NWS Calibration Workshop, LMRFC March, 2009 slide - 1
Strategy for Interactive Calibration
of theSacramento Model
Using NWSRFSInteractive Calibration Program (ICP)
NWS Calibration Workshop, LMRFC March, 2009 slide - 2
General Considerations when using ICP • Change duration/scale depending on flow components/parameters being examined• Remove large errors in parameter values whenever detected• Periodically return to previous steps to recheck results• Primary statistics to periodically examine:
• annual bias• seasonal bias• flow interval bias
• Remain flexible• Think! Ask questions before you view the simulation
• What do I expect to see?• What secondary effects could I see?• Why didn’t I see what I expected?
NWS Calibration Workshop, LMRFC March, 2009 slide - 3
NWS Calibration Workshop, LMRFC March, 2009 slide - 4
NWS Calibration Workshop, LMRFC March, 2009 slide - 5
NWS Calibration Workshop, LMRFC March, 2009 slide - 6
Sacramento Model Calibration Strategy
Start with a priori parameters
1. Remove large errors, especially volume
2. Adjust low flow parameters to get a reasonable baseflow simulation
3. Adjust tension water capacities
4. Adjust parameters that primarily affect storm runoff
5. Final adjustments to improve seasonal and flow interval bias patterns
NWS Calibration Workshop, LMRFC March, 2009 slide - 7
Snow-17 Calibration Strategy
• Remove large errors, – timing of snowmelt runoff. – form of precipitation
• Adjust major parameters– MFMAX, MFMIN– SCF– UADJ– SI
NWS Calibration Workshop, LMRFC March, 2009 slide - 8
Sacramento Model Calibration
• Each parameter is designed to represent a specific portion of the hydrograph under certain moisture conditions
• Concentrate on having each parameter serve its primary function rather than overall goodness of fit.
NWS Calibration Workshop, LMRFC March, 2009 slide - 9
• Check water balance – annual bias should be less than 10-20%
• Check for large timing errors, most common:– Storm runoff/baseflow ration in error (raise or lower
entire percolation curve)– Amount of surface runoff incorrect (adjust UZFWM)– Improper channel response function for Sacramento
Model (remove interflow from channel response unit hydrograph)
Sacramento Model Calibration StrategyStep 1 – Remove Large Errors
NWS Calibration Workshop, LMRFC March, 2009 slide - 10
Sacramento Model Calibration StrategyStep 2 – Reasonable baseflow simulation
• Identify primary baseflow component
• Adjust key parameters:– LZPK, LZSK, LZFPM, LZFSM, PFREE
• May need to adjust ZPERC and REXP
• Determine if riparian evaporation exists and determine general magnitude of RIVA (then set RIVA back to zero)
NWS Calibration Workshop, LMRFC March, 2009 slide - 11
Sacramento Model Calibration StrategyStep 3 – Tension water capacities
• Determine UZTWM and LZTWM based on periods when maximum soil moisture deficits occur.
• While examining UZTWM, check and adjust value of PCTIM
NWS Calibration Workshop, LMRFC March, 2009 slide - 12
Sacramento Model Calibration StrategyStep 4 – Storm runoff simulation
• Get proper division between surface runoff and interflow by changing UZFWM
• Adjust UZK to get correct timing of interflow.• Refine percolation curve over large range of
LZDEFR values– Primarily adjust ZPERC and REXP– Use ICP percolation analysis feature
• Determine if ADIMP is needed. If so, determine proper value.
NWS Calibration Workshop, LMRFC March, 2009 slide - 13
Sacramento Model Calibration StrategyStep 5 – Final Adjustments
• Determine value of RIVA if riparian evaporation exists
• Adjust ET-Demand values to improve seasonal bias pattern (alter by changing monthly PE adjustment curve).
• Refine timing of peaks by modifying channel response (unit hydrograph)
• Raise or lower percolation curve to improve flow interval bias pattern by changing LZFSM and LZFPM by the same ratio.
NWS Calibration Workshop, LMRFC March, 2009 slide - 14
Parameter Relationships when Watershed Divided into Sub-Areas
• Keep parameter values the same, except when the hydrograph from one sub-area can be isolated (then can modify parameters for the sub-area influencing that response).
• Relationships can be established between certain values based on soils, vegetation, etc, (then maintain that ration (ratio, diff) as parameter values are adjusted.
NWS Calibration Workshop, LMRFC March, 2009 slide - 15
Calibration of HL-RDHM
NWS Calibration Workshop, LMRFC March, 2009 slide - 16
Calibration of SAC Parameters with Scalar Multipliers
• Use of scalar multipliers (assumed to be uniform over a basin) greatly reduces the number of parameters to be calibrated. This assumes the spatial distribution of a-priori parameters is realistic.
