分布式降雨径流模型率定的多目标分析方法

利用多目标方法，对自主研制的具有物理基础的分布式降雨径流模型进行了参数的多目标分析和模型有效性探讨．通过对参数的敏感性分析，确定糙率系数和土壤饱和水力传导率为率定的主要参数，选择效率系数、模拟洪峰的相对误差和洪峰达到时间的绝对误差为目标函数．在模型率定过程中，首先利用多目标最优化方法获得模型参数的非劣解集，然后又利用离散型协调规划从非劣解集中选出满意的模型参数，最后通过对非劣解集模拟的过程线的范围及形状的分析，对模型的有效性进行了评价．研究结果表明，多目标模型率定方法综合考虑了水文过程的各种要素，获得的满意解优于传统的单目标优选结果，非劣解集的聚合程度及模拟的过程线的范围能够反映模型的有效件．

Multi-objective Approach to the Calibration of a Distributed Rainfall Runoff Model

The multi-objective approach was applied to the calibration and validation of distributed rainfall-runoff model developed by the authors.By sensitivity analysis of model's parameters,the Manning's roughness of surface and soil saturated hydraulic conductivity were determined to be calibrated.The Nash-sutcliffe coefficency,relative error of simulated and observed flood peak discharge and absolute error of simulated and observed peak times were selected as objectives.In the process of model calibration,the non inferior solution sets were obtained firstly by using multi-objective approach,then a prefer solution from sets was selected by using discrete compromise programming,finally,the bounds and properties of simulated hydrographs of non inferior solution sets were analyzed and the models performance was valuated.The results show that the parameters determined by the multi-objective approach through trade-off analysis between objectives and hydrographs are more rational than single one,the bounds and properties of simulated discharge processes of non inferior solution sets can reveal the adequacy or inadequacy of the model.