A new UC Irvine and NASA JPL study finds that Thwaites Glacier (West Antarctica) could contribute up to ~30 mm (1.2 inches) of global sea- level rise by 2100, roughly twice prior estimates, even under constant recent climate conditions (no additional warming).

The research, led by Mattia Poinelli (then at UC Irvine, now UCLA) with Eric Rignot (UC Irvine) and colleagues from NASA’s Jet Propulsion Laboratory and other institutions, was published in Nature Communications (October 2026). The paper is titled “Sea- level contribution from Thwaites Glacier doubles in an observation- constrained ice- sheet model.”

Key advance: modeling a “grounding zone” instead of a simple “grounding line”

Traditional ice- sheet models treat the transition from grounded ice to floating ice shelf as a narrow, discrete grounding line. Satellite radar data (including from the ICEYE constellation) show a broader grounding zone, often several kilometers wide, where warm, pressurized seawater repeatedly intrudes under ice that models previously treated as firmly grounded. This causes vigorous basal melting during tidal cycles.

By incorporating a grounding zone (typically 2- 5 and more km wide) into the Ice- sheet and Sea- level System Model (ISSM) and calibrating it against more than 20 years of observations (ice speed, thinning, and grounding-line position from the mid-1990s to 2023), the model much better reproduces the glacier’s rapid historical retreat, thinning, and acceleration. Without the zone, historical mass loss is under-reproduced by a factor of about four.

Projections

Under climate conditions held at the recent average:

• No grounding zone: ~8.3 mm by 2100.
• 1 km zone: ~14.3 mm (near the ISMIP6 multi- model ensemble mean of ~15.2 mm).
• 3 km zone (statistically strong fit): ~19.4 mm.
• 5 km zone (consistent with current radar observations): ~30.1 ± ~6 mm, about twice the ensemble average.

Adding a realistic grounding zone increases the projected contribution substantially (more than extra ocean warming of 0.5- 0.7 °C in the model). The main trunk of Thwaites is currently stabilized by the Mouginot Ridge; simulations with wider zones give roughly a 60% probability of unpinning from it by 2100 (as early as the late 2030s in some runs), after which retreat can continue more readily over a deepening seafloor.

Context and caveats

Thwaites (sometimes called the “Doomsday Glacier”) holds enough ice for ~65 cm of global sea- level rise if fully melted and sits in a sector responsible for roughly one-third of Antarctic ice loss.

The new 30 mm figure is for Thwaites alone by 2100 under no further climate change; it is a lower- bound scenario that does not include additional warming, calving, or evolving zone width. Full collapse is not projected this century.⁠

Rignot noted that earlier projections struggled to match observed mass- loss rates due to missing physics; the updated model matches the record and yields higher projections, implying greater coastal flooding risk in coming decades than previously anticipated.

The authors recommend incorporating actively melting grounding zones into broader ice- sheet models and collecting more oceanographic data beneath the ice shelf.⁠

This builds on earlier 2024 UC Irvine- led satellite observations of kilometer- scale seawater intrusions under Thwaites.

_____________________________________________________________________________________

The grounding zone is a transitional band (often several kilometers wide) at the edge of a marine-terminating glacier where ice shifts between resting on the seabed and floating on the ocean. 

It replaces the simpler traditional concept of a sharp “grounding line”.

Traditional grounding line vs. observed grounding zone

In classic ice- sheet models, the grounding line is treated as a narrow, discrete boundary: upstream of it, ice is firmly grounded on bedrock (supported by the solid Earth), and downstream it floats as an ice shelf. Ocean melting is typically assumed to drop to zero at or just beyond that line.

Satellite radar observations (especially high- resolution synthetic aperture radar from constellations like ICEYE) show a broader, dynamic grounding zone. At Thwaites Glacier, this zone is typically 2- 6 km (or more) wide along the main trunk, depending on the local bed slope. In shallower- slope areas it can reach ~6 km; on steeper flanks it is narrower (~2 km). Within this zone the ice is not permanently grounded or floating, it migrates and flexes with the tides.

How the mechanics work

1. Tidal flexure and lift- off
   Ocean tides raise and lower the water level. Because ice is elastic and the bed is not perfectly flat, the floating ice shelf transmits this pressure inland. The ice sheet periodically lifts a few centimeters to tens of centimeters off the seabed over a multi- kilometer distance. When the tide falls, the ice settles again. This creates a migrating “hinge” or transition zone rather than a fixed line.
2. Seawater intrusion
   As the ice lifts, pressurized seawater from the ocean cavity under the ice shelf is forced inland beneath the previously grounded ice. Intrusions can be only 5- 10 cm thick but can extend several kilometers upstream of the mean grounding position, especially into bed depressions or channels beyond stabilizing ridges (such as the Mouginot Ridge at Thwaites). The water is relatively warm (often several degrees above the in- situ freezing point) because it originates from Circumpolar Deep Water or modified versions of it that have been steered toward the Antarctic coast by climate- altered currents.
3. Vigorous basal melting
   The intruding seawater contacts the underside of the ice over a large area. Melting rates in this zone can reach tens to over 100 meters of ice thickness per year, far higher than typical rates under fully grounded ice or even under many floating shelves. The meltwater and tidal pumping help flush the zone, continually renewing the heat supply. This basal melting thins the ice from below, reduces the effective weight of the ice column, and lowers basal friction (the ice is no longer fully “stuck” to the bed).
4. Feedback on ice flow and stability
   Thinning and reduced basal resistance allow the ice to accelerate and stretch. Faster flow draws more ice from the interior. Because many West Antarctic beds slope downward inland (retrograde), once the grounding zone retreats past a stabilizing topographic high (a ridge or bump), the glacier can un- pin and continue retreating more readily. The wider the active melting zone, the easier and earlier this detachment becomes.

Why this matters for projections

Most earlier models that treated the transition as a thin line under- predicted the observed rates of thinning, acceleration, and grounding- line retreat at Thwaites. When models incorporate a multi- kilometer actively melting grounding zone and are calibrated to decades of satellite observations, they reproduce the historical changes far better and project roughly twice as much sea- level contribution by 2100 (up to ~30 mm from Thwaites alone under constant recent climate forcing).⁠

In short, the grounding zone acts as a dynamic “leaky hinge” where tidal pumping delivers warm ocean water deep under the ice sheet, driving enhanced melting, reducing stability, and accelerating mass loss in ways a simple grounding- line assumption cannot capture.

_____________________________________________________________________________________

Sea-level contribution from Thwaites Glacier doubles in an observation-constrained ice-sheet model

Abstract summary

Thwaites Glacier is a major contributor to Antarctic mass loss. Most models treat the grounded-to-floating transition as a sharp boundary and ignore kilometer-scale seawater intrusions that create an actively melting grounding zone. Using a calibrated Ice-sheet and Sea-level System Model (ISSM), the authors show that actively melting grounding zones 2–5 km wide are required to reproduce nearly three decades (1996–2023) of observed retreat, thinning, and acceleration. Projecting the calibrated model forward under constant recent climate forcing yields up to 30 ± 6 mm of global mean sea-level rise by 2100—approximately double previous projections—because grounding-zone processes control detachment from stabilizing seafloor ridges (unpinning possible as early as 2038).

Key scientific points

Problem with traditional models

Standard ice- sheet models struggle to simultaneously match the observed rapid retreat (~25 km on the main trunk since 1996 at ~800 m yr⁻¹), thinning, and acceleration of Thwaites. They typically treat the grounding line as a discrete boundary and force melt rates toward zero there. Satellite data (especially ICEYE SAR) show a dynamic grounding zone several kilometers wide where warm modified Circumpolar Deep Water repeatedly intrudes under ice, lifts it during tidal cycles, and drives vigorous melting.

Model setup

• Model: ISSM with a regularized Coulomb friction law.
• Actively melting grounding zone of length L is explicitly parameterized; melt is allowed inland of the flotation line.
• Initialized to a mid-1990s state.
• Calibrated via Bayesian methods against three independent observational records (1996- 2023): surface velocity, ice thickness change, and grounding- line position.
• A small, localized bed- elevation correction (within map uncertainty) was applied to satisfy flotation consistency with the 1996 grounding line.

Historical reconstruction results

• Simulations without a grounding zone require unrealistically high melt rates (~150 m yr⁻¹) and still under-predict later discharge, thinning, acceleration, and retreat.
• Grounding zones of 2- 5 km with melt rates of 90-120 m yr⁻¹ (consistent with satellite and ocean-model estimates) successfully reproduce concurrent retreat of both the main trunk and the eastern ice shelf, the observed ~32 m of grounded-ice thinning, and the acceleration.
• Simulations lacking a grounding zone are four times less likely to match the historical record.

Future projections (constant 1996- 2023 climate forcing, a lower- bound scenario)

• The main trunk is currently stabilized by the Mouginot Ridge (~950 m below sea level).
• Probability of unpinning from this ridge by 2100:
  • No zone or 1 km zone → essentially 0% (stays pinned).
  • 2- 3 km zone → ~45%.
  • 4- 6 km zone (consistent with recent observations) → ~60%, with unpinning as early as 2038.
• Once unpinned, the glacier retreats more readily across retrograde seafloor toward the South Ridge.
• Sea- level contribution by 2100 (melt rate fixed near the observationally preferred ~110 m yr⁻¹):
  • No grounding zone: 8.3 ± 3.0 mm
  • 1 km zone: 14.3 ± 3.1 mm (near the ISMIP6 ensemble mean of 15.2 mm)
  • 3 km zone: 19.4 ± 5.2 mm
  • 5 km zone: 30.1 ± 5.7 mm (twice the ISMIP6 mean)

Simply including a grounding zone raises the contribution by ~51%; each extra kilometer of zone width changes it by ~22% on average. Moderate additional ocean warming (+0.5- 0.7 °C) adds only ~15% more loss, so grounding-zone extent is the stronger control under the tested conditions.

Discussion and limitations

The authors emphasize that representing actively melting grounding zones, combined with observational calibration, substantially improves both hindcast skill and projection reliability. Limitations include:

• Fixed (time- averaged) grounding- zone length rather than a prognostic, evolving one.
• No explicit reduction in basal friction from seawater lubrication.
• No evolving sub- shelf cavity geometry or full ocean circulation.
• No calving or interannual ocean variability.
• The projections are deliberately a lower- bound scenario (no further climate warming).

They conclude that actively melting grounding zones should be incorporated into regional and continental-scale ice- sheet models and call for more oceanographic measurements beneath the ice shelf.

The full paper (including methods, figures, and supplementary material) is available at the link you provided; a PDF download is offered on the page.

Journal: Nature Communications on 12 October 2026 (open access).

DOI: 10.1038/s41467-026-77976-1

Provided: University of California, Irvine 

Authors: Mattia Poinelli,
Eric Rignot,
Hélène Seroussi,
Joshua Cuzzone,
Tyler Pelle &
Eric Larour