Seawater under the 'Doomsday Glacier' doubles its 2100 sea-level estimate, model finds
- A new Nature Communications study says Thwaites Glacier alone could add up to 30 mm (about 1.2 inches) to global seas by 2100.
- That is about twice the average of earlier models. It assumes the 1996–2023 climate holds, which the authors call a lower-bound scenario.
- The key input, how far seawater reaches under the ice, comes from satellite radar; a 2028 trip aims to place instruments at the grounding line.
In a low-lying street in Jakarta, Lagos or Miami, a few centimetres of sea can decide whether a high tide reaches someone’s front door. A new study says one Antarctic glacier alone could add up to about 3 centimetres to global seas by 2100.
That glacier is Thwaites. The study, published on 12 October in Nature Communications, puts its share of sea level rise at up to 30 ± 6 millimetres (about 1.2 inches). That is about twice the average of earlier models run under the same assumptions. It also assumes the climate stays as it was between 1996 and 2023, which the authors call a lower bound.
So what is the Doomsday Glacier? It is the nickname for Thwaites, a West Antarctic glacier that CNN describes as roughly the size of Florida.
Tides push seawater kilometres under the ice
The team, led by Mattia Poinelli at UC Irvine and NASA’s Jet Propulsion Laboratory, studied the “grounding line”. That is where ice leaves the seabed and starts to float. Models usually draw it as a single clean edge. Radar from Finland’s ICEYE satellites shows a grounding zone kilometres wide, where tides push seawater under the ice and melt it from below.
The model could only recreate the glacier’s retreat since the mid-1990s once it allowed a zone 2 to 5 km wide. Take the zone out, and the paper finds the model’s odds of matching that history fall fourfold.
With a 1 km zone, Thwaites adds about 14 mm by 2100, close to the average of ISMIP6, the main international comparison of ice-sheet models. The width the model itself found most likely, about 3 km, gives about 19 mm. With a 5 km zone, which fits recent satellite data, it adds about 30 mm.
Why 2038 is the earliest case, not a forecast
A seabed bump called the Mouginot Ridge holds back the glacier’s main trunk. With 4 to 6 km zones, the model gives about a 60% chance that the ice slips off it by 2100. The glacier would then retreat faster toward the next ridge. Only in the widest runs does unpinning come as early as 2038.
The study does not predict a Thwaites glacier collapse by 2100. Its 30 mm figure covers Thwaites alone, not total sea-level rise.
The width of the seawater zone still comes from satellites
The result moves by about 22% for each extra kilometre of grounding zone. Robert Larter, a marine geophysicist at the British Antarctic Survey who was not involved, told CNN: “there is a lot we still need to find out about grounding zone processes to be able to make confident predictions about how fast ice will be lost.”
Ted Scambos, a senior research scientist at the University of Colorado Boulder, also not involved, told CNN the findings could be “a real gamechanger for sea level forecasting from Thwaites”, but cautioned that direct data from beneath the ice is still needed.
The authors list limits, like a fixed zone width; some could make the estimate too low. In one test, an ocean 0.5 to 0.7 °C warmer raised ice loss by about 15%.
What 30 mm of Thwaites glacier sea level rise means on the coast
The 30 mm adds to every other source of rising water. A higher starting sea lets storm surges and high-tide floods reach further inland. If you live by the coast, check your city’s flood maps and sea-wall plans; that is usually where numbers like this show up.
Co-author Eric Rignot said in the university’s release: “We must do what we can now to curb greenhouse gas emissions, but even if we do, we can expect higher sea levels in the future resulting from the damage that has already been done.” Some of this melting is already locked in. Cutting emissions still limits what comes after.
Better physics makes the warning sharper, not louder
On balance, this is careful work. The model earned its projection by first matching the past, and the authors say plainly that their scenario is a floor. It is also a quiet tech story: commercial satellite radar helped narrow a hard question. We will follow it in our science news.
The weak point is that the biggest input still comes from satellites, not instruments under the ice. The university’s “worse than thought” headline is also stronger than the paper, which says its results “lie toward the upper end of previously reported uncertainty ranges”.
The likelier reading is that this is a planning signal, not an alarm: a reason for coastal cities to plan for more water, not to panic.
Rignot told CNN the team plans to visit Thwaites in 2028 to place instruments at the grounding line itself. Until then, the ice keeps its own count.
Prefer to watch? Follow us on YouTube
The tech stories that matter, explained in plain words. Subscribe so you never miss one.
Sources (4): every key fact checked against at least two
- Nature Communications: Sea-level contribution from Thwaites Glacier doubles in an observation-constrained ice-sheet model (Poinelli et al., 2026)
- UC Irvine: Melting of Antarctica's Thwaites Glacier worse than thought (12 October 2026)
- CNN: Thwaites melt study on the 'Doomsday Glacier' (15 October 2026)
- Gizmodo: Ocean water is seeping under the Doomsday Glacier (14 October 2026)
Related stories
Harvard study: 403 US data centers used up to as much electricity as Michigan
Data center electricity consumption, estimated: a Harvard-led study of 403 US hyperscale sites finds 68 to 99 TWh in a year, near Michigan's use.
Vulcan rocket fails 10 minutes into flight, and 40 Amazon Leo satellites do not reach orbit
Vulcan rocket failure on 15 November: ULA's rocket suffered an anomaly 10 minutes in and did not deliver 40 Amazon Leo satellites to orbit.
A flash of laser light briefly lines up the hidden electric order inside a quantum crystal
Auburn physicists used laser pulses to briefly line up tiny electric regions in a quantum ferroelectric metal, for about a trillionth of a second.