| Author: | Wu, Songbo |
| Title: | Model parameterization and sequential estimation in multi-temporal InSAR deformation monitoring |
| Advisors: | Ding, Xiaoli (LSGI) |
| Degree: | Ph.D. |
| Year: | 2022 |
| Subject: | Synthetic aperture radar Remote-sensing images Image processing -- Digital techniques Hong Kong Polytechnic University -- Dissertations |
| Department: | Department of Land Surveying and Geo-Informatics |
| Pages: | xvii, 170 pages : color illustrations |
| Language: | English |
| Abstract: | Spaceborne interferometric synthetic aperture radar (InSAR) is one of the most powerful geodesy techniques for mapping ground deformation. Compared with traditional techniques such as leveling, InSAR has proven its advantages in broad coverage, and high spatial resolution. It can provide ground deformation maps with up to subcentimeter precision. InSAR retrieves the ground deformation by analyzing the interferometric phase of SAR images (termed an interferogram [IFG]) that cover a common area. The IFG is composed of the phase components related to, for example, the deformation, topography, atmosphere, and decorrelation noise. Compensation of non-deformation phase being challenging with a single IFG, the accuracy of InSAR-derived deformation is then limited by atmosphere delay and decorrelation noise. To overcome the limitations, the IFG stacking technique, known as multi-temporal InSAR (MT-InSAR), was developed to retrieve more accurate deformation. Three types of MT-InSAR methods have been developed. The first type is Persistent Scatterer InSAR (PS-InSAR), which analyzes the point targets with high phase quality and single-prime baseline network. The second type is the Small BAseline Subset (SBAS) technique, which is designed to retrieve the deformation using unwrapped IFGs with multi-prime baseline network. The third type is a joint framework of PS-InSAR and SBAS, which features improved spatial resolution of the deformation. Since all of these methods share a similar data processing framework, many studies have been devoted to optimizing coherent point selection, spatiotemporal phase unwrapping, model parameterization, and sequential data processing among others. Despite the great efforts have been made to advance the MT-InSAR method, many challenges remain to be overcome, such as coherent point selection and spatiotemporal phase unwrapping, which are key tasks involved in MT-InSAR and are usually conducted separately with the empirical threshold. In the selection of coherent points, due to discrepancies in threshold settings, even the same MT-InSAR method with the same datasets will result in different results. In addition, an improper deformation model employed in traditional MT-InSAR methods may result in the removal of coherent points undergoing unmodeled deformations, thus biasing the parameter estimation. Furthermore, due to the increase in urban infrastructure density and the discontinuity of ground features, the phase unwrapping of MT-InSAR is more complicated and its accuracy is biased, which may even lead to the failure of phase unwrapping. On the other hand, with an increase in the number of SAR datasets, InSAR tends to be a near-real-time (NRT) monitoring technique by regularly updating the deformation map. It is, however, difficult to reach the goal due to the entire dataset having to be reprocessed each time a new image is obtained. To resolve these issues, the development of advanced sequential approaches has become necessary. In this thesis, we conduct the research works to improve the applicability of MT-InSAR from three perspectives. To select the coherent points, we propose a pixelwise method for estimating ground deformation with an IFGs stack. The method can simultaneously select the coherent point and retrieve the unwrapped phase from the wrapped IFGs. It can also avoid the empirical setting of thresholds and pre-assumed deformation models, thus simplifying the InSAR data processing steps. The proposed method uses iterative least squares to calculate the unwrapped phase. The closure value of the closing triangular network is performed to enhance the accuracy of the phase unwrapping. Based on the linear inversion system, the proposed estimator is suitable for a small subset of IFGs and can be extended to sequential estimation for extensive SAR data. To overcome the influence of the increase in the infrastructure density and the discontinuity of ground objects on MT-InSAR urban phase unwrapping, we propose a height-guided network strategy for isolating the propagation path of the solution in network adjustments. This network strategy can avoid the impact of an improper propagation path caused by the discontinuity of ground infrastructures. Importantly, two quality controllers, namely network refinement and unwrapping error correction, are deployed to improve phase unwrapping accuracy. The proposed method is examined using both simulated and real experiments, proving its promising performance in urban areas with dense infrastructures such as Hong Kong. To efficiently process and sequentially estimate the large amount of InSAR datasets, we present an sequential estimator for MT-InSAR continuous ground deformation monitoring. The method processes extensive SAR data in batch sequences, and the results of each batch are constrained with an overlap to ensure consistency between each batch estimation. Phase optimization and closure value detection are deployed to isolate the observations that likely suffer from ambiguity and considerable noise. The Kalman filter is used to update the deformation and height with the new observation of the unwrapped phase. The simulated and real experiments in various scenes are designed to evaluate the feasibility of the proposed methods. A study of the decade settlement of Hong Kong International Airport (HKIA) is presented using multi-platform SAR datasets. The results prove the effectiveness of the proposed methods and indicate that the serious heterogeneous ground settlement on the HKIA platform is mainly associated with the fill materials, alluvial deposits, and construction stages. |
| Rights: | All rights reserved |
| Access: | open access |
Copyright Undertaking
As a bona fide Library user, I declare that:
- I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
- I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
- I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.
By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.
Please use this identifier to cite or link to this item:
https://theses.lib.polyu.edu.hk/handle/200/11974