中文摘要：

海洋的声速结构对水下声传播有重要影响，在印度洋中北部复杂多变的海洋物理和水文环境中，获取声速剖面的时空统计分布规律对水下目标探测和水下声通信有重要意义．由于垂直梯度法在声速结构分析中的局限性及其在印度洋中北部海域的适用性问题，采用多元统计分析中的最优分割法对声速跃层进行分析，并应用最近10年的地转海洋学实时观测阵数据对印度洋中北部海域声速剖面的特征量进行了计算，获得了声速跃层的垂直结构特征和时空变化规律；还利用经验正交函数（EOF）表示方法，分析了印度洋中北部声速剖面拟合精度随EOF阶次的分布特点．根据印度洋的海洋物理特征，揭示了声速剖面特征量时空演变的内在物理机理．研究结果表明：最优分割法是适合印度洋声速结构的跃层判断方法，并提出了相应的判断准则参数：声速剖面拟合精度随EOF阶次变化的区域性分布特征较明显，其季节性变化较小；印度洋中北部的深海声道轴只在5°s以南明显存在，在15°S-25°S附近海域存在三个跃层；印度洋中北部声速剖面结构可分为单跃层、双跃层I型、双跃层II型和三跃层四种类型以及春夏秋冬四个季节模态．声速剖面的分析结果对于水声传播和声纳系统的使用具有一定参考意义．

英文摘要：

The structure of sound speed in ocean has a strong influence on underwater sound propagation. For underwater target detection and underwater acoustic communication, it is of great significance to obtain the temporal and spatial distribution of sound speed profile. Since the limitations and applicability of vertical grads method in structural analysis of sound speed in the Indian Ocean, optimal partition method is developed to calculate the sound spring layer. The study focuses on analysing the eigenvalues of sound speed profiles （SSPs） in the noah-central Indian Ocean based on the last 10 year data of array for real-time geostrophic oceanography. The vertical structure characteristics of sound speed are investigated in the Indian Ocean, and the laws of temporal and spatial variation are obtained. The distribution of the fitting accuracy with the order of empirical orthogonal functions （EOF） are found. The physical mechanisms of the temporal and spatial variation of SSPs are revealed with the marine physical environment in Indian Ocean. The simulation results show that optimal partition method is suitable for the area to judge the structure of spring layer, and the parameters of the corresponding criterion is also proposed. The regional distributions of the fitting accuracy with the order of EOF are more obvious than with the seasonal variations. The deep channel axis exists at south 5~S and there are three spring layers between 15~S and 25~S. The structures of SSPs in the Indian Ocean can be classified into four types： single spring layer, double spring layer type I, double spring layer type II and three spring layer, and for seasonal models： spring model, summer model, autumn model and winter model. The analysis results of the SSPs can provide some reference value for acoustic propagation and the sonar systems.