ABSTRACT Permafrost is a widespread thermal phenomenon in Arctic and high mountain regions that, in combination with steep terrain, exerts a key control on slope stability. Its degradation in response to climatic warming can modify the likelihood and timing of rockfall, debris flow and related mass movements. Robust information on the current state and short‐ to mid‐term evolution of permafrost in bedrock and unconsolidated deposits is therefore essential for natural hazard assessment in high‐mountain regions. This study presents a virtual monitoring framework that enables continuous simulation of the hydrothermal state of permafrost bodies and its temporal evolution. The framework is driven by a physical snow and ground energy‐balance model (SNOWPACK) and can be forced either with in situ meteorological observations or with medium‐range numerical weather forecasts, thereby permitting seasonal‐scale projections of permafrost conditions including an explicit representation of forecast uncertainty via ensemble input. The system is applied to two sites in the Bernese Oberland (Swiss Alps). At Rottal, where permafrost occurs in massive bedrock, simulated ground temperatures show good agreement with borehole observations, indicating that the model setup is suitable for relatively homogeneous lithological conditions. At Schilthorn, by contrast, jointed micaceous shales overlain by fine‐grained sediments form a much more complex subsurface. These conditions cannot be adequately captured with current SNOWPACK parameterizations, which leads to reduced model skill and prevents a realistic representation of heat and water fluxes in the lower part of the profile, even though simulations remain robust for the shallow subsurface. Using the Rottal site as an example, the study demonstrates the potential of seasonal forecasts of permafrost evolution when medium‐range weather prediction ensembles are used to drive the model. This approach allows the propagation of atmospheric forecast uncertainty into the simulation of hydrothermal states within permafrost. Overall, the results highlight that a virtual monitoring system relying on a physically based model and meteorological data is fundamentally feasible and promising for hazard‐relevant permafrost applications, but they also underline the need for improved parameterizations of subsurface physical properties where stratified or highly heterogeneous ground conditions prevail.
Mani et al. (Sun,) studied this question.