基于SVM的ASAR波模式海浪波高与周期提取研究

发布时间：2018-01-05 22:15

本文关键词：基于SVM的ASAR波模式海浪波高与周期提取研究　出处：《内蒙古大学》2017年硕士论文　论文类型：学位论文

【摘要】：星载合成孔径雷达(Synthetic Aperture Radar,SAR)作为一种微波遥感探测手段,可以不受日照、云雾等外界环境因素影响实现对地观测。ENVISAT卫星搭载的先进合成孔径雷达(Advanced Synthetic Aperture Radar,ASAR)工作于C波段,具有5种工作模式,其中波模式全天时开通,数据量丰富。本文提出了基于支持向量机(Support Vector Machine,SVM)回归模型的ASAR波模式数据有效波高(Significant Wave Height,SWH)和平均波周期(Mean Wave Period,MWP)提取新方法。论文详细阐述了 SAR数据反演海浪参数的背景及意义,并对现有的海浪参数反演技术的进展进行了扼要介绍。分析了 SAR海表面成像机制并指出了现有MPI方法、参数化方法(PARSA)、半参数化方法(SPRA)等海浪谱反演存在的不足,进而提出使用SVM提取海浪参数的新方法。本文完成工作如下:1.SVM模型的建立。详细介绍了 ASAR波模式图像预处理步骤、ASAR图像谱分解过程及特征参数提取过程,并且从空间域和频域分析了特征参数与SWH和MWP之间的关系。然后,给出了 SVM回归机模型的理论推导过程,使用30771景与全球气候再分析数据(ERA-Interim)匹配的ASAR波模式样本进行SVM模型训练。2.SVM模型验证与分析。本文分别采用浮标和ERA-Interim数据对SVM方法提取的SWH和MWP进行了印证,SWH均方误差分别为0.49米和0.4米,相关度分别为0.8和0.93。MWP均方误差分别为1.08秒和0.62秒,相关度为0.76和0.87。由此说明,基于SVM回归模型的ASAR SWH和MWP提取是一种有效的方法。3.SVM模型进一步评估,对台风案例和2011年全球海况季节性变化特征进行了分析。论文结果表明,本文提出的SVM模型对高海况区域的海洋风暴预警、监测及海浪参数的业务化运行具有重要的意义。
[Abstract]:Spaceborne synthetic Aperture radar (SAR) as a microwave remote sensing method, it can not be exposed to sunlight. The advanced synthetic Aperture Radar (ASAR) carried by the Earth observation. ENVISAT satellite is affected by external environmental factors such as clouds and fog. Advanced Synthetic Aperture Radar. ASAR) operates in C-band and has five modes of operation, in which the wave mode is switched on all day. This paper presents support Vector Machine based on support vector machine. The effective wave height of ASAR wave model data is significant Wave Height. In this paper, the background and significance of wave parameters inversion from SAR data are described in detail. The development of ocean wave parameter inversion technology is briefly introduced, the imaging mechanism of SAR sea surface is analyzed and the existing MPI method, parameterized method, is pointed out. The demerit of the semi-parameterization method, such as SPRA, in the inversion of ocean wave spectrum. Then a new method of extracting ocean wave parameters using SVM is proposed. The main work of this paper is as follows: 1. The establishment of ASAR model is completed. The preprocessing steps of ASAR wave mode image are introduced in detail. ASAR image spectral decomposition process and feature parameter extraction process, and from the spatial and frequency domain analysis of the relationship between feature parameters and SWH and MWP. Then. The theoretical derivation process of SVM regression model is given. ERA-Interim using 30771 View and Global Climate Reanalysis data. The matching ASAR wave pattern samples are trained by SVM model. 2.SVM model verification and analysis. This paper uses buoy and ERA-Interim data to extract SWH and SVM from SVM method, respectively. MWP confirmed it. The mean square error (MSE) of SWH was 0.49m and 0.4m, and the correlation degree was 0.8,0.93.MWP mean square error was 1.08s and 0.62s, respectively. The correlation is 0.76 and 0.87. It shows that ASAR SWH and MWP extraction based on SVM regression model is an effective method for further evaluation. 3. The characteristics of seasonal variation of global sea conditions in 2011 are analyzed. The results show that the SVM model proposed in this paper is an early warning method for ocean storms in high sea conditions. Monitoring and operational operation of wave parameters are of great significance.
【学位授予单位】：内蒙古大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN957.52

【相似文献】