基于排列熵的径流突变识别

Application of permutation entropy to runoff mutation analysis

针对变化环境下的径流序列表现出来的复杂性,采用排列熵对径流序列的突变性进行分析。采用相空间重构建立新的向量,计算序列的动力学熵值变化特征,通过熵值的变化规律进行序列突变点位置和数目的识别。利用该方法对黄河流域花园口站1919-1998年共79年的天然月径流序列的突变进行了实际验证。结果表明,排列熵能够有效地对序列的突变进行检测,检测到的突变年和气候的突变有着很好的响应关系。排列熵值的变化可以很好地刻画序列的突变特征,从序列的复杂性入手可以有效地对序列的突变进行分析。

Permutation entropy is adopted to analyze the mutation of runoff series that could be complicated under changeable environments. New vectors are constructed by phase space reconstruction, and their dynamic characteristics of entropy variation are calculated. Then by the variation trends of entropy, the location and number of the mutation points can be identified. This method was verified using a natural monthly runoff series from 1919 to 1998 of the Huayuankou hydrometric station on the Yellow River. The results indicate that the mutation can be efficiently tested and identified with permutation entropy, and the detected mutation shows a very good response to climate mutation. Mutation characteristics of a runoff series can be well depicted by the changes of permutation entropy, so the mutation can be effectively analyzed according to the complexity of the series.