The base is not smooth. It is a landscape no model was built to explain.
Somewhere beneath the Dotson Ice Shelf in West Antarctica, a torpedo-shaped robot named Ran spent 27 days threading through darkness, scanning the ice ceiling above it with sonar pulses, and sending back the first detailed portrait anyone has ever seen of a glacier's underside. What the maps revealed is both illuminating and unsettling: the assumptions that have underpinned sea level rise projections for years may not hold.
Dotson Ice Shelf sits in West Antarctica, a slab of floating glacier ice roughly 350 meters thick. Scientists have long flagged it as a significant contributor to future sea level rise, given its size and position. But until now, the cavity beneath it — the cold, pressurized space between the ice and the seafloor — was essentially unreachable. Satellites can read the surface. Ice cores can sample the material itself. Neither can tell you what is happening at the base, where ocean water meets ice and the real melting begins.
Ran changed that. The autonomous underwater vehicle, programmed by a team led by Anna Wåhlin, a professor of oceanography at the University of Gothenburg, dove into the cavity and traveled more than 1,000 kilometers back and forth beneath the glacier over those 27 days, reaching as far as 17 kilometers into the shelf. Wåhlin described the experience of seeing the resulting maps as something like seeing the far side of the moon for the first time — a place you knew existed but had never actually looked at.
Some of what Ran found confirmed what researchers had suspected. The western portion of Dotson melts faster than the rest, and the maps showed why: strong underwater currents erode the ice from below, and the effect is especially pronounced at vertical fractures running through the shelf. Where the currents are strongest, the melt is most aggressive. That part of the story made sense.
What did not fit the existing models was the terrain itself. The underside of the glacier is not a flat, featureless plane. It is a landscape — ridges, plateaus, and formations that resemble sand dunes, sculpted into the ice in ways that current scientific models cannot fully account for. The researchers believe these structures were shaped by meltwater flowing under the influence of Earth's rotation, the same Coriolis force that steers hurricanes and ocean gyres. David Holland, a professor at NYU's Courant Institute of Mathematical Sciences and one of the paper's authors, noted that Earth's rotation appears to be shaping these under-ice formations just as it shapes the most dramatic weather systems on the planet's surface.
The study, published in Science Advances, combined Ran's wide-area snapshot with a second, complementary data stream. In 2022, Holland led a team that drilled a hole through the ice shelf — a thousand feet deep and just one foot wide, bored using a hot water drill — and installed a mooring beneath it. That mooring has been collecting continuous measurements ever since, building a time series of how the melt rate at the base of the shelf changes over time. Together, the two methods gave researchers something they had never had before: broad spatial coverage from the submarine, and continuous temporal coverage from the mooring.
Karen Alley, a glaciologist at the University of Manitoba and a co-author on the paper, described the maps as a major leap forward. Scientists had suspected the base of ice shelves was more complex than models assumed, but Ran produced a picture far more detailed and complete than anything previously available. Crucially, that imagery now gives researchers a way to calibrate what satellites are seeing from above — to ground-truth the remote sensing data that most climate projections depend on.
The implications for sea level forecasting are significant. Holland was direct about the stakes: the ability to project how coastlines will change in a warming world depends on data gathered from beneath Antarctic ice shelves. If the models cannot explain what Ran found — and Wåhlin was clear that current models fall short — then the projections built on those models carry an unknown margin of error. That is not a comfortable position when the question being asked is how much ocean cities around the world should expect at their doorsteps in the coming decades.
The researchers are treating this not as a conclusion but as an opening. There is, Wåhlin said, a wealth of processes still waiting to be discovered beneath the glaciers. The next missions will need to look more closely at what Ran uncovered — and begin the harder work of building models that can actually explain it.
Notable Quotes
Many previous assumptions about melting of glacier undersides are falling short. Current models cannot explain the complex patterns we see.— Anna Wåhlin, professor of oceanography, University of Gothenburg
Our ability to project the future of the global coastline from rising sea levels critically depends on data we obtain from beneath Antarctic ice shelves.— David Holland, professor, NYU Courant Institute of Mathematical Sciences
The Hearth Conversation Another angle on the story
Why does it matter so much that we see the underside of the glacier specifically? Can't we infer what's happening from the surface?
The surface tells you the glacier is changing. The underside tells you why. Melt happens at the base, where ocean water makes contact with ice — and that process is invisible from above.
So what was the biggest surprise in what Ran found?
The terrain. Researchers expected the base to be relatively smooth. Instead it's a landscape — ridges, dune-like formations, plateaus. Things that current models simply weren't built to account for.
What's causing those formations?
The leading theory is that meltwater flowing beneath the ice is being shaped by Earth's rotation — the same Coriolis effect that steers hurricanes. It's a force we associate with large-scale atmospheric events, not the underside of a glacier.
Does that mean the existing sea level rise projections are wrong?
It means they're built on incomplete information. The models can't explain what Ran observed. That's not the same as saying the projections are off by a specific amount — it's more that the uncertainty is larger than we thought.
What did the mooring add that the submarine couldn't provide on its own?
Ran gave a wide snapshot — a lot of space, one moment in time. The mooring gives the opposite: one fixed point, but two years of continuous data. Together they let you see both the shape of the problem and how it's changing.
How difficult was it to install that mooring?
They drilled a hole a thousand feet deep and just one foot wide through the ice shelf using a hot water drill, then dropped the mooring through before the ice refroze. It's a narrow window, literally and figuratively.
What does this mean for how we study Antarctica going forward?
It opens a door. The researchers are explicit that there's a wealth of processes still undiscovered beneath these glaciers. Ran proved the method works — now the question is what else is down there.