Calibration and uncertainty analysis of a hydrological model based on cuckoo search and the M-GLUE method

Abstract

The watershed hydrological model is regarded as a powerful tool for simulating streamflow, but it is subject to many uncertainties. TOPMODEL (TOPography-based hydrological MODEL) is used as hydrological modeling in this paper; general likelihood uncertainty estimation (GLUE) and multi-criteria GLUE (M-GLUE) methods are applied to evaluate the uncertain effect of model parameters on streamflow simulation, and three climate models are used to investigate the uncertain effect of meteorological input data. A new parameter calibration method (cuckoo search algorithm) is proposed in this study. Taking Beiluo River basin as a study case, analysis of the simulation results reveals that the cuckoo search algorithm is applicable and effective in optimizing the model parameters. The Morris and GLUE methods are employed to analyze the sensitivity of the parameters, and the two methods consistently demonstrated that there are three sensitive parameters in TOPMODEL. Additionally, the results of M-GLUE method are superior to the GLUE method, and both methods can effectively analyze the uncertainty of parameters. The precipitation and potential evaporation predicted by the three climate models exhibit an increasing trend, and the simulated average annual streamflow of the climate system model of the Beijing Climate Center (BCC-CSM1.1) is optimal and followed by Centre National de Recherches Meteorologiques Earth system model (CNRM-CM5) and Canadian Earth System Molde (CanESM2). However, results obtained by all the three methods are greater than the baseline period value, indicating that the diverse input data of the hydrological model lead to uncertainty in the streamflow simulation.