Thwaites Glacier sea level projection doubles to 30 mm by 2100 in new ice sheet model

Thwaites Glacier is the broad, fast moving ice stream that drains a vast portion of the West Antarctic Ice Sheet into the Amundsen Sea, and it is the most closely watched single glacier on Earth. Its sea level equivalent is about 65 centimeters, meaning the ice it holds today would raise global mean sea level by that amount if it all reached the ocean. The Amundsen Sea Embayment sector where Thwaites sits already accounts for roughly one third of the mass loss from the Antarctic Ice Sheet.

For decades, ice sheet models have treated the boundary between ice resting on the seabed and ice floating on the ocean as a sharp line, known as the grounding line, and retreat in that framework happens when the line migrates inland. The new study argues that this picture is wrong in a way that matters: the transition is a grounding zone several kilometers wide, and warm seawater moves into it on tidal cycles, lifting and melting ice from below that models had assumed was firmly grounded.

The study was published in Nature Communications on October 12, 2026 under the title Sea-level contribution from Thwaites Glacier doubles in an observation-constrained ice-sheet model. It appears as volume 17, Article number 10364, and is open access. Its central result is that Thwaites will contribute up to 30 ± 6 millimeters, about 1.2 inches, to global mean sea level by 2100, approximately double previous projections, and that this happens even without any additional climate warming beyond conditions already observed.

The work was led by glaciologists at the University of California, Irvine and NASA's Jet Propulsion Laboratory. Lead author Mattia Poinelli, a UC Irvine postdoctoral scholar in Earth system science when the study was conducted and now a UCLA project scientist and NASA JPL research affiliate, described a process of progressive detachment from the ridges that currently slow the glacier down.

Key Facts

phys.org reported on October 13, 2026 that the model projects Thwaites will add up to 30 millimeters (1.2 inches) of global sea level rise by 2100, twice prior estimates, and that the simulations hold future climate forcing constant so that the influence of grounding zone processes can be isolated. The outlet quoted Poinelli explaining that the Thwaites main trunk is being temporarily stabilized by the Mouginot Ridge, and that once that ridge is unpinned the glacier will continually retreat across a predominantly retrograde seafloor toward its next topographic high, the South Ridge.

Nature Communications reported on October 12, 2026 that actively melting grounding zones are required to reproduce nearly three decades of observed retreat, thinning and acceleration of Thwaites in a calibrated ice sheet model, and that a successful reconstruction occurs only when grounding zones are represented as 2 to 5 kilometer wide regions. The glacier has experienced about 25 kilometers of grounding line retreat in its western, fastest flowing sector since at least the mid 1990s. Extending the calibrated model forward produces grounding zone driven unpinning as early as 2038 and up to 30 ± 6 millimeters of global mean sea level rise by 2100.

Scienmag reported on October 13, 2026 that the team used the Ice sheet and Sea level System Model, whose mesh resolves features as small as 500 meters near the grounding line, running an ensemble of 105 simulations spanning melt rates of 10 to 150 meters per year and grounding zone lengths of zero to six kilometers. A Bayesian calibration with 10,000 samples against four observables found best fits centered on grounding zones 2 to 5 kilometers wide and melt rates of 90 to 120 meters per year. Simulations lacking a grounding zone were four times less likely to reproduce the historical trend, and with no grounding zone represented the model yields just 8.3 ± 3.0 millimeters by 2100.

The observational backbone of the study comes from synthetic aperture radar readings collected by the Finnish ICEYE satellite constellation, which detected kilometer scale grounding zone fluctuations. As tides rise and fall, the grounding line migrates and warm modified Circumpolar Deep Water penetrates several kilometers beneath nominally grounded ice, episodically lifting and melting it from below. Pakistan Tribune reported on October 13, 2026 that the team compared simulations with records of retreat, thinning and acceleration between 1996 and 2023, and that grounding zones several kilometers wide helped reproduce those changes together in the same model run.

The forecast of detachment rests on probability rather than certainty. phys.org reported on October 13, 2026 that simulations factoring in a 4 to 6 kilometer grounding zone predict a 60 percent probability of unpinning the Mouginot Ridge by the end of the century, and that at 5 to 6 kilometers the glacier can unpin as early as 2038. Pakistan Tribune reported on October 13, 2026 that this is an outcome within the model ensemble rather than a forecast that detachment will definitely occur that year.

Analysis

What this really means is that a large part of the uncertainty in sea level projections for Thwaites was never about how fast the atmosphere warms. It was about whether models were resolving the physics at the glacier's edge. The comparison in this study is stark. The same ice sheet model, run without a grounding zone, produced 8.3 ± 3.0 millimeters of sea level contribution by 2100. Run with a physically grounded, observationally constrained grounding zone of 2 to 5 kilometers, it produced about 30 ± 6 millimeters. That gap is not a difference in emissions scenarios. It is a difference in whether the model lets the ocean reach ice it had assumed was protected.

The bigger picture here is that Thwaites is not being destabilized only at its front. The grounding zone mechanism means the ocean works inland, episodically, on tidal timescales, beneath ice that appears from the surface to be resting on solid ground. Satellite radar from the ICEYE constellation made that visible at kilometer resolution, and the calibration exercise shows that the historical record of retreat, thinning and acceleration cannot be reconstructed without it. About 25 kilometers of grounding line retreat in the western, fastest flowing sector since the mid 1990s is the fingerprint the model had to match, and only the grounding zone version matched it.

The emphasis on the Mouginot Ridge gives the projection a topographic logic that is easy to follow. A ridge pins the glacier in place. Once grounding zone processes unpin it, the bed behind the ridge slopes in the retrograde direction, deepening inland, so retreat toward the next high point, the South Ridge, has no natural resting place along the way. The 60 percent probability attached to unpinning the Mouginot Ridge by 2100, and the model ensemble's ability to reach detachment as early as 2038 in the widest grounding zone cases, are the numbers that carry the result. It is a projection from a calibrated model rather than a measurement of water already in the ocean.

Why It Matters

Sea level projections feed directly into coastal planning, infrastructure standards and adaptation budgets, and a doubling of the Thwaites contribution changes the input to those calculations rather than the political conversation around emissions. Because the study holds climate forcing constant, part of the committed sea level rise from Thwaites is already embedded in the climate state we have, and it does not depend on how quickly emissions fall over the coming decades.

The finding also reorders the research agenda. If grounding zone processes, which depend on tidal cycles and on the inland penetration of warm modified Circumpolar Deep Water, govern when Thwaites detaches from stabilizing ridges, then observing those processes becomes as important as measuring surface melt or air temperature. The 2024 UC Irvine led investigation that used satellite radar to document extensive seawater intrusion beneath Thwaites is the direct ancestor of this modeling work, and the sequence shows how a single observation can move a process from a footnote to the center of a projection.

Next Up

The near term question raised by the paper is whether the grounding zone representation can be exported to other glaciers and to the wider Antarctic Ice Sheet, since the calibration here is specific to Thwaites and the grounding zone widths that fit best are 2 to 5 kilometers rather than a universal constant. The 2038 unpinning date that appears in the widest grounding zone simulations is an outcome inside a model ensemble, not a forecast that detachment will definitely occur that year, and monitoring the Mouginot Ridge is the obvious way to test it.

Closer observation of tidal grounding zone migration, expanded satellite radar coverage of the Amundsen Sea Embayment and continued comparison against the 1996 to 2023 record will determine whether the 30 ± 6 millimeter figure holds, grows or is revised downward. The 60 percent probability of unpinning the Mouginot Ridge by 2100 is the number to watch, because it is the hinge on which the doubled sea level contribution turns.