Seasonal preconditioning and triggering of alpine mass movements at an active rock slope from machine-learning seismic monitoring

Rock slopes in high-mountain environments are increasingly affected by climate-driven changes in thermal and hydrological conditions, yet links between environmental forcing, deep-seated deformation, and surface mass-movement activity remain poorly constrained at the event scale. Here, we present a continuous three-year seismic catalog of nearly 13,000 detected mass-movement events at the Spitze Stei rock slope in the Swiss Alps, generated using a Random Forest classifier applied to continuous seismic data. The catalog reveals a bimodal seasonal pattern, with enhanced activity in early summer and autumn. Statistical analyses identify rainfall intensity as the dominant short-term trigger during the warm season, whereas the duration of above-freezing periods controls activity during winter and transitional freeze-thaw conditions. Beyond immediate triggers, seasonal preconditioning affects mass movement activity. Elevated spring activity coincides with snowmelt-driven pore-pressure increases and material weakened by winter freeze-thaw cycles, while the autumn peak occurs during maximum seasonal thaw depth, when near-surface layers are mechanically weakened and highly responsive to short-term meteorological forcing. Comparison with GPS displacement data shows that deep-seated deformation is seasonally controlled by hydrological forcing in spring and early summer, whereas surface mass movements do not systematically scale with deep creep later in the year. These results demonstrate that thaw depth strongly impacts seasonal susceptibility, while short-term hydrological and thermal perturbations initiate individual failures. Continuous ML-based seismic monitoring therefore provides a quantitative framework for linking event-scale mass-movement activity with evolving environmental forcing in high-mountain terrain.