NOAA Efforts to Improve Precipitation and Flood Forecasting by Studying American River

Robert K. Hartman, Hydrologist-in-Charge NOAA, National Weather Service California-Nevada River Forecast Center 3310 El Camino Avenue, Suite 227 Sacramento, CA 95821 Tel: 916-979-3056, x322 Fax: 916-979-3067 Email: Robert.Hartman@noaa.gov Web: http://www.wrh.noaa.gov/cnrfc

BIOGRAPHICAL SKETCH Robert Hartman is the Hydrologist-in-Charge at the California-Nevada River Forecast Center. Mr. Hartman has 23 years of federal flood and water supply forecasting experience including 9 years in his current position. Mr. Hartman’s experience includes all aspects of RFC operations, development, and management obtained in positions at three National Weather Service River Forecast Centers, the USDA/Soil Conservation Service, and the NWS Office of Hydrology. Mr. Hartman has a B.S. degree in Forestry from Utah State University and an M.S. degree in Watershed Management from the University of Arizona.

ABSTRACT

The National Oceanic and Atmospheric Administration (NOAA) plans to undertake two related research efforts to improve precipitation, flood, and water resources forecasting. The presentation will briefly describe the Hydrometeorological Testbed (HMT) and the Distributed Model Intercomparison Project: Phase 2 (DMIP 2) with emphasis on the research being proposed for the North Fork of the American River. The HMT program is being developed for the purpose of advancing water resources data assimilation. The general strategy of this effort is to conduct research and development to deploy advanced systems for observed information to support critical decision making for flood mitigation, hydropower energy generation, water resources control, and fisheries management. More specifically, high resolution atmospheric and hydrometeorologic observations (precipitation, soil moisture, snowpack, winds, temperature, and moisture) will be collected and analyzed for several key water resource applications such as distributed hydrologic model validation, quantitative precipitation forecast (QPF), and estimation (QPE) validation. DMIP 2 continues the work begun under DMIP 1 to conduct research into advanced hydrologic models for river and water resources forecasting. Twelve groups from the hydrologic research community participated in DMIP 1, resulting in a wealth of knowledge for the scientific community and valuable guidance for the National Weather Service (NWS) research program. DMIP 2 is designed around two themes: 1) continued investigation of science questions pertinent to the DMIP 1 test sites, and 2) distributed and lumped model tests in hydrologically complex basins in the mountainous Western US.

112

NO AA Efforts to Im prove Precip itation and F lood Forecasting

by S tudying Am erican R iver

Robert K . Hartm anNOAA/NW S/CNRFC

A Two Part Episode…

•• NO AA H ydrom eteorological NO AA H ydrom eteorological TestbedTestbed• David K ingsm ill

•• University of Colorado / C IR ES and NO AA Environm ental University of C olorado / C IR ES and NO AA Environm ental Technology Laboratory Technology Laboratory -- Boulder, COBoulder, C O

•• D istributed M odel D istributed M odel In tercom parisonIntercom parison Project 2Project 2• M ike Sm ith

•• NO AA NO AA –– NW S NW S –– Office of H ydrologic Developm ent Office of Hydrologic D evelopm ent –– Silver S ilver Spring, M DSpring, M D

113

W hy an HMT?•• Integrated P lanning Team for Hydrological Integrated Planning Team for Hydrological

Services Report (Septem ber 2002)Services Report (Septem ber 2002)• Key Inform ation Gaps

•• Quantitative Precipitation Inform ation (QPE & QPF)Quantitative Precipitation Inform ation (QPE & QPF)•• Unbiased Precipitation Estimates with Reliability Unbiased Precipitation Estimates with Reliability

InformationInformation•• Reliable 0Reliable 0 --6 Hour 6 Hour PrecipPrecip . . NowcastsNowcasts•• LongerLonger-- term, Unbiased Model term, Unbiased Model QPFsQPFs

•• Hydrologic ForecastingHydrologic Forecasting•• HighHigh --resolution Flash Flood Forecasts & W arningsresolution Flash Flood Forecasts & W arnings•• Probabilistic R iver Forecasts Using 0Probabilistic R iver Forecasts Using 0 --3 day QPF and 3 day QPF and

Seasonal Precipitation ForecastsSeasonal Precipitation Forecasts

• Outstanding R&D Needs•• Im plem ent Hydrom eteorological Im plem ent Hydrom eteorological TestbedsTestbeds to to

Dem onstrate & Evaluate NextDem onstrate & Evaluate Next--Generation Datasets, Generation Datasets, Forecast Techniques, & M odelsForecast Techniques, & M odels

•• System atically evaluate prom ising new m ethods that can System atically evaluate prom ising new m ethods that can influence both NW P and influence both NW P and nowcastingnowcasting using the m anusing the m an --m achine m ix forecasting parad igm .m achine m ix forecasting parad igm .

