Changing friction at the base of an Alpine glacier

Repeating earthquakes are a global phenomenon of tectonic faults. Multiple ruptures on the same fault asperities lead to nearly identical waveforms characteristic for these seismic events. We identify their microseismic counterparts beneath an Alpine glacier, where basal sliding accounts for a significant amount of ice flow. In contrast to tectonic faults, Alpine glacier beds are subject to large variations in sliding velocity and effective normal stresses. This leads to inter- and sub-seasonal variations in released seismic moment from stick–slip asperities, which we explain with the rate-and-state friction formalism. During summer, numerically modelled effective normal stresses at asperities are three times higher than in winter, which increases the local shear resistance by the same factor. Stronger summer asperities therefore tend to form in bed regions well connected to the efficient subglacial drainage system. Moreover, asperities organise themselves into a state of subcriticality, transferring stresses between each other. We argue that this seismic stick–slip behavior has potentially far-reaching consequences for glacier sliding and in particular for catastrophic failure of unstable ice masses.


Figure S 5: Comparison of waveforms within a cluster.
Vertical seismograms of events within a stick-slip cluster for various seismic stations. All events within the same cluster arise from the same asperity. Each event is plotted in a different color of a rainbow colormap. In addition to the stations used for the data analysis in this paper, the record of a borehole seismometer RA64 close to the stick-slip asperity is shown. Note the mixed P-wave polarities.      [15].

Recurrence Time -Seismic Moment Scaling Velocity Normalization
We derive a theoretical relation for the event recurrence time and the seismic moment from the standard definition of seismic moment [16] where γ is the shear modulus, Δ #$%# is the seismic slip, and is the radius of a circular rupture. Combined with the static stress drop [17,18] and assuming all slip is released seismically Δ ./.01 = Δ #$%# = -1 this results in a scaling [19]: For the case of purely seismic slip, we expect that the event recurrence time is proportional to the inverse of the loading velocity 1 . Thus, if we assume Δτ # = . similar to [20], we can normalize our summer recurrence time measurements by summer loading velocities However, for 1/0* ( ) > 1/0* ( ) (as measured), this only leads to even larger recurrence times in summer when accounting for changes in loading velocity. Smaller summer event recurrence times would be needed though, to adapt our measurements from a scaling of -∝ ! !.CD±!.!+ to a lower power of -∝ ! +/, as observed for microseismic tectonic strike-slip faults [20].

Interpretation of Posterior Inversion Parameter Values
Best fitting values of the inversion parameters from Table S4 give indications about subglacial conditions at basal stick-slip asperities: o E: Approximately 10-15% of the ice overburden pressure for the winter season indicating conditions close to flotation. Approximately threefold value in summer, similar to borehole water pressure measurements.
o : Realistic sliding velocity (~5% of surface velocity) that is in accordance with direct measurements of sliding velocity further down-glacier [21].
o : Friction coefficient posterior only weakly constrained, but higher than for pure ice at the pressure melting point on hard bedrock and till. The friction coefficient posterior is lower than for rock-on-rock sliding, but comparable to debris laden ice on till, which points towards the existence of a frozen fringe over a soft bed at stick-slip asperities [22,23].
o k: Spring constant in the 1D spring-slider model corresponds to the shear modulus in 3D. Posterior values range between shear moduli expected for subglacial till in the MPa range [5,6], and ice in the GPa range [10].
o a-b: The low value of 0.008 indicates only mild rate-weakening as expected for ice at the pressure melting point [24].
o L: Values in the micron range point towards subglacial till with a high clay content [25].
Best fit posterior inversion values do not allow for a strict interpretation of material properties of the sliding interfaces. The stiff spring constant is pointing towards hard bedrock conditions, whereas low friction and characteristic slip distance point towards a soft bed. Borehole camera observations at a stick-slip asperity show fine grained till which reacts on changes in subglacial water pressure, and ice with rock intrusions up to a few meters above the ice-till interface, pointing towards spatially limited soft bed conditions at the asperities [26]. The bedrock of Rhonegletscher can be assessed from the glacier forefield, where the glacier recently retreated. Undulations of bedrock filled with till between humps confirm the existence of confined till patches at the glacier bed [27].