1. Comparing SVI spatial distribution
with nutrient load data in the
Mark Twain watershed in Missouri
Sapana Lohani, University of Missouri
Claire Baffaut, USDA-ARS, CSWQ Columbia, MO
Allen Thompson, University of Missouri
2. Validation
• Question: Does SVI help to explain annual nutrient
load in water quality measurement ?
• Hypothesis:
𝐴𝑛𝑛𝑢𝑎𝑙 𝐿𝑜𝑎𝑑 𝑆𝑡𝑟𝑒𝑎𝑚 = 𝑓(𝑅𝑎𝑖𝑛 𝐹𝑎𝑐𝑡𝑜𝑟, 𝐶𝑟𝑜𝑝 𝐹𝑎𝑐𝑡𝑜𝑟)
• Everything else being same, watershed with SVI-
low will contribute less than watershed with SVI-
high.
5. Crop
Factor
Land Cover Types
Corn/Sorghum
Corn, Dbl Crop Barley/Corn, Dbl Crop Barley/Sorghum, Dbl Crop
Corn/Soybeans, Dbl Crop Durum Wht/Sorghum, Dbl Crop Oats/Corn, Dbl Crop
WinWht/Corn, Dbl Crop WinWht/Sorghum, Pop or Orn Corn, Sorghum, Sweet
Corn
Soybean Soybeans
Other Ag
Sunflower, Winter Wheat, Dbl Crop WinWht/Soybeans, Oats, Millet, Alfaalfa,
Clover/Wild flowers, Other Small Grains, Misc Vegs & Fruits, Rye, Rice,
Pasture Grass/Pasture, Other Hay/Non Alfaalfa, Switchgrass
Forest Deciduous Forest, Evergreen Forest, Forest, Mixed Forest, Woody Wetlands
Other Non-Ag
Aquaculture, Background, Barren, Clouds/No Data, Developed, Developed/High
Intensity, Developed/Low Intensity, Developed/Med Intensity, Developed/Open
Space, Fallow/Idle Cropland, Herbaceous Wetlands, Nonag/Undefined, Open
Water, Perennial Ice/Snow, Shrubland, Water, Wetlands
SVI Classes
High
Moderately
High
Moderate
Low
Land Cover Types
Corn/Sorghum
Corn, Dbl Crop Barley/Corn, Dbl Crop Barley/Sorghum, Dbl Crop
Corn/Soybeans, Dbl Crop Durum Wht/Sorghum, Dbl Crop Oats/Corn, Dbl Crop
WinWht/Corn, Dbl Crop WinWht/Sorghum, Pop or Orn Corn, Sorghum, Sweet
Corn
Soybean Soybeans
Other Ag
Sunflower, Winter Wheat, Dbl Crop WinWht/Soybeans, Oats, Millet, Alfaalfa,
Clover/Wild flowers, Other Small Grains, Misc Vegs & Fruits, Rye, Rice,
Pasture Grass/Pasture, Other Hay/Non Alfaalfa, Switchgrass
Forest Deciduous Forest, Evergreen Forest, Forest, Mixed Forest, Woody Wetlands
Other Non-Ag
Aquaculture, Background, Barren, Clouds/No Data, Developed, Developed/High
Intensity, Developed/Low Intensity, Developed/Med Intensity, Developed/Open
Space, Fallow/Idle Cropland, Herbaceous Wetlands, Nonag/Undefined, Open
Water, Perennial Ice/Snow, Shrubland, Water, Wetlands
4 Cover Definitions
Corn/Sorghum +
Soybean + Other Ag
Corn/Sorghum +
Soybean
Corn/Sorghum
Only
Soybean
Only
Land Cover Data from NASS (National Agricultural Statistics Service): 2006-2010
6. SVI
Nutrient
Load
Sediment Load for WAG-SVI Sediment
Load for
WAG-ILow Moderate Moderately High High
SVI-Runoff
TNITR 9 11 15 22 14.25
TPHOS 0.7 1.1 1.7 2.9 1.6
SEDMT 0.7 1.2 2.2 5.1 2.3
DNITR 5.4 2.7 5.5 4.9 4.625
DPHOS 0.26 0.27 0.27 0.24 0.26
SVI-Leaching DNITR 1 2 4 8 3.75
Crop
Factor
𝑊𝐴𝐺 − 𝑆𝑉𝐼 =
𝑆𝑉𝐼=1
4
𝐿𝐶𝐹𝑟𝑆𝑉𝐼
𝑌𝑒𝑎𝑟
∗ 𝑆𝑒𝑑𝑖𝑚𝑒𝑛𝑡 𝐿𝑜𝑎𝑑 𝑆𝑉𝐼 𝑊𝐴𝐺 − 𝐼 = 𝐿𝐶𝐹𝑟 𝑌𝑒𝑎𝑟
∗ 𝑆𝑒𝑑𝑖𝑚𝑒𝑛𝑡 𝐿𝑜𝑎𝑑 𝐴𝑔
With SVI
Without SVI
-Chan et al., 2017
7. Regression
• Multiple regression using R programming
• Threshold used – 10% based on p-values
