5 Conclusions
Despite Crane Glacier being invoked in many conversations of unstable ice cliff failure, the limited data collected during the LBIS collapse never directly measured a cliff failure event. Previous analysis of Crane Glacier’s behavior from 2002-2004 is predicated on an inaccurate 2002 grounding line position. Our analysis shows that retreat in 2002 and 2003 occurred in a remnant, floating ice shelf, but in the two years following LBIS collapse, calving rates accelerated, consistent with a positive feedback and geometric instability. This includes the final phase of retreat, which may have occurred by the failure of a grounded ice cliff with an initial height of 111 m, with a minimum calving rate of ~8.50 km/a.
The glacier terminus retreated to a configuration where calving rates slowed, and the system maintained a constant terminus position for 7 years with no further retreat. Meanwhile, the mass dis-equilibrium induced by terminus changes drove widespread thinning in the glacier interior. Finally, as sea-ice filled the fjord in 2011, Crane Glacier’s terminus advanced and an ice shelf was re-established, supporting a thickening and slowing (but overall thinner than pre-LBIS collapse) glacier profile.
Given a maximum terminal cliff height of only 111 m, available process models indicate that terminus stresses would not exceed the failure threshold of undamaged ice. However, using model values consistent with damaged ice, cliff failure at Crane Glacier becomes consistent with the process models discussed here. This highlights the importance of better understanding ice shelf damage in projecting future ice sheet behavior – if high damage is plausible for Crane Glacier, it means that cliff failure likely did govern the final stage of glacier retreat, and future ice sheet margins may be more susceptible to fracture than anticipated.
The rapid nature of ice shelf break-up makes observing ice cliff failure an inherently challenging problem. Brittle processes are fast and therefore require high temporal sampling to observe, best accomplished in situ. Measuring cliff failure processes will require having field sensors in the right place at the right time, something not easily done at scale. This means that observational evaluation of ice cliff models must, in part, rely on the temporally limited (and therefore at times, ambiguous) remote sensing record. While studies like this one cannot end the debate over model realism and the role of cliff failure in the future evolution of the Antarctic Ice Sheet, they contribute to the larger corpus of evidence required to justify any novel treatment of cliffs in continental scale models of Antarctica and Greenland.