基于CEEMD的希尔伯特-黄变换算法在森林边界层湍流中的应用  

Application of Hilbert Huang Transform Method Based on CEEMD in Forest Boundary Layer Turbulence

作　　者：王彦淇 张宇 苏有琦 张茜 叶敏 WANG Yanqi;ZHANG Yu;SU Youqi;ZHANG Qian;YE min(College of Atmospheric Sciences,Chengdu University of Information Technology/Chengdu Plain Urban Meteorology and Environment Sichuan Provincial Field Scientific Observation and Research Station,Chengdu 610225,Sichuan,China)

机构地区：[1]成都信息工程大学大气科学学院/成都平原城市气象与环境四川省野外科学观测研究站,四川成都610225

出　　处：《高原气象》2025年第2期445-461,共17页Plateau Meteorology

基　　金：国家自然科学基金项目(42275131);第二次青藏高原综合科学考察研究项目(2019QZKK010203)。

摘　　要：为解决传统经验模态分解算法(Empirical Mode Decomposition,EMD)中存在的模态混叠现象,引入了互补集合经验模态分解算法(Complementary Ensemble Empirical Mode Decomposition,CEEMD)和镜像延拓算法对EMD算法分解中存在的缺陷进行改进。本文选取四峨山人工森林区的湍流观测的个例数据,首先对比分析了两种方法的差异,明确CEEMD算法的优势;随后选取了不同高度下稳定层结和不稳定层结的个例数据,应用希尔伯特-黄变换算法对该个例下的风速u和温度T序列的湍流特征进行了分析,探讨了希尔伯特黄变换算法的应用。结果表明:CEEMD的算法分解结果更加精细,模态函数的模态混叠缺陷得到了更好的压制,模态能量分布更集中,希尔伯特边际谱存在更多的能量尖峰,能量分布更加清晰。不同的模态函数存在有各自的特征频率,分解所得的模态函数中包含着不同尺度的运动,其中包含有满足-2/3斜率的惯性副区的湍流运动,以及对应着含能区的低频大尺度模态,且CEEMD分解得到的边际谱能量尖峰很好的反映了各模态函数的含能特征。个例分析表明:CEEMD算法可以作为一个典型的二分滤波器,经CEEMD分解后,该个例湍流信号中u风的各模态函数中存在有3~6 min的阵风波动,不同高度、不同稳定层结下湍流特征表现有所差异,正午不稳定层结下相比夜间稳定层结下希尔伯特边际谱幅值更高,三维风速在各个高度混合更好,且较低高度由于冠层的作用,存在对大尺度湍涡的破碎作用,边际谱相比其他高度表现出低频小而高频大的特征,而温度T在该个例下与三维风速表现有所不同:稳定层层结下不同高度湍涡混合更好,而不稳定层结下由于不同高度热力吸收的差异,较低高度边际谱幅值较高,并随着高度的升高而减小。To address the modal aliasing phenomenon in traditional Empirical Mode Decomposition(EMD)algorithms,the Complementary Ensemble Empirical Mode Decomposition(CEEMD)and Mirror Extension algorithm were introduced to improve the shortcomings in EMD algorithm decomposition.This article selects case data from turbulence observations in the artificial forest area of Mount Si'e.Firstly,the differences between the two methods are compared and analyzed to clarify the advantages of the CEEMD algorithm;Then,case data of stable and unstable layers at different heights were selected,and the Hilbert Huang transformation method was applied to analyze the turbulent characteristics of the wind speed U and temperature T series under this case,exploring the application of the Hilbert Huang transformation method.The results indicate that the algorithm decomposition of CEEMD is more detailed,the mode aliasing defect of the modal function is better suppressed,the modal energy distribution is more focused,the Hilbert marginal spectrum has more energy spikes,and the energy distribution is clearer.Different modal functions have their own characteristic frequencies,and the decomposed modal functions contain motion of different scales,including turbulent motion in the inertial sub region with a slope of-2/3,and low-frequency large-scale modes corresponding to the energy containing region.The marginal spectral energy peaks obtained from CEEMD decomposition well reflect the energy containing characteristics of each modal function.Individual case analysis shows that the CEEMD algorithm can act as a typical binary filter.After CEEMD decomposition,there are gust fluctuations of about 3~6 minutes in the various modal functions of the U-wind in the turbulence signal of this case.The turbulence characteristics vary at different heights and stable layers.The Hilbert marginal spectral amplitude is higher in the unstable layer at noon compared to the stable layer at night,and the three-dimensional wind speed is better mixed at various heights.Moreover,

关 键 词：希尔伯特黄 CEEMD 湍流 森林下垫面 镜像延拓 

分 类 号：P425.2[天文地球—大气科学及气象学]