On August 14 th 2015 a large debris flow initiated by the occurrence of two slope failures at the front of the Lou rock glacier flooded part of the town of Lanslevillard, France. The present study aims to understand the meteorological and geomorphological context that led to these failures. Investigations were conducted by combining meteorological data, surface movements and geophysical transects. The analysis indicates that the Lou rock glacier is directly connected to an active torrential channel and has a natural predisposition to frontal failure due to the steepness of its front. The slope failures were triggered after a heat wave followed by a three-week period of almost continuous rainfall. Water flowing on top of the permafrost table was observed in the two head scarps, suggesting that regressive erosion consecutive to this concentrated subsurface water flow triggered the failures.For one of the slides, traces of previous failures were observable on historical aerial imagery dating back to the 1950's, while the second slide corresponded to a novel event and developed on the frontal slope of a fast moving and destabilized rock glacier lobe. We also discuss the increase in local predisposition to failure related to the remarkable morphological modifications such as frontal advance and development of surface cracks associated with the lobe destabilization.

Abstract. Bedrock permafrost is a feature of cold mountain ranges that was found responsible for the increase of rock fall and landslide activity in several regions across the globe. In Greenland, bedrock permafrost has received so far little attention from the scientific community, despite mountains are a predominant feature on the ice-free coastline and landslide activity is significant. With this study, we aim to move a first step towards the characterization of bedrock permafrost in Greenland. Our study area covers 100 km2 of mountain terrain around the town of Sisimiut – 68° N on the West Coast. We first acquire surface ground temperature data from 2020–2021 to model bedrock surface temperatures time series from weather forcing on the period 1850–2022. Using a topographical downscaling method based on digital elevation model, we then create climatic boundary conditions for 1D and 2D heat transfer numerical simulations at the landscape level. In this way we obtain permafrost distribution maps and ad-hoc simulations for complex topographies. Our results are validated by comparison with temperature data from two lowland boreholes (100 m depth) and geophysical data describing freezing/unfreezing conditions across a mid-elevation mountain ridge. Finally, we use regional carbon pathway scenarios 2.6 and 8.5 to evaluate future evolution of ground temperatures to 2100. Our results indicate a sporadic permafrost distribution up to roughly 400 m.a.s.l., while future scenarios suggest a decline of deep frozen bodies up to 800 m.a.s.l., i.e. the highest summits in the area.