Amplitude and Phase InSAR Analysis of Large Landslide Datasets



Journal Title

Journal ISSN

Volume Title



Landslides are common natural phenomena that can have catastrophic consequences, with significant economic, environmental, and social impacts worldwide. These events can occur in a matter of seconds or evolve over geologic timescales. As such, there is a growing need for reliable and effective techniques to monitor and assess landslides over different time intervals. One promising approach is the use of Synthetic Aperture Radar (SAR), which can penetrate through clouds and can operate day or night, making it an ideal tool for landslide detection and monitoring. In this thesis, we explore the use of SAR-based methodologies to observe landslides over various timescales, ranging from fast-moving (meters/second) to slow-moving (cm/year). To assess the effectiveness of SAR-based methodologies, we conducted several case studies using amplitude and phase-based SAR techniques. For instance, in the aftermath of a storm event in Hiroshima, Japan, and an earthquake followed by a storm in Haiti, we employed SAR intensity-based failure detection methods to locate landslides. Furthermore, we leveraged the processing power and catalog of satellite radar images from Google Earth Engine to develop an algorithm that detects landslides over multi-year periods, which we validated using optical imagery and the Normalized Difference Vegetation Index (NDVI). In addition to these case studies, we investigated the movement of slow-moving landslides in Western Oregon, a region prone to recurring landslide movements and long-term accelerating trends due to coastal erosion. We used ESA Sentinel-1 C-band (5.6 cm wavelength) multi-temporal interferometric SAR data and PRISM climate data from 2017–2021 to analyze the displacements of a large landslide inventory over time. We also explored the relationship between precipitation, landslide morphology, and deformation rates. Finally, we used NASA AirMOSS InSAR data acquired at P-band (70 cm wavelength) over several landslides and compared it to the shorter-wavelength Sentinel-1 C-band multi-temporal SAR data to assess the advantages and drawbacks of longer wavelengths sensors. Our findings highlight the potential of SAR-based methodologies to detect and monitor landslides over various time scales. This study offers valuable insights into the development of reliable and effective tools for landslide management, with practical applications for disaster response, urban planning, and risk assessment.



Landslides, Synthetic Aperture Radar, SAR, InSAR, Amplitude, Deformation, Displacement, Failure, Monitoring, Detection, Machine learning, Change detection, C-band, P-band, Time series, Earthquake, Haiti, Japan, Oregon, Precipitation, Natural hazards