A torpedo-shaped robot has been making its way through water so dark and frigid that it hardly seems like the same planet somewhere beneath the Southern Ocean. It was sent there to measure the warm tidal water pushing under pressure as far as ten kilometers beneath the Thwaites Glacier—past the point where the ice lifts off the seafloor and starts to float, past the thin, cold layer that had been serving as a sort of natural insulation.
This is something that scientists had already suspected but had not directly confirmed. When it arrived, what it discovered was unsettling. There is a disruption in the insulating layer. The glacier is retreating more quickly than predicted by the models. Furthermore, the models were already frightening.
Thwaites Glacier (“Doomsday Glacier”) — Antarctica
| Location | West Antarctica — part of the West Antarctic Ice Sheet (WAIS), a basin of ice nearly three times the size of Texas, sitting below sea level |
| Size | Roughly the size of Florida — more than 2 kilometres thick in places; functions as the primary barrier holding back the West Antarctic Ice Sheet |
| Current sea level contribution | Accounts for 4% of global sea level rise; losing approximately 50 billion tons of ice per year; oceans currently rising 4.6 mm annually |
| Collapse scenario | Thwaites collapse alone: 65 cm (26 inches) of sea level rise. Full WAIS loss: 3.3 metres — threatening major coastal cities and all low-lying island nations |
| Key finding (2024) | Tidal action is pumping warm seawater up to 10 km under the glacier’s grounding zone, disrupting the cold insulating layer and accelerating retreat beyond current model predictions |
| Research method | Torpedo-shaped autonomous robot deployed beneath the glacier; combined with high-resolution satellite imagery and hydrological modelling by UC Irvine, University of Waterloo, Dartmouth, and University of Edinburgh teams |
| Retreat history | Retreating for over 80 years; pace has significantly accelerated in the past 30; grounding zone — where glacier meets seabed — is moving further back each decade |
| Geoengineering debate | Proposals include giant submarine curtains of fabric or air bubbles to block warm tidal currents; researchers caution 15–30 years of study needed before any intervention could be recommended |
| Reference / research body | International Thwaites Glacier Collaboration (ITGC) |
More than two kilometers thick in some places, Thwaites is larger than Florida and has been losing ice for more than 80 years. Rob Larter, a marine geophysicist who contributed to the most recent study, summed up this trend with the bluntness that scientists seldom use in formal communications: “Our findings indicate it is set to retreat further and faster.” In the last thirty years, that process has clearly accelerated. The glacier is already responsible for about 50 billion tons of ice loss annually, or 4% of the total rise in sea level worldwide. In terms of climate tipping points, it is not a risk for the future. It’s a current one. How quickly is the issue being urgently renegotiated.
Here, geography is important, but it often gets lost in the abstraction of millimeters and percentage points. Sitting in a basin beneath sea level is the West Antarctic Ice Sheet, the enormous bowl of ice that Thwaites is essentially keeping in place. The only thing keeping warmer ocean water from flooding that basin, shattering the ice and causing a collapse that would raise global sea levels by more than three meters is the glaciers along its marine-facing rim.
If Thwaites went alone, the seas would rise by about 65 centimeters. Large portions of Miami, Mumbai, Shanghai, and dozens of smaller coastal cities that haven’t had much time to adapt could be permanently submerged. Melodrama is not the reason why scientists refer to it as the Doomsday Glacier. It describes what occurs downstream quite literally.
In overlapping but separate studies, research from UC Irvine, the University of Waterloo, Dartmouth, and the University of Edinburgh has demonstrated that the mechanism causing melt is more complicated and, in certain aspects, more aggressive than previously thought. In ways that no one had sufficiently considered, tidal action is the culprit.
The rhythmic push and pull of ocean tides is pushing warm saltwater under the ice at high pressure as parts of the glacier lift off the seabed and start to float, upsetting the cold freshwater layer that had been slowing melt. With every tidal cycle, the grounding zone—the crucial line where ice meets the seabed—is being eaten away from beneath in pulses. The stakes were clear, according to Christine Dow, a glaciologist at the University of Waterloo who co-authored one of the studies: scientists had been hoping for a timeline of a century or even five hundred years. “A big concern right now,” she replied, “is if it happens much faster than that.”
The worst forecasts have a counterweight, albeit a small one. According to a different study from Dartmouth and Edinburgh, Thwaites might be less vulnerable to a phenomenon known as marine ice cliff instability than the most disastrous hypotheses, which hold that tall ice cliffs created by retreating glaciers quickly collapse and cause runaway loss. Thinning ice may actually stabilize those cliffs rather than hasten their collapse, according to the evidence. It’s not exactly good news. The majority of the actual risk resides in a broad and extremely uncertain middle ground, which is more akin to a slight softening of the worst-case end of the probability range.
As we watch this debate unfold in real time, it is now impossible to avoid talking about geoengineering. In a report published by the University of Chicago’s Climate Systems Engineering Initiative, researchers urged immediate funding for the study of glacial geoengineering, which includes interventions such as the installation of enormous submarine curtains made of fabric or pressurized air that are intended to prevent warm tidal currents from reaching the base of the glacier.
The scientists involved are the first to admit that the proposals sound a little bit like science fiction. The report’s co-author, John Moore, pointed out that it might take fifteen to thirty years of research to decide whether any of these interventions are worth trying. This means that work on something that might never be used needs to start now because waiting until the situation is obviously critical leaves no time to determine whether a response is possible.
Geoengineering is, at best, a painkiller rather than a cure, according to Gernot Wagner, a climate economist at Columbia. This may be the most honest way to frame the issue. It might maintain the patient’s functionality long enough to deal with the fundamental issue, which is still the rate at which greenhouse gases are being added to the atmosphere by human activity. Time is bought by a curtain beneath the ice. It doesn’t reduce the difference between what the climate can absorb and current emissions trajectories. Even the most ambitious geoengineering plans may not be able to outpace tidal forcing, which is already functioning at a depth and scale that robots are just now starting to map. The glacier is not waiting for the study to be finished.
There is currently no definitive answer to the most difficult question, which is whether Thwaites’ loss is already irreversible. Antarctic snowfall does help replenish some of what the glacier loses, according to Michelle Maclennan, a climate scientist at the University of Colorado. However, the math is working against the glacier because ice is disappearing more quickly than snow can replenish it.
The glacier is compressed from above while the ocean works on it from below when a warming atmosphere eventually switches that dynamic from snowfall to rainfall and surface melting. Like almost everything in climate science, how quickly that occurs depends on decisions made today in energy policy and emissions agreements, in rooms far from any ice sheet, by people who have probably never given much thought to the underside of a glacier in the dark waters of the Southern Ocean.
