近海卫星测高波形分类与判别

Study on the methods of coastal altimetry waveform classification and discrimination

针对近海测高数据可用率低下以及传统重跟踪技术的局限性,该文提出了基于动态聚类分析的测高波形分类方法以及基于移动最小距离的波形判别方法,并给出了确定最佳分类数的有效性指标。实现了海洋波形与非海洋波形正确识别,为处理海量测高波形以及根据波形研究最佳重跟踪算法提供了便利。以穿过台湾海峡和台湾近岸的TOPEX/Poseidon波形数据为例进行试验,结果表明,测高波形分为海洋波形、尖锥波形和其他散射波形3类时聚类效果最佳。与OCOG、移动相关系数和β-5等方法的分类结果对比分析表明,本文提出的方法具有一定优势。

In view of low availability of altimetry data in coast and limitation of traditional waveform retrackers, a waveform classification scheme based on dynamic cluster analysis and the discrimination method based on moving minimum distance were presented, meanwhile the validity index for determining the optimal cluster number was introduced. The method realized the recognition of ocean waveform and non-ocean waveform correctly, and provided convenience to process mass altimetry data and study the best algorithm in accordance with waveforms. The experiment was implemented with the samples of coastal Topex/Poseidon waveforms over Taiwan Strait and coastal zones. The results showed that best clustering performance could be achieved when the wave forms were divided into three categories of ocean waveform, peaky waveform and the other scattering waveform. Compared with the traditional classification methods such as OCOG, moving correlation coefficient and beta-5, the new method was significantly improved.