                 |
GIS-Based Water Quality Modeling with SWAT
Collaborators
PI:
Sabine Grunwald, Soil and Water Science Department, University of Florida
Co-PI:
Kevin Czaijkowski, Department of Geography and Planning, University of Toledo, OH
Collaborators:
Jeff Arnold, United States Department of Agriculture (USDA) - Agricultural Research Service
(ARS), Temple, TX and
A. van Griensven
Graduate student:
Kevin Alicia, Department of Geography and Planning, University of Toledo, OH
Chen Qi, Civil Engineering, University of
Florida
Time
09/2000 to 06/2003
Funding Source
Ohio Lake Erie Commission
Soil and Water Assessment Tool (SWAT)
SWAT is a quasi physically-based water quality simulation model that operates on a daily time step.
It is a basin-scale model developed by Dr. Jeff Arnold, USDA-ARS, in Temple, TX. SWAT was
developed to predict the impact of land management practices on water, sediment and agricultural
chemical yields in complex watersheds with varying soils, land use and management conditions over
long periods of time. The SWAT model components include:
(i) hydrology,
(ii) weather,
(iii) sedimentation,
(iv) soil temperature,
(v) crop growth,
(vi) nutrients,
(vii) pesticides, and
(viii) agricultural management (Neitsch et al., 1999).
To accurately predict movement of sediment, nutrients and pesticides, the hydrologic cycle as
simulated by the model must conform to what is happening in the watershed. For simulations the
watershed of interest is subdivided into simulation elements, either grid cells, virtual basins or
subbasins. For each simulation element water flux and transport of sediment and agrichemicals
are simulated and then routed through a watershed, i.e., water and chemicals are transported
from one simulation element to the next depending on flow characteristics. An ArcView based
interface is available to input GIS into SWAT (DiLuzio et al., 1997).
More information about SWAT is available at: http://www.brc.tamus.edu/swat/
References
DiLuzio, M., R., Srinivasan and J.G., Arnold. 1997. An integrated user interface for SWAT using
ArcView and Avenue. ASAE Meeting, Minneapolis, MN Aug. 10-14, 1997; Paper No. 972235.
Neitsch, S.L., J.G., Arnold and J.R., Williams. 1999. SWAT - Soil and Water Assessment Tool -
user's manual Version 99.2. Grassland, Soil and Water Research Laboratory & Blackland
Research Center, USDA-ARS, Temple, TX.
Objectives
(1) Assemble a high-quality Geographic Information System (GIS) dataset for the Sandusky Watershed
using readily available data, which includes soils, land use, management, geology, topography, and
climate. Our goal is to develop a new GIS land use layer utilizing Landsat TM images.
(2) Simulate transport processes such as infiltration, sediment, nutrient (N, P) and pesticide transport to
assess point and non-point source pollution in the Sandusky Watershed utilizing SWAT.
(3) Utilize existing detailed loading data to test model predictions.
(4) Identify areas that show large loads of sediments and agrichemicals.
(5) Describe and visualize the spatial distribution of water quality indicators over time in the Sandusky
Watershed.
Study Area: Sandusky Watershed, Ohio
The Sandusky Watershed, with a drainage area at Fremont of 3,240 km2, is located within the Lake
Erie Watershed and Great Lakes basin. The Sandusky River is the second largest of the Ohio rivers
draining into Lake Erie. Analysis of 1994 LANDSAT data indicates that 84% of the land is used for
agriculture, 12.6% is wooded, 1.2% is urban and 1.1% is non-forested wetlands. Major crops
based on county-level estimates in 1985 were corn (Zea mays L.) with 35.6% of cropland acreage,
soybeans (Glycine max L.) with 44.9% and wheat (Triticum aestivum L.) with 19.5%. Crop
production was similar in 1995 with 32.1% in corn, 49.1% in soybeans and 18.7% in wheat.
Tillage practices shifted from 86% conventional management in 1985 to 50.5% in 1995, as farmers
replaced conventional with conservation tillage practices. Tile drainage is used extensively throughout
the watershed. Urban areas within the Sandusky Watershed are Bucyrus, Fremont, Tiffin, and Upper
Sandusky, and numerous smaller communities. The river and its major tributaries support important
recreational uses for watershed residents. Bedrock underlying the watershed is primarily Silurian
dolostone and Devonian limestone. In the eastern portion of the watershed, Devonian shale and
Mississippian sandstone are present. Surface features of the Sandusky Watershed reflect the effects
of the Wisconsinian glaciers, which retreated approximately 13,000 years ago. This resulted in two
physiographic regions in the watershed, the Lake Plains in the northern portion and the Till Plains in
the central and southern portions. The landscape of the Lake Plains is an extremely flat plain of fine
clay sediments, formed by wave action of glacial meltwater lakes that preceded Lake Erie. Sandier
soils are present in the remnants of occasional beach ridges from these lakes. The Till Plains consists
of flat to gently rolling plains with heavy till soils. Most of the relief within the Till Plains is located in
three end moraines that lie in an east-west orientation. The majority of the Till Plaines consists of flatter,
ground moraines which lie between the end moraines. Besides glacial till, lacustrine sediments and
alluvial deposits along the drainage system of the Sandusky River are found. Dominant soils are
Hapludalfs, Ochraqualfs, Fragiaqualfs, Medisaprists, Fluvaquents, and Argiaquolls. Textures are
mainly silt loam and silty clay loam. Average annual precipitation ranges from 881 mm at Fremont
to 964 mm at Bucyrus. Historic precipitation data for the watershed show highest amount for July
(99 mm) and smallest for February (48 mm). Annual mean discharge for the Sandusky Watershed
at Fremont is 29.1 m3s-1, Honey Creek at Melmore is 3.8 m3s-1, and Rock Creek at Tiffin is
0.88 m3s-1.
