Presentation

Simulating Yield using STATSGO
and SSURGO soil datasets
R. Butalia 1 , L. Negm 2 , S. Tian 2
North Carolina State University
1
Department of Forestry
2
Department of Biological and Agricultural Engineering

Outline
1. Introduction and Objectives
2. Soil Datasets – STATSGO and SSURGO
3. Methods
• Model Selection and Application
• Subbasin Selection
• Validation
4. Results
5. Summary and Conclusion

Introduction
Study Area
• Robeson Creek
Watershed
• Piedmont, Carolina
Slate Belt
• Impaired
waterbody, active
TMDL for P
• Section 319 funded
• 7 Sampling points
• Mixed land use

Introduction
SWAT
• SWAT is a hydrologic modeling tool developed
at Texas A&M University.
• Capable of predicting the effect of different land
uses on water yield and quality in large complex
watersheds.

Objectives
• Determine the influence of soil input dataset
spatial resolution on model behaviour
• Determine the variation in this influence for two
basins – smaller basin nested in larger
STATSGO
SSUGRO
WS A
WYLD AST
WYLD ASR
WS B
WYLD BST
WYLD BSR
WS A -Entire Basin WS B – Subbasin 6

Soil Datasets
• STATSGO – lower spatial resolution
• SSURGO – more detailed geography and more
recent data (survey conducted 1990s).
• VBA code created for pre-processing SSURGO 2.2.1
• Only one soil type per polygon was used in
SWAT

Methods
Model Selection and Application
• SWAT
• Commonly used, tested at various spatial scales
• Allows use of spatial data, Arc SWAT
• Simulation experiment
• Two different soil datasets
• Two different watershed sizes
• Uncalibrated simulations

Methods
Model Selection and Application
Soil Water Balance (Arnold et al 1999)
R – Rainfall
Q - Runoff
ET – Evapotranspiration
P - Percolation
QR – Return Flow
Calculated for each HRU for each time step, typically daily

Methods
Name
Source
Date
STATSGO
Simulation
SSURGO
Simulation
Spatial Resolution/
Map Scale
Flow
NCSU
WQ
2002 -
2006
Elevation
NCDOT
8/15/07
X
X
20 foot cell,
rounded down to
nearest foot
Land Use /
Land Cover
MRLC
8/15/07
X
X
30 meter cell,
minimum 1 acre
Unit
Soil
STATSGO
SWAT
(modified)
8/15/07
X
1: 250, 000
Soil SSURGO
NRCS
3/20/07
X
1: 24,000

Methods
• Subbasins delineated using sampling
locations as outlets
• Subbasins characteristics from reports
created by SWAT
• Attempt to normalize for Land Use

Soil Datasets
STATSGO
Watershed Outlet
Subbasin 6 Outlet

Soil Datasets
SSURGO
Watershed Outlet
Subbasin 6 Outlet

Sample Profiles
Cid
Badin
Georgeville
• Clay and Silty Clay Loams
• Variation in depth of profile
• Most commonly low permeability
STATSGO - 62 in.
SSURGO – bottom of C 80 in.

Methods
Subbasin Selection: Yield ratio (measured flows)
ratio
Ratio
5
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
sub_1 Sub1
sub_2 Sub2
sun_3
Sub3
sub_4 sub_6 sub_7
Sub4
Sub6
Sub7
May-02 Dec-02 Jun-03 Jan-04 Aug-04 Feb-05 Sep-05 Mar-06
Date
time

Methods
Subbasin characteristics
(value in %) WATERSHED SUBBASIN 6
NC061 2.8 0
Soil Type NC064 22.2 23.8
NC068 75 76.2
0 - 2 53.6 52.7
2 - 6 44 45.4
Land Slope 6 - 10 2 1.6
10 - 15 0.15 0.14
> 15 0.02 0.04
URBN 9.3 5.9
WATR 0.39 0.62
SWRN 0.15 0.1
Land Use FRST 69 64.8
RAG 9.6 11.95
HAY 11.4 15.1

Methods
Validation
Compared simulated flow to measured data
I. Nash Sutcliffe (E)
II. Pearson Product Momentum Coefficient (r)

Results
Monthly Flow Entire Watershed
water Yield Flow (1000 yield (1000 m 3 ) /m)
m3/M)
9000
8000
7000
6000
5000
4000
3000
2000
1000
0
Results from SSURGO data
Results from STATSGO data
Measured results
Sep-02 Mar-03 Oct-03 Apr-04 Nov-04 May-05 Dec-05 Jul-06
Date
time

Results Monthly Flow Subbasin 6
water yield (1000 m3/M)
Flow (1000 m 3 )
1000
900
800
700
600
500
400
300
200
100
0
Sep-02 Mar-03 Oct-03 Apr-04 Nov-04 May-05 Dec-05
time
Date
Results from SSURGO data
Results from STATSGO data
Measured results

Results
STATSGO
SSUGRO
E r E r
Watershed -0.75 -0.1 -1 -0.24
Sub basin 6 0.09 -0.07 -1.45 -0.16
Not Indicative of accurate modelling

S&R1
Results
Ratio = Q 6
/ Q ww
3
SSURGO ratio STATSGO ratio Measured ratio
2.5
2
Ratio
1.5
1
0.5
0
0 5 10 15 20 25 30 35 40
Time

Slide 21
S&R1
this should have been measured ratio and then shown statsgo and ssurgo
Sharon & Raj, 2/20/2008

Results
• In our case SSURGO yields were lower
than STATSGO. Others also running
uncalibrated models (Peschel et al,
2006) have found different results

Results
Entire Watershed
Differences between simulations using SSURGO and
STATSGO data for key hydrologic processes: A
400
350
Percolation Soil Water Water Yield (q)
300
250
Difference (mm
200
150
100
50
Initial Conditions
0
-50
-100
0 5 10 15 20 25 30 35 40 45
Time Indicator

Results
Entire Watershed
Differences between simulations using SSURGO and
STATSGO data for key hydrologic processes: B
30
Surface Runoff Groundwater Flow ET Water Yield (q)
20
10
Difference (mm)
0
-10
-20
-30
-40
0 5 10 15 20 25 30 35 40 45
Time Indicator

Results
T-test Analysis Comparing Yield (mm) Results
from STATSGO and SSURGO
Values in cells are
t (P)
STATSGO
WW
SSURGO
SB6
STATSGO
SB6
-0.79 (0.433)
-0.17
(0.8687)
SSURGO
WW
-2.97
(0.0054)
3.67 (0.0008)

Conclusions
• Results using SSURGO data showed
differences between the entire watershed and
the nested subasin.
• Need to perform further sensitivity analysis
and compare calibrated results
Future Steps
• Calibration and sensitivity analysis may lead
to different conclusions
• Factor in saprolite and restrictive layers

Acknowledgements
NCSU Water Quality Group
Dan Line
Karen Hall
Hugh Devine, CNR, NCSU
Hoffman Fund, NCSU