Kosta Georgakakos and Nick Graham HYDROLOGIC RESEARCH …€¦ · 25-02-2010 · GEORGIA WATER...

02/25/2010HRC-GWRI NCEP 1

HYDROLOGIC RESEARCH CENTER

GEORGIA WATER RESOURCES INSTITUTE

Kosta Georgakakos and Nick Graham

Aris Georgakakos


The INFORM Demonstration ProjectIntegrated Forecast and Reservoir Management

Phase I (2003 – 2009) Development and ImplementationDemonstration

Phase II (2009 – 2012)EnhancementsTransition to Operations


Increase efficiency of water use in Northern California using climate, hydrologic and decision science


Implement an integrated forecast-management system for the Northern California reservoirs using real-time data and operational forecast models

Perform tests with actual data and with management input

Demonstrate the utility of climate and hydrologic forecasts for water resources management in Northern California

Transition to Operations


123.5 123 122.5 122 121.5 121 120.5

38.5

39

39.5

40

40.5

41

41.5

0 500 1000 1500 2000 2500 3000 3500 4000

Elevation (meters)

Degrees West Longitude

Degr

ees N

orth

Lat

itude

Major Resevoirs in Nothern California

TrinityShasta

Orovil le

Folsom

Trinity River

Pit River

Sacramento River

Feather River

N. Fork American RiverOroville

Folsom

TrinityShasta


Sponsors:CALFED Bay Delta Authority California Energy CommissionNational Oceanic and Atmospheric Administration

Collaborators:California Department of Water Resources California-Nevada River Forecast CenterSacramento Area Flood Control Agency U.S. Army Corps of Engineers U.S. Bureau of ReclamationNational Centers of Environmental Prediction

SPONSORS-COLLABORATORS


HRC‐GWRI, 2006. Integrated Forecast and Reservoir Management (INFORM) for Northern California: System Development and Initial Demonstration. California Energy Commission, PIER Energy‐Related Environmental Research. CEC‐500‐2006‐109, 244pp. and 9 Appendices

http://www.energy.ca.gov/pier/final_project_reports/CEC-500-2006-109.html

http://www.hrc-lab.org/projects/dsp_projectSubPage.php?subpage=inform


Generate consistent real-time forcing sequences of rainfall and temperature.

River/ReservoirPlanning & Management

Simulate soil moisture, evapotranspiration, runoff, and streamflow.

GFS & CFS INPUTDownscaling

Simulate current and adaptive mgt. policiesand assess impactson water uses.

Watershed Hydrology

Economic and otherbenefits


San Joaquin River

San Luis

Clair Engle Lake

Trinity Power Plant

Lewiston

Lewiston

JF Carr

Whiskeytown

Shasta

Keswick

ShastaSpring Cr

Keswick

Oroville

Thermalito

Folsom

Natoma

New Melones

Tulloch

Goodwin

Oroville

Folsom

Nimbus

Melones

Tracy Pumping

Banks Pumping

San Joaquin River

Amer

ican

Riv

er

Feat

her R

iver

Sacramento River

Trinity River Clear Creek

Yuba River

Bear River

Delta-Mendota Canal

California Aqueduct

O’Neill Forebay

To Dos Amigos PP

To Mendota Pool

Sacramento San Joaquin River DeltaReservoir/

Lake

Power Plant

Pumping Plant

River Node

Reservoir/Lake

Power Plant

Pumping Plant

Reservoir/Lake

Power Plant

Pumping Plant

River Node

ISV

IFT

IES,IMC,IYB,ITI

DDLT,DBS,DCCWD,DNBA

DDM

DFDM

DDA

DSF

DSB

Black Butte

New Bullards Bar

Trinity River System (Clair Engle Lake, Trinity Power Plant, Lewiston Lake, Lewiston Plant, JF Carr Plant, Whiskeytown, Clear Creek, and Spring Creek Plant);

Shasta Lake System (Shasta Lake, Shasta Power Plant, Keswick Lake, Keswick Plant, and the river reach from Keswick to Wilkins);

Feather River System (Oroville Lake, Oroville Power Plants, Thermalito Diversion Pond, Yuba River, and Bear River);

American River System (Folsom Lake, Folsom Plant, Natoma Lake, Nimbus Plant, Natoma Plant, and Natoma Diversions);

San Joaquin River System (New Melones Lake, New Melones Power Plant, Tulloch Lake, Demands from Goodwin, and Inflows from the main San Joaquin River); and

Bay Delta (Delta Inflows, Delta Exports, Coordinated Operation Agreement--COA, and Delta Environmental Requirements).

