基于改进模拟优化方法的水文模型的参数异参同效性及径流模拟研究  

The Study on Equifinality of Hydrological Model Parameters and Runoff Simulation Based on the Improved Simulation-optimization Algorithm

作　　者：邢贞相 王丽娟 王欣 付强 纪毅 李衡 刘亚娟 XING Zhenxiang;WANG Lijuan;WANG Xin;FU Qiang;JI Yi;LI Heng;LIU Yajuan(School of Water Conservancy and Civil Engineering,Northeast Agricultural University,Harbin 150030,China;Collaborative Innovation Centre of Promote Grain Production in Heilongjiang Province,Harbin 150030,China;Key Laboratory of Yellow River Sediment of the Ministry of Water Resources,Yellow River Institute of Hydraulic Research,Zhengzhou 450003,China)

机构地区：[1]东北农业大学水利与土木工程学院,黑龙江哈尔滨150030 [2]黑龙江省粮食产能提升协同创新中心,黑龙江哈尔滨150030 [3]黄河水利科学研究院水利部黄河泥沙重点实验室,河南郑州450003

出　　处：《应用基础与工程科学学报》2020年第5期1091-1107,共17页Journal of Basic Science and Engineering

基　　金：国家重点研发计划(2017YFC0406004);国家自然科学基金项目(51979038,51179032);黑龙江省自然科学基金项目(E2015024);教育部高等学校博士点新教师资助基金项目(20112325120009);黑龙江省领军人才梯队后备带头人资助项目(500001);黑龙江省博士后资助基金(LBH-Q12147);黑龙江省水利厅科技开发项目(201402,201404,201501);东北农业大学SIPT创新训练项目(202010224105)。

摘　　要：为研究异参同效问题对水文模型预报精度影响,提高模型预报精度,本文利用SCE-UA的模拟优化模式,计算Nash模型参数,并对其异参同效性进行研究,并利用多准则似然权重对传统的SCE-UA算法改进来减弱参数的异参同效性,以期进一步提高水文模型的预报精度.首先,将SCE-UA算法与Nash模型进行耦合,通过SCE-UA多次迭代来求解多个Nash模型最优参数值,组成最优参数集合,并分析其参数异参同效性的特征.其次,利用多重准则筛选方法对模拟优化算法得出的最优参数组进行二次优选,然后再根据其对应的似然权重加权求得最终的唯一最优参数.本文将这种参数优选方法命名为基于似然权重的多准则参数优选法(SMMLW).最后,将最优参数代入Nash模型,用于实际洪水预报,并依据最优参数集合给出90%的置信区间,而后,通过选定的4个评价指标对预报结果的不确定性进行了分析和评价.此外,为对比本文方法的有效性,将基于SMMLW方法的Nash模型预报结果与基于传统的SCE-UA算法和AMMCMC算法的Nash模型计算结果的精度进行对比分析.结果表明:(1)异参同效现象对Nash模型参数的影响特征表现为两方面,即对较优似然值的参数组取值范围的影响和最优似然值的参数组数量的影响;(2)异参同效的两方面特征的影响程度均与输入洪水的洪量以及洪峰的大小有关,即随着输入洪水的洪量和洪峰的增加,异参同效现象越明显;(3)对同一场洪水,与基于传统的SCE-UA算法和AM-MCMC算法的Nash模型预报结果相比,本文构建的基于SMMLW的Nash模型预报具有更高的精度;(4)从径流预报结果的不确定性度来看,基于SMMLW的Nash模型的预报结果能够较好地考虑由于Nash模型参数不确定性而导致的预报结果的不确定性,从而能取得较高精度的预报结果.In order to study the influence of equifinality phenomenon on the accuracy of hydrological forecasting and improve the accuracy,the Nash model parameters were calculated by using the simulation-optimization model,and the equifinality phenomenon was studied in this paper.The traditional SCE-UA algorithm was improved by using the multi-criteria likelihood weight to reduce the equifinality of the parameters and improve the prediction accuracy of the Hydrological Model.Firstly,the SCE-UA algorithm was coupled with the Nash model to build a Simulation-optimization algorithm.The optimal parameter sets of the Nash model were obtained by a large number of cycles of the Simulation-optimization algorithm,and then the characteristics of the equifinality were analyzed.Secondly,the optimal parameter sets were screened again by a Screening Method based on Multi-criteria Likelihood Weight(SMMLW),and then a final optimal parameter set was obtained.Finally,the final optimal parameter set was substituted into the Nash model flood forecasting,and then the floods in verification period was simulated.The 90%confidence interval of flood forecasting was calculated from the Nash model with the optimal parameters sets.The uncertainty of the forecast results was analyzed by 4 evaluation indexes.To compare the effectiveness of the method proposed in this paper,the simulation results based on the SMMLW was compared with that of the SCE-UA algorithm and the AM-MCMC algorithm.The results were shown as below:(1)The influence of equifinality on the Nash model parameters was shown in two aspects,that is,the influence of the value range of the optimal parameters and the influence on the number of optimal parameter sets;(2)The influence degree of the two characteristics of equifinality was both associated with the values of the flood volume and the flood peak.Specifically,the equifinality phenomenon was more obvious with the increasing of the values of the flood volume and the flood peak.(3)For a flood,the simulation accuracy based on the SMMLW wa

关 键 词：SCE-UA 异参同效 参数率定 似然权重 不确定性 Nash模型 径流模拟 

分 类 号：P333[天文地球—水文科学]