NWS Calibration Workshop, LMRFC March, 2009 Slide 1 Sacramento Model Derivation of Initial Parameters.

NWS Calibration Workshop, LMRFC March, 2009

Slide 1

Sacramento ModelDerivation of Initial Parameters


Slide 2

Methods

• Derived through hydrograph analysis– Observed hydrographs contain a lot of information!– See handouts for manual methods

• Derived from GIS-soils information– From STATSGO (NRCS) state level data

• Developed by Victor Koren of HL• Delivered to RFCs in CAP

– From SSURGO (NRCS) county level data• Shows improvement over STATSGO based parameters


Slide 3

1

10

100

1000

26583 26633 26683 26733 26783 26833

Dis

char

ge (

m3 s-1

day)

Flint Creek @ Kansas, OK. Illinois River @ Watts, OK.

Elk River @ Tiff City, MO. Baron Fork @ Eldon, OK.

Illinois River @ Tahlequah, OK. Blue River @ Blue, OK.

Observed hydrographs contain a lot of information


Slide 4

Initial SAC-SMA Parametersfrom STATSGO Data

• Objective estimation procedure that can produce spatially consistent and physically realistic parameter values (Koren et al. 2000, 2003)

– Improve starting point estimates of conceptual model parameters based on national database of soil property data

– Constrained calibration so that parameter adjustment occurs within range of values to maintain conceptual consistency

• Provide spatial distributions of parameters to make simulation possible as NWS moves to smaller scale distributed and flash flood modeling


Slide 5

Parameter Estimation Methodologyusing STATSGO Data

• Physically-based approach to quantify relationships for 11 major parameters of SAC-SMA (16 total)– STATSGO dominant soil texture grids for 11 soil layers for

conterminous U.S. are used.

– Hydraulic soil properties (e.g., θs, θfld, θwlt, Ks) estimated for each USDA textural class using empirical relationships (Campbell 1974, Clapp and Hornberger 1978, Cosby et al. 1984)

– Split between upper and lower layers based on NRCS curve number (function of soil type, land use, hydrologic condition, moisture status)

– SAC-SMA storages are defined using these assumptions. Additional physical reasoning, empirical relationships establish remaining estimated parameters (Koren 2000).

Current NWS Approach, contd.


Slide 6

Current Limitationsof using STATSGO Data

• Limitations of present methodology (some noted in Koren et al. 2003)

– Data Resolution• Uses STATSGO (1:250K) data intended for multi-state, regional

scale analysis (~100 – 200 km2 polygons) leading to limitations on reliability of parameters due to spatial sampling

• Measurements typically don’t extend below 152 cm, so unable to account for certain local conditions, e.g., areas of deep groundwater

• Assumed a single generic land use type and condition (“range” land use, “fair” condition), leading potentially to bias

– Other uncertainties and scientific questions related to derived property-parameter and property-property relationships

Current NWS Approach, contd.


Slide 7

Strategies for Improvement

– Using data of finer resolution• Use county level (SSURGO) field surveys (~1:24K) with

polygons typically consisting of a single soil component type• Use USGS land cover grids at 30m scale

– Incorporating data on impervious layers in soil column leading to local restrictions on parameter values

– Applying base flow analysis to develop more accurate estimates of lower layer dynamic (free water) storage

GOAL: Explore potential to improve parameter estimates by:


Slide 8Time, days

Comparison of Hydrographs: STATSGO vs. SSURGO Data

(coarse vs fine scale)Catchment 3: 2–7/1975

Linear scale

Semi-Log scale

0

50

100

150

200

250

300

Dis

char

ge

(cm

sd)

OBS

STAT

SSUR

0.1

1

10

100

1000

1 30 59 88 117 146 175

Dis

char

ge

(cm

sd)

OBS

STAT

SSUR

0

50

100

150

200

250

OBS

STAT

SSUR

0.1

1

10

100

1000

1 30 59 88 117 146 175

OBS

STAT

SSUR

Catchment 2: 2–7/1974

Time, days

Results, contd.


Slide 9

Demonstration of scale difference between STATSGO and SSURGO

SSURGO

STATSGO

Background


Slide 10

Objective estimation procedure that can produce spatially consistent and physically realistic values for 11 of the 16 SAC-SMA parameters

• STATSGO + Assumed spatially constant “pasture or range land use” under “fair” hydrologic conditions, (Koren et al. 2000, 2003)

• STATSGO + Spatially variable land use land cover data

• SSURGO + Spatially variable land use land cover data, (Anderson et al., 2005, Zhang et al., 2008)

Background


Slide 110

0.2

0.4

0.6

0.8

1

0 0.2 0.4 0.6 0.8 1

Rmod (STATSGO+uniform lulc)

Rm

od

STATSGO+LULC

SSURGO

small basins

0

0.2

0.4

0.6

0.8

1

0 0.2 0.4 0.6 0.8 1


Rm

od

STATSGO+LULC

SSURGO

large basins

0

0.2

0.4

0.6

0.8

1

0 0.2 0.4 0.6 0.8 1


Rm

od

STATSGO+LULC

SSURGO

all basins

Results

Rm: Modified correlation coefficient. It

is calculated by reducing normal correlation coefficient by the ratio of the standard deviations of the observed and simulated hydrographs.


Slide 12

Conclusions• Use of land cover data and higher resolution soil data results

in different a-priori SAC-SMA parameters.

• Overall simulation results for three sets of a-priori parameters are similar.

• The effect of using higher resolution soil data and land use land cover data is different between smaller basins and larger basins. Improvements are mainly for smaller basins when SSURGO data are used. Generally similar results for large basins for three sets of a priori parameters.

• Improvement from using detailed soil data is greater than using gridded land cover data.

• Results suggest that the SSURGO based parameters are preferable for smaller scale applications.


Slide 13

Status of SSURGO – Based SAC-SMA Parameter Derivation

NWS Calibration Workshop, LMRFC March, 2009 Slide 1 Sacramento Model Derivation of Initial Parameters.

Documents

nws calibration workshop

lot of information slide

data resolution

statsgo nrcs state level

soil layers

reliability of parameters

flash flood modeling

soil column