Objectives:Water SupplyEnergy GenerationEnvironmentEcologyRecreation


ATM FORECASTS

Adjustments

HYDROLOGIC MODEL

Ensemble Forecast

DECISION MODEL MANAGER

Release Schedule

ASSESSMENT SYSTEM

Management Benefits

ATM DATA


Decision Maker

ACTUAL VIRTUAL

Real Time Input

System

Characteristics

Flexible

Trade-offs

Measurable

Benefits

Other

BenefitsMeasurable

Benefits

Other

Benefits

Fixed Operation

Rules


UPSTREAM ANDLATERAL FLOW

HYDROLOGIC MODELS

DOWNSCALINGBIAS ADJUSTMENT

DECISION MODELSDOWNSTREAM FLOW MODELS

CNRFC

DWR/USBR

INFORM SYSTEM

RESERVOIRRELEASES

TRADE-OFFS/MANAGEMENTINPUT

GFS & CFS FORECASTS

NCEP


1-16


Distributed Tributary Basin System for Oroville Example of INFORM Hydrology Modeling


Phase I - Assessment Summary:

1. Orographic model downscaling without bias adjustment tends to overestimate (underestimate) precipitation in high (low) elevations (for areas of order 500 km). Temporal variability is reasonably well represented.

2. Temperature downscaling exhibits generally small bias with good diurnal variation.

3. Simulation with adaptations of the operational snow-soil models exhibits generally good performance with some late lags for high events of FNF.

4. Forecasts of 2-day inflow volume appear reasonable but require adjustment for bias for the range of GFS-driven ensembles to include the observed FNFs.


FolsomSimulation


OrovilleSimulation


ShastaSimulation


TrinitySimulation


MF American Lwr


Pit RiverUpr


Sacr. River@ Delta Lwr


Shasta Inflow


Folsom Inflow


San Joaquin River

San Luis

Clair Engle Lake

Trinity Power Plant

Lewiston

Lewiston

JF Carr

Whiskeytown

Shasta

Keswick

ShastaSpring Cr

Keswick

Oroville

Thermalito

Folsom

Natoma

New Melones

Tulloch

Goodwin

Oroville

Folsom

Nimbus

Melones

Tracy Pumping

Banks Pumping

San Joaquin River

Amer

ican

Riv

er

Feat

her R

iver

Sacramento River

Trinity River Clear Creek

Yuba River

Bear River

Delta-Mendota Canal

California Aqueduct

O’Neill Forebay

To Dos Amigos PP

To Mendota Pool

Sacramento San Joaquin River DeltaReservoir/

Lake

Power Plant

Pumping Plant

River Node

Reservoir/Lake

Power Plant

Pumping Plant

Reservoir/Lake

Power Plant

Pumping Plant

River Node

ISV

IFT

IES,IMC,IYB,ITI

DDLT,DBS,DCCWD,DNBA

DDM

DFDM

DDA

DSF

DSB

Black Butte

New Bullards Bar

System SchematicTrinity River System (Clair Engle Lake, Trinity Power Plant, Lewiston Lake, Lewiston Plant, JF Carr Plant, Whiskeytown, Clear Creek, and Spring Creek Plant);

Shasta Lake System (Shasta Lake, Shasta Power Plant, Keswick Lake, Keswick Plant, and the river reach from Keswick to Wilkins);

Feather River System (Oroville Lake, Oroville Power Plants, Thermalito Diversion Pond, Yuba River, and Bear River);

American River System (Folsom Lake, Folsom Plant, Natoma Lake, Nimbus Plant, Natoma Plant, and Natoma Diversions);

San Joaquin River System (New Melones Lake, New Melones Power Plant, Tulloch Lake, Demands from Goodwin, and Inflows from the main San Joaquin River); and

Bay Delta (Delta Inflows, Delta Exports, Coordinated Operation Agreement--COA, and Delta Environmental Requirements).

03/20/2008HRC-GWRI INFORM 30

Water DistributionFlow RegulationHydro Plant OperationEmergency Response

Monthly Decisions• Releases/EnergyTarget Conditions• State Variables

Planning Tradeoffs

• Water Supply/Allocation• Energy Generation• Carry-over Storage• Env.-Ecosystem Management

Development Tradeoffs

• Urban/Industrial • Agriculture• Power System• Socio-economic & Ecological

Sustainability

Operational Tradeoffs

• Flood Management• Water Distribution• Energy Generation• Env.-Ecosystem Management

Benefit/Impact Functions• Water Supply• Energy• Flood Damage• Env.-Ecosystem

Scenario/Policy Assessment

Monthly / Several Decades

Actual Hydrologic Conditions

Actual Demands

Climate-Hydrologic Forecasts

Demand Forecasts• Water• Food • Energy• Env.-Ecosystem

Climate-Hydrologic Forecasts

Demand Forecasts• Water Supply• Power Load/Tariffs• Flood Damage• Env.-Ecosystem Targets

