摘要: 水文事件一般具有多个方面的特征属性，而各个特征属性之间普遍具有相关性，因此需要采用特定的方法对水文变量的相关性进行分析。本文综述了现有的相关性计算方法，指出了现有方法的不足和缺陷；引入Copula熵的概念，用以衡量复杂水文现象的相关结构，并给出Copula熵与互信息的关系和计算方法。最后，以神经网络预报因子的选择为例，验证Copula熵的适用性。比较研究表明：基于Copula熵因子选择的BP神经网络预报结果最好，好于常用的线性相关系数法，为探讨水文相关性分析提供了一条新的途径。
Abstract:
Hydrological events are usually characterized by several correlated variables. There is a great need to estimate the correlation of hydrological variables. In this study, the current hydrologic correlation analysis methods were reviewed, the disadvantages of which were also discussed. The concept of copula entropy was introduced to estimate the dependences. The relationship between copula entropy and mutual information was discussed and the calculation procedures of copula entropy were given. Finally, the proposed method was used for selecting the inputs of artificial neural network for flood forecasting. The comparative study results show that the proposed method performs better than conventional linear regression method and provides a new way for hydrological correlation analysis.