吕宋海峡海面涡度场与地转流场的特征分类

摘要

The method of SOM is applied to the characteristic analysis of sea surface vortex field and geostrophic current field in the Luzon Strait, based on the satellite altimetry data.The result shows that: in the experiment of SOM 2×2, the spatial distributions of the 4 patterns are in good agreement with the seasonal mean climatology, among which the winter pattern lasts from November to the next March and the summer pattern lasts from May to September. The BMU time series shows that interannual variations exist. When the ENSO signal is intensive, the seasonal variation is more apparent.The SOM 3×4 result implies that the 12 patterns can also be divided into 4 categories, that is, the winter type, the spring type, the summer type and the fall type. Patterns of the same category vary both spatially and temporally. Each pattern exists for several months, and appears most in one or two specific months.The time series implies that the interannual variation and the seasonal variation both exist.%基于 AVISO 卫星高度计数据计算得到的地转流场，利用 SOM 神经网络对吕宋海峡海面涡度场与地转流场进行特征分析，研究吕宋海峡上层流场的变化特征。结果表明：在 SOM 2×2的网络结构实验中，4个分类结果的空间形态类似于气候态的季节分布形式，其中冬季型可以由11月延续至次年3月，夏季型在5月至9月都会存在。时间序列显示分类结果存在年际变化信号，其中在 ENSO 较强的年份，季节变化信号明显。SOM 3×4网络结构的结果显示，12个分类结果可以分为4类，即冬季型、春季型、夏季型、秋季型。每种类型内的分类图形形态存在差别；在时间上，分类图形存在连续性，并集中出现在某个月份。时间序列显示季节变化信号与年际变化信号同时存在。