Date post: | 03-Jan-2016 |
Category: |
Documents |
Upload: | taylor-dillard |
View: | 24 times |
Download: | 3 times |
Arkansas Red Basin River Forecast Center
An Operational Forecast Office Perspective An Operational Forecast Office Perspective of the of the
National Weather Service National Weather Service Hydrologic Distributed Modeling System Hydrologic Distributed Modeling System
(HDMS)(HDMS)
Presented by:
Diane CooperArkansas-Red Basin River Forecast Center Hydrologist
2Arkansas-Red Basin River Forecast Center October 18, 2006
ObjectivesObjectives
Overview of HDMS
Preliminary Statistical Analysis
3Arkansas-Red Basin River Forecast Center October 18, 2006
HDMS, HDMS, What is it?What is it?
A Hydrologic Model that takes into account: Spatial distribution of hydrologic characteristics across a
drainage basin including soils, vegetation, land use, slope
Temporal and spatial distribution of rainfall…4X4 km HRAP grids
Many 4x4 km grid cells embedded in the main basin.
Maintains spatial Precipitation footprint which impacts hydrologic response.
4Arkansas-Red Basin River Forecast Center October 18, 2006
Features Of HDMSFeatures Of HDMS
Performs routing simulations using the kinematic wave technique
– flow velocity in each element is dependent on flow level
Separates the runoff components– Surface flow: includes impervious, surface and
direct runoff
– Subsurface flow: includes interflow and baseflow
5Arkansas-Red Basin River Forecast Center October 18, 2006
Advantages of HDMSAdvantages of HDMS
Potential for improved simulations for:– Basins with non-uniform rainfall.– Basins with relatively impervious areas (surface runoff
occurs quickly…i.e. highly populated regions). Prediction of hydrologic variables at interior
points. Output of Gridded soil moisture states. Potential to improve flash flood forecasting. Can account for Land-use change (i.e. burn
areas).
6Arkansas-Red Basin River Forecast Center October 18, 2006
Statistical AnalysisStatistical Analysis
Comparisons of HDMS and NWSRFS SQIN to the observed flow data was performed for 14 of the 21 test basins from Early 2002 through August 2005.
For a more “consistent” comparison, the Lumped Model SQIN crest time and peak discharge were compared to the hourly observed time series. This is referred to as the Adj. NWSRFS SQIN.
Note: Seven Basins were not included in the analysis due to:
• 4 HDMS test basins are not identified in the NWSRFS Lumped model
• 2 basins had a very short NWSRFS SQIN timeseries.
• 1 Basin the SQIN has not yet been generated.
How well is HDMS performing compared to How well is HDMS performing compared to the NWSRFS the NWSRFS lumpedlumped model? model?
7Arkansas-Red Basin River Forecast Center October 18, 2006
Statistical AnalysisStatistical Analysis
For 7 of the 14 test basins, the overall flow percent bias is lower with the HDMS simulations
For 8 of the 14 basins, the overall Correlation Coefficient ”R” is closer to 1 with most of them at 0.85 and higher.
Note: Due to limitations with the statistical analysis software, the Overall simulations do not use the same observed timeseries dataset. HDMS is compared to the 1-hour observed discharge while NWSRFS is compared to a 6-hour timeseries. Hence, flow is “lost” in the 6- hour timeseries.
Multi-Year Overall FlowMulti-Year Overall Flow
8Arkansas-Red Basin River Forecast Center October 18, 2006
Statistical AnalysisStatistical Analysis
For only 3 of the 14 test basins, the HDMS model performed better than the Adj. NWSRFS on the Timing of the Crest. However when taking the standard deviation into account, HDMS had a lower deviation on half of the basins.
