📄 earth_science

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

By incorporating actively melting, kilometer-scale grounding zones into a calibrated ice-sheet model, researchers demonstrate that Thwaites Glacier's retreat is driven by these previously neglected processes, potentially doubling its projected sea-level contribution by 2100 compared to earlier estimates.

Original authors: Mattia Poinelli, Eric Rignot, Helene Seroussi, Josh Cuzzone, Tyler Pelle, Eric Larour

Published 2026-06-25

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Mattia Poinelli, Eric Rignot, Helene Seroussi, Josh Cuzzone, Tyler Pelle, Eric Larour

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine Thwaites Glacier as a massive, slow-moving river of ice in Antarctica, acting like a giant cork holding back a vast amount of frozen water. Scientists have long known this "cork" is getting loose, but they've been struggling to figure out exactly how fast it will pop out and how much sea level will rise as a result.

This paper presents a new way of looking at the problem that suggests the situation is much more urgent than previously thought. Here is the breakdown in simple terms:

The Old View vs. The New Discovery

For a long time, scientists modeled the edge of the glacier (where the ice touches the ocean) as a sharp, clean line—like a wall. On one side is solid ground ice; on the other is floating ice. They assumed the melting happened mostly under the floating part.

However, new satellite observations revealed something surprising: The "wall" isn't a wall at all. It's actually a wide, muddy zone where seawater sneaks several kilometers underneath the grounded ice.

Think of it like this:

The Old Model: Imagine a boat docked at a pier. The water only touches the hull once the boat is fully floating.
The New Reality: It's more like the water is seeping under the dock itself, lifting the wood and melting the supports while the boat is still technically "on land."

The Experiment: Tuning the Model

The researchers built a sophisticated computer simulation of Thwaites Glacier. They tried to make the model match what actually happened over the last 30 years (how fast the ice retreated, how much it thinned, and how fast it sped up).

The Failure: When they treated the edge as a sharp line (the old way), the model failed. To make it match reality, they had to invent impossible melting rates that were too high to be real.
The Success: When they added the "wide zone" where water intrudes under the ice (2 to 5 kilometers wide) and applied a realistic melting rate of about 110 meters per year inside that zone, the model worked perfectly. It matched 30 years of satellite data almost exactly.

The Big Prediction: Doubling the Risk

Once they had a model that worked for the past, they used it to predict the future (up to the year 2100), assuming the climate stays exactly as it is today (no extra warming).

Here is the startling result:

Previous Projections: Earlier models suggested Thwaites would contribute about 15 millimeters to global sea-level rise by 2100.
This Study's Projection: With the new "wide melting zone" included, the contribution jumps to 30 millimeters.

The Analogy: Imagine you are driving a car. Previous models said you would arrive at your destination in 10 minutes. This new study says, "Actually, we found a shortcut in the road that speeds up the engine. You're going to arrive in 5 minutes." The paper claims the sea-level rise contribution doubles simply because we finally accounted for the water sneaking under the ice.

The "Peg" That Might Break

The glacier is currently held in place by a rocky underwater hill called the "Mouginot Ridge." Think of this ridge as a peg holding a heavy rug in place.

If the melting zone is narrow, the rug might stay pinned for a while.
But with the wide, active melting zone, the "peg" is being dissolved much faster.

The study predicts that with the new model, there is a high chance the glacier will detach from this stabilizing ridge as early as 2038 (just 12 years from now). Once that peg breaks, the glacier can slide backward rapidly across a deep, sloping ocean floor, leading to massive ice loss.

The Bottom Line

The paper argues that we have been underestimating the danger because we didn't realize how far the ocean water reaches under the ice. By fixing the model to include this "sneaky" melting zone, the projected sea-level rise from Thwaites Glacier doubles, even without any additional global warming. The authors conclude that future models must include this wide melting zone to get accurate predictions.

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

Problem Statement
Thwaites Glacier in West Antarctica is a primary contributor to Antarctic mass loss and a critical source of uncertainty in future sea-level rise projections. Its rapid retreat is driven by ocean-induced melting at the grounding line, where grounded ice transitions to floating ice shelves. While previous studies have highlighted the role of warm Circumpolar Deep Water (CDW) in eroding the ice base, existing ice-sheet models struggle to simultaneously reproduce historical trends of glacier retreat, thinning, and acceleration. A key deficiency identified is the treatment of the grounding line as a sharp, discrete boundary between grounded and floating ice. This approach neglects kilometer-scale seawater intrusions beneath grounded ice, recently observed via satellite, which create a dynamic "grounding zone" characterized by active melting. Furthermore, standard melt parameterizations often force melt rates to zero at the grounding line, contradicting observations that suggest melting is substantial and potentially dominant within this zone.

