Flood Forecasting and Stream Flow Simulation of the Upper Awash River Basin, Ethiopia Using Geospatial Stream Flow Model (GeoSFM)

Abstract

The Geospatial Stream Flow Model (GeoSFM) has been widely applied in data scarce regions for flood forecasting and stream flow simulation with remotely acquired data. GeoSFM was applied in the Upper Awash River basin (UARB) with observed input data set. GeoSFM sensitivity to observed input data quality, subbasin partition, and change in parameter were investigated. Results demonstrated that GeoSFM is sensitive to the size and number of subbasins. Among the eight model parameters, the basin loss and curve number are the most sensitive in UARB. GeoSFM evaluation using a split sample of 10 years observed daily discharge showed satisfactory performance, Nash-Sutcliff Efficiency 0.67 and 0.70, coefficient of determination, 0.60 and 0.65 for calibration and validation, respectively. The monthly average simulation captured 76 % of the observed variability over 10 years. Comparative analysis suggested increasing partitions improves performance in capturing flooding events and the single basin scenario can potentially be used for flood forecasting or resource assessment purposes. The 60 % coverage of vertisol in the basin and low quality of observed data affected model performance. Further evaluation of GeoSFM in heterogeneous soil type and land use/cover can help to identify the influence of dominant physical characteristics. In general, GeoSFM offers a competent platform for stream flow simulation and water resource assessment in data scarce regions.