Models and Calibratio n
-
Upload
quemby-stanley -
Category
Documents
-
view
22 -
download
0
description
Transcript of Models and Calibratio n
Models and Calibration
CBRFC 2011 Stakeholder Forum
November 3, 2011
2
CBRFC Forecast Groups
White Yampa
White Yampa Basin
ENMC2
White Yampa - Elk near Milner (ENMC2)
Upper (10000-11970)
Middle (8500-10000)
Lower (7205-8500)
White Yampa - Elk near Milner (ENMC2)
In reality the 3 areas (upper, middle and lower) are represented (simulated) by only 3 points (Lumped Model)
The inputs our model needs for calibrations and operations (at these 3 points) are:• precipitation• temperature• freezing level
Calibrations/Simulations - inputs
ENMC2 upper areaElevation = 10569 ftArea = 64 sqmi
ENMC2 middle areaElevation = 9146 ftArea =159 sqmi
ENMC2 lower areaElevation = 7692 ftArea =246 sqmi
For Elk River at Milner (ENMC2)
Each area (upper, middle and lower) MAP is built using precipitation stations that (hopefully) have similar characteristics to that area
For the ENMC2• Upper area – DRLC2 .46, ELKC2.46• Middle area - DRLC2 .46, ELKC2.46• Lower area - DRLC2 .46, ELKC2.46
These weights were chosen to guarantee water balance in each area. The water balance in each area was calculated using the PRISM sets
Calibrations/Simulations - Precipitation
White Yampa - Elk near Milner (ENMC2)Precipitation Gages
Nearby stations (whose climatology is known) area used to calculate the temperature at the mid-point elevation of the area (whose climatologies are calculated using the climatology of the nearby stations)
Temperature is calculated by using the difference in station and area climatology
For the ENMC2• Upper area – CRSC2 0.009, ELKC2 0.011• Middle area - SBTC2 0.009, CRSC2 0.013, ELKC2
0.045• Lower area - WBSW4 0.009, SBTC2 .02,HAYC2
0.013, ELKC2 0.019
Calibrations/Simulations - Temperature
White Yampa - Elk near Milner (ENMC2)Temperature Gages
Precipitation and temperature are calculated every six hours at each area within the basin
30 year historical record calculation• Used to calibrate hydrologic models Operationally done in a similar way• Ensures our forecasts will have similar
quality/characteristics to calibration For the Elk at Milner this is done for the upper,
middle and lower areas
Calibrations/Simulations - Inputs
A snow model is first run for each area in the basin• accumulates/ablates snow A soil moisture model is then run for each area• Controls amount of water from the snow model • retained in the soil• evaporates or• ends up in the stream Evaporation is a calibrated amount:• E=P-Q
Calibrations/Simulations - Models
SNOW ACCUMULATIONAND ABLATION MODEL (SNOW-
17)
Precipitationand
Air Temperature
Rainor
Snow
Energy Exchangeat
Snow-AirInterface
Snow CoverHeat Deficit
GroundMelt
SnowCover
OutflowRain plusMelt
Areal Extentof the
Snow Cover
Liquid WaterStorage
Transmissionof
Excess Water
AccumulatedSnow Cover
Rainon
Bare Ground
Deficit = 0
Bare ground or
snow cover
Sacramento Soil MoistureAccounting Model
TENSION WATER STORAGE
FREE WATER STORAGE
PRIMARYFREE
WATERSTORAGE
TENSIONWATER
STORAGE
TENSIONWATER
STORAGE
SUPPLEMENTARYFREE WATER
STORAGE
LOWER ZONE
UPPER ZONEDIRECTRUNOFF
INTERFLOW
SURFACERUNOFF
BASEFLOW
SUBSURFACEOUTFLOW
Calibrations - Results
Simulations - Real Time (no mods)
Simulations - Results (with mods)
19
Reservoir modeling is difficult as they are not physically based. However, we calibrate the reservoir models assuming two different modes:• Irrigation (use average releases)• Spillway/passflow
Operationally we can do the following:• Assume the current release • Input a schedule• Allow the spill/passflow rules
Calibrations/Simulations - Reservoirs
Reservoir Calibrations
Reservoir Simulations – Assume Constant Outflow
Reservoir Simulations – Use Rules
Reservoir Simulations – Use Rules
Reservoir Simulations – Release Schedule
Unregulated flow = Observed flow + Diversions (measured) + Storage
Natural flow = Unregulated flow + Consumptive Use Consumptive use (in basin irrigation) can only be estimated• In our simulations, we simulate natural flow but subtract
out the consumptive use so the output is always unregulated flow
So:• We simulate “natural flow”• We remove the in-basin irrigation (consumptive use)• This is the simulated unregulated flow. It simulates the
actual flow plus the measured diversions (adjusted flow) Operational considerations• Observed flow = Unregulated flow - Diversions - Storage
Adjustments to Flow
http://www.cbrfc.noaa.gov/wsup/doc.php
463 basins• 1139 sub areas (1-3 per basin)
88 reservoirs 22 routes (no snow model or soil model)
Calibrations/Simulations
30 year averages are updated once every 10 years
WY2011 Used:• 1971-2000 for averages• 1971-2000 for statistical prediction• 1976-2005 for ESP
Update for WY2012 will be based on 1981-2010 time series• New averages based on 1981-2010• New statistical prediction equations based on 1981-2010
period• Recalibration of 463 river basin models to the 1981-2010
period• Better effort made to create MAP and MAT time series• CBRFC will debut all of this for January 2012 forecasts
Recalibration
• Preliminary Data• 18% reduction in
mean
Effect on Forecasts
WY2012 forecasts will be based on 1981-2010 inputs in both forecast models• ESP and SWS will both use the same period
SNOTEL network much stronger for 1981-2010 period than in 1970s. This network is critical for forecast skill.
All things equal, these forecasts will be lower since input data sets are drier in the 30 year average• Especially true in early season forecasts• Later season forecasts more controlled by observed
snowpack
Percent of normal forecast values should remain largely unchanged (since normals AND forecasts will be lower)
Example: ESP forecastsMean influence to early season forecasts somewhat less
32
Example: ESP forecastsMean influence to late season forecasts small
33
Questions
34