Evaluating SWAT Model Performance for Runoff, Percolation, and Sediment Loss Estimation in Low-Gradient Watersheds of the Atlantic Coastal Plain

Abstract

With predicted alterations in climate and land use, managing water resources is of the utmost importance, especially in areas such as the United States (U.S.) Coastal Plain where extensive connections exist between surface and groundwater systems. These changes create the need for models that effectively assess shifting hydrologic regimes and, in that context, we examine the performance of the Soil and Water Assessment Tool (SWAT) in a low-gradient, shallow-aquifer-dominated watershed of the U.S. Coastal Plain using a gridded reanalysis dataset. We evaluate accuracy, uncertainty, and parameter sensitivity by comparing observed and predicted streamflow at two gaging stations and assess model predictions for yearly average runoff (SURQ), percolation (PERC), and sediment loss (SYLD). Streamflow performance was acceptable during calibration (NSE = 0.67 and 0.60) and very good during validation (NSE = 0.84 and 0.91). Model predictions for SURQ, PERC, and SYLD coincided with expected ranges for this region. Parameters related to shallow aquifer properties or groundwater were highly sensitive, which indicates the need for continued study of spatial and temporal variability within the sub-surface components of these hydrologic systems. Our findings highlight the applicability of this reanalysis dataset for modeling hydrologic processes in poorly gaged watersheds and adds to the body of research that seeks to develop effective assessment tools for shallow-aquifer-dominated systems. Our methodology can effectively assist watershed managers in establishing baseline rates of hydrologic processes as is crucial with future predicted shifts in hydrologic regimes due to land-use alteration and climate change.