Swat-based Streamflow and Embayment Modeling of Karst-affected Chapel Branch Watershed, South Carolina

SWAT is a GIS-based basin-scale model widely used for the characterization of hydrology and water quality of large, complex watersheds; however, SWAT has not been fully tested in watersheds with karst geomorphology and downstream reservoir-like embayment. In this study, SWAT was applied to test its ability to predict monthly streamflow dynamics for a 1,555 ha karst watershed, Chapel Branch Creek, which drains to a large embayment and is comprised of highly diverse land uses. SWAT was able to accurately simulate the monthly streamflow at a cave spring (CS) outlet draining mostly agricultural and forested lands and a golf course plus an unknown groundwater discharging area, only after adding known monthly subsurface inputs as a point source at that location. Monthly streamflows at two other locations, both with multiple land uses, were overpredicted when lower lake levels were prevalent as a result of surface water flow to groundwater (losing streams). The model underpredicted the flows during rising lake levels, likely due to high conductivity and also a deep percolation coefficient representing flow lost to shallow and deep groundwater. At the main watershed outlet, a wide section performing as a reservoir embayment (R-E), the model was able to more accurately simulate the measured mean monthly outflows. The R-E storage was estimated by using a daily water balance approach with upstream inflows, rainfall, and PET as inputs and using parameters obtained by bathymetric survey, LiDAR, and downstream lake level data. Results demonstrated the substantial influence of the karst features in the water balance, with conduit and diffuse flow as an explanation for the missing upstream flows appearing via subsurface conveyance to the downstream cave spring, thus providing a more accurate simulation at the embayment outlet. Results also highlighted the influences of downstream lake levels and karst voids/conduits on the watershed hydrologic balance. Simulation performance of hydrology could be improved with more accurate DEMs obtained from LiDAR for karst feature identification and related modification of SWAT parameters. This SWAT modeling effort may have implications on nutrient and sediment loading estimates for TMDL development and implementation in karst watersheds with large downstream embayments that have significant changes in water level due to adjoining lakes.