基于MOMM-GLUE法月水量平衡模型不确定性分析

Uncertainty Analysis of Monthly Water Balance Model Based on MOMM-GLUE Algorithm

以两参数月水量平衡模型为例,建立了用于模型不确定性分析的MOMM-GLUE方法,针对单目标参数率定的不足,采用基于MCMC方法的多目标SCEM-UA算法推求参数后验分布,应用GLUE法分析参数的敏感性与预报不确定性。并针对蒸发规律明显的季节性,将两参数模型的蒸发参数C按汛期与非汛期分解为三参数模型。结果表明,不同目标函数分析参数敏感性结果不同,较高后验密度的参数集并不十分敏感,有利于参数寻优;就区间估计而言,三参数模型优于两参数模型,但相应增加了参数不确定性。

The method of MOMM-GLUE is developed for analyzing the uncertainty of two parameters water balance model（P2M） and MCMC-based multi-object SCEM-UA algorithm is applied to infer parameter of posterior distribution.And the sensitivity and forecasting uncertainty of model parameters is studied with GLUE method.Due to the season changes of watershed evaporation,the evaporation parameter C is divided into two parameters corresponding to the flood season and non-flood season,which is called three-parameter model（P3M）.It shows that the results of parameter sensitivity analysis rely on the selected objective function;the higher posterior density of parameter sets has little sensitivity;as far as the interval estimation property is concerned,P3M is superior to P2M while increasing the parameter uncertainty.