Soil erosion is one of the most important of today’s global environmental problems. Over the past few decades, soil conservation practices were implemented to reduce soil erosion in the United States. However, excessive sediment still remains among the most prevalent water quality problems. Agricultural fields and in particular ephemeral gullies (EGs) are considered to be a major contributor of sediment. The overall goal of this study was to improve modeling utility to identify and quantify sources of sediment. Specific objectives were:​ (1) to develop and demonstrate a method of field-scale targeting using Soil and Water Assessment Tool (SWAT) and to use this method as a targeted, flexible approach to pay explicitly for sediment-yield reductions;​ (2) to evaluate topographic index models (Slope Area [SA], Compound Topographic Index [CTI], Slope Area Power [SAP] and Wetness Topographic Index [WTI]) and a physical-based model (Overland Flow Turbulent [OFT]) in predicting spatial EG location and lengths.

Black Kettle Creek watershed was the focus of an innovative project to pay for modeled field sediment reductions. An Arc-Geographical Information System (GIS) tool bar was developed that post processed SWAT hydrologic response unit output to field boundaries and prepared maps of high-priority fields by sediment, total nitrogen, and total phosphorus and was demonstrated to be useful for field-scale targeting. Calibrated SWAT model was used to establish baseline sediment yields. Various Best Management Practices (BMPs) were simulated and payments to implement each BMP for a given field were calculated. This study helped to guide determination of appropriate farmer support payments and quantified the important influence of BMP type and site-specific conditions for use in targeting conservation practice funding to achieve maximum soil-loss reductions per dollar spent. Extreme care should be used in selecting the source of spatial model input data when using SWAT for field-level targeting.

Automated geospatial models were developed in a GIS environment to spatially locate and derive length of EGs using topographic index and physical based models. EG predictions were better for the SA model among the four topographic index models tested. Individual calibration of topographic index model threshold for each application site was needed. An OFT model (physical based model), which utilized topography, precipitation, soil, landuse/landcover and SWAT-based runoff estimates, did not need individual site calibration, and may have broader applicability than empirical based models.