GRIEROM: GRID–BASED VARIABLE SOURCE AREA SOIL–WATER EROSION AND DEPOSITION MODEL

Abstract

Erosion in a variable source area and the sediment in surface runoff have a major impact on soil conservationand water quality. It is necessary to locate the soil erosion areas and the paths of sediment transport efficiently to alleviatepotential soil loss problems and protect water quality. The objective of this study was to predict temporal variations and spatialdistributions of sediment transport in a variable source area for storm events. A GRIdbased soilwater EROsion anddeposition Model (GRIEROM) is described. This model uses ASCIIformatted map data supported by the GRASS GeographicInformation Systems (GIS) and generates distributed results such as sediment concentration and flux in overland flow areas.GRISTORM was adopted for the hydrological processes. A processbased soil erosion concept was adopted to simulatesoilwater erosion and deposition. The model deals separately with rill and sheet flow and handles vegetation in terms of soilcover. The model was applied to a 170 ha watershed located in the Northern Catskill region of New York. The initial soilmoisture conditions for storm events were based on calibrated values from a Soil Moisture Routing (SMR) model and wereadjusted to include soil moisture variations within a day. Predicted flows from four storm events in 1993 were compared withobserved flows at the watershed outlet for hydrologic calibration. The average NashSutcliffe efficiency for predicting flowat the outlet using calibrated data was 0.847. Parameters related to sheet/rill flow on bare soil were more sensitive than thoseon vegetated soil. Sediment depositability (
) proved to be the most sensitive parameter, followed by the low value of soilerodibility (). The NashSutcliffe efficiencies for predicting sediment yield at the outlet for the two storms were negativebecause of additive prediction and time error of sediment concentration, respectively. The average NashSutcliffe efficiencyfor another two storms was 0.476. The temporal variations and spatial distributions of overland flow and sediment areas werepresented using GRASS. These presentations suggest that most sediment was transported to the stream from the areas thathad high overland flow velocities.