Deep-seated landslides: multiscale investigation for hazard assessment integrating geomorphology, geophysics, and remote sensing

Deep-seated landslides are complex slope movements whose characterization requires the integration of multiple investigation methods. However, the synthesis of these observations into coherent conceptual models remains challenging, especially in complex geological settings. To address these limitations, this thesis presents four case studies that combine geological, geomorphological, geophysical, and remote sensing methods with process-based analyses. Two contributions focus on passive seismic techniques, in particular the Horizontal-to-Vertical Spectral Ratio (HVSR) method, for subsurface characterization of large landslides. In San Vito Romano (central Italy), an integrated geological and geophysical model is developed for a poorly documented landslide, leading to revised estimates of its geometry, thickness, and volume, and demonstrating the effectiveness of HVSR in supporting landslide hazard assessment. In a second case, a recently activated rock block slide in Cavatore (northwestern Italy) is used to test passive seismic methods in a well-constrained geomorphological context. The results show that seismic attributes, combined with clustering techniques, enable an internal zoning of the unstable area and the identification of sectors with different degrees of mechanical degradation and instability. A further analysis addresses rainfall-induced landslide occurrence in northwestern Italy through the development of an extensive open-access landslide inventory of the extreme autumn 2019 events that also triggered the Cavatore landslide. Geostatistical analyses highlight a strong lithological control on landslide distribution and non-linear relationships between rainfall forcing and landslide occurrence, emphasizing the role of geological context in modulating rainfall thresholds along the boundary separating the Alpine Ligurian Units from the Tertiary Piedmont Basin. The fourth case study focuses on the western Corinth Rift (Greece), where Sentinel-1 InSAR time series are integrated with field mapping to identify active deep-seated landslides. Time-series clustering separates gravitational deformation from other tectonic signals and shows that rainfall predominantly controls long-term displacement, while seismicity induces localized and delayed kinematic responses. Overall, this thesis integrates field observations, passive seismic data, and satellite-based monitoring for the characterization of deep-seated landslides. It evaluates the role of predisposing factors (lithology and structural setting) and triggering factors (rainfall and seismicity) in controlling slope movements, with important implications for landslide risk management.