Analysis of the spatio-temporal characteristics of retreat caused by rockfall on a cliff with interbedded soft and hard rock layers

Retreat of rock cliffs due to rockfall is a common geological phenomenon in various environments. Both the rockfall events and the subsequent retreat pose potential risks to the infrastructure located both above and below the cliff. Extensive research on rock wall retreat has been conducted in both high alpine and coastal environments, addressing single rock wall instability, as well as linear and catchment-scale rock cliff dynamics under climate change. However, investigations into rock wall retreat in sub-alpine regions have primarily relied on inventories at the linear or regional scale. More attention is still required to understand the retreat of individual rock cliffs in sub-alpine environments. In this study, we focus on a subalpine molasse (sandstone-marls) cliff located at La Cornalle, Vaud, Switzerland, as the case study.
Using monthly Structure-from-Motion (SfM) photogrammetry surveys, meteorological data from a weather station and the Swiss Meteorological Office, and rock temperature obtained from thermal couples installed in the subsurface of the rock from December 2019 to September 2023, we established the rockfall inventory, calculated the retreat rate for the cliff, and analyzed the spatial and temporal features of the retreat with the support of the correlation between meteorological parameters and rockfall data. We found that 4051 rockfalls were documented during the nearly four-year survey, resulting in an average retreat rate of 19.6 mm/year. Specifically, for the marl layers, the retreat rate is about 21.1 mm/year, and for the sandstone layers, it is about 23.7 mm/year. Regarding the newly formed and flat vertical cliff face resulting from a rock collapse between December 17, 2021, and January 11, 2022, the retreat rates vary significantly between the marl and sandstone layers. The retreat rate for the marl layer from the new face is measured at 45.5 mm/year, while for the sandstone layer, it is 22.5 mm/year. This discrepancy is reasonable due to the inherent weakness of marl compared to sandstone. Additionally, no sandstone overhang existed when the face was newly created. As time progresses, the retreat of the marl layer weakens the support for the overlying sandstone, leading to subsequent rockfalls from the sandstone layer. Rainfall played a crucial role in the retreat evolution, while freezing-thaw cycles did not show a clear impact on the occurrence of rockfalls. However, snow melting could be a triggering factor for rockfalls detected during the winter. Interestingly, extreme hot weather during the last two summers did not immediately trigger many rockfalls. 
Our study introduces a comprehensive rockfall inventory from the sub-alpine Molasse cliff and investigates potential contributing factors to rockfall events. Additionally, we propose a model to elucidate the historical evolution of rock cliff retreat.