基于PCA和vbICA的时序InSAR大气延迟改正方法研究

InSAR is one of the key techniques in geodetic survey. The tropospheric delay is one of the main error sources of InSAR. The accurate correction of tropospheric delay is the key to improve the accuracy of InSAR measurements. However, the current tropospheric delay correction methods have many shortcomings. For instance, the a priori knowledge of deformation is either not easy to obtain or inaccurate. The selection of experience models is flat. Most methods don’t take into account the variability of signal distribution in different interferograms. In addition, the ability to reduce turbulent components of troposphere is poor. All above limits the application of the phase-based methods..In this project, a new tropospheric delay correction technique based on SAR data itself by using PCA (Principal Component Analysis) and vbICA (Variational Bayesian Independent Component Analysis) is proposed: First, the potentials and advantages of PCA and vbICA methods for constructing the InSAR deformation model are explored. The theory and methodology of collaborative establishment of deformation model based on PCA and vbICA are studied. Then, taking into account the variability of signal distribution in different interferograms, an adaptive empirical model is constructed. An iterative algorithm is proposed to finely separate the stratified and the turbulent tropospheric delay. Finally, the potential of the new method for high-precision InSAR deformation monitoring is tested by applying it to different test areas. The effectiveness and applicability of the method are also evaluated comprehensively. The project is conducive to the breakthrough of the research bottleneck of InSAR atmospheric delay correction method. It will promote the development of InSAR technique in the field of high-precision surface deformation monitoring, which has important theoretical and practical value.

InSAR是大地测量领域的关键技术之一。大气对流层(简称大气)延迟是InSAR主要误差源之一，其精确改是提高监测精度的关键。但目前大气延迟改正方法存在诸多问题，先验形变知识不准确、经验模型选择单一、未考虑不同干涉图信号分布差异性、对湍流成分削弱能力差等问题限制了基于SAR数据自身方法的应用空间。项目引入主成分分析(PCA)和变分贝叶斯独立成分分析(vbICA)，研究一种InSAR大气延迟改正新方法：挖掘PCA和vbICA性能优势，研究协同构建时序InSAR地表形变模型的理论方法；顾及不同干涉图信号分布特征，构建自适应经验模型，提出基于循环迭代的大气垂直分层和湍流成分精细分离方法；选择测试区域，挖掘新方法用于InSAR高精度形变监测的潜力，综合系统地评价方法有效性及适用性。项目有利于突破InSAR大气延迟改正方法的研究瓶颈，促进InSAR高精度地表形变监测技术的发展，具有重要理论和实际意义。

InSAR是大地测量领域的关键技术之一。大气对流层(简称大气)延迟是InSAR主要误差源之一，其精确改是提高监测精度的关键。但目前大气延迟改正方法存在诸多问题，先验形变知识不准确、经验模型选择单一、未考虑不同干涉图信号分布差异性、对湍流成分削弱能力差等问题限制了基于SAR数据自身方法的应用空间。本研究基于PCA/vbICA方法构建了时序InSAR数据处理模型，并深入分析了时序InSAR信号，构建了地表形变模型，基于多项式模型，实现了时序 InSAR 大气延迟信号的提取，解决了大气延迟信号与形变信息耦合问题，达到提高地表形变监测精度的目的。通过对干涉信号的多次分解，实现了干涉相位中多种误差源的提取，较精细地实现大气湍流成分和分层成分的分离，同时在矿区、城市、火山等研究区的应用中，相较于未进行大气延迟改正的干涉图平均精度提高近50%，实现了高精度地表形变信息的提取。