Uncertainty estimation of rainfall- runoff calibration process using the Generalized likelihood method (GLUE) in HBV model

It is not possible to define the exact amount of runoff in the watersheds generated by rainfall due to contribution of different variables such as evaporation, transpiration and infiltration and it contains error.There is uncertainty in simulating streamflow because of complexity of the process. Hence, it is necessary to consider uncertainty in calibration of applied models which are categorized in three parts. They are associated with input data, model structure and parameters of the model. Using statistics and Bayesian analysis is a useful approach for calculating the uncertainty of simulation processes. The mixture of Bayesian analysis and Monte Carlo approach regarding the goodness of fit measures has been used for quantification of uncertainty in this research. This method is called generalized likelihood uncertainty estimation (GLUE). It has applied for calculating the uncertainty of modeling daily stream flow in upstream of Shahid Rajaee dam in Tajan watershed with HBV model. The results have shown that it is possible to quantify the uncertainty with the proposed method. Moreover, the Nash suttclife measure varied between 0.4 and 0.68. The results have shown that the GLUE approach is a suitable method for both quantification of uncertainty and also investigation of equifinality theory which states that different parameter sets can result to a similar goodness of fit index.