•• Assess their va lue in term s of im proved regional Assess their va lue in term s of im proved reg ional perform ance on F lood/F lash F lood W arning and Q PF perform ance on F lood/F lash F lood W arning and Q PF G PRA m easures. G PRA m easures.

•• Use these resu lts as an objective basis for decisions on Use these results as an objective basis for decisions on transitions to operations both in the test region and transitions to operations both in the test region and nationally. nationally.

HM T G oals

114

HM T Concept

M arty RalphNO AA/ETL-PACJET

Profiler/RASS/Met (ETL)Profiler/RASS (Other Agency)X-band/Met/Rain gaugeS-band/Met/DisdrometerWSR-88D (NEXRAD)

HMTMicrophysics Array

H M T - 2 0 0 4

BB Y = B odega B ay

BSC = B ig Sulfu r Creek

CVD = C loverdale

CZC = Cazadero

FRS = Fort Ross

G RK = G oat R ock

HM T-2004 Russian R iver

HBG = Healdsburg

HLD = H oplan d

LSN = Lake S on om a

RO D = R io D ell

SP T = S alt P oin t

-123.8 -123.6 -123.4 -123.2 -123.0 -122.8 -122.6

38.2

38.4

38.6

38.8

39.0

1002003004005006007008009001000110012001300

Elev. (m)

S O N O M A

M E N D O C I N O

M A R I N

L A K E

N A P A

Pt. Arena

PacificOcean

BBY

CZC

FRS

Pt.Reyes N.S.

50 km

HMT-2004 Microphysics Array

Sfc. met. + rain gaugeAdditional rain gaugesSoil-moisture probesGPS IPWX-band polarimetric scanning radarS-band precip profiler915-MHz wind profilerRaindrop disdrometerGPS rawinsondesRiver gaugeOcean buoy

SanPablo Bay

CVD

HLD

LSN

BSC

SPT

GRK

Gualala

HBG

ROD

115

III

B a nd I

B a nd II

O nset o f Ban d II

O nset o f B and I

ra in

sn ow

onset of band I

B a nd I

B a nd II

St at ion E leva tio n

III

O nset o f Ban d II

O nset o f B and I

sn ow level

Russian R iver HM TBodega Bay vs. G rass Valley

W inder Profiler Data

X-band

0 5 10 15 20 25 30Reflectivity (dBZ) at 0.5 deg e levation

W SR-88D radar at KM UX0102 UTC 14 Jan 03

KMU X

10

20

30

40

50

60 km

NO AA/ETL X-band radar at FRS0059 UTC 14 Jan 03

Reflectivity (dBZ) at 1.0 deg e levation

0 10 20 30 40 50

G ap-F illing X-band Radarvs. W SR-88D

• Nearest NEXR AD radar sees no significant echoes approaching flood-prone w atershed

• NO AA/ETL’s Coastal X-band radar fills NEXR AD gap

116

Proposed HM T Tim eline

Am erican R iver Basin HM T2005-2008

117

Am erican R iver Basin HM T2005-2008

•• Focus on Q PE (observations) and shortFocus on Q PE (observations) and short -- term term Q PF (forecasts). A lso freezing level and snow Q PF (forecasts). A lso freezing level and snow level inform ation.level inform ation.

•• Focus on North Fork Am ericanFocus on North Fork Am erican• Concentration of data collection effort to develop a

“truth set” upon which solid research, developm ent, and direction can be based.

HM T Participation

•• Project M anagem entProject M anagem ent• David K ingsm ill – NO AA/O AR/ETL

•• NO AA PartnersNO AA Partners• O AR(ETL,NSSL,FSL,CDC,AO M L)• NW S(O HD,HPC,NO HR SC,W R,ER,EM C)• NESDIS(O RA)

• O ther Partners

118

Distributed M odelIntercom parison Project G O ALS

Engage the hydro logic m odeling com m unity to:Engage the hydro logic m odeling com m unity to:

•• Im prove our understanding of the capabilities and Im prove our understanding of the capabilities and lim itations of distributed hydrologic m odels when lim itations of d istributed hydrologic m odels when forced by operational quality m eteorological dataforced by operational quality m eteorological data

•• G uide research and developm ent in support of the G uide research and developm ent in support of the NO AA’s NW S m ission to forecast river and stream NO AA’s NW S m ission to forecast river and stream conditions. conditions.