Inflow Scenarios

Development/Demand Scenarios• Water/Energy• Water/Benefit Sharing• Environmental Sustainability

Daily Decisions• Releases/EnergyTarget Conditions• State Variables

Benefit/Impact Functions• Water Supply• Energy• Flood Damage• Env.-Ecosystem

Near Real Time Decision Support

Hourly / 1 Day

Mid/Short Range Decision Support

Daily, 6-Hourly, or Hourly / 1 Month

Long Range Decision Support

Weekly, 10-Day or Monthly / 1-2 Years

Infrastructure Develpmnt.Water Sharing CompactsSustainability Targets

Management Policy

INFORM DSS: OverviewMultiple Objectives, Time Scales, & Decision Makers

Man

agem

ent

Agen

cies

/Dec

isio

ns

Plan

nin

g A

gen

cies

/Dec

isio

ns

Oper

atio

nal

Pla

nnin

g a

nd M

anag

emen

tO

ff-lin

eA

sses

smen

ts

02/25/2010 30HRC-GWRI NCEP


Set- up:

• Forecasted inflows were provided by HRC with start date March 1st (112 traces, 9 month horizon, and five locations: Clair Engle Lake, Shasta, Oroville, Folsom, and Yuba);

• Historical monthly average values are used for locations where forecasted inflows are not available;

• Monthly reservoir parameters and constraints (max, min, and target storage levels; evaporation rates);

• Minimum river flow and Bay Delta requirements; and

• Base monthly demands at all locations;

DSS objective: Develop the tradeoff between water supply deliveries and carry over storage that meets all other stated requirements.

Spring 2006, 2007, and 2008 Case Studies



Forecasted Inflow Mean Comparison

1727

6988

5108

3139

2511

9184

5174

3578

2009

7195

40673482

1802

4805

35883334

0

1000

2000

3000

4000

5000

6000

7000

8000

9000

10000

Trinity Shasta Oroville Folsom

Inflo

w (c

fs)

Historical F2006F2007F2008

Mean 9- month Inflow Forecasts Comparison: 2006, 2007, 2008



Mean Monthly Inflow Forecasts Comparison: 2006, 2007, 2008 Forecasted Inflow Means - Trinity

0

1000

2000

3000

4000

5000

6000

7000

8000

Mar-01 Apr-01 May-01 Jun-01 Jul-01 Aug-01 Sep-01 Oct-01 Nov-01

cfs

His.F2006F2007F2008

Forecasted Inflow Means - Shasta

0

5000

10000

15000

20000

25000


cfs

His.F2006F2007F2008

Forecasted Inflow Means - Oroville

0

2000

4000

6000

8000

10000

12000

14000

16000

18000


cfs

His.F2006F2007F2008

Forecasted Inflow Means - Folsom

0

2000

4000

6000

8000

10000

12000


cfs

His.F2006F2007F2008



March 1 Reservoir Storages: 2006, 2007, 2008

Reservoir Initial Storages On March 1st

2013

3872

2992

463

2020 20191902

3786

2997

594

20021895

1490

2660

1456

375

1532

1774

0

500

1000

1500

2000

2500

3000

3500

4000

4500

Trinity Shasta Oroville Folsom New Melones San Lius

Stor

age

(100

0 AF

)

Y2006Y2007Y2008



Water Deliveries vs. Carry over Storage vs. Energy Tradeoffs Total Demand Fraction vs. Terminal Storage Tradeoff

8871

8040

7210

6373

5533

88808372

8078

7478

6718

4000

5000

6000

7000

8000

9000

10000

0 0.1 0.2 0.3 0.4 0.5 0.6

Demand Fraction

Term

inal

Sto

rage

(100

0 A

F)

Y2008

Historical Mean

Total Demand Fraction vs. System Energy Tradeoff

3736

4164

4556

4923

5254

47554932

5233

5515

5802

3000

3500

4000

4500

5000

5500

6000

0 0.1 0.2 0.3 0.4 0.5 0.6

Demand Fraction

Ener

gy (G

WH

)

Y2008

Historical Mean



Mean Carry Over Storage and Energy Comparisons: 50% Base Demand

Simulated Total Terminal Mean Storage Comparison

95378995

5533

0

2000

4000

6000

8000

10000

12000

Y2006 Y2007 Y2008

1000

AF

Simulated System Mean Energy Comparison

7841

6946

5254

0

1000

2000

3000

4000

5000

6000

7000

8000

9000

Y2006 Y2007 Y2008

GW

H



Lake Level Forecast Ensembles: 2008

10% Base Demand 50% Base Demand



X2 Location Forecast Ensembles: 2008

10% Base Demand



System Carryover Storages of Major Reservoirs

0

2000

4000

6000

8000

10000

12000

2006 2007 2008

TAF

Initial

Carryover9537(INFORM)

