Speaker: Alexander Handwerger, JPL/CalTech

Landslides are major hazards with significant impacts on both natural and built environments worldwide. A primary objective in remote sensing of landslides is to monitor their behavior and extent, ideally providing early warning before they cause damage or loss of life. In this study, we utilize multiple remote sensing datasets—including satellite radar from Sentinel-1, airborne radar from NASA’s Uninhabited Aerial Vehicle Synthetic Aperture Radar (UAVSAR), and optical satellite imagery from PlanetScope—to investigate two prominent landslides in California, USA: (1) the Mud Creek landslide in Big Sur and (2) the Portuguese Bend landslide in Rancho Palos Verdes. Leveraging these diverse data sources, we mapped active landslide extents and quantified kinematics. In the case of Mud Creek landslide, we found that InSAR detected slow movements (dm/year) over an 8-year period but was limited in capturing the higher rates (>10 m/year) that preceded catastrophic failure. In contrast, pixel offset tracking of PlanetScope optical images revealed the landslide accelerated, in a predictable way, several weeks before failure. For Portuguese Bend landslide, we found that satellite InSAR detected 7 years of slow motion (cm/yr) but became unreliable when the landslide accelerated significantly starting in Summer 2023. Using satellite InSAR coherence time series and new UAVSAR data acquired at weekly intervals, we were able to identify the active slope extent and better quantify its sliding rate. Our work demonstrates the importance of multi-sensor approaches for capturing the complex dynamics of landslides, offering critical insights for hazard assessment and mitigation.