DM IP 1 Partic ipants

1)1) M assachusetts Institute of M assachusetts Institute of Technology (M IT )Technology (M IT)

2)2) Hydrologic Research C enter H ydrologic Research C enter (HRC)(HR C)

3)3) DH I W ater and Environm entD HI W ater and Environm ent4)4) Univ . of ArizonaU niv. of Arizona5)5) National C enters for N ational C enters for

Environm ental Prediction Environm ental Prediction (NCEP)(NC EP)

6)6) Univ . of O klahom aU niv. of O klahom a7)7) Univ . of W aterloo, O ntario, U niv . of W aterloo, O ntario,

CanadaC anada8)8) U tah State Univ . and National U tah State Univ . and National

Institute of W ater Research Institute of W ater Research (N IW R), New Zealand(N IW R), N ew Zealand

9)9) U .S. NO AA National W eather U .S. NO AA National W eather Serv ice O ffice of Hydrologic Serv ice O ffice of Hydrologic D evelopm ent (O HD)Developm ent (O HD)

10)10) U .S. Departm ent of Agriculture U .S. Departm ent of Agriculture (USDA) (U SDA) B lacklandBlackland Research Research C enterCenter

11)11) Univ . of California at BerkeleyUniv . of California at Berkeley12)12) The Hydraulic and E lectrica l The Hydraulic and E lectrica l

C ollege of College of W uH anW uHan U niv., C hinaUniv., China

119

DM IP 1 Basins and T im elinehttp://www.nws.noaa.gov/oh/hrl/dm ip/

1

ab

1. Blue R. at B lue, O K2. Ill. R. a t Tahlequah, OK3. Baron Fork at Eldon, OK4. Peacheater Cr. at Christie, OK5. Flint Cr. a t Kansas, O K6. Ill. R. a t Watts, OK7. Ill. R. a t S avoy, OK8. Elk R. a t Tif f City, MO.

Ungaged LocationsGaged Locations

a. B lup1b. B lup2c. E ldp1d. Wt tp1e. Tifp1

BlueRiver

2 3

56 7

4

8 e

d

c2 3

56 7

4

8 e

d

c23

56 7

4

8 e

d

c

Elk River

Illinois River

8 G auged Basins Ranging in S ize from 65 – 2500 km 2

Jan – Jun : DM IP plan developed

2000

Jan : Data availab le on web

M arch : Participants subm itted sim ulations

August: DM IP workshop at NO AA’s NW S/O HD

O ct: DM IP Special Issue,Journal of Hydrology, 298

2001Dec : DM IP officially begins,web set open

2002

2003

2004

Interior P oints

-80

-60

-40

-20

0

20

40

60

80

% Im

prov

emen

t

P arent B as ins

-50

-40

-30

-20

-10

0

10

20

A RS HRC OHD UTS

UWO OU A RZ A v g

DHI WHU MIT

DM IP 1 ResultsPercent improvement in flood peak prediction from

calibrated distributed hydrologic models compared to the lumped model

C hristie Kansas Savoy Eldon Blue W atts T iff C ity Tahlequah

120

(See Journal of Hydrology, Volume 298, 2004, Specia l Issue for details)

DM IP 1 Conclusions

• Som e distributed hydro log ic m odels can produce sim ulations com parable to or better than lum ped m odels. M any cannot.

• D istributed m odel param eters derived from physica l data provide reasonable in itia l guesses but param eter calibration significantly im proves outlet s im ulations.

• The m ost significant ga ins from distributed m odels over lum ped m odels were seen for a long, narrow basin, the B lue R iver (1233 km 2), and a sm all interior gage, Peacheater Creek (65 km 2).

DM IP 1 Conclusions

• Distributed m odels produced reasonable results at interior points but not as good statistically as those from parent basins.

• M odels that com bined conceptual ra infall-runoff techniques with physica lly-based routing techniques showed the best perform ance (HL-R M S is an exam ple).

