第41卷第6期2022年12月Vol.41,No.6Dec.,2022声学技术TechnicalAcoustics基于K-means聚类分析与单经验正交函数回归法的南海声速剖面估计方法研究欧圳翼,屈科(广东海洋大学电子与信息工程学院,广东湛江524000)摘要:基于遥感参数和Argo历史数据对水体声速剖面(SoundSpeedProfile,SSP)进行重构,对单经验正交函数回归(singleEmpiricalOrthogonalFunction-regression,sEOF-r)法在南海的适用性进行了研究。由于南海动力活动的复杂性,SSP扰动相对复杂,同时海域内SSP样本稀疏,相关的SSP统计学估计方法在南海区域还难以有效应用。文章基于K-means对样本进行聚类分析,讨论南海海域正交经验函数模态的一致性。通过扩大重构实验网格解决样本稀疏的问题。利用经典的sEOF-r对南海SSP进行反演,对重构SSP的误差分析说明了该方法在南海海域应用的有效性。SSP重构的均方根误差为2.3411m·s-1,较大误差主要出现在深度40~200m,其原因是海域内混合层深度发生变化。实验证明在南海区域内利用遥感参数可以有效地估计SSP。关键词:声速剖面;聚类分析;海面遥感参数;南海;单经验正交经验函数(sEOF-r)中图分类号:TB556文献标志码:A文章编号:1000-3630(2022)-06-0821-06SoundspeedprofileinversionintheSouthChinaSeabasedonK-meansclusteranalysisandsingleempiricalorthogonalfunctionregressionOUZhenyi,QUKe(CollegeofElectronicandInformationEngineering,GuangdongOceanUniversity,Zhanjiang524000,Guangdong,China)Abstract:Thesoundspeedprofile(SSP)isreconstructedbyremotesensingparametersandArgopreviousdataintheSouthChinaSea,andtheapplicabilityofsingleempiricalorthogonalfunctionregression(sEOF-r)intheSouthChinaSeaisstudied.DuetothecomplexityofhydrodynamicactivitiesintheSouthChinaSea,thecorrespondingSSPdisturbanceisrelativelycomplex,andmeantimetheSSPsamplesintheseaareaaresosparsethattherelatedSSPestimationmethodsarestilldifficulttobeeffectivelyapplied.BasedontheK-meansclusteranalysisofsamples,theconsistencyoftheorthogonalempiricalfunctionmodesisdiscussedinthispaper.Expandingtheinversiongridcansolvetheproblemofsparsesamples.TheclassicsEOF-risusedtoinverttheSSPintheSouthChinaSea,andtheerroranalysisofthereconstructedSSPisusedtoprovetheeffectivenessofthemethod.Therootmeansquareerrorof...