Study On Hydrological Model Parameter Calibration Method And Characteristics Of Equifinality Phenomenon

The watershed hydrological model is the approximate expression of the hydrological real process by hydrologists using mathematical formulas and physical and chemical principles based on the preliminary understanding of the random and complex hydrological processes and water circulating laws.It is an important tool for hydrological process simulation and hydrological rules research.The key problems to build and run hydrological models are the calibration of the internal parameters and the determination of the hydrological model structure.In the actual calibration of hydrological model parameters,the influence of various uncertain factors,such as mutual influence of parameters,the error of input data,the error of the model structure and the multi-extremum phenomenon of the objective function in parameter calibration,made it difficult to obtain the true value of the parameter directly by the method of parameter calibration.In the process of parameter calibration,it existed many different groups of parameters had the same value of likelihood function,which was called equifinality phenomenon.The equifinality phenomenon in parameter calibration has become a hot research issue in rencent years.It is of great significance for current hydrological model parameter calibration and uncertainty analysis to explore the causes of equifinality phenomenon and to formulate solutions to avoid the equifinality phenomenon in parameter calibration.In this paper,SCE-UA algorithm and SCEM-UA algorithm were used as parameter optimization algorithms,and Nash model parameters were taken as the research object.The Nash model parameters were optimized based on the observed rainfall and runoff data of the BaoAn hydrologic station in Naoli River Basin of the Sanjiang Plain.According to the background of two kinds of input data(ideal data and actual data),the performance characteristics of equifinality phenomenon were analyzed respectively.In order to weaken the effect of equifinality phenomenon on the result of parameter calibration,the Screenig Method based on Multi-creteria Likelihood Weight was proposed.The main contents and conclusions of this paper are as follows.(1)Construction of a simulation optimization model based on different optimization algorithms and parameter calibrationSCE-UA algorithm and SCEM-UA algorithm were used to build simulation optimization models,and Nash model parameters were calibrated respectively.A new parameter optimization method was proposed based on the optimization results of parameters according to the simulation optimization model of SCEM-UA algorithm,which is called the Screenig Method based on Multi-creteria Likelihood Weight(SMMLW).Through the comparison of the simulation accuracy of the three algorithms,the rationality of the new parameter optimization method proposed in this paper was verified.(2)Analysis of the characteristics of equifinality phenomenon based on different input data situationsBased on the background of two different input data(ideal data and actual data),the generality and difference of performance characteristics of equifinality phenomenon were analyzed.The results were showen that: 1)For the same flood,under the ideal data background,the effect of the equifinality phenomenon was less than that of the real data.2)For different floods,under the ideal data or the actual data background,the obvious degree of the equifinality phenomenon was related to the flood peak and flood volume of the input flood,and all in accordance with the certain law,and all in accordance with the certain law.Specifically,under the background of ideal data or actual data,with the increase of flood peak and flood volume of input floods,the effect of equifinality phenomenon is increasing.On the contrary,with the reduction of flood peak and flood volume of input floods,the effect of equifinality phenomenon were significantly decreased.