1. Re-conceptualizing SWAT for Variable Source Area Hydrology
Zachary Easton1*, Daniel Fuka1, Todd Walter1, Dillon Cowan2, Elliot Schneiderman3, Tammo Steenhuis1
1Dept. Biological and Environmental Engineering, Cornell University
2School of Civil and Environmental Engineering, Cornell University
3New York City Dept. Environmental Protection, Kingston, NY

2. Re-conceptualizing SWAT
Variable Source Area (VSA) Hydrology
Curve Number and VSA hydrology
Convincing SWAT it recognizes VSAs

3. Common Perception of Runoff
Rain
Infiltration
“Runoff”
Infiltration Excess
a.k.a. Hortonian Flow
(Horton 1933, 1940)

4. Is Hortonian Flow Common?
New York
March
April
May
June
July
August
September
October
November 1 10
100 20 30 50 40 70 60 80 90
Return Period (yr)
% of Area Generating Hortonian Flow
Walter et al ASCE J. Hydrol. Eng. 8:

5. Saturation Excess Runoff
Rain
Subsurface
water
rises
Some areas saturate to the surface

6. Saturation Excess Runoff
Rain
Upland interflow may exfiltrate
Rain on saturated areas becomes overland flow
Dunne and Black Water Resour. Res. 6:
Dunne and Black Water Resour. Res. 6:

7. Variable Source Areas
Flow Path
Current Water Quality Models were not Intended to Capture this Complexity
General Watershed Loading Function (GWLF)
Soil Water Assessment Tool (SWAT)
Agricultural Nonpoint Source Pollution Model (AGNPS)

8. Re-conceptualizing SWAT
Variable Source Area (VSA) Hydrology
Curve Number and VSA hydrology
Convincing SWAT it recognizes VSAs

9. USDA-NRCS Curve Number Model
“Runoff”=Pe2/(Pe+S)
S=25400/CN-254
Tables link CN to land use and
soil infiltration capacity

10. Curve Numbering VSA hydrologyP
Q = Pe2/(Pe+S)
Watershed Q Af
DQ = AfDPe
Unsaturated
Saturated
dQ/dPe = Af
Af = f(S,Pe)
Steenhuis et al ASCE Div Drain. & Irr. 121:

11. We know how much area is contributing…
Af = f(S,Pe)
…but from where in the landscape?

12. Soil Topographic Index Wetness Index Classes10 1
s8=f(S)
s7=f(S)
33.10
3.52
s9=f(S)
s10=f(S)
s1=f(S)
s6=f(S)
si = local storage
s2=f(S)
s5=f(S)
s3=f(S)
Easton et al., J. Hydrol. 348:
Lyon et al Hydrol. Proc. 18(15):
Schneiderman et al Hydrol. Proc. 21:
s4=f(S)

13. Re-conceptualizing SWAT
Variable Source Area (VSA) Hydrology
Curve Number and VSA hydrology
Convincing SWAT it recognizes VSAs

14. Revisit the HRU concept
Landuse
Define HRUs as the coincidence of soil type and landuse
Hydrological/chemical properties are defined at the HRU
So runoff/P loss is the same here (lowland pasture)
As here (upland pasture)
Is this a good assumption?
Soils
HRUs

15. VSA concept defines HRUs as the coincidence of soil topographic index and landuse
So runoff/P loss is now not the same here (lowland pasture)
As here (upland pasture)
Better Assumption?
STI
Landuse
SSURGO
HRUs

16. Distributing CN-values
Average of Standard CNs = 73.1 distributed according to landuse/soils
Average of VSA CNs = 73.1, distributed according to a wetness index
Wetness Index
Classes 10 1

17. Modify the Available Water Content
High runoff prone area = high moisture content (in general)
We relate local soil water storage, e,i, to AWC with the following:
ρb = soil bulk density (g cm-3)
clay = soil clay content (cm3 cm-3).
Wetness Class 10
Wetness Class 9
Wetness Class 2
Wetness Class 1

18. Test Results: Streamflow
-Standard

19. Test Results: Runoff from Pastures
SWAT-VSA SWAT-Standard
Easton et al., J. Hydrol. 348:

20. Test Results: Runoff SWAT-VSA SWAT-Standard
Easton et al., J. Hydrol. 348:

21. Test Results: Soil water
SWAT-VSA SWAT-Standard
Easton et al., J. Hydrol. 348:

22. Test Results: Soil Water
Data curtsey of :
Lyon et al Adv. Water Resour. 29(2):
Lyon et al HESS. 10:
Wetness Index Classes
Test Results: Soil Water 10
100 m N 1
585 m 1m
Cont.
Water
Level
Loggers
600 m

23. SWAT-VSA
Movie courtesy of Dr. Steve Lyon

24. Test Results: Soil waterb
Index
Easton et al., J. Hydrol. 348:

25. Test Results: Phosphorusa
r2 = 0.76
E = 0.71 b
-Standard
r2 = 0.68
E = 0.47

26. Test Results: Phosphorus
SWAT-VSA SWAT-Standard

27. Take-home Messages
Storm runoff is generated from small parts of the landscape
Areas prone to saturate – e.g., toe slopes, shallow soils, topographically converging areas
Variable Source Areas – they expand and contract
We can predict where and when storm runoff will be generated
We can improve integrated and distributed predictions by considering VSAs
Implications for watershed management

28. Curve Numbering VSA hydrologyP
Q = Pe2/(Pe+S)
Af = dQ/dPe
Watershed
Af = 1-(S2/(Pe+S)2 Q Af
where dQ is the equivalent depth of excess rainfall generated during a time period over the whole watershed area, and dPe is the incremental depth of precipitation during the same time period. Thus, by differentiating the CN eq with respect to Pe, the fractional contributing area for a storm can be written as:
Af = 1-(S2/(Pe+S)2 Eq 3
According to above, runoff only occurs from areas that have a local effective available storage, e, (mm), less than Pe. Therefore, by substituting e for Pe in Eq. 3 we have a relationship for the fraction of the watershed area, As, that has a local effective soil water storage less than or equal to e for a given overall watershed storage of Se:
As = 1-(S2/(σe+S)2 Eq 4
Solving for σe gives the maximum effective local soil moisture storage within any particular fraction, As, of the overall watershed area:
σe=S(√(1/(1-As))-1) Eq 5
For a given storm event with precipitation P, the location of the watershed that saturates first (As = 0) has local storage e = 0, and runoff from this location will be P – Ia. Successively drier locations retain more precipitation and produce less runoff according to the moisture – area relationship of Eq. 5. The driest location that saturates during a storm defines the total contributing area (Af).
As average effective soil moisture storage (Se) changes through the year, the moisture-area relationship will shift accordingly (Eq. 5); Se is constant once runoff begins.
As = 1-(S2/(σe+S)2
Unsaturated
Saturated
σe=S(√(1/(1-As))-1
Af = f(S,Pe)
Steenhuis et al ASCE Div Drain. & Irr. 121: