Durations and propagation patterns of ice sheet instability events

Continued atmospheric and ocean warming places parts of the West Antarctic Ice Sheet at risk for collapse through accelerated ice flow and grounding line retreat over reversed bed slopes. However, understanding of the speed and duration of ice sheet instability events remains incomplete, limiting our ability to include these events in sea level rise projections. Here, we use a first-order, empirical approach, exploring past instability events in the Fennoscandian (FIS) and Laurentide (LIS) ice sheets to establish a relationship between catchment size and the duration of instability events. We also examine how instabilities propagate through ice sheet catchments, and how this propagation is controlled by topography and existing flow organisation at the onset of an event. We find that the fastest documented paleo-collapses involved streaming or surging in corridors that are wide compared to their length, and in which fast flow did not resume after the event. Distributed ice stream networks, in which narrow ice streams were intertwined with slow-flow interstream ridges, are not represented among the fastest documented events. For the FIS and LIS, there is geological evidence for instability events covering areas of ∼100,000 km2, with durations between 100 and 300 yr. Comparison of the spatial patterns and topographic contexts of Lateglacial collapse events in former Northern Hemisphere ice sheets and the current WAIS suggest that only a minor part of the WAIS area may be at risk for unimpeded collapse, and that negative feedbacks will likely slow or halt ice drawdown in remaining areas. The Pine Island Glacier (PIG) and Thwaites Glacier (TG) catchments in West Antarctica are likely to respond in very different ways to possible further grounding line retreat. The PIG may experience a minor collapse over its main trunk, but the bed topography favours a less dramatic retreat thereafter. The TG is probably not as close to a threshold as PIG, but once efficient drainage has progressed inwards to reach the Bentley Subglacial Basin (BSB) and Bentley Subglacial Trench (BST), a full collapse of the area may occur. The likely time perspective for a BSB–BST collapse is the time required for 100–200 km of grounding line retreat in the TG system plus 100–300 years for an actual collapse event.