| Description: |
Abstract Background The slope deformation of the Eggerbach slope above the Gradenbach river in Carinthia, Austria, has been monitored since 1966 following a critical acceleration event that caused damming and erosion of the Gradenbach river, resulting in devastating floods and debris flows in Putschall village. This study analyses several decades of comprehensive monitoring data, including time series from continuous, non-continuous GNSS stations (2009–2023), analog wire extensometer measurements (1980–present), borehole water level recordings, and meteorological data from the SPARTACUS v. 2.1 and SNOWGRID-CL gridded climate datasets. Distinct displacement patterns observed across four GNSS stations necessitated revision of the previously assumed single-domain geological model. Results Engineering geological analysis reveals that the Eggerbach slope constitutes a rock compound slide comprising four kinematically distinct domains: two upper translational slides (controlled by foliation-parallel surfaces at 60–80 m depth) and two lower rotational slides (controlled by fault plane ruptures along the Mölltal fault system at 120–140 m depth). Movement vector analysis shows systematic \alpha-angle progression from 14° to 34° across domains, reflecting the transition from compression-dominated vertical zones to rejoin to extension-dominated translational sliding in the upper slope. Annual horizontal displacements range from 4.11 to 6.74 m across domains since 2009, with significant velocity increases in 2018, 2020, and 2021 (peak values 0.54–1.15 m/yr) that deviate substantially from the 40-year baseline trend (0.04 to 0.19 m/yr). Statistical analysis establishes snowmelt—not precipitation—as the primary trigger for acceleration events, with cross-correlation analysis revealing r = 0.67 (p < 0.001) at 15–22 day lags between snow water equivalent (SWE) decreases and velocity peaks. Spring snowmelt (April–May) accounts for 89% of total annual displacement, with SWE reductions >150 mm within 30-day periods triggering velocity increases exceeding 2× background rates in 73% of analyzed events. Machine learning models (Support Vector Machines, Random Forest, XGBoost) achieved 78–85% variance explanation for velocity prediction and successfully forecasted 2018–2021 acceleration events despite training only on lower-velocity pre-2010 data. Feature importance analysis quantifies that borehole water levels (42%) and SWE (25%) dominate displacement controls, while precipitation contributes marginally (8%), enabling operational 15–30 day velocity forecasting for early warning applications. Discussion and Conclusion A critical temporal anomaly—where comparable SWE levels in 1991–1993 versus 2018–2021 produced 3–8× different velocity responses—suggests potential degradation of drainage infrastructure installed in the 1970s through mechanical shearing, calcareous precipitation in carbonate-rich domains, or pathway blockage. However, this hypothesis requires validation through targeted drainage system investigation. The stability number (Ns = 1.08 >> 0.25) calculated from uniaxial compressive strength analysis indicates ductile plastic behavior, suggesting gradual rather than catastrophic failure potential provided consolidation barriers function and toe-erosion is prevented. The lower rotational domains function as compressional buttresses absorbing gravitational loading from upper translational masses, creating mechanical coupling where system stability depends critically on maintaining toe confinement. |