Simulating nonpoint source pollutant loading in a karst basin: A SWAT modeling application

Highlights: • Simulating water quantity and quality is especially challenging in karst watersheds • SWAT model performance was quantified in a HUC-8 karst basin of the central U.S. • A baseflow separation technique captured rapid delivery of groundwater to streams • Pasture land use dominated 62% sediment, 75% TN, and 66% of TP loading's • Legacy TP contributions from wastewater treatment plants were 33% of TP loading Karst terrain confounds efforts to parameterize and calibrate The Soil and Water Assessment Tool (SWAT). The purpose of this study was to use SWAT to simulate long-term natural (e.g. climate) and human (e.g. land use) impacts to streamflow, sediment and nutrient loading in a karst catchment. Sub-objectives were to: (1) perform a rigorous model performance assessment of SWAT, and (2) apply SWAT to highlight critical source areas of NPS pollutant loading in a large (HUC-8 spatial scale) karst basin of the central USA. Long-term (25-years) of daily streamflow data collected at seven nested U.S. Geological Survey flow gaging sites were used to calibrate and validate SWAT. Annual estimates of spring flow contributions from dominant springs were disintegrated to daily timeseries' dependent on baseflow separation results calculated from observed streamflow data. Model calibration results of daily streamflow showed percent bias (PBAIS) ranged from 1.1 to 19.3%, and Nash-Sutcliff efficiency (NSE) values ranged from 0.25 to 0.63. Model validation results of daily streamflow showed PBIAS ranged from −5.1 to −32.8%, and NSE values ranged from 0.44 to 0.78. Results highlighted potential critical source areas of pollutant loading from pasture sources. Simulated overland (i.e. not including channel processes) sediment, TN, and TP yields were 108 Mg km⁻² yr⁻¹, 14.1 kg ha⁻¹ yr⁻¹, 0.68 kg ha⁻¹ yr⁻¹, respectively, at the HUC-8 spatial scale. The greatest N and P nutrient yields were associated with a waste water treatment plant (WWTP), and WWTP loading contributions persisted over 100 km downstream ultimately accounting for approximately 37% of TP loading contributions at the basin outlet. Results have implications for end users tasked with creating physically realistic models of karst watershed hydrology using SWAT, and point to an ongoing need for improved sub-surface storage and flow routing SWAT modeling routines.