Identification of the Sliding Surface of a Dormant Landslide Using UAV-Borne Low-Frequency Ground Penetrating Radar: The Melizzano Case Study (Southern Italy)

Landslide analysis in vegetated and inaccessible areas represents a key challenge for hazard assessment and risk mitigation. This study presents the application of a UAV-borne low-frequency Ground Penetrating Radar (GPR) system to investigate the internal structure and deep geometry of an earth slide located in southern Italy. The GPR system is based on a Cobra Plug-in Sub-Echo 70 antenna operating in the 20–140 MHz frequency range and mounted on a multirotor UAV. The survey consisted of an east–west transect crossing the landslide body, flown at approximately 1 m above ground level due to dense vegetation about 60–70 cm high. To ensure precise navigation and stable flight altitude, the UAV was equipped with a high-precision GNSS system and a radar altimeter, enabling accurate terrain-following acquisition in complex topographic conditions.  The acquired radargram shows a coherent subsurface response down to an investigation depth of approximately 15 m. A laterally continuous, high-amplitude reflector is clearly visible at around 10 m depth and is interpreted as the main sliding surface controlling the gravity-driven process. Above this surface, zones of chaotic and heterogeneous reflections indicate disturbed stratigraphic units and reworked debris involved in deformation processes across the main body. The analysis highlighted a roto-translational kinematics in the upper part of the source area, as often observed in this type of landslide, evolving downslope into a predominantly translational movement. The central portion of the radargram exhibits more continuous sub-parallel reflectors, suggesting partial preservation of the original stratigraphic organization within the displaced material.  Despite the limited acquisition geometry, our results demonstrate that UAV-borne low-frequency GPR provides critical subsurface information for understanding gravity-driven processes. The identification of depth and geometry of landslide failure  surfaces  represents a key contribution to landslide susceptibility analysis, hazard evaluation for definition of effective risk mitigation and monitoring strategies.