Methodology
The authors utilized the Ice-sheet and Sea-level System Model (ISSM) to simulate the evolution of Thwaites Glacier from 1996 to 2023 and project its behavior through 2100. The study introduced a novel parameterization to explicitly represent an "actively melting grounding zone" extending inland from the flotation line. In this framework, the grounding zone is defined as a region of length $L$ where basal melt rates decrease linearly from a maximum value at the grounding line to zero at the intrusion distance.

The model was initialized to represent the mid-1990s state, incorporating a localized +100 m correction to bed topography in floating ice regions to ensure consistency with the observed 1996 grounding line position. The study employed a Bayesian calibration approach to evaluate the model against three independent observational records: surface velocity, ice thickness, and grounding line position. The calibration varied two key parameters: the submarine melt rate ( $\dot{m}$ ) and the grounding zone length ( $L$ ). A hindcast ensemble of 105 simulations was performed to identify the joint parameter space that minimizes misfit with historical trends. Following calibration, the best-performing configurations were extended to 2100 under a constant climate forcing baseline (averaged from 1996–2023) to isolate the impact of grounding-zone physics from future ocean warming uncertainties.

Key Results

Reconstruction of Historical Trends: The model successfully reproduced the observed retreat, thinning, and acceleration of Thwaites Glacier over the 25-year period only when the grounding zone was represented as a 2–5 km wide region with melt rates of approximately 110 m yr⁻¹. Simulations treating the grounding line as a discrete boundary required unrealistically high melt rates (150 m yr⁻¹) and still failed to capture the magnitude of thinning and acceleration observed in the mid-2000s.
Parameter Trade-offs: Bayesian analysis revealed a compensatory relationship between grounding zone length and melt rate. However, simulations without a grounding zone were found to be four times less likely to reproduce historical trends than those with a multi-kilometer grounding zone.
Future Projections and Unpinning: Under constant climate forcing, the length of the grounding zone emerged as the dominant predictor of whether the glacier would detach from the stabilizing Mouginot Ridge.
- Models with no grounding zone or a narrow 1-km zone did not retreat beyond the ridge by 2100.
- Models with a 2–3 km grounding zone predicted a 45% probability of unpinning by 2100.
- Models consistent with recent satellite observations (4–6 km grounding zone) showed a ~60% probability of crossing the ridge by 2100, with unpinning potentially occurring as early as 2038.
Sea-Level Contribution: The study projects that a grounding zone length of 5±1 km (consistent with recent high-resolution observations) results in a global mean sea-level contribution of 30.1 ± 5.7 mm by 2100. This is approximately double the mean projection of the Ice Sheet Model Intercomparison Project for CMIP6 (ISMIP6) under the same constant forcing scenario (15.2 ± 21.8 mm) and represents a 130% increase over the ISMIP6 mean.

Significance and Claims
The paper claims that explicitly resolving kilometer-scale, actively melting grounding zones fundamentally alters both hindcast and forecast modeling skills for Thwaites Glacier. By calibrating the model against multiple independent observational constraints, the authors demonstrate that grounding-zone processes govern the detachment from stabilizing topographic ridges, a mechanism previously underrepresented in models.

The authors assert that their findings establish a strong benchmark for projection skill, showing that the extent of the grounding zone exerts a stronger control on retreat and unpinning than moderate variations in thermal forcing. Consequently, the study argues that actively melting grounding zones must be systematically incorporated into regional and continental-scale ice-sheet models, including international intercomparison frameworks, to narrow the uncertainty in Antarctica's contribution to future sea-level rise. The paper emphasizes that while fully coupled ice-ocean simulations are ideal, the current approach of calibrating effective melt rates against observed state changes provides a reliable path to reducing projection uncertainty.

The Old View vs. The New Discovery

The Experiment: Tuning the Model

The Big Prediction: Doubling the Risk

The "Peg" That Might Break

The Bottom Line

More like this