Time to Peak ErrorTime to Peak Error
HDMS Time to Peak Error
-60
-50
-40
-30
-20
-10
0
10
20
30
40
50
60
0 5000 10000 15000 20000 25000 30000 35000
Peak Discharge(cfs)
Tim
e t
o P
ea
k E
rro
r(h
ou
rs)
ELDO2
ELMA4
SVYA4
KNSO2
TALO2
BSGM7*
INCM7*
WTTO2
BLKO2
BLUO2
CBNK1
CVSA4*
ELTT2
MLBA4
SLSA4*
SPRA4
TIFM7
WSCO2*
9Arkansas-Red Basin River Forecast Center October 18, 2006
Statistical AnalysisStatistical Analysis
For 8 of the 14 basins, the normalized Mean Peak Discharge Error is better with HDMS.
When evaluating the Standard Deviation and the normalized Peak Discharge “Adj” error, 9 of the 14 basins performed better with HDMS.
Peak DischargePeak Discharge
Plots of normalized HDMS peak discharge errors for Events identified
between 4/02 through 8/05.
HDMS Peak Discharge Error
-2.50
-2.00
-1.50
-1.00
-0.50
0.00
0.50
1.00
1.50
0 10000 20000 30000 40000 50000 60000
Peak Discharge (cfs)
No
rmal
ized
Cre
st E
rro
r (2
-yea
r fl
oo
d f
req
uen
cy p
eak
dis
char
ge)
ELDO2
ELMA4
SVYA4
KNSO2
TALO2
BSGM7*
INCM7*
WTTO2
BLKO2
BLUO2
CBNK1
CVSA4*
ELTT2
MLBA4
SLSA4*
SPRA4
TIFM7
WSCO2*
10Arkansas-Red Basin River Forecast Center October 18, 2006
A Closer Look at a A Closer Look at a Calibrated BasinCalibrated Basin
Minor improvement from 0.84 (Lumped) to 0.86 (HDMS) in the correlation coefficient “R”.
Improved simulation of the higher flow events, typically these are under simulated.
Time to crest error showed a decrease with a smaller variability:
– HDMS error was 1.9 hrs - most crests late – NWSRFS error was 3.3 hrs – most crests early
Statistical analysis for Corbin, KS (CBNK1) – period 1/2002 though 8/2005.
11Arkansas-Red Basin River Forecast Center October 18, 2006
Focusing on the EventsFocusing on the Events
CBNK1 Crest Time Error
-20
-15
-10
-5
0
5
10
15
20
0 2000 4000 6000 8000 10000
Peak Discharge (cfs))
Cre
st
Tim
e E
rro
r (H
ou
rs)
HDMS
NWSRFS
AdjNWSRFS
CBNK1 Peak Discharge Error
-1.2
-1
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0 3000 6000 9000 12000 15000 18000
Peak Discharge (cfs))
Nor
mal
ized
Pea
k Er
ror
(2-y
ear f
lood
freq
uenc
y p
eak
disc
harg
e)
HDMS
NWSRFS
AdjNWSRFS
Plot of the Crest Time Error for the 26 events identified at CBNK1. HDMS
tended to be late in its timing of the crest while NWSRFS tended to be early.
Plot of the Peak Discharge Error for the 26 events identified at CBNK1. The Normalizing Factor is the 2-year flood frequency which is 8990 cfs. Both models give a mix of over and under simulations for events that are 6000 cfs and less. However for the larger events, both models dramatically undersimulate the crest.
12Arkansas-Red Basin River Forecast Center October 18, 2006
SummarySummary
HDMS appears to perform as well or better than lumped model for most of test basins…especially in the simulation of peak discharge.
HDMS shows promise as a more advanced Hydrologic Model for NWS operations.
