基于EEMD的光学湍流廓线确定项与随机项分析  

Background and Stochastic Terms of Optical Turbulence Profile Based on Ensemble Empirical Mode Decomposition

作　　者：陈小威[1,2] 李学彬[1] 孙刚[1] 刘庆[1] 朱文越[1] 翁宁泉[1,3] 

机构地区：[1]中国科学院安徽光学精密机械研究所大气成分与光学重点实验室,安徽合肥230031 [2]中国科学技术大学研究生院科学岛分院,安徽合肥230031 [3]中国科学技术大学环境科学与光电技术学院,安徽合肥230026

出　　处：《光学学报》2017年第1期1-9,共9页Acta Optica Sinica

基　　金：国家自然科学基金(41205023;41375017)

摘　　要：为了可靠地评估大气的光束传播效应,必须确定大气光学湍流的路径分布。采用高分辨率无线电探空仪探测光学湍流强度的垂直分布,利用集合经验模态分解将其分解为不同尺度的本征模态分量,并分析了不同分量变化的周期性及对整体变化的贡献。结果表明,部分本征模态函数分量具有周期性,并通过了周期显著性检验;方差贡献率表明整体趋势变化和随机强噪声是大气光学湍流廓线随高度变化的主要原因。利用基于连续均方误差准则的滤波方法实现了大气光学湍流确定项和随机项的分离,相关分析得到背景水平和统计平均值相关系数大于0.99。并分析随机项得出光学湍流随机项是非平稳序列且具有多重分形结构。Knowledge of distribution of atmospheric optical turbulence in propagation path is indispensable to evaluate the optical propagation effects accurately.Vertical profiles of optical turbulence strength are obtained from balloon flights.The data are decomposed into intrinsic mode functions on different scales by ensemble empirical mode decomposition.The quasi-cyclical variations on different components as well as their contribution to entire profile are analyzed.The results show that the cyclical scale within some intrinsic mode components is confirmed passing the statistical significance.Variance contribution rates show that the overall trend and stochastic intense noise are the main cause for the optical turbulence variation.A signal-filtering method based on consecutive mean square error is used to separate the profile into background and stochastic term.Correlation analysis of background and statistical mean show that the correlation coefficient is larger than 0.99.Analysis of stochastic term indicates that the optical turbulence profile is nonstationary series with multifractal structure.

关 键 词：大气光学 湍流 集合经验模态分解 周期尺度 背景水平 

分 类 号：P427.1[天文地球—大气科学及气象学]