8995(INFORM)

5633(INFORM)


Carryover Storage Comparisons

0

2000

4000

6000

8000

10000

12000

2006 2007 2008

TAF

ObservedDSS AVGDSS MinDSS MAX

Energy Generation from Major Plants

0

1000

2000

3000

4000

5000

6000

7000

8000

9000

2006 2007 2008

GW

H

ObservedDSS AVGDSS MINDSS MAX

Relevant and decision worthy – Energy Commission funded Phase II

High Visibility and Impact Project for Forecast and Management Agencies in California


Objective: Test Utility of using dynamic downscaling beyond 16 days

Why?So that physically consistent and coherent events produced by the CFS are downscaled in a way that preserves significant precipitation or warming episodes in the watersheds of interest

Potential Benefit:Reduce uncertainty bounds for significant episodes for forecast lead times out to 30 – 45 days.


NCEP Data (Ken Mitchell and Rongqian Yang – thank you):

Two case studies:16 CFS runs with start dates at twelve-hourly intervals extending 45-days each

Case A:CFS run period 12/02/2005-12/09/2005INFORM forecast evaluation period 12/09/2005 06Z – 01/16/2006 00Z

Case B:CFS run period 01/30/2006-02/06/2006INFORM forecast evaluation period 02/06/2006 06Z – 03/16/2006 00Z


Upgrade and implementation of INFORM multicomputer platform to allow timely processing of 20 GFS ensembles for better statistical estimation of uncertainty

Development of bias adjustment procedures that combine physical model improvements and statistical methodologies (benefits to all leads)

Utilization of 3-D CFS fields to produce downscaling of precipitation and temperature products commensurate with those from GFS (benefits to 16 – 30/45 day forecasts)

Continued validation of system forecasts and demonstration assessments


Available data not directly usable for dynamic downscaling (e.g., onlyPrecipitable water is available for downloading)

For INFORM we developed a post-processor to convert the available data to level data usable with INFORM dynamic downscaling models (very Approximate and temporary until 3-D CFS data becomes available)

Issues of transmission from NCEP to HRC and CNRFC in California


Z T Q U V1000 X X X

850 X X X X X

700 X X X X X

500 X X X X X

400 X X X X X

200 X X X X

Z U V P.WAT1000 X

850 X X X

700 X X X

500 X X X

200 X X X

N/A X

OAKLAND RADIOSONDE DATAMONTHLY 1981-2004

BY MONTH:

- CALCULATE & SAVE MONTHLY MEANS, STD. DEV.- REMOVE MEANS, STANDARDIZE DATA -> Y* & Z*- CALCULATE EOFS OF YOB & ZOB

- GET α,e s.t. Y* = α et and β,f s.t. Z* = β ft

LINEAR REGRESSION: = A Y*

with A = et Ct f and C=<α β>/(<α2>1/2<β2>1/2) (orthogonality)

SAVE A (monthly regression coefficients;

hindcast skill ~ 0.8-0.9 for Q 850 and 700

YOB as in CFS ZOB (for downscaling)

EOF > EOFregression

Z

Z U V P.WAT1000 X

850 X X X

700 X

500 X

200 X

N/A X

CFS DATA - MONTHLY 1981-2004LEAD TIMES OUT TO 60 days, 12 hourly

BY MONTH:- CALCULATE & SAVE MONTHLY MEANS, STD. DEV. OF Ycfs

Then, for each sounding out to 60 days lead time:

-REMOVE MEANS, STANDARDIZE -> Y*cfs

- USE REGRESSION COEFFICIENTS *cfs = A Y*cfs

-TRANSFORM *cfs to observations using OBS means and STD DEVs

AVAILABLE CFS AUGMENTED CFS = YcfsZ U V P.WAT

1000 X

850 X X X

700 X X X

500 X X X

200 X X X

N/A X

GEOSTROPHY

Z

Z


Operational GFS and CFS forecast information is shown to have beneficial impactfor regional water resources planning and management in Northern Californiaif used within the integrated forecast and reservoir management proceduresof INFORM

But there are improvements needed for adoption in operational use:

Availability of CFS forecasts out to 45 days in form suitable for use withdynamic downscaling procedures

Improved method of delivery to CNRFC of real-time ensemble forecast products (both GFS and CFS) for use in the INFORM system

Kosta Georgakakos and Nick Graham HYDROLOGIC RESEARCH …€¦ · 25-02-2010 · GEORGIA WATER...

Documents

Transcript of Kosta Georgakakos and Nick Graham HYDROLOGIC RESEARCH …€¦ · 25-02-2010 · GEORGIA WATER...