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

The accurate determination of the amount of runoff resulting from precipitation on the surface of the watersheds is accompanied by errors due to the effect of various components, such as soil moisture, evaporation and transpiration, infiltration, and the impossibility of accurately measuring them. Therefore, the simulation of the precipitation-runoff process is associated with uncertainty. Uncertainty in calibrating models is caused by input information, model structure, and used parameters. Quantifying uncertainty is necessary for making decisions in water resource plans. One of the methods of calculating the uncertainty in the simulation process is to use Bayes's theory as the basis of calculations. In this research, an innovative method, which is a combination of Bayes analysis and the Monte Carlo method, taking into account the goodness of fit criteria, under the title of generalized similarity function, was used to calculate uncertainty. To determine the uncertainty of the parameters used in the calibration of the HBV rainfall-runoff model, the equation of daily flow entering the Shahid Rajaei dam in the Tajen catchment was used. The results showed that the mentioned method can detect uncertainty in the model. So the Nash index was obtained in the range of 0.4 to 0.68. The mentioned method is effective in identifying and introducing the co-termination theory, using a set of different parameters in the calibration of the model. So that by using the set of parameters, the same value of the goodness of fit index is obtained.