基于T检验的水文时间序列HHT分析方法及应用

Hilbert-Huang变换方法(HHT)是针对非线性、非平稳序列的一种新的时间序列分析方法.针对前人在应用中存在的忽略统计检验的问题,提出了对本质模态函数(IMF)的周期进行T检验的改进方法,设计了基于T检验的HHT模型计算步骤和算法流程,可检验IMF周期的有效性.将改进后的HHT方法用于长江宜昌水文站50年的逐日径流量的分析中,得到了14个反映径流序列波动规律的IMF、一个反映径流变化趋势的趋势项r及Hilbert谱.研究表明,宜昌站径流量存在1~3a、4~5a、9a及准14a的年际变化周期,年内波动对整个径流序列的影响亦较大.对比分析表明了所设计的基于T检验的HHT方法的有效性.

Hilbert-Huang transform based on T-test for hydrological time series and its application

Hilbert-Huang transform (HHT) is a new way for analyzing nonlinear, non-stationary time series. In previous applications of this method, statistical test was ignored. So the model of HHT based on T-test was designed, and T-test was used for testing the rationality of the average period for intrinsic mode functions (IMF). The improved HHT method was performed on the daily water discharge data sets at Yichang hydrological station in Yangtze river for 50 years. The data was decomposed into fourteen IMFs and one residual. The result shows that the discharge has periods of 1～3 years, 4～5 years, about 9 years and about 14 years, while fluctuate within a year shows great influence on the whole discharge time series. Comparative analysis shows the effectiveness of HHT method based on T-test.