Colorado Basin River Forecast Center Water Supply Forecasting Method Michelle Stokes Hydrologist in Charge Colorado Basin River Forecast Center April 28,

Page 1

Colorado Basin River Forecast CenterWater Supply Forecasting Method

Michelle StokesHydrologist in Charge

Colorado Basin River Forecast CenterApril 28, 2014

Page 2

Outline

• Overview of Colorado Basin River Forecast Center (CBRFC)

• CBRFC Hydrologic Model• Water Supply Forecasts Products• Improving the Forecasts

Page 3

Streamflow ForecastHydrographs:

• Flood forecasts • Recreational use• Deterministic

• Updated daily• 1-6 hour time step• 10 days into the future

www.cbrfc.noaa.gov

Page 4

Water Supply Forecasts:

• Forecast volume• Probabilistic

• Usually monthly, or seasonal time step

• Official forecast updated monthly – Guidance updated daily

• 1 to 5 years into the future

www.cbrfc.noaa.gov

Page 5

• Continuous – runs all the time, not just during events.• Conceptual – physically based, but uses parameters in

place of hard-to-get data.• Lumped – uses mean areal inputs (temperature,

precipitation); not distributed.

CBRFC Hydrologic Model

Page 6

CBRFC Hydrologic ModelConceptual Model

• RFC forecast uses a snow model and a rainfall-runoff model:

– SNOW-17: Temperature index model for simulating snowpack accumulation and melt

– Sacramento Soil Moisture Accounting Model: Hydrologic model used to generate runoff

Page 7

Ensemble Streamflow Prediction (ESP)• generate ensemble of hydrographs• generate probabilistic forecasts

CBRFC Hydrologic ModelComposed of three major interrelated components.

Calibration System (CS)• determine model parameters• store historical data

Daily Operational Forecasts (DOF)• generate short term deterministic

river forecasts• maintain model states

Page 8

Calibration System (CS)

• Choose from a variety of models and processes– Snow accumulation and ablation– Soil moisture model– Unit Hydrograph– Channel routing– Reservoir operations

• Determine the optimal set of parameters for each model to best simulate flow

• Store historical precipitation, temperature and flow time series for the basin

ESPDOF

Page 9

Daily Operational Forecast(DOF)

• Keeps track of model states, including soil moisture and snowpack

• Quality Controlled Inputs– Observed precipitation, temperature, and streamflow – Forecast precipitation (5 days) and temperature (10 days)

• Model adjusted by forecasters in real time• Run multiple times per day so there is continual quality control,

updating and adjusting• Outputs 10 day regulated deterministic streamflow forecast

CS

ESP

Page 10

Ensemble Streamflow Prediction (ESP)

• Uses model states from DOF as starting point and can also use forecast precipitation (5 days) and temperature (10 days) inputs

• Uses historical precipitation and temperature time series from CS and statistical distributions to derive probabilistic flow forecasts

• Can adjust output for model bias

CS

DOF

Page 11

Ensemble Streamflow Prediction Probabilistic Forecasts

198119821983….2010

Current hydrologic states : River / Res. LevelsSoil MoistureSnowpack

-> Future Time Past <-

• Start with current conditions (from the daily model run)• Apply precipitation and temperature from each historical year

(1981-2010)• A forecast is generated for each of the years (1981-2010)

as if, going forward, that year will happen• This creates 30 possible future streamflow patterns.

Each year is given a 1/30 chance of occurring

Page 12

ESP Probabilistic Forecasts

1. The flows are summed into volumes for the period of interest (typically April 1 – July 31)

2. The statistics are simplified3. 50% exceedance value approximates

the most probable forecast

# EMPIRICAL SAMPLE POINTS# Cond. #Trace Year Data Exceed. # year Weight Point Prob. # --------------------------------- 1981 0.033 10583427.0 0.290 1982 0.033 8372498.00 0.806 1983 0.033 12646544.0 0.065 1984 0.033 11904022.0 0.129 1985 0.033 11402967.0 0.161 1986 0.033 10406237.0 0.355 1987 0.033 8369501.00 0.839 1988 0.033 8719326.00 0.742 1989 0.033 7605042.50 0.935 1990 0.033 9761623.00 0.452 1991 0.033 9690117.00 0.484 1992 0.033 9298360.00 0.613 1993 0.033 10987106.0 0.226 1994 0.033 9395003.00 0.548 1995 0.033 14388755.0 0.032 1996 0.033 8611564.00 0.774 1997 0.033 10736442.0 0.258 1998 0.033 10159611.0 0.419 1999 0.033 12520652.0 0.097 2000 0.033 8252478.50 0.871 2001 0.033 9312369.00 0.581 2002 0.033 6439105.00 0.968 2003 0.033 9439112.00 0.516 2004 0.033 8867351.00 0.710 2005 0.033 10415361.0 0.323 2006 0.033 8235550.00 0.903 2007 0.033 8964843.00 0.645 2008 0.033 8954274.00 0.677 2009 0.033 11320183.0 0.194 2010 0.033 10185848.0 0.387

# Exceedance Conditional # Probabilities Simulation # ----------------------------- 0.900 8237243.000 0.800 8420311.000 0.700 8893428.000 0.600 9303964.000 0.500 9564614.000 0.400 10175353.000 0.300 10533006.000 0.200 11253565.000 0.100 12458982.000

12

3

Page 13

Water Supply Forecasts• Target is April-July unregulated runoff volume.• Mostly ESP; sometimes adjusted based on hydrologist

expertise or additional model guidance• Official forecast issued once a month January through June• ESP guidance run daily

Page 14

Forecast Evolution Plot

10%

30%

50%70%90%

RAW ESP OFFICIAL FORECAST

10%

50%

90%

Page 15

Water Year Outlooksfor Reclamation’s 24-Month Study

• Forecasts of monthly unregulated volumes through the end of the water year (September).

