Remote Monitoring of Geotechnical Infrastructure Slopes Using InSAR: Challenges and Solutions

Geotechnical infrastructure slopes are critical elements of the transportation and flood defence networks. Accurate and timely monitoring of such elements is important in order to enable proactive intervention strategies and to avoid failures. Monitoring geo-infrastructure slopes can be done using both in-situ measurements and remote sensing observations using satellite data. In-situ measurements offer precise and localized data but are time-consuming, expensive, and limited in terms of spatial coverage. Over recent years, Interferometric Synthetic Aperture Radar (InSAR) technique has emerged as a powerful and promising tool for monitoring geo-infrastructure slopes. The technique can provide results with comparable precision to ground-based approaches, but with higher spatial coverage. In addition, archives of previously acquired images can be used to reveal past displacements.
The primary aim of this thesis is to utilize InSAR for monitoring geo-infrastructure slopes while developing effective solutions to overcome its limitations. The first phase of this thesis is to assess and explore the potential and limitations of this technique in monitoring ground motion at both individual slope and network-wide scales. In this regard, the feasibility and effectiveness of InSAR from C-Band sensors, such as Sentinel-1, as a monitoring tool for railway embankments constructed on peatlands is assessed. Its results are validated by comparing them with in-situ geophysical and geotechnical data. This case study highlights how remote detection of embankment motion, coupled with targeted in-situ investigation, can underpin proactive intervention to improve safety. At a broader scale, InSAR data, as provided by the European Ground Motion Service (EGMS), is utilized to identify zones of potential hazard/risk from ground motion along the road and railway networks in Ireland. The results highlight both the power of remotely measured surface motion data from satellite images to support cost-effective hazard assessment at wide spatial network scales and the importance of local (hydro)geological considerations in infrastructure management.
The second phase of this thesis focuses on addressing InSAR limitations in monitoring areas with a low density of coherent scatterers, i.e., measurement points. The purpose of this phase is to investigate how machine learning and optimization techniques can be leveraged in InSAR processing chain to enhance its results in monitoring such areas. In this regard, a novel framework is proposed to identify Coherent Pixels (CPs) using Deep Learning (DL) methods. Results demonstrate that the proposed framework identifies more CPs, while excluding more non-CPs, compared to amplitude-based selection approaches.
The last phase of this thesis is to develop cost effective Corner Reflectors (CRs) for monitoring small and densely vegetated slopes with no coherent scatterers. It focuses on investigating the use of low-cost and lightweight materials for making cost effective CRs. The microstructure of these materials is analysed by using Scanning Electron Microscopy (SEM) technique in a laboratory. Their visibility and backscattering properties are assessed in Sentinel-1 images. Furthermore, two CRs are installed in a landslide to investigate their performance in a real InSAR application.
Overall, this research demonstrates the significant potential of InSAR technology for monitoring geo-infrastructure slopes, while also addressing its limitations through innovative solutions. It highlights the effectiveness of this technique, particularly when integrated with in-situ data, for monitoring ground motion.