InSAR数据缺失拟合探讨

发布时间：2018-01-19 19:29

本文关键词： InSAR 缺失数据拟合 Kriging 拟合推估各向异性　出处：《长安大学》2015年硕士论文　论文类型：学位论文

【摘要】：InSAR监测技术因为它具有高精度、高分辨率等诸多优点,已逐步成为进行地面变化监测的重要方法手段之一,并在地震监测、煤矿监测、火山监测等地质灾害监测研究以及地面沉降监测等诸多方面得到了广泛的应用。但是,由于难以避免的受到时间、空间上各种去相干现象的影响,InSAR技术提取的形变信息中经常存有缺失现象,因此需要后期对其进行填补处理。对缺失的数据进行拟合填补,通常采用空间数据的插值方法来进行。论文在对比分析常用的数据插值方法基础上,重点讨论了既能顾及结构性成分,又能较好表达地表局部变化的拟合推估,并就各向异性随机信号协方差函数的拟合进行了研究,提出了基于各向异性的自适应拟合推估法,并将其连同其他方法应用于地质灾害监测缺失数据拟合之中。取得主要成果如下:1.讨论了影响InSAR监测成果的误差源以及影响InSAR相干性、造成数据存有缺失的各去相干源;分析了影响缺失数据拟合精度的因素,发现除拟合方法外,缺失数据区域大小以及缺失数据变化起伏程度是影响拟合精度的关键因子。2.探讨了多项式曲面拟合法、反距离加权法、克里金插值法和拟合推估法在缺失数据拟合应用中的特性。考虑到随机信号通常表现出各向异性的特点,而在常规拟合推估方法中,常被认为各向同性。基于此,论文重点研究了各向异性协方差函数的拟合问题,提出了基于各向异性的自适应拟合推估法,并将其应用于缺失数据的拟合之中,通过实例验证了基于各向异性的自适应拟合推估法的有效性。3.以地面沉降监测数据以及地震形变监测数据作为实例,对比并且分析了不同的拟合方法的拟合精度,发现多项式曲面拟合法精度偏低,不适合于缺失数据的拟合;反距离加权尽管方法简单,但对于大区域数据拟合,易导致拟合过于平滑的现象;克里金插值与拟合推估方法所具有的拟合精度较高。对于存有明显区域特征的形变区域,采用分区拟合具有更高的填补精度。
[Abstract]:Because of its advantages of high precision and high resolution, InSAR monitoring technology has gradually become one of the important methods for ground change monitoring, and has been used in seismic monitoring and coal mine monitoring. Geological hazard monitoring, such as volcanic monitoring, and land subsidence monitoring have been widely used. However, due to the inevitable influence of time and space, there are various decoherence phenomena. The deformation information extracted by InSAR technology often has the phenomenon of missing, so it needs to be filled in the later stage, and the missing data is fitted to fill. Based on the comparison and analysis of the common data interpolation methods, this paper focuses on the fitting and estimation of the local changes of the ground surface, which not only takes into account the structural components but also can better express the local changes of the surface. The fitting of anisotropic random signal covariance function is studied, and an adaptive fitting method based on anisotropy is proposed. The main results are as follows: 1. The error sources that affect the InSAR monitoring results and the InSAR coherence are discussed. The various decoherence sources that cause the data to be missing; The factors influencing the fitting accuracy of missing data are analyzed, and it is found that except the fitting method. The size of missing data area and the fluctuation degree of missing data are the key factors to affect the fitting accuracy. 2. The polynomial surface fitting method and inverse distance weighting method are discussed. The characteristics of Kriging interpolation and fitting estimation in the application of missing data fitting, considering that random signals usually exhibit anisotropy, but in the conventional fitting and estimation methods. This paper focuses on the problem of anisotropic covariance function fitting, and puts forward an adaptive fitting and estimation method based on anisotropy, and applies it to the fitting of missing data. The validity of the adaptive fitting method based on anisotropy is verified by an example. 3. Taking the data of ground subsidence monitoring and seismic deformation monitoring as an example. By comparing and analyzing the fitting accuracy of different fitting methods, it is found that the fitting accuracy of polynomial surface fitting method is low, which is not suitable for the fitting of missing data. Although the inverse distance weighting method is simple, it is easy to make the fitting too smooth for large area data fitting. The method of Kriging interpolation and fitting and estimation has higher fitting accuracy and better filling accuracy for deformation regions with obvious regional characteristics.
【学位授予单位】：长安大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：P225.1

【参考文献】