13Arkansas-Red Basin River Forecast Center October 18, 2006
ContactsContacts
OHD– Lee Cajina [email protected]– Seann Reed [email protected]
ABRFC – Diane Cooper [email protected]
WGRFC– Paul McKee [email protected]
14Arkansas-Red Basin River Forecast Center October 18, 2006
Appendix AAppendix AHDMS Statistical Analysis DataHDMS Statistical Analysis Data
Basin 2 year Flood Frequency(Normalizing Factor)
CorrelationCoefficient
"R"
Overall PercentBias
Mean Time toPeak Error
(hr)
Mean Time to peak ST Dev
Mean Normalized Peak Discharge
Error
Mean NormalizedPeak St Dev
CVSA4 1980 0.75 -23.6 8.8 3.7 -0.41 0.36
ELMA4 5480 0.81 7.84 7.2 4.0 -0.29 0.35
MLBA4 21100 0.92 17.74 5.6 5.6 0.01 0.18
SLSA4 27900 0.79 58.51 11.6 10.3 -0.07 0.27
SPRA4 794 0.69 16.09 3.2 2.9 -0.07 0.56
SVYA4 10700 0.88 -0.02 5.7 6.7 -0.23 0.44
TIFM7 23400 0.95 5.07 4.9 7.4 -0.03 0.14
CBNK1 8990 0.87 -23.09 1.9 7.1 -0.18 0.31
BLKO2 22400 0.86 -29.71 13.1 18.3 -0.24 0.26
BLUO2 8770 0.82 14.78 8.8 17.0 -0.03 0.19
ELDO2 16200 0.92 0.26 -1.4 13.0 -0.18 0.34
KNSO2 3580 0.89 7.98 1.5 5.3 0.02 0.22
TALO2 20300 0.94 38.82 -2 11.5 0.08 0.17
WSCO2 1280 0.83 -21.46 3.5 2.6 0.19 0.34
WTTO2 20700 0.88 60.69 9.3 11.1 0.01 0.21
ELTT2 2610 0.59 41.7 -1.7 39.3 0.07 0.21
ELTT2a 7.2 15.0 0.01 0.15
BSGM7 5020 0.9 0.34 2.3 4.7 -0.05 0.20
INCM7 3980 0.94 20.03 -0.7 11.1 0.09 0.41
15Arkansas-Red Basin River Forecast Center October 18, 2006
Appendix BAppendix BNWSRFS Statistical Analysis DataNWSRFS Statistical Analysis Data
Basin Correlation Coefficient “R”
Overall Percent Bias
Mean Time to Peak Error (hr)
Mean Time to peak ST Dev
Mean Normalized Peak Discharge Error
Mean Normalized Peak St Dev
CVSA4 N/A N/A N/A N/A N/A N/A
ELMA4 0.63 -22.43 -2.8 3.8 -0.13 0.4
MLBA4 0.79 -3.83 -2.7 5.8 -0.02 0.15
SLSA4 0.82* -32.99* -2.6* 4.5* –0.14* 0.23
SPRA4 N/A N/A N/A N/A N/A N/A
SVYA4 0.79 0.51 2.0 7.0 -0.08 0.12
TIFM7 0.87 -20.28 1.6 8.5 -0.06 0.17
CBNK1 0.84 0.83 -3 8.2 -0.12 0.34
BLKO2 0.9 -15.3 -1.2 14.1 -0.12 0.18
BLUO2 0.85 -47.29 -6.2 17.9 -0.07 0.15
ELDO2 0.78 -28.22 -3.6 9.0 -0.24 0.38
KNSO2 0.86 -4.46 -4 8.3 0.02 0.26
TALO2 0.95* -27.96* -2.0* 14.1* –0.12* 0.12*
WSCO2 0.52* -41.34* -5.2* 5.0* –0.14* 0.18*
WTTO2 0.96* -22.49* -1.4* 7.8* –0.08* 0.10*
ELTT2 0.68 56.5 -6 39.1 0.02 0.15
ELTT2a 2.7 16.9 0.07 0.21
BSGM7 0.65† -70.42† -6† 0.7† 0.15† 0.05†
INCM7 0.70† -68.97† 0† 0† –0.29† 0.24†
16Arkansas-Red Basin River Forecast Center October 18, 2006
Appendix CAppendix CAdj. NWSRFS Statistical Analysis DataAdj. NWSRFS Statistical Analysis Data
Basin Adj Mean Time to Peak Error (hr)
Adj. Mean Time to peak ST Dev
Adj. Mean Normalized Peak Discharge Error
Adj. Mean Normalized Peak St Dev
CVSA4 N/A N/A N/A N/A
ELMA4 -2.2 2.7 -0.27 0.51
MLBA4 -1.5 5.7 -0.04 0.16
SLSA4 -3.9* 7.3* –0.16* 0.24*
SPRA4 N/A N/A N/A N/A
SVYA4 2.7 6.6 -0.15 0.26