• The m ean of selected m odel ensem bles com pares favorably w ith that of the single best distributed m odel

(See Journal of Hydrology, Volume 298, 2004, Specia l Issue for details)

121

DM IP 2 Science G oals

•• Confirm basic DM IP 1 conclusions w ith a longer Confirm basic DM IP 1 conclusions w ith a longer validation period and m ore test basinsvalidation period and m ore test basins

•• Im prove our understanding of distributed m odel Im prove our understanding of d istributed m odel accuracy for sm all, interior point sim ulationsaccuracy for sm all, interior point sim ulations

•• Evaluate new reEvaluate new re --analysis forcing data setsanalysis forcing data sets•• Evaluate the perform ance of distributed m odels Evaluate the perform ance of d istributed m odels

in prediction m ode in prediction m ode •• Use available soil m oisture data to evaluate the Use available soil m oisture data to evaluate the

physics of d istributed m odels physics of distributed m odels

DM IP 2 Science G oals

•• Im prove our understanding of the w ay routing Im prove our understanding of the w ay routing schem es contribute to the success of d istributed schem es contribute to the success of d istributed m odels by designing an experim ent that uses a m odels by designing an experim ent that uses a com m on routing m odelcom m on routing m odel

•• Continue to gain insights into the interplay Continue to gain insights into the interplay am ong spatia l variability in ra infall, am ong spatia l variability in ra infall, physiographic features, and basin response, physiographic features, and basin response, specifically in m ountainous basins specifically in m ountainous basins

•• Im prove our understanding of scale issues in Im prove our understanding of scale issues in m ountainous area hydrologym ountainous area hydrology

•• Im prove our ability to characterize sim ulation Im prove our ability to characterize sim ulation and forecast uncerta inty in different hydrologic and forecast uncerta inty in different hydrologic regim esregim es

122

Nevada

Californ ia

Texas

Oklahom a

Arkansas

M issouriKansas Elk R iver

IllinoisR iver

B lue R iver

AmericanR iver Carson

R iver

Additional Tests in DM IP 1 Basins1. Routing2. Soil Moisture3. Lumped vs. D istributed4. Prediction mode

Tests w ith C om plex Hydrology1. Snow, Rain/snow events2. Soil Moisture3. Lumped vs. D istributed4. Mountain Terra in

H M T

DM IP 2 - Scope

A mericanRiver

FolsomDam

Ame rican RiverBa sin

North Fo rk A mericanRiver Basin

Nevad a

Ca lifornia

St ud ya re a s

Carso n Rive rDrainag e B asin

Nevada

California

E ast Fork CarsonRiver Ba sin

Sa cra me nt o

C a rsonC ity

R en o

L ak eTa h oe Ca rson

Rive r

S ierra Nevada

Mtns

50 m iles

50 kilo me te rs0

Scale

M id dle Fo rk

So uth Fork

DM IP 2 - S ierra Nevada Basins

-Reasonable existing data-P lanned instrumentation-W idely studied-CUAHSI in itia tive-HMT Linkage

123

Proposed DM IP 2 T im eline

•• M arch 2005 M arch 2005 ---- D istribute Science P lan for reviewD istribute Science P lan for review•• O ctober 2005 O ctober 2005 –– Project start, basic data available Project start, basic data available

on w eb (not ETL data)on w eb (not ETL data)•• O ctober 2006 O ctober 2006 –– O klahom a sim ulations dueO klahom a sim ulations due•• July 2007 July 2007 –– S ierra N evada basic sim ulations dueS ierra N evada basic sim ulations due•• O ctober 2007 O ctober 2007 ---- Sum m ary w orkshopSum m ary w orkshop

DM IP 2The follow ing have already shown interest in participating:

•• KonstantineKonstantine G eorgakakos, HRCG eorgakakos, HRC•• Kenneth M itchell, N O AA/NW S/NCEPKenneth M itchell, N O AA/NW S/NCEP•• M ario M ario D iLuzioD iLuzio , Texas A&M, Texas A&M•• W itoldW itold Krajewski, U . IowaKrajewski, U . Iowa•• Praveen Kum ar, U . IllinoisPraveen Kum ar, U . Illinois•• E ldhoE ldho IypeIype , Indian Tech. Institute, Bom bay, Indian Tech. Institute, Bom bay•• Sandra G arcia, U . Sandra G arcia, U . C artegenaC artegena , Spain, Spain

124