【作者】 陈艳玲；

【导师】 黄珹； 丁晓利；

【作者基本信息】 中国科学院研究生院（上海天文台） ， 天体测量与天体力学， 2007， 博士

【摘要】 星载SAR(Synthetic Aperture Radar，合成孔径雷达)及InSAR技术(Interferometric Synthetic Aperture Radar，合成孔径雷达干涉测量)是近三十年来发展起来的空间对地观测新技术，是未来三维测图与区域地形形变监测领域最具潜力的新技术之一，在农林业、海洋、地形及形变测绘、地质、生态等地球科学领域发挥着日益重要的作用。考虑到我国该领域的研究起步较晚，远远落后美国、欧洲等发达国家。2006年，我国第一颗SAR小卫星“遥感卫星一号”发射成功，更促使我国加速开展这方面的研究，以尽快赶上国际先进水平。在这样的背景下，本文从星载SAR原始数据(Level 0)出发，阐述了星载SAR成像的机理以及成像质量评定指标，提出了InSAR对成像算法及成像产品的几点要求，紧接着系统、全面地概括了(D-)InSAR技术的基本原理及局限性。在这些理论的支撑下，本文围绕星载SAR成像以及星载SAR、InSAR技术的应用主要做了以下工作：(1)星载SAR成像处理。成像处理是SAR技术的核心部分。第三章分析了星载SAR点目标回波信号模型以及几种常用的数字成像处理算法，接着介绍了距离—多普勒算法的数据处理流程及几个关键的问题；然后在第五章采用ERS-1、Radarsat-1以及JERS-1卫星数据利用EV-APP软件做了成像实验，并对成像质量做了评价。(2)星载SAR在海洋气象学中的应用—星载SAR图像反演海面风矢量研究。首先介绍了星载SAR估算海面风场的基本原理、算法及数据处理步骤。接着，以2002年5月7日香港地区ERS—2 SAR海洋图像为例，对经典的SWDA(SAR Wind Direction Algorithm)—谱分析方法加以改进，求得具有180°模糊度的风向，并用香港天文台气象浮标实测数据消除了风向不确定性。最后，利用CMOD4 GMF(Geophysical Model Function，地球物理模式函数)计算得到海面上10m高的风速。与浮标实测资料相比，文中两个20km×20km试验区域ROI1与ROI2风向误差分别为23.71°、7.00°，平均风速误差分别为0.18m／s、—0.12m／s，精度均较高。由此可以得出结论：如果对SAR预先进行ADC(Analog Digital Converter)改正以及精确校准，结合改进的SWDA和CMOD4，可以获得高精度的风矢量。(3)二轨差分干涉(D-InSAR)技术监测抚顺市矿山沉降。文中利用二轨差分干涉测量技术，使用ERS-1／ERS-2卫星1993年5月22日至2000年6月13日所获取的辽宁省抚顺地区6幅SAR图像以及SRTM 3×3弧秒分辨率的DEM，研究了抚顺市的沉降情况，监测精度达cm级。需要指出的是，文中采取了一种全新的干涉相位图频域滤波方法，使用这种算法使得差分干涉图的质量平均改善了28％。实验证明，D-InSAR技术可以有效地监测矿区沉降情况。论文的创新性或特点表现在以下几个方面：(1)从公开的文献看，本文在国内首次同时对不同卫星的SAR原始数据进行成像试验，并对成像的结果进行了定量评价，为InSAR数据处理中某些问题的解决提供了参考思路；(2)其次，星载SAR在海洋学、气象学方面的应用研究是极具挑战性和发展潜力的课题，本文从星载SAR图像提取了香港地区的风向和风速，并对反演算法进行了改进，从而为海洋及气象服务提供更加可靠的资料，这是具有前瞻性的研究，；(3)我国矿区InSAR沉降监测可应用性研究并不多见。本文利用D-InSAR这种新兴的技术对国内普遍存在的矿山沉降问题进行深入研究，并在数据处理中将新的频域滤波算法用于干涉相位图，取得了相当好的结果。更多还原