• Results with 90% confidence interval (significance level of 0.1)
were significant
• Results with 95% and 99% confidence interval were also noted
𝐴𝑛𝑛𝑢𝑎𝑙 𝐿𝑜𝑎𝑑 𝑆𝑡𝑟𝑒𝑎𝑚 = 𝑓(𝐶𝑟𝑜𝑝 𝐹𝑎𝑐𝑡𝑜𝑟, 𝑅𝑎𝑖𝑛 𝐹𝑎𝑐𝑡𝑜𝑟)
8. Regression Runs
Annual
Load
Cover
Definition
Sites Years Total
Soil Runoff
All Combined 5 4 all all 20
Year-wise 5 4 all 5 100
Site-wise 5 4 7 all 140
260
Soil Leaching
All Combined 1 4 all all 4
Year-wise 1 4 all 5 20
Year-wise 1 4 7 all 28
52
• All Combined: 35 observations / 154 variables
• Year-wise: 7 observations / 154 variables
• Site-wise: 5 observations / 154 variables
9. Regression Results
SVI Land Cover
SVI is better
than Non-SVI
Non-SVI is
better than SVI
Only SVI
Significant
SVI
improved
significance
Total Runs
SVI
Runoff
All Combined 45 % 10 % 10 % 35 % 20
Ag lands only 0 0 0 0 5
Corn/Sorghum
+Soybean
40 % 0 % 20 % 20 % 5
Corn/Sorghum
Only
40 % 0 20 % 20 % 5
Soybean Only 40 % 40 % 0 40 % 5
SVI
Leaching
All Combined 50 % 0 0 50 % 4
• All Combined: 5 output variables, 4 land cover definitions, 5 years and 7 sites
(35 observations / 154 variables)
10. Regression Results
SVI
SVI is better
than Non-SVI
Non-SVI is
better than SVI
Only SVI
Significant
SVI
improved
significance
Total Runs
SVI
Runoff
All Combined 45 % 10 % 10 % 35 % 20
Ag lands only 0 0 0 0 5
Corn/Sorghum
+Soybean
40 % 0 % 20 % 20 % 5
Corn/Sorghum
Only
40 % 0 20 % 20 % 5
Soybean Only 40 % 40 % 0 40 % 5
SVI
Leaching
All Combined 50 % 0 0 50 % 4
• All Combined: 5 output variables, 4 land cover definitions, 5 years and 7 sites
(35 observations / 154 variables)
11. Conclusion
• SVI was significant in helping explain pollutant loads in 45% cases
for soil runoff and 50% cases for soil leaching.
• There were cases (35% for soil runoff and 50% for soil leaching)
where even not using SVI was significant but by using SVI
component in the crop factor, the significance was improved.
14. Summary – SVI Runoff Potential
** SVI is reported to have over-estimated runoff risk in those watersheds, which could be
improved if drained condition of HSG were used.
Efficient Could be Efficient ** Inefficient due to steep
slope
Little River Experimental
Watershed
South Fork of Iowa River Goodwin Creek Watershed
Goodwater Creek – Mark
Twain Lake Watershed
Walnut Creek WE-38
Upper Big Walnut Creek Choptank Watershed
Cedar Creek Watershed Beasley Lake Watershed
Delta Water Research
Management Center
15. Summary – SVI Leaching Potential
• SVI is useful for in identifying areas inherently vulnerable leaching.
• Experts from almost all sites agreed that SVI maps represented leaching in the
watersheds well.
• However, there are fewer monitoring data for leaching (unlike erosion monitoring,
which is more common).
• Few sites (South Fork of Iowa River, Walnut Creek, Little River, Upper Big Walnut
Creek) have monitoring data for leaching.
• Leaching might also benefit from using the drained hydrologic soil group when the
land is drained (Delta Water).