基于最大熵谱估计的水文周期分析

最大熵谱估计法是以AR模型为基础的一种参数谱估计方法。该方法通过频谱分析，可以计算出全局性的主要周期，从而反映出水文时间序列的整体特性。本文介绍了最大熵谱估计法的基本概念和理论，并论述了用Burg算法作为求解AR模型参数的方法。针对岷江紫坪铺水文站1937-2004年的年径流实测资料，本文将最大熵谱估计理论应用于水文时间序列周期的提取上，并采用Fisher检验方法对提取出的周期进行了检验。通过分析发现，岷江紫坪铺站年径流序列存在4年左右的变化准周期，对揭示岷江中上游水文周期的变化规律具有一定的探讨意义。在研究中同时也发现，最大熵谱估计也存在一定的缺陷，在今后的研究中还需进一步建立熵与水文过程之间的联系，从而更好的揭示出水文时间序列的变化规律。

Application of Maximum Entropy Method in the Analysis of Hydrologic Period

The Maximum Entropy Method is a kind of parameter spectrum estimation method based on AR model. By spectral analyzing, the overall period can be calculated, thus the holistic characteristics of hydrological time series can be reflected. This paper introduces basic concepts and theories of the Maximum Entropy Method and discusses the method of solving the parameters of AR model with Burg algorithm, With the measured annual runoff data from 1937 to 2004 at the Zipingpu Hydrologic Station, the Maximum Entropy Method is used to extract hydrological time series period, and the Fisher method is used to test the extracted period. By analyzing the result we find that there is a hydrologic period of almost 4 years in hydrological annual runoff series at the Zipingpu Hydrologic Station, which is meaningful for revealing the changes in hydrologic period in the upper reaches of the Min River. However, it is found that the Maximum Entropy Method has drawback. So the relationship between entropy and hydrologic process should be built to reveal the changes in hydrologic time series in the future.