Application of Generalized Likelihood Uncertainty Estimation Approach in Real Time Updating of Flood Forecasting Model

Document Type : Research Article



In this paper, the generalized likelihood uncertainty estimation (GLUE) method is evaluated for updating rainfall-runoff model parameters in Gharehsoo-Karkheh sub basin. In two scenarios, observed discharges in continuous (from the beginning of the flood hydrograph) and discrete periods were used to change the parameters and update the forecasts in each time step. The continuous scenario showed better performance. Therefore, lead time and forecast error were computed in the continuous scenario. It was assumed that precipitation ends in the next time step while rainfall intensity and observed discharge in the current time step were used for updating. The results showed that GLUE was capable and flexible in correcting the forecast discharge error due to no forecast of rainfall. Lead time was represented by the difference in the time at which the peak discharge distribution becomes stable and the actual observed time to peak discharge. A lead time of 8 hours was achieved for two floods in Gharehsoo sub-basin with some 20 hours of time of concentration. The forecast error remains the same after the time when the peak discharge distribution becomes stable. The error in the simulation of the peak discharges of two floods were 21 and 16 percent which are caused by the uncertainty in the model structure and input data.