TIFM7 -0.6 8.5 -0.07 0.18
CBNK1 -3.3 7.6 0.14 0.35
BLKO2 -0.9 13.6 0.12 0.18
BLUO2 -5.7 18.1 0.08 0.16
ELDO2 -3.7 8.2 0.32 0.42
KNSO2 -3.6 8.2 0.03 0.3
TALO2 -1.9* 15.0* –0.12* 0.12*
WSCO2 -3.1* 5.6* –0.27* 0.30*
WTTO2 -1.4* 8.0* –.12* 0.11*
ELTT2 -4.9 39.0 0.01 0.15
ELTT2a 3.9 16.3 0.03 0.13
BSGM7 -5.5† 0.7† –0.16† 0.06†
INCM7 1.5† 3.5† –0.33† 0.28†
17Arkansas-Red Basin River Forecast Center October 18, 2006
Reference Information for Reference Information for Appendices A, B and C.Appendices A, B and C.
Appendix A, B and C are summaries of selected statistical parameters. The Correlation Coefficient “R” and Percent Bias were derived from the multi-year time series analysis. The HDMS simulation was compared to the one hour observed and the NWSRFS simulation was compared to the six hour observed discharge. The NWSRFS “Adj.” information is a comparison of the six hour NWSRFS simulations to the one-hour instantaneous discharge time series. The ELTT2a Peak Error averages, excludes 2 events which both models performed very poorly.
Note: “†” indicates the NWSRFS multi-year analysis began in March 2005, and “*” indicated the analysis began in the summer of 2003.) Elsewhere, the multi-year period was from April 2002 through August 2005.
The basins shaded in Pink, the Annual Peak Discharge’s period of record is less than 10 years. So the accuracy of the 2-year frequency peak discharge normalization factor is suspect.
18Arkansas-Red Basin River Forecast Center October 18, 2006
Appendix DAppendix DTable of the HDMS Test BasinsTable of the HDMS Test Basins
NWS Handbook 5 ID USGS Site Number Gauge Location Basin Size mi2 (km2)
CVSA4 07194880 Osage Creek at Cave Springs, AR 35 (90)
DMLA4 07196900 Barron Fork Creek at Dutch Mills, AR 40 (105)
ELMA4 07195000 Osage Creek at Elm Springs, AR 130 (337)
MLBA4 07252000 Mulberry River near Mulberry. AR 373 (966)
SLSA4 07195430 Illinois River at Siloam Springs, AR 575 (1489)
SVYA4 07914800 Illinois River at Savoy, AR 167 (433)
SPRA4 07195800 Flint Creek at Springtown, AR 14 (37)
CBNK1 07151500 Chikaskia River at Corbin, KS 794 (2056)
BSGM7 07188653 Big Sugar Creek at Pineville, MO 141 (365)
INCM7 07188885 Indian Creek at Anderson, MO 239 (619)
TIFM7 07189000 Elk River at Tiff City, MO 872 (2258)
BLKO2 07152000 Chikaskia River at Blackwell, OK 1859 (4815)
BLUO2 07332500 Blue River near Blue, OK 476 (1233)
CPCO2 07196973 Peacheater Creek at Christie, OK 25 (65)
ELDO2 07197000 Barron Fork River at Eldon, OK 307 (795)
KNSO2 07196000 Flint Creek at Kansas, OK 110 (285)
TALO2 07196500 Illinois River at Tahlequah, OK 959 (2484)
WSCO2 07195865 Sager Creek at West Siloam Springs, OK 19 (49)
WTTO2 07195500 Illinois River at Watts, OK 635 (1645)
AMAT2 07227500 Canadian River at Amarillo, TX 19445 (50363)
ELTT2 07312200 Beaver Creek near Electra, TX 10298 (26672)