CAJ | 학술논문

传统的突变检验方法 Mann-Kendall 法主要基于线性和概率统计理论，对于高度复杂和非线性的水文时间序列检测能力不足。采用基于非线性动力学参数的样本熵方法，结合滑动技术和滑动移除技术，利用湘江流域1961~2009年多个气象站点日降雨资料和控制站湘潭站日径流资料，对降雨和径流时间序列进行突变检验。结果表明：日尺度降雨和径流时间序列样本熵值最小，复杂度最低，可预测性最高；径流分别在1966年、1983年、1992年以及2002年发生突变，降雨在1966年、1987年以及2002年发生突变；滑动样本熵与滑动移除样本熵相结合不仅能有效地识别突变，更能发现突变前后时间序列的动力学变化特征。
전통적돌변검험방법 Mann-Kendall 법주요기우선성화개솔통계이론，대우고도복잡화비선성적수문시간서렬검측능력불족。채용기우비선성동역학삼수적양본적방법，결합활동기술화활동이제기술，이용상강류역1961~2009년다개기상참점일강우자료화공제참상담참일경류자료，대강우화경류시간서렬진행돌변검험。결과표명：일척도강우화경류시간서렬양본적치최소，복잡도최저，가예측성최고；경류분별재1966년、1983년、1992년이급2002년발생돌변，강우재1966년、1987년이급2002년발생돌변；활동양본적여활동이제양본적상결합불부능유효지식별돌변，경능발현돌변전후시간서렬적동역학변화특정。
The traditional mutation test method including Mann-Kendall method,which is mainly based on the theories of linear,probability and statistic,is insufficient for testing the highly complex and nonlinear hydrological time series.Combined with moving and moving cut data technologies,sample entropy based on nonlinear dynamic parameter was applied for the mutation test on rainfall and runoff time series in Xiangjiang river basin.The test data included daily rainfall in several meteorological stations and daily runoff in Xiangtan control station of Xiangjiang river basin from 1 961 to 2009.The results show that the sample entropy of daily rainfall and runoff time series is smallest,the complexity is lowest,and the predictability is highest;the mutations of runoff occur in 1 966,1 983,1 992 and 2002,the mutations of rainfall occur in 1 966,1 987 and 2002;the combination of moving sample entropy and moving cut data sample entropy can not only recognize the mutation effectively, but also discover the characteristics of dynamic variation before and after the mutation of time series.