Riftquakes: Recording and Modeling Seismic Signals of Rifting at Pine Island Glacier

Abstract:

Nearly 50% of Antarctic ice discharge into the ocean occurs via iceberg calving (Depoorter et al, 2013). Large tabular icebergs calve from ice shelves along large fractures called rifts, but the physics of rifting are poorly understood. How fast does rift propagation occur? What portion of the ice shelf thickness fractures during rift propagation? We investigate these questions using data from seismometers and GPS sensors deployed on Pine Island Glacier ice shelf (PIG) from January 2012 to December 2013 surrounding the calving of iceberg B31, which exceeded 700 km² in area and calved in November 2013 along a large rift. Using TerraSAR-X imagery, we identify a large 7 km rifting event between May 8 and May 11, 2012. We identify the seismic signal generated by this rifting event, which occurred on May 9, 2012. The signal is broadband, containing energy at frequencies higher than 1 Hz and lower than 0.01 Hz, and exhibits dispersion characterized by high frequencies arriving before low frequencies. We use features of the May 9 “riftquake” to detect similar events, which we classify using K-shape clustering. We hypothesize that the observed signals are flexural gravity waves generated by a bending moment applied to the ice shelf during fracture. To test this hypothesis, we model the ice shelf as a dynamic beam supported by an inviscid, incompressible ocean. We find that the model reproduces observed riftquake waveforms when forced with a point moment. We then use a Markov Chain Monte Carlo inversion to model representative waveforms from each cluster of observed events. The inversion reveals that source durations on the order of seconds have the highest likelihood of explaining observed riftquake waveforms, suggesting that rifting occurs on elastic timescales. The inversion also reveals that fractures spanning only a small fraction of the ice shelf thickness have the highest likelihood of explaining observed riftquake waveforms, suggesting that large-scale rift propagation is the result of many small fracture events. This means that the surface expression of large rifting events like the May 9 riftquake may represent fracture of only the top portion of the ice shelf, and that the majority of the ice shelf thickness had already fractured via basal crevassing.