面向中尺度涡提取的SLA去噪方法研究

发布时间：2018-03-20 19:40

本文选题：包络面去噪　切入点：中尺度涡提取　出处：《地球信息科学学报》2016年09期 　论文类型：期刊论文

【摘要】：噪声去除一直是基于卫星高度计资料的海洋中尺度涡提取研究的难点和热点,然而无论是卷积滤波器还是信息滤波器都存在对海面高度异常(SLA)数据的局部过处理现象。鉴于此,本文提出一种基于包络面去噪的海洋中尺度涡提取方法。该算法可利用分离层内的信息稳定性和层间的信息完备性,很好地改进了卷积运算没有考虑局部噪声的不足,进而有效地提高去噪能力。其具体流程为:首先,对初始化的原始数据场进行上下包络面构造,形成原始数据子场;然后,根据子场内部和子场间的稳健性,把原始数据场转换为子场集合;其次,利用子场极差和标准差,对子场集合进行信息重组,形成噪声去除后的信息场;最后,利用去噪后的信息场数据,采用Winding-Angle(WA)和泛克立金中尺度涡提取算法在西北太平洋进行对比验证实验。实验结果表明,本文提出的方法较前人的方法有较大的提升,准确率为91.23%,取得了较好的应用效果。
[Abstract]:Noise removal has always been a difficulty and hotspot in the research of ocean mesoscale vortex extraction based on satellite altimeter data. However, both convolution filter and information filter have the phenomenon of local overprocessing of sea surface height anomaly SLAs. In this paper, a mesoscale vortex extraction method based on envelope surface denoising is proposed. The algorithm can make use of the stability of information in the separation layer and the completeness of the information between the layers, and can improve the convolution operation without taking into account the deficiency of local noise. The process is as follows: firstly, the initial data field is constructed by the upper and lower envelope surface to form the original data subfield, and then, according to the robustness of the subfield and the subfield, The original data field is transformed into a subfield set. Secondly, the information field after noise removal is formed by reorganizing the information of the subfield set by using the subfield extreme deviation and the standard deviation. Finally, the de-noised information field data is used. Winding-Anglege WAA) and Pan-krikin mesoscale vortex extraction algorithm are used to compare and verify the experimental results in the Northwest Pacific Ocean. The experimental results show that the proposed method has a greater improvement than the previous methods, and the accuracy is 91.23.The results show that the proposed method has a good application effect.
【作者单位】：中国科学院遥感与数字地球研究所数字地球重点实验室;中国科学院大学;
【基金】：国家自然科学基金项目(41371385、41476154)
【分类号】：P715.6

【相似文献】