Comparison of semi-distributed, GIS-based hydrological models for the prediction of streamflow in a large catchment

Abstract

Predicting streamflow in a large arid and semi-arid basin is of great importance in understanding the availability of water for spatial planning and water resource management. In this study, two geographic information system-based (GIS-based) semi-distributed hydrological models were compared for predicting flow. TOPMODEL and SWAT require the use of a GIS to process input data obtained from various sources, such as the digital elevation model (DEM), topographic index (TI), hydrologic response unit (HRU), meteorological stations, and soil- and land-use maps. Daily hydro-meteorological data were collected from 1989 to 2007, and 90-m resolution of DEM was considered. The models were compared, and their performances for the prediction of peak flows and runoff volumes were discussed. TOPMODEL and SWAT obtained good coefficient values for the validation period, i.e., 0.61 and 0.68, respectively. According to relative error percentage (RE %) criteria, TOPMODEL provided a promising value for the validation period (64 %) for peak flows, whereas SWAT provided about 70 %. TOPMODEL provided 5-year overestimation and 1-year underestimation for runoff volume; SWAT provided 2-year underestimation and 4-year overestimation. For this study, both models obtained promising simulation results for surface flow.