【摘要】 本文利用涡旋内温度异常剖面对黑潮延伸体区域的涡旋进行分类与分析。基于卫星高度计识别的涡旋数据和与涡旋时空匹配的Argo浮标温度异常剖面数据,本文提出了一种优化的高斯混合模型无监督聚类方法实现涡旋内Argo浮标温度异常剖面的自动聚类,进而实现与剖面时空匹配涡旋的自动分类。实验结果表明,相同类型的Argo温度异常剖面呈现出明显的空间聚集效应,而不同类型的Argo温度异常剖面则在空间上相互分离,与表层(次表层)温度异常剖面相匹配的涡旋主要分布在黑潮延伸体的北部(南部)。相较于直接利用区域范围对涡旋进行分类的方法,本文所提出的涡旋分类方法更为客观,合成分析特定类型涡旋的水下结构也更为合理。更多还原

【Abstract】 Mesoscale eddies are widespread in all oceans of the world. The Argo temperature anomaly profiles(TAPs) inside eddies can describe the main underwater features. This paper classifies and analyzes the mesoscale eddies by using the TAPs inside eddies in the Kuroshio Extension(KE) region. Based on the eddy data identified by the satellite altimeter and the Argo TAPs data matched with eddies space-time, an improved unsupervised clustering method for Gaussian Mixture Modeling(GMM) is proposed to realize the automatic clustering of the Argo TAPs inside the eddies, so as to achieve the automatic classification of eddies matching the space-time profiles. The experimental results show that the same type of Argo TAPs exhibit obvious spatial aggregation results, while the different types of Argo TAPs are spatially separated from each other. Cyclonic and anticyclonic eddies with large anomalies are also found along the southern and northern flank of the Kuroshio path. Compared with the method of directly using a specific area to classify eddies, the eddy classification method proposed in this paper is more objective, and the synthetic analysis of the underwater structure of a specific type of eddy is also more reasonable.更多还原