水文模型参数优选中率定与校核目标函数的关系研究

在水文模型参数优选过程中,通常很难找到一组参数值使得率定阶段和校核阶段的径流模拟误差同时达到最小,为此,我们需要采用基于Pareto关系的多目标优化方法来寻求Pareto最优解的集合.本文采用一种基于马尔可夫链蒙特卡洛（MCMC）方法的多目标优化方法来搜寻水文模型参数优选问题中Pareto解集,并以三水源新安江模型为例,给出了由率定阶段的目标函数和校核阶段的目标函数所构成的Pareto锋面.结果证明,率定阶段和校核阶段的目标函数是相互冲突的,不可能同时取最小值,由于这种Pareto关系的存在,使得我们在选择水文模型的全局最优参数值时存在很大的不确定性.因此,如何减少这种不确定性是水文模型研究中一个很重要的问题.

Relationship Between Calibration and Verification Objective Functions in the Optimization of Hydrological Model Parameters

In the optimization of the hydrological model parameters,it is very difficult to obtain the minimum discharge simulation errors for both the calibration period and the verification period simultaneously.In such a case,the multi-objective optimization methods based on the Pareto relationship are often adopted to search for the set of the Pareto optimal solutions.In this paper,a multi-objective optimization method based on the Makov Chain Monte Carlo method is employed to search for the Pareto optimal solutions.This method is applied to the Xinanjiang model of the three water sources,and the Pareto front formed by the objectives of both the calibration and verification periods is presented.The results have shown that the objective of the calibration period is conflicted with the one for the verification period,and they cannot be minimized simultaneously.Because of the existence of such a Pareto relationship between the objectives of the calibration and verification periods,a great deal of uncertainty will arise in the selection of the so-called "global optimum parameter values" for the hydrological models.Hence,how to reduce this kind of uncertainty in the hydrological modeling remains a very important problem to be solved.