Hektoria Glacier Collapse: Anatomy of a Two-Phase Retreat

What this research found

A short policy brief, synthesising published observations rather than producing new ones, on what the 2018 to 2024 collapse of Hektoria Glacier on the eastern Antarctic Peninsula means for coastal planning. It separates the 2022 breakup of the floating tongue, which does not raise sea level, from the 2023 to 2024 runaway of the grounded trunk, which does, then benchmarks three neighbouring glaciers that share the same geometry. The headline is timescale rather than volume: an ice plain on a landward-deepening bed went from stable to fully ungrounded in roughly two years.

• The second phase is the sea-level event. Between 2023 and 2024 Hektoria's grounded trunk accelerated from about 1.2 to 40 metres per day, a 33-fold speed-up, thinned at roughly 14 metres per year, and its grounding line retreated about 25 kilometres inland in a single year.
• The 2022 loss of the floating tongue was hydrostatically neutral. That ice was already displacing its own weight in seawater, so its calving raised no sea level, and the brief flags conflating the two phases in public messaging as a risk to public trust.
• The bed slopes the wrong way, deepening inland from −600 metres at the 2018 grounding line to −780 metres by 2024, which removes the braking force that would normally arrest a retreat. Warm Circumpolar Deep Water had already thinned the tongue at 2 to 10 metres per year over 2018 to 2021.
• Three neighbouring glaciers carry the same vulnerability signature of floating tongue, retrograde bed, and ocean-driven melt. Jorum has sped up 11-fold from 1.1 to 12 metres per day on a bed deepening to −700 metres, Crane sits on the deepest bed at −900 metres after losing its tongue in 2002, and Cadman on the western Peninsula reproduces the mechanism independently with a doubling from 1.5 to 3.0 metres per day.
• The volumes are regional rather than global. The whole Larsen B catchment holds roughly 18 millimetres of global sea-level equivalent in grounded ice, and about 361.8 gigatonnes of land ice must be lost per millimetre of sea-level rise. What is unprecedented is the pace: two years, against the decades to centuries assumed in earlier risk models.

How it was done

Values throughout are literature-derived rather than newly measured. The retreat history comes from the 2024 Christie and colleagues analysis, bed elevations from BedMachine Antarctica version 3, ice velocities from Sentinel-1 synthetic-aperture radar as served through NASA's ITS_LIVE products, ocean-melt forcing from Cook and Naughten, and continental mass balance from IMBIE. These were assembled into a labelled two-phase retreat timeline, a schematic of grounding-line retreat on a retrograde bed, and a table scoring four Peninsula glaciers on bed depth at the grounding line and on frontal speed-up between 2018 and 2024. The result is a two-page brief closing on four concrete policy actions, among them continuous monitoring of the analogue glaciers and keeping the two collapse phases distinct in public communication.

Data sources

• Christie et al., Geophysical Research Letters 51:e2024GL110592 (2024) — two-decade evolution of Hektoria and its record-setting retreat
• BedMachine Antarctica version 3 (Morlighem et al., Nature Geoscience 13:132, 2020) — bed elevation
• Copernicus Sentinel-1 synthetic-aperture radar and NASA ITS_LIVE regional ice-velocity products (NSIDC MEaSUREs version 1)
• IMBIE mass balance of the Greenland and Antarctic ice sheets, 1992 to 2020 (Otosaka et al., 2023)
• Naughten et al., Nature Climate Change 13:1003 (2023) and Cook et al., Science 353:283 (2016) — ocean forcing of glacier retreat
• Rott et al., The Cryosphere 5:125 (2011) — glacier imbalance after Larsen B ice-shelf disintegration
• Wallis et al., Nature Communications 14:7535 (2023) — Cadman Glacier, western Antarctic Peninsula