东篱科研大数据发现系统（DRDS）

位置：成果数据库 > 期刊 > 期刊详情页

基于Copula函数的非一致性洪水峰量联合分析

ISSN号：0559-9350
期刊名称：水利学报
时间：2013.10.15
页码：1137-1147
分类：P333[天文地球—水文科学;水利工程—水文学及水资源;天文地球—地球物理学]
作者机构：[1]天津大学水利工程仿真与安全国家重点实验室,天津300072
相关基金：国家自然科学基金资助项目（51279123、51179117）;国家自然科学基金创新研究群体科学基金资助项目（51021004）
相关项目：供水水源地突发性水污染事件风险评估与应急管理问题研究-以引滦入津水源地于桥水库为例

作者：冯平,李新|

关键词：非一致性, 混合分布, COPULA函数, 联合分布, non-stationary, mixed distribution, copula method, joint distribution

中文摘要：

传统的洪水单变量频率分布形式不能反映在变化环境下洪水序列的真实分布情况，且不适合构建洪水多变量联合分布进行洪水特征的多变量联合分析。以大清河南支沙河上游王快水库入库年洪峰序列和年最大6日洪量序列为基本数据，基于各序列变异点诊断结果，运用混合分布法确定各序列的理论分布，并以此为边缘分布，采用Copula函数法构建其二维联合分布，对两变量重现期及特定条件下的洪峰和洪量条件频率进行了分析，计算了两变量联合分布设计值。结果表明，非一致性洪水单变量重现期介于二维联合重现期与二维同现重现期之间；当峰量具有较高相关性时，发生超过某一频率洪峰设计值的洪峰，会有较大可能发生超过同频率洪量设计值的洪量。基于两变量联合分布得到的洪峰、洪量设计值比单变量同频率法得到的设计值偏大，从工程设计角度偏于安全，对防洪控制是有利的。

英文摘要：

As amounts of flood control projects have been constructed in the Daqing River Basin, the un derlying surface conditions and the characteristics of the runoff yield and concentration are changed, which leads to the inconsistency of the reservoir inflow flood time series, the univariate frequency distribution re sults derived from the traditional methods is unable to reflect the true frequency distribution under changing environment and is not proper to be used as the marginal distribution to construct the joint distribution for multivariate frequency analysis. In this paper, the annual flood peak series and the 6day annual maximum flood volume series of Wangkuai Reservoir which lies in the Shahe, the south branch of the Daqing River, were used as the basic data. Based on the change point diagnosis resuhs of the data, the mixed distribu tion method and the Copula method are used to deduce the univariate marginal distribution and the joint distribution of the flood time series, respectively. Finally, the joint return periods and the conditional fre quency of the flood peak and the flood volume under special conditions were calculated and analyzed, and the design value was also derived on the basis of the joint distribution and the joint return periods. The re sults reveal that the univariate return period lies between the two kinds of joint return periods which repre sent either the peak or the volume, or both the peak and the volume values, are exceeded; given flood peak exceeding the design value of a certain frequency, the probability of flood volume exceeding the de sign value of the same frequency is large on condition that the flood peak and volume are highly correlat ed; the design value of the joint distribution is larger than the design value based on the traditional univar iate samefrequency method, which is more safe and beneficial within the perspective of the flood design projects and flood control.

同期刊论文项目