• Parameters from 1-hour, lumped model calibrations can be a good starting point. Lumped model parameters, if derived at the 1 hour time scale, can be used to derive initial scalar multipliers, i.e.
multiplier = [lumped model parameter]/[basin average of gridded a-priori parameter values]
• Scalar multipliers are adjusted using similar strategies and objectives to those for lumped calibration
• Both manual and a combination of automatic and manual calibration on scalar multipliers have proven effective
NWS Calibration Workshop, LMRFC March, 2009 slide - 17
1 Get observed streamflow data
Same
2 Get outlet lat-lon, HRAP coordinates, lat-lon, drainage area
Get drainage area and lat-lon boundaries (or line segment definition, for MAPX)
3 Estimate channel routing parameters at outlet and generate parameter grids
Estimate unit graph (manually from hydrograph or using empirical method, e.g. IHABBs)
4 Add outlet to connectivity file
Not required
5 Adjust pixel areas to match USGS areas
Not required
6 Prepare HL-RDHM input deck
Prepare MCP3 input deck
7 Not required Run CAP to get mean a-priori parameter estimates (SAC and PE); enter values into MCP3 deck
8 Not required Run MAPX
9 Run HL-RDHM Run MCP3
10 Iteratively adjust scale factors for selected parameters option to meet calibration objectives (manual or automatic).
Iteratively adjust parameter values to meet calibration objectives (manual or automatic)
11 Plot mean precipitation, simulated, and observed flow and runoff time series to assist with parameter adjustments (e.g. using ICP).
Same.
12 No equivalent. Can output and plot time series of mean states or states at selected points but this information does not necessarily provide clear guidance on parameter changes. An R script is provided to assist with routing parameter adjustment.
Using ICP, examine the time variation of model states, percolation curve, and unit graph values to help with parameter adjustments.
13 Examine simulation statistics using STAT-QME and/or STAT_Q.
Same
14 Visually examine the spatial patterns of inputs, parameters, and model results (XDMS or GIS software).
No equivalent.
Comparison Between Calibration Steps for Distributed and Lumped Modeling
Distributed Lumped Distributed Lumped
NWS Calibration Workshop, LMRFC March, 2009 slide - 18
Manual Headwater Calibration• Follow similar strategies to those used for lumped
calibration except make changes to multipliers, e.g. from Anderson (2002):
– “Remove large errors– Obtain reasonable simulation of baseflow– Adjust major snow model parameters, if snow
is included\– Adjust tension water capacities– Adjust parameters that primarily affect storm
runoff– Make final parameter adjustments”
Can still use PLOT-TS and STAT-QME
• Stat-Q event statistics summarize how well you do on bias, peaks, timing, and RMSE, etc over any # of selected events. • R scripts assist with routing parameter adjustment.
See HL-RDHM User Manual for a detailed example.
NWS Calibration Workshop, LMRFC March, 2009 slide - 19
HL-RDHM
SAC-SMA, SAC-HT
Channel routing
SNOW -17
P, T & ET
surface runoff
rain + melt
Flows and state variables
base flowHillslope routing
AutoCalibration
Execute these components in a loop to find the set of scalar multipliers thatminimize the objective function
NWS Calibration Workshop, LMRFC March, 2009 slide - 20
km
iiksiko
k k
XqqJ1
2
,,,,1
2
1n
Multi-Scale Objective Function (MSOF)
• Minimize errors over hourly, daily, weekly, monthly intervals (k=1,2,3,4…n)
• q = flow averaged over time interval k
• n = number of flow intervals for averaging
• mk = number of ordinates for each interval
• X = parameter set
k1
-Assumes uncertainty in simulated streamflow is proportionalto the variability of the observed flow-Inversely proportional to the errors at the respective scales. Assume errors approximated by std.
Emulates multi-scale nature of manual calibration
k1
Weight =
NWS Calibration Workshop, LMRFC March, 2009 slide - 21
Beforeautocalibrationof a prioriparameters
After autocalibration
Observed
Auto Calibration: Case 2Example of HL-RDHM Auto Calibration: ELDO2 for DMIP 2
Arithmetic Scale
NWS Calibration Workshop, LMRFC March, 2009 slide - 22
Possible Strategy
• Start with best a-priori or scaled lumped parameters
• Run automatic calibration
• Make manual adjustments (particularly for routing parameters) to get the preferred storm event shapes
NWS Calibration Workshop, LMRFC March, 2009 slide - 23