3D Discrete Fracture Network Modelling from UAV Imagery Coupled with Tracer Tests to Assess Fracture Conductivity in an Unstable Rock Slope: Implications for Rockfall Phenomena

The stability of a rock slope is strongly influenced by the pattern of groundwater flow through the fracture system, which may lead to an increase in the water pressure in partly open joints and the consequent decrease in the rock wall strength. The comprehension of the fracture pattern is a challenging but vital aspect in engineering geology since the fractures’ spatial distribution, connectivity, and aperture guide both the water movement and flow quantity within the rock volume. In the literature, the most accepted methods to hydraulically characterise fractured rocks in situ are the single borehole packer test, the high-resolution flow meters for fractures, and the artificial tracer tests performed in boreholes. However, due to the high cost a borehole requires and the general absence of wells along coastal cliffs, these methods may not be appropriate in rockfall-prone areas. In this study, an unsaturated rocky cliff, strongly affected by rockfalls, was investigated by combining kinematic analysis, Discrete Fracture Network (DFN) modelling, and artificial tracer tests. The DFN model and potential rock block failure mechanisms were derived from high-resolution 3D virtual outcrop models via the Structure from Motion (SfM) photogrammetry technique. An artificial tracer was injected using a double ring infiltrometer atop the recharge zone of the slope to determine the infiltration rate and validate the DFN results. The DFN and tracer test methods are frequently used at different spatial scales and for different disciplines. However, the integration of digital photogrammetry, DFN, and tracer tests may represent a new step in rockfall and landslide studies. This approach made possible the identification of groundwater flow patterns within the fracture system and revealed about a 10-day tracer transit time from the injection area and the monitored slope, with similar conductivity values gathered from both the DFN and tracer test. Planar and wedge failures with volumes ranging from 0.1 and 1 m3 are the most probable failure mechanisms in the areas. The results were consistent with the delay between the intense rainfall and the slope failures previously documented in the study area and with their mechanisms.