Quantifing model uncertainity to improve stream flow prediction geba cathment, upper tekeze River basin, Ethiopia
- International Journal of Hydrology
This study was conducted to predict stream flow at the outlet of the gauged Geba watershed and analyze the associated uncertainty that can affects its accurate prediction. Semi distributed HBV light model was applied to estimate stream flow of the Geba catchment and associated uncertainty through Monte Carlo Simulation procedure. The calibration and validation of the model was found satisfactory as performance rating criteria value of coefficient of correlation (R2) and Nash-Sutcliffe simulation efficiency (ENS) is found to be 0.74 and 0.72 for calibration and 0.73 and 0.70 for validation ,respectively. In the same order from the model uncertainties analysis the percentage of the simulated data within the uncertainty bound is only 35% for calibration and 29% for validation, which shows that there is uncertainty in the process. Then using Monte Carlo Simulation procedure in HBV light model parameter uncertainty was tested and found with ENS value of 0.76 for calibration and 0.79 for validation. And this shows that the overall associated uncertainty come from either conceptual or input or a combination of them but not from parameter identification.Even though the predicted amount of flow of 955.33MCM is almost equivalent to the latest study, the uncertainty might come due to either neglected abstractions or poor quality of input data. Therefore, this simulated amount should not be used for any water resource development works unless the correction of these cause of uncertainties are reduced as uncertainty in estimation of simulated flow will lead to wrong water resource management decision.
flow prediction, HBV light, ERDAS, model uncertainty, geba cathment