a basin scale, this method will assume uniform coverage and depletion in a pixel

19. (c) CONT –API Parameter Grids A priori parameters for 11 parameters derived from lumped model Use lumped model parameters for others Use the Evaporation index only No frozen ground option Parameters can be replaced by a lumped value or adjusted by a factor

20. (d) Routing Parameter Grids  Hillslope routing parameter grids: Hillslope slope (S h ) Hillslope roughness (n h ) Channel density (D)  Channel routing: Channel slope (S c ) Channel roughness (n c ) Channel width and shape parameters (a, b) -OR- Specific discharge (a) and shape parameter (b) from a discharge cross-sectional area relationship (a, b)

21. METHOD TO ESTIMATE CHANNEL ROUTING PARAMETERS Momentum equation describing steady, uniform flow: – Q is flow [L3/T] – A is cross-section area [L2] Parameters a and b must be estimated for each model grid cell. Basic Idea: (1) Estimate channel parameters at basin outlet using USGS flow measurement data. (2) Estimate parameters in upstream cells using relationships from geomorphology and hydraulics. Two methods are being tested: 3. METHOD TO ESTIMATE CHANNEL ROUTING PARAMETERS Momentum equation describing steady, uniform flow: Q is flow [L 3 /T] A is cross-section area [L 2 ] Parameters a and b must be estimated for each model grid cell. Basic Idea: (1) Estimate channel parameters at basin outlet using USGS flow measurement data. (2) Estimate parameters in upstream cells using relationships from geomorphology and hydraulics. Two methods are being tested: 3. METHOD TO ESTIMATE CHANNEL ROUTING PARAMETERS Momentum equation describing steady, uniform flow: Q is flow [L 3 /T] A is cross-section area [L 2 ] Parameters a and b must be estimated for each model grid cell. Basic Idea: (1) Estimate channel parameters at basin outlet using USGS flow measurement data. (2) Estimate parameters in upstream cells using relationships from geomorphology and hydraulics. Two methods are being tested: Parameters a and b must be estimated for each model grid cell.

22. Channel Shape Method: (Tokar and Johnson 1995) (Gorbunov 1971) Channel Shape Method: 1.Assume simple channel shape. (B = width, H = depth) 2.From USGS data, estimate , , and channel roughness (n) at the outlet 3.Using an empirical equation, estimate local parameter n c using channel slope (S o ) and drainage area (F o ) at the outlet. Estimate n i at upstream cells. 4.For a selected flow level at the outlet, estimate spatially variable  i values (for each cell i) using Q i and A i estimates derived from geomorphological relationships (see below) 5. Assume  is spatially constant within a basin and compute a i and b i at each cell using  i , and n i, Rating Curve Method: 1.Determine a o and b o at the outlet directly from regression on the flow measurement data. 2.Using the same geomorphological relationships as in the channel shape method, equations for estimating a i and b i can be derived: Geomorphological Assumptions: 1.On average, flow is a simple function of drainage area and downstream flow. Leopold (1994), Figure 5.7 suggests  may vary from 0.65 to 1 in different parts of the U.S. Results shown here use  = 1 and  = 0.8. 2.On average, cross-sectional area of flow can be related to stream order. R l is Horton’s length ratio, k is stream order Assume simple channel shape. (B = width, H = depth) From USGS data, estimate α, β, and channel roughness (n) at the outlet Using an empirical equation, estimate local parameter nc using channel slope (So) and drainage area (Fo) at the outlet. Estimate ni at upstream cells. For a selected flow level at the outlet, estimate spatially variable ai values (for each cell i) using Qi and Ai estimates derived from geomorphological relationships (see below) Assume β is spatially constant within a basin and compute ai and bi at each cell using ai b, and ni, (Tokar and Johnson 1995)

23. Rating Curve Method: Determine ao and bo at the outlet directly from regression on the flow measurement data. Using the same geomorphological relationships as in the channel shape method, equations for estimating ai and bi can be derived: –Geomorphological Assumptions: On average, flow is a simple function of drainage area and downstream flow. Leopold (1994), Figure 5.7 suggests g may vary from 0.65 to 1 in different parts of the U.S. Results shown here use g = 1 and g = 0.8. On average, cross-sectional area of flow can be related to stream order. Rl is Horton’s length ratio, k is stream order (Tokar and Johnson 1995) (Gorbunov 1971)

24. 6) Forcings a)Gridded Precipitation b)Temperature c)Evaporation

25. (a) Gridded precipitation Gridded products archived: http://dipper.nws.noaa.gov/hdsb/data/nexrad/nexrad.html http://dipper.nws.noaa.gov/hdsb/data/nexrad/nexrad.html -available products: –GAGEONLY –RMOSAIC –MPE (XMRG) –One file for one hour for the entire RFC

26. (b) Gridded Temperature Gridded products archived are available: Hydrometeorology group: David Kitzmiller Use of the MAT for the basins to generate grid products Requires –A program to generate grids –Basin definitions (connectivity file) –MAT for each basin –Elevation map –Regional lapse rate

27. (c) Gridded Evaporation Evaporation is essential for CONT-API Only the evaporation option is tested For now we will use seasonal evaporations Monthly adjustments are used Maps are available in CAP (Calibration Assistant Program)

28. (7) Case study Juniata River Basin (11 subbasins)

29. First HL-RMS Run for Juniata Outlet, Juniata at Newport Saxton, Interior point Williamsburg, Interior point - Model resolution 4km x 4km - Total number of pixels =497 - Watershed area = 8687 km 2 - Model parameters = a priori - Channel parameters are derived from USGS measurements at New port.

30. Comparison of simulation

31. Performance Statistics

32. Summary Introduced distributed hydrologic modeling Develop skill in handling distributed data, parameter, and output Distributed model complements the existing operation Opportunities in future to apply to small basins, interior points for flash flood