Hydrologic units and monitoring stations in the Sandusky Watershed.
Impairment of the Sandusky Watershed
Ohio EPA's 1998 field survey of stream segments of the Sandusky Watershed indicated the following
causes for impairment:
(1) Lower Sandusky Watershed (incl. Wolf, Green, Indian, and Muddy creeks)
84.4 stream miles assessed from total of 279.3 stream miles
(2) Middle Sandusky Watershed (incl. Sugar, Willow, Rock, Honey, and Broken Knife creeks)
118.6 miles assessed from total of 256.0 stream miles
(3) Tymochtee Creek
21.4 miles assessed from total of 187.8 stream miles
(4) Upper Sandusky Watershed (incl. Sandusky River, Negro Run, Rock Run, Broken Sword Creek,
Paramour creeks, Crestline Tributary)
99.4 miles assessed from a total of 222.7 miles
| |
Impairments |
Miles impaired |
| Lower Sandusky Watershed |
Habitat alterations |
22.0 |
| |
Flow alterations |
21.0 |
| |
Nutrient enrichment |
17.0 |
| |
Siltation |
18.0 |
| |
Ammonia |
17.0 |
| Middle Sandusky Watershed |
Habitat alterations |
6.5 |
| |
Nutrient enrichment |
2.5 |
| |
Siltation |
13.0 |
| |
Total organics |
2.5 |
| |
Cause unknown |
18.0 |
| Tymochtee Creek |
Habitat alterations |
21.4 |
| |
Siltation |
21.4 |
| Upper Sandusky Watershed |
Habitat alterations |
18.0 |
| |
Nutrient enrichment |
41.0 |
| |
Siltation |
75.0 |
| |
Ammonia |
29.0 |
| |
Metals |
21.0 |
| |
Oil and greese |
5.0 |
Water Quality Monitoring
Monitoring results of the Water Quality Laboratory (WQL) indicate that unit area loads in
the Sandusky Watershed are highest for total phosphorus, suspended sediments, and nitrates
out of seven major watersheds in Ohio and higher than most other locations. In the Sandusky
Watershed point sources of total phosphorus averaged 5.5% and do not constitute more
than 15% of the annual loads in any year of the period 1975-1995.
Results
van Griensven, A, T. Meixner, S. Grunwald, and R. Srinivasan. 2008.
Fit-for-purpose uncertainty versus calibration uncertainty in model-based
decision making, Hydrological Sciences Journal.
Grunwald S. and C. Qi. 2006. GIS-based
water quality modeling in the Sandusky Watershed. J. of the American Water
Resources Association, 42(4): 957-973.
van Griensven A., T. Meixner, S.
Grunwald, A. Di Luzio and R. Srinivasan. 2006. A global sensitivity
analysis tool for the parameters of multi-variable watershed models. J. of
Hydrology, 324: 10-23.
Qi C. and S. Grunwald. 2005. GIS-based
hydrologic modeling in the Sandusky Watershed. Transactions of the ASAE
48(1): 169-180.
van Griensven A., T. Meixner, S. Grunwald and R. Srinivasan. 2005.
Evaluation methods for SWAT models. SWAT 3rd Int. Conference, Zuerich,
Switzerland, July 11-15, 2005.
Qi C. and S. Grunwald. 2004. GIS-based spatially-distributed water
quality modeling in the Sandusky Watershed. ASA-CSSA-SSSA Meeting, Seattle,
WA, Oct. 31 - Nov. 4, 2004. (poster)
|
Hydrologic units and monitoring stations in the Sandusky Watershed.
Impairment of the Sandusky Watershed
Ohio EPA's 1998 field survey of stream segments of the Sandusky Watershed indicated the following
causes for impairment:
(1) Lower Sandusky Watershed (incl. Wolf, Green, Indian, and Muddy creeks)
84.4 stream miles assessed from total of 279.3 stream miles
(2) Middle Sandusky Watershed (incl. Sugar, Willow, Rock, Honey, and Broken Knife creeks)
118.6 miles assessed from total of 256.0 stream miles
(3) Tymochtee Creek
21.4 miles assessed from total of 187.8 stream miles
(4) Upper Sandusky Watershed (incl. Sandusky River, Negro Run, Rock Run, Broken Sword Creek,
Paramour creeks, Crestline Tributary)
99.4 miles assessed from a total of 222.7 miles