【作者】 任永政；

【导师】 贺明霞； Susanne Lehner；

【作者基本信息】 中国海洋大学 ， 海洋信息探测与处理， 2009， 博士

【摘要】 德国于2007年6月15日发射的TerraSAR-X卫星是国际上第一颗具有全极化、高空间分辨率、沿轨干涉工作模式的X波段合成孔径雷达(SAR)业务化卫星。目前，TerraSAR-X卫星仅提供多模式高质量地球观测图像产品，尚无反演的海洋环境参数例如海面风场、海表流场等被开发。本文旨在利用TerraSAR-X卫星的独有特性，发展高分辨卫星海面风场和卫星海表流场的反演方法。后者是目前卫星海洋学的空缺，并且其反演方法与海面风场有关。目前成熟的从微波散射计和微波合成孔径雷达后向散射系数获取海面风场的反演模式多基于C、Ku和L波段，普遍采用海面归一化后向散射系数(NRCS)与风矢量的经验关系作为卫星SAR海面风场反演的地球物理模式函数。Masuko基于机载实验曾提出X波段SAR反演海面风场的地球物理模式函数。本文利用SIR-C／SAR(X)NRCS数据和ECMWF ERA-40模式风场数据检验该模式，发现存在明显的系统误差。在此基础上，针对X波段VV极化SAR，提出一个新的基于线性方法的海面风场反演经验模式(XMOD)，以描述NRCS与海面10m风速、风向及雷达入射角之间的关系，并利用SIR-C／SAR(X)NRCS数据以及ECMWF ERA-40风场数据确定XMOD的线性系数。XMOD模拟出的NRCS与SIR-C／SAR(X)的NRCS相关性为0．91，均方根误差为2．68dB。将XMOD应用于不同工作模式的VV极化TerraSAR-X海表图像以反演海面风场，反演的风速与QuickScat风速相比，二者相关性为0．97，均方根误差为0．92 m／s；与HIRLAM细网格模式风场结果相比，TerraSAR-X风速平均偏高1．10m／s。其中风向采用风条纹或模式结果。上述XMOD仅适用于VV极化，为将其扩展到TerraSAR-X HH极化，需要建立极化率模型。基于Thompson和Mouche的极化率模型，本文建立了二个极化率模型PR1和PR2，并分别应用于海面风场反演，PR1和PR2误差分别为0．56m／s和0．09m／s，PR2优于PR1。将极化率模型PR2应用于HH极化TerraSAR-X图像反演海面风场，其结果与QuickScat结果比较，二者相关性为0．96，均方根误差为0．86 m／s。TerraSAR-X卫星具有沿轨干涉工作模式，可以获得卫星海表流场数据。由于TerraSAR-X卫星尚未公开ATI数据，本文使用Romeiser 2002年提出的反演方法，采用X-波段机载沿轨干涉SAR数据(ATI)以及X波段航天飞船SRTM数据进行海表流场反演方法研究。利用机载ATI数据进行相位方位向位移校正，相位转换流场，二维流场合成，海浪运动校正后获得反演的海表流场。其中使用了Romeiser开发的SAR正向成像模型M4S软件。反演的海表流场与现场ADCP测量结果和德国航道工程研究院提供的模式结果进行比较，误差分别为0．16m／s和0．04m／s，与Romeiser的结果吻合。用于三维地形测量的SRTM的两个天线在沿轨方向上有7m位移，Romeiser提出可用于ATI海表流场反演。本文采用2000年丹麦小贝尔特海峡的SRTM数据进行相位校正，平地效应校正，非零干涉相位校正，相位转换流场后获得反演的海表流场。反演结果有待印证。更多还原

【Abstract】 Satellite TerraSAR-X launched on June 15, 2007 by Germany is the first operational Satellite, which is quipped with X-Band synthetic aperture radar (SAR) with the following characteristics: multi polarizations, high spatial resolution and along-track interferometric operation mode. So far only the high quality multi operation mode of TerraSAR-X images over earth can be offered but no the retrieved marine environmental parameters such as sea surface wind fields and sea surface current fields. The aim of this thesis is to develop the retrieval algorithms of high resolution satellite sea surface wind fields (SSW) and sea surface current fields (SSC) using the unique properties of TerraSAR-X.So far, the well-rounded retrieval algorithms of SSW from microwave scatterometer and SAR are mainly based on C-, Ku-, and L-band. Generally, the empirical relationship between the normalized radar cross sections (NRCS) and wind vectors is established as the geophysical model function (GMF) to retrieve SSW from SAR images. Masuko proposed an X-band SAR GMF based on airborne experiment. In this thesis, the Masuko model is validated with SIR-C/SAR(X) NRCS and ECMWF ERA-40 data and find that there is an obvious system error. Therefore, a linear X-band VV polarization GMF (XMOD) is developed to describe the relationship between NRCS, 10m height wind speed, wind direction and incidence angle. The global ocean SAR data from SIR-C/SAR(X) missions on April and October 1994 and wind fields From ECMWF ERA-40 are used to tune the linear parameters of XMOD. The correlation and RMS between the simulated NRCS by XMOD and SIR-C/SAR(X) NRCS are 0.91 and 2.68dB respectively. Then the SSW is retrieved from different operation mode of VV polarization TerraSAR-X images. The direct comparison of XMOD results to QuickScat shows that the correlation and the RMS between them are 0.97 and 0.92 m/s respectively. The mean wind speed from TerraSAR-X is 1.10m/s bigger than the results of High Resolution Limited Area Model (HIRLAM). The wind direction used here is extracted from wind streak or model results. XMOD mentioned above can only be applied with VV polarization SAR images, to be use the HH polarization SAR data, a polarization ratio model is needed. On the base study of Thompson and Mouche, two polarization ratio models PR1 and PR2 are developed and used to retrieve SSW. The errors of the retrieved results are 0.56m/s and 0.09m/s respectively, i.e. PR2 model is better than PR1 model. SSW is also retrieved from TerraSAR-X images using PR2 model and the correlation and the RMS against the QuickScat results are 0.96 and 0.86 m/s respectively.The ATI mode of TerraSAR-X enables SSC measurements. Because the ATI data of TerraSAR-X are not disclosed yet, so the X-band airborne ATI data and SRTM data are used to study the retrieval algorithms of SSC with the method proposed by Romeiser in 2002. The airborne ATI data are processed with the following steps: the correction of ATI phase, conversion of ATI phase into first guess current fields, composition of two dimensional SSC, and then correction for contributions of wave motion with the software of SAR imaging model M4S developed by Romeiser in 2000. Then the retrieved SSC is compared to the results of Acoustic Doppler Current Profiler (ADCP) in situ and German Federal Waterways Engineering and Research Institute (BAW), the errors are 0.16 m/s and 0.04 m/s respectively, which are consistent with Romeiser’s results. Because there is 7m separation along-track of the cross-track antennas of SRTM, so Romeiser proposed that the SRTM data can be used to study on SSC retrieval. In this thesis, the SRTM data acquired over Lille Baelt in Denmark in 2002 are processed with the following steps: phase correction, phase correction of the effect of flat earth, the non-zero interferometric phase correction and then the corrected phase is converted to SSC. However, the retrieved SSC is needed to be validated.更多还原