Antarctic Ice Cliffs Calve Faster Than Models, Then Stop

Only 5 per cent of the Antarctic ice cliffs mapped in a new survey stand taller than 52.2 metres, according to a study published in Science on 15 October 2026. The figure matters because glaciologists had proposed roughly 100 metres as the height at which an exposed cliff of solid ice collapses under its own weight, exposing a taller cliff inland that collapses in turn. Very few cliffs in the survey came close to that threshold.

Caroline Needell of the MIT-WHOI Joint Program and two co-authors, Catherine C. Walker of the Woods Hole Oceanographic Institution and Jeremy N. Bassis of the University of Michigan, combined airborne and satellite laser altimetry gathered between 2003 and 2023 to locate and measure cliffs along the margins of both Antarctica and Greenland. In Greenland the 95th percentile was 76.6 metres. The two ice sheets differ in almost every other respect, yet their cliffs came out at similar average and maximum heights, with the tallest concentrated where glaciers run fastest towards the sea.

Four glaciers that retreated and then did not

For the behaviour rather than the inventory, the team used NASA's Ice, Cloud and Land Elevation Satellite-2, which carries a laser altimeter over the polar regions. Data from 2019 to 2023 gave measurements every three months of more than 15,000 Antarctic ice cliffs, MIT said in its account of the work. The average cliff was about 30 metres. The tallest found was 127 metres.

The authors then narrowed to 16 glaciers whose cliffs passed 90 metres and identified four with both detailed repeat observations and measurable retreat over the study window. Those four calved more rapidly than models of marine ice cliff instability assumed, and then they stopped.

What governs the stopping, on the authors' reading, is the gradient of ice thickness inland. Where thickness changes gradually, the ice never builds the height needed to sustain a cascade. The land beneath does similar work: where bedrock rises, the ice over it is thinner than its surface slope suggests, acting as a speed bump. Two features in front of the cliff also showed up. Deeper water buoys a cliff and pushes back against it, and so does thicker melange, the slush of broken bergs and sea ice at a glacier's front, which Needell said taller cliffs carry more of and which could be a built-in stabilising effect.

What it says about Thwaites

The authors apply the result to Thwaites Glacier in West Antarctica, the catchment that has driven most of the alarm about rapid sea level rise, and argue it would be unlikely to undergo immediate instability-driven retreat if a shelf failure exposed tall cliffs at its present grounding line, according to the AAAS summary of the paper. NASA, the US National Science Foundation and the DOMINOS project, part of the International Thwaites Glacier Collaboration, funded the work.

The honest reading is that this narrows a tail risk rather than removing one. Marine ice cliff instability has never been directly observed, which is why it has sat in sea level projections as a plausible mechanism rather than a measured rate, and a four-year record of cliffs at today's grounding lines cannot say how ice behaves at heights nobody has yet seen. The faster-than-modelled calving points the other way, towards more ice loss, and it is the pairing with short-lived retreat that softens it.

Cliff height has been the single variable carrying this threshold in predictive models, a narrowness Parallax Nexus has noted in climate projections before. After 15 October it is three or four.