基于HHT的滨海降水多尺度时频特征研究

Multi-scale Time-frequency Characteristics Analysis of Coastal Precipitation Based on HHT

基于EEMD的改进型HHT方法,深入剖析了滨海围垦滩涂区域降水的多尺度周期特征进行。结果表明,滨海地区赣榆、东台、上海测点年降水和季节降水变化的周期振荡主要体现在2-3a、5-7a左右的年际尺度上和10-15a、25-30a左右的年代际尺度上,与江淮地区的整体特征和气候背景基本一致,但滨海地区的地理位置与环境特殊性使其降水周期振荡的频域相对较宽且离散。与离散小波变换固定频带下的信号分解相比,HHT方法的EEMD分解具有更优的自适应性;与连续小波变换相比,HHT方法在滨海地区降水信号高频周期振荡特征的解译和信号分解中频域混叠现象的避免方面更具优势。

The precipitation multi-scale time-frequency characteristics in the coastal reclamation land were analyzed in detail by means of the improved Hilbert-Huang Transform （HHT） method, which was based on the Ensemble Empirical Mode Decomposition （EEMD） algorithm. The results showed that the domi- nant inter-year periodicities were at about 2--3 years and 5-7 years, and the dominant inter-decadal perio- dicities were at about 10-15 years and 25-30 years for the annual and seasonal precipitation dynamics at Ganyu, Dongtai and Shanghai sites. The revealed coastal precipitation periodicities matched the overall precipitation periodicities in the whole Jianghuai area, However, because of the special geography location and environmental situation in the coastal region, the frequency domain of the coastal precipitation periodic oscillations was broader and discrete. Comparing with the Discrete Wavelet Transform method, which de- composed signals at the fixed frequency bands, the improved HHT method showed better self adaptability at the signal decomposition because of the EEMD algorithm. Comparing with the Continuous Wavelet Transform method, the improved HHT method showed more advantages in revealing the signal periodic oscillations at the high-frequency scales because of its capacity in avoiding the frequency aliasing phenome- na involved in the signal decomposition.