STREAMFLOW ESTIMATION IN UNGAUGED CATCHMENTS USING REGIONALIZATION TECHNIQUE, AWASH RIVER BASIN, ETHIOPIA

Abstract

Many streams in Ethiopia are ungauged and do not have flow records. Even when streamflow gauges are available, records are too short. In Awash River Basin, which is one of the largest Ethiopian rift valley Basins, many catchments are ungauged. Consequently, in estimating streamflow, applying regionalization techniques for ungauged or poorly gauged catchments is crucial. This paper deals with estimating streamflow at ungauged catchments using HBV-light model in Awash River Basin. Hydrological and meteorological data for the selected gauged catchments were collected and the respective catchment characteristics were derived for both gauged and ungauged catchments. To optimize the model parameters for gauged catchments, model calibration were performed using automatic and manually techniques for a period of 2000 – 2010. The model was then validated for a period of 2011-2015. A sensitivity analysis was performed for model parameters. Two regionalization methods which are multiple regression (regional model) and similarity based were applied to transfer model parameter values from gauged to ungauged catchments. The calibration and validation result shows that all the seven selected gauged catchments have a good agreement since the Nash-Sutcliffe Efficiency (NSE) value greater than 0.60 and Relative Volume Error (RVE) lies between ±10%. Sensitivity analysis of HBV light model parameters evaluation shows that with respect to NSE parameters like; MAXBAS, K0, BETA, K1 and FC are the most sensitive. Besides this, with respect to RVE parameters like; BETA, LP and FC are the most sensitive for all catchments. Both the regional model and similarity based methods have shown good performance in estimating streamflow at ungauged catchments, but the performance comparison of the regionalization methods indicates that estimation of streamflow at ungauged catchments using similarity based method yields better performance. Finally, on the bases of parameter transfer; similarity based method gives the total average annual flow at ungauged catchments as 3.07BMC and this estimated flow can be used as data for different water resource applications.