花园口站年最大洪峰流量时间序列的HHT分析

Analysis on Annual Maximum Peak Flow Series at Huayuankou Station Based on Hilbert-Huang Transform

Hilbert-Huang变换能够定量描述非线性、非平稳复杂时间序列的时频特性,较传统分析方法更具优势。通过对时间序列进行EMD分解,得到变化过程的内在模态函数和趋势项函数,而后对各内在模态函数进行Hilbert-Huang变换,从而揭示出时间序列的多时间尺度特征。以黄河花园口站1952-2009年的年最大洪峰流量时间序列为例,对其进行多时间尺度分析,得到不同波动周期的振荡分量及趋势分量,具体分析了各分量的变化特征。结果表明,花园口年最大洪峰流量变化过程中存在准3.2a、准6.4a、准11.8a和准31.0a周期的波动,其中准3.2a和准6.4a的周期波动是引起原序列波动的主要原因,近60年来花园口年最大洪峰流量变化呈递减趋势,由此揭示了年最大洪峰流量变化过程的多时间尺度特征。在此基础上,探讨了各波动分量变化的影响因素,其变化与大气低频振荡、ENSO、太阳活动及气候变迁等因素有关。

Compared with traditional methods for time series, Hilbert-Huang Transform （HHT） has advantages to quantitative- ly describe time-frequency characteristics for nonlinear, unsteady and complex time series. Empirical Mode Decomposition meth- od is adopted to analyzed time series,and Intrinsic Mode Function （IMF） and trend term are obtained. Then HHT is made for each IMF. Taking annual maximum peak flow series at Huayuankou Station in lower Yellow River from 1952 to 2009 as exam- ple, different oscillation components with different fluctuation periods and residual component of annual maximum peak flow se- ries are obtained,and variation of each component is analyzed in detail. The result shows as follows that the trend of annual maximum peak flow series at Huayuankou Station is stepwise decreasing, there are four periodic oscillations of 3.2 a, 6.4 a, 11. 8 a and 31.0 a for the process of annual maximum peak flow variation, and the variation of annual maximum peak flow is caused mainly by the periodic oscillations of 3.2 a and 6.4 a in fact. Then multiple time-scale structure Of the variation of annual maxi- mum peak flow series can be revealed. Based on the above, the reasons and infection factors for the variation of each fluctuation component are discussed. The variation of eaeh IMF may be caused by QBO, ENSO, solar activity and climate change etc.