【作者】 汤益先；

【导师】 王超； 张红；

【作者基本信息】 中国科学院研究生院（遥感应用研究所） ， 地图学与地理信息系统， 2006， 博士

【摘要】 目前在我国16个省、市的近50个城市已经发生了地面沉降，每年造成的经济损失过亿，并已成为阻碍我国城市可持续发展的制约因素之一，因此必须采取有效的监控手段，对地面沉降区域进行全面监测，及时了解地面沉降的现状，准确预测地面沉降的发展趋势，从而提出有效的遏制或防治的对策。另一方面随着对地空间观测技术的发展，尤其是上世纪木发展的合成孔径雷达差分干涉测量(Differential SAR Interferometry DInSAR)技术，为人们提供了全天时全天候获取高精度的地表地形信息以及地表形变信息的空间技术手段，使得利用遥感技术进行地面沉降监测成为可能，但是地面沉降由于属于微小地表形变，一般时间跨度数年，因此利用传统的差分干涉测量技术将会严重受时间，几何去相干以及大气效应的影响，而使DInSAR对地面沉降的监测失效。 本论文在分析传统差分干涉测量的基础上，针对长时间序列地表形变监测的需要，利用传统干涉处理所获取的差分干涉相位图，积极开展基于相干目标的DInSAR的方法研究，选择在长时间序列上保持相干的像素点作为研究对象，利用所选相干点上的可靠相位信息，建立线性形变模型，进行精密SAR差分干涉测量相位分析，完成地表形变反演，并在此基础上，对苏州和沧州地区两个实验区域进行了地面沉降的实际应用研究，利用本文所介绍的基于CT的反演方法，获得了近8年的区域地表形变变化，并与实际的水准数据结果比较，显示了很好的一致性。 本文在开展基于CT的DInSAR研究中，主要进行了如下研究： 1、相干点的选择，作为长时间序列地表形变的研究对象，对相干点选择的准确度直接影响了最后反演的结果。本文通过分析相干目标在时空的统计特性基础上，说明了利用幅度时间序列的离散指数与利用时间序列相干系数阈值方法的一致性及差异性，针对大数据集的情况下，采用幅度时间序列的离散指数可以在保持图像分辨率的情况下获得准确的相干点集，而在有限的数据集情况下，由于数据量的有限使得图像的幅度统计特性无法完全正确进行估计，而采用了相干系数法，保证了CT点的正确性的同时，牺牲了一定的图像分辨率；更多还原

【Abstract】 At present there are about 50 cities in 16 provinces suffered from subsidence, and the economic losses are more than 100 million each year, which have become one of the limiting factors of the urban sustainable development. In order to know the status of subsidence in time, forecast the trends of subsidence exactly and make the prevent policy, the efficient way to monitor subsidence must be carried out. On the other hand, with the development of the technique of spaceborne earth observation (EO), specially the development of the Diffferential SAR Interferometry in last century, it is possible to get the information of the topography and deformation of the earth surface with high resolution independent of the weather and sun illumilation. Despite of the tremendous potential of DInSAR technique, limitations due to temporal decorrelation, spatial decorrelation and atmospheric dishomogeneities prevent this technique from wildly being applied in crustal deformation monitoring, especially in the low-rate deformation cases.In this paper based on the analysis of traditional DInSAR, the CT-DInSAR research is carried out to meet the need of the deformation monitoring on long time scale. The coherent targets are selected from the time scale as the research object. By using linear deformation model, the phase of the selected pixels is analyzed to obtain the result of deformation. In the end, the algorithm proposed in this paper is applied to Suzhou and Cangzhou citys, which are in Jiangsu and Heibei provinces respectively. And the results of deforamation of 8 years are presented and are compared with the leveling measurements showing a good agreement.In this paper, we focus on the following points:1. The selection of the Coherent Target (CT) As the objects we study on the long time scale, the accuracy of the selection effects the result of this algorithm directly. Based on the analysis of the spatial and temporal statistical characteristics of the coherent targets, the difference of selection method between temporal Dispersion Index of amplitude on time and the correlation threshold is presented. For a stack with sufficient images, the method of Dispersion Index of amplitude can get the CTs with the origin resolution, and for a small number of images set, the correlation threshold is adopted instead of the Dispersion Index of amplitude, which promises the accuracy at the cost of the resolution, because the estimation of the statistic characteristic of amplitude is not good on the condition of the limited number of images.更多还原