【Abstract】 Space borne SAR (Synthetic Aperture Radar) and InSAR (Interferometric SAR)are two kinds of new modern remote sensing space geodesy technologies, which havegot rapid development during the last 30 years. The two technologies providebrand-new tools for the 3D topographic mapping and earth surface deformationmonitoring, and play more and more important roles in earth science, such asagriculture and forestry, oceanology, topography and deformation surveying, geology,and ecology etc. Compared with America and European developed countries, studieson SAR and InSAR in China have relatively fallen behind. Until 2006, Chinese firstsmall SAR satellite was launched successfully, which spurred us to carry out morerelated researches as soon as possible so as to shorten the distance between China anddeveloped countries.Under this background, this paper introduces the principles of space borne SARimaging processing and several indexes which can reflect imaging quality. And thenilluminates the principles and limitations of InSAR and D-InSAR techniques.Subsequently, this paper focuses on the SAR imaging and the applications of SARand InSAR, and carries out some researches summarized as follows:(1) Space-borne SAR imaging processing. Imaging processing is the kernel of SARtechnique. Chapter 3 analyzes the echo signal model of point target; introducesseveral SAR imaging algorithms, illuminates the essential of range compressing andazimuth compressing during the Range-Doppler algorithm, discusses some keyproblems, such as Doppler parameters estimation, range migration correction andspeckle noise elimination in detail; and then adopts SAR raw data of Radarsat-1,ERS-1, and JERS-1 to do some imaging tests in chapter 5. Finally, we getencouraging imaging results.(2) The application of Space-borne SAR on the marine meteorology-sea windinversion by SAR image, seen in chapter 6. We first introduce the principle and threemain algorithms of ocean wind retrieval with SAR images. Then we provide the flowchart of retrieval procedures. As an example, ERS-2 SAR images covering HongKong region acquired on May 7, 2002 are used to carry out test of wind vectorretrieval. The data processing includes pre-processing of SAR image, ADC (AnalogDigital Converter) compensation, accurate calibration and speckle removal, and theapplication of classical SAR Wind Direction Algorithm for wind direction retrieval.Data collected on a buoy by the Hong Kong Observatory are then used to resolve thewind direction uncertainty. Finally, the GMF (Geophysical Model Function)-CMOD4is adopted to estimate the wind speed at the height of 10m above sea level. Comparedwith the data recorded by Hong Kong Observatory, the errors in the wind direction inthe areas of ROI1 and ROI2 are 23.71°and 7.00°, respectively, and the errors inaverage wind speed are 0.18 m/s and—0.12 m/s. The results show that betterpre-processing, together with the use of classical spectral analysis algorithm andCMOD4 model can offer high quality wind vector results.(3) Monitoring mining subsidence with two-pass D-InSAR technique in Fushun city,China. We use 6 ERS-1 and ERS-2 SAR images acquired from May 22, 1993 to Jun13, 2000 and covering whole Fushun city, and the SRTM DEM with3 arc-secondresolution to conduct two-pass differential SAR interferometry. They can form 15differential interferograms in total, but only 5 with satisfactory baseline and coherence.By investigating the 5 interferometric pairs, we obtain the subsidence rate of Fushuncity. During the data processing, we use a new modified Goldstein filtering algorithmin frequency field to filter the differential and greatly improved the interferogramquality. Finally, the case study verifies that D-InSAR technique is able to detectcm-level subsidence of the mining area.Some points of the former works deserve to be mentioned: (1) First, this paperprocesses the raw data of several satellites received by different satellite stations andcarry out related imaging tests, which provides possible answer to the uncertainproblems in InSAR data processing. (2) Second, as predicted the applications ofspace-bome SAR on the oceanology and meteorology will be one of the mostchallenging and valuable projects in the near future, meanwhile similar works inChina have not got enough attention relatively. This paper estimates the sea windvector using space-borne SAR image, and have done little improvement for theinversion algorithm, thus offered more reliable data for meteorology services. (3)Studies on mining subsidence using D-InSAR are very rare in the field of earth sciences.Therefore this paper adopts new technique-D-InSAR to investigate the miningsubsidence, and uses a brand-new modified Goldstein filtering algorithm in frequencyfield to filter the interferogram phase maps and get very good results.更多还原