Modification of the SWAT Model to Simulate Hydrologic Processes in a Karst-influenced Watershed

TR Number
Date
2009-08-12
Journal Title
Journal ISSN
Volume Title
Publisher
Virginia Tech
Abstract

In the United States, karst ecosystems cover approximately 20 percent of the country and karst aquifers provide 40 percent of the water used for drinking. In karst-influenced watersheds, karst features such as sinkholes and sinking streams act as rapid pathways for carrying water and pollutants into streams and groundwater. Human activities on karst landscapes can present some special problems such as alterations to hydrologic regime, contamination of groundwater, ground subsidence, and damage to cave ecosystems. Modeling a karst-influenced watershed can provide a better understanding of the interactions between surface and ground water and how water quality is affected by human activities.

Several models were evaluated to determine their ability to model both discharge and nutrient transport in karst watersheds. The Soil Water Assessment Tool (SWAT) model was found to be appropriate due to its capability to represent almost all of the hydrological processes, its user-friendliness, and its ability to generate most of the parameters from available data. Moreover, SWAT can represent nitrogen transformations and transport processes and calculate nitrogen loadings, which is critical for karst watersheds. While it has been widely used and found to be an appropriate prediction tool, it does not explicitly include the capacity to represent specific features characteristic of karst-influenced basins. Baffaut and Benson (2008) modified the SWAT 2005 code to simulate faster aquifer recharge in karst environments, and this version was further modified here in the SWAT-Karst to represent karst environments at the HRU scale. A new parameter sink allows simulating the hydrology and nitrate transport in a sinkhole representing its unique landuse and soil characteristics, and a new parameter ss partitions nitrate transported with water that is lost from sinking streams.

The SWAT-KARST model was used to simulate discharge and nitrogen loadings within the Opequon Creek karst-influenced watershed, located in the Potomac and Shenandoah River basin in Virginia and West Virginia. In the Opequon Creek watershed, SWAT-karst using the HRU to represent sinkholes had a more notable impact in the watershed hydrology than SWAT-B&B using a pond to represent sinkholes.

Results of statistical evaluation show that SWAT-karst and the Baffaut and Benson (2008) version performed better than SWAT in predicting streamflow in a karst-influenced watershed. Although SWAT-karst showed almost the same performance as SWAT-B&B, SWAT-karst model offers the flexibility to represent the unique relationship between surface and ground water in karst features in an HRU.

Using an HRU to represent sinkholes can depict the associated variability of a karst landscape. The new variables sink and ss provide a mechanism to represent the nutrient transport through sinkholes and sinking streams. Sensitivity analysis showed that SWAT-karst was sensitive to the new parameter sink which can be used for model calibration and to represent water recharge and nutrient transport to aquifers outside the watershed boundary.

Description
Keywords
karst, sinkholes, sinking streams, nitrate, SWAT, watershed modeling
Citation
Collections