• Uses the raw ESP monthly volumes along with the official April-July water supply forecast.– Ensure individual month volumes for April through July

equal official April-July total volume.

• Issued the beginning of each month.– Begin issuing the following water year in June.

Page 16

24-MonthStudy ModelUnregulated Inflow Forecast

Upper Basin Inflows and Model Run Duration (Most Probable)

16

Month Issued Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov

Jan RFC RFC RFC

Official

A-JOfficial

A-JOfficial

A-JOfficial

A-JESP Jan

ESP Jan

inter-polate

inter-polate

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

Feb RFC RFC RFC

Official

A-JOfficial

A-JOfficial

A-JESP Feb

ESP Feb

inter-polate

inter-polate

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

Mar RFC RFC RFC

Official

A-JOfficial

A-JESP Mar

ESP Mar

inter-polate

inter-polate

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

Apr RFC RFC RFC

Official

A-JESP Apr

ESP Apr

inter-polate

inter-polate

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

May RFC RFC RFCESP May

ESP May

inter-polate

inter-polate

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

Jun RFC RFC RFCESP Jun

ESP Jun

ESP Jun

ESP Jun

ESP Jun

ESP Jun

ESP Jun

ESP Jun

ESP Jun

ESP Jun

ESP Jun

ESP Jun

ESP Jun

inter-polate

inter-polate

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

Jul RFC RFC RFCESP Jul

ESP Jul

ESP Jul

ESP Jul

ESP Jul

ESP Jul

ESP Jul

ESP Jul

ESP Jul

ESP Jul

ESP Jul

ESP Jul

inter-polate

inter-polate

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

Aug RFC RFC RFCESP Aug

ESP Aug

ESP Aug

ESP Aug

ESP Aug

ESP Aug

ESP Aug

ESP Aug

ESP Aug

ESP Aug

ESP Aug

inter-polate

inter-polate

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

Sep RFC RFC RFCESP Sep

ESP Sep

ESP Sep

ESP Sep

ESP Sep

ESP Sep

ESP Sep

ESP Sep

ESP Sep

ESP Sep

inter-polate

inter-polate

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

Oct RFC RFC RFCESP Oct

ESP Oct

ESP Oct

ESP Oct

ESP Oct

ESP Oct

ESP Oct

ESP Oct

ESP Oct

inter-polate

inter-polate

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

Nov RFC RFC RFCESP Nov

ESP Nov

ESP Nov

ESP Nov

ESP Nov

ESP Nov

ESP Nov

ESP Nov

inter-polate

inter-polate

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

Dec RFC RFC RFCESP Dec

ESP Dec

ESP Dec

ESP Dec

ESP Dec

ESP Dec

ESP Dec

inter-polate

inter-polate

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

81-10 Avg

* Light grey text indicates that the model is run in this month, however, only results for the first 24 months of the model run (black text) are published in the 24 Month Study report

ESP monthly values are generated using the RFC ESP forecasted volume for the water year using the current month's initial hydrological conditions. The RFC provides monthly volumes consistent with the 3-month forecast and the water year ESP volume.

April-July Unregulated Inflow

Interpolated values are calculated by UCBOR and are based on percent of the 81-10 average. The method takes the precent of average of the previous month's forecast value and interpolates over two months to the percent of average for the month following the interpolation period. This is done to smoothly transition between the end of the current water year and the next water year.

WY 2013 Source of Monthly Unregulated Inflow for Upper Colorado Reservoirs in the 24 Month Study

RFC values are issued by the Colorado Basin River Forecast Center (RFC) as the official forecast values for the next three-month period of time. The values are calculated using Statistical Water Supply (SWS) modeling and/or Ensemble Streamflow Predictions (ESP) modeling. This official forecast has the least amount of error associated with it.

Official A-J values are official forecast values issued by the RFC for the April-July runoff period using SWS and ESP, with some coordination with the NRCS. Apr-Jul water supply forecast volume is disaggregated by the RFC.

81-10 Avg values are the monthly average inflow values generated from water years 1981-2010 calculated using the database maintained by the Bureau of Reclamation Upper Colorado Region (UCBOR). A water year begins October 1 and ends September 30.

Most Probable

Page 17

Page 18

Improving CBRFC Forecasts

• Improved snow modeling– Collaboration with Utah State University– Collaborations with Jet Propulsion Lab to use NASA snow

cover grids to inform model

• Incorporating climate information– Hydrologic Ensemble Forecasting System (HEFS)– Climate informed ESP

• Post adjusting ESP

Page 19

Thank you!

Michelle StokesHydrologist in Charge

Colorado Basin River Forecast [email protected]