Detection and quantification of permafrost change in alpine rock walls and implications for rock instability

Abstract

The perennial presence of ice in permafrost rock walls alters thermal, hydraulic and mechanic properties of the rock mass. Temperature-related changes in both, rock mechanical properties (compressive and tensile strength of water-saturated rock) and ice mechanical properties (creep, fracture and cohesive properties) account for the internal mechanical destabilisation of permafrost rocks. Two hypothetical ice-/rock mechanical models were developed based on the principle of superposition. Failure along existing sliding planes is explained by the impact of temperature on shear stress uptake by creep deformation of ice, the propensity of failure along rock-ice fractures and reduced total friction along rough rock-rock contacts. This model may account for the rapid response of rockslides to warming (reaction time). In the long term, brittle fracture propagation is initialised. Warming reduces the shear stress uptake by total friction and decreases the critical fracture toughness along rock bridges. The latter model accounts for slow subcritical destabilisation of whole rock slopes over decades to millennia, subsequent to the warming impulse (relaxation time).<br /> To gain further understanding of thermal, hydraulic and mechanic properties of permafrost rocks, multidimensional and multi-temporal insights into the system are required. This Ph.D. adopted, modified and calibrated existing ERT (electrical resistivity tomography) techniques for the use in permafrost rocks. Laboratory analysis of electrical properties of eight rock samples from permafrost summits brought upon evidence that the general exponential temperature-resistivity relation, proposed by McGinnis (1973), is not applicable for frozen rocks, due to the effects of freezing in confined space. We found, that separate linear temperature-resistivity (T- ρ) approximation of unfrozen, supercooled and frozen behaviour offers a better explanation of the involved physics. Frozen T-ρ gradients approach 29.8 ±10.6 %/°C while unfrozen gradients were confirmed at 2.9 ±0.3 %/°C. Both increase with porosity. Path-dependent supercooling T-ρ behavior (3.3 ±2.3 %/°C) until the spontaneous freezing temperature -1.2 (±0.2) °C resembles unfrozen behavior. Spontaneous freezing subsequent to supercooling coincides with sudden self-induced temperature increases of 0.8 (±0.1) °C and resistivity increases of 2.9 (±1.4) kΩm. As temperature-resistivity gradients of frozen rocks are steep, temperature-referenced ERT with accuracies in the range of 1 °C is technically feasible in frozen rock. Technical design for field measurements in permafrost-affected bedrock was developed from 2005 to 2008 in consecutive measurements at a rock crest in the Swiss Alps (Steintaelli, 3150 m a.s.l., Matter Valley) and in a gallery along a north face in the German/ Austrian Alps (Zugspitze, 2800 m a.s.l.). 2D measurements in the Steintaelli along S-, NE-, NW- and Wfacing rock walls showed that ERT provides information on temporal and spatial patterns of thawing, refreezing, cleftwater flow and permafrost distribution in a decameter scale. Monthly, annual and multiannual data were compared using a time-lapse inversion technique and showed consistent results. Seasonal thaw at the Zugspitze was observed in February and monthly from May to October 2007 with high-resolution ERT (140 electrodes). An error model based on the measured offset of normal-reciprocal measurements was operated to empirically fit inherent error. A smoothness-constrained, error-controlled inversion routine (CRTomo) was applied to gain quantitatively reliable ERT data. Application of temperature-referenced laboratory data is consistent with temperature data observed in the adjacent borehole and with temperature logger data. Calculated temperature changes are in accordance with slow thermal conduction away from the rock surface and subsequent refreezing from the rock face in September/October. Smoothness-constrained, error-controlled inversion was transferred to pseudo-3D measurements collated from five 2D-transects with an offset of 4 m across a NE-SW facing ridge in the Steintaelli. In spite of the enormous topography, ERT transects were capable of resolving permafrost and thaw dynamics at the NE facing slope and along ice-filled crevices as well as disclosing unfrozen rock on the SW-facing rock slope. Consecutive measurements of 2006, 2007 and 2008 provide coherent results in line with temperature logger data.<br /> ERT measurements confirm that aspect is the most important control of permafrost distribution in rock walls, for a given altitude. At 3150 m a.sl., rock permafrost was found in NE-, NW- and E-facing rock walls in the Steintaelli but not in S-facing transects. Multiannual 3D data show that all NE-facing rock slopes still comprise decameter large permafrost bodies, but the 10^4.5 Ωm (31.6 kΩm) line which represents a definite transition to the –2 °C range is not reached in any of the transects apart from the surrounding of ice-filled clefts or at the surface. Semiconductive effects of centimetre to decimetre wide frozen fractures significantly cool ambient bedrock and have a dominant influence on the spatial distribution of permafrost under the crestline. Multiannual 2D data reveal that cleftwater inundation in two fracture systems can effectively prevent a decametre large rockwall from cooling below –1 °C (20 kΩm) during two years with permafrost aggradation (August 2005 to August 2007) in sheltered positions. An adjacent rockwall with similar surface characteristics but no hydraulic interconnectivity cooled significantly below –3 °C (> 60 kΩm) in the same time. Steep, highly dissected rock masses can create local permafrost occurrences of meter size even on SW-facing rock slopes.<br /> Seasonal thaw of rock permafrost occurs much faster than expected. Monthly measurements at the Zugspitze showed that maximum thaw depth in 2007 was already reached in July/August. In May, rapid warming of permafrost rocks with a resistivity increase equivalent to 1.5 °C warming and more was observed along a fracture zone with active cleftwater flows up to 30 m away from the rock face.<br /> Eighteen extensometer transects along the 3D-ERT array in the Steintaelli indicate that rock deformation on the permafrost-affected crest line and in the NE-facing slope is 3-4 times higher than in the non-perennially-frozen SW-facing slope. The velocity of rock displacements in late summer is 20 times higher than in all-season measurements. Velocities along a directly ERT-approved permafrost rock slope respond exponentially to mean air temperature during observation period with an R²; of 0.86. These findings support the hypothesised rapid sliding response to temperature change due to enhanced ice-creep and failure of ice in fractures.