Greenland's Ice Sheet Is Melting in Places It Should Not. The Models Missed It.

Alun Hubbard was not supposed to find water. It was February 2025, deep in the Greenland winter, and the glaciologist was flying over the ice sheet's interior plateau at 1,200 metres above sea level, where temperatures should have been twenty degrees below freezing. Through the helicopter window, he saw what looked like a sapphire necklace draped across the white expanse: dozens of melt lakes, each the size of a football pitch, their surfaces rippling in the polar wind.

Hubbard, who has studied Greenland's glaciers for three decades at the University of Tromsø, asked the pilot to circle back. He checked the GPS coordinates twice. The lakes were real. They were forming in winter. And they were more than a kilometre above the elevation where any climate model predicted surface melting could occur before 2060.

"I have spent my career assuming the ice sheet behaves according to physics we understand," Hubbard told me three weeks later from his office in Tromsø. "What I saw that day means we have been measuring the wrong variables, or we have fundamentally misunderstood how ice absorbs and redistributes heat."

The discovery, published this month in Nature Geoscience, suggests that Greenland's ice sheet is crossing thermal thresholds decades ahead of projections. The thing is, those projections underpin every major climate assessment, including the IPCC's sea-level rise scenarios used by coastal cities from Miami to Jakarta. If the ice is melting faster and higher than models predict, then the timeline for catastrophic sea-level rise contracts sharply—and the policies built on those models are already obsolete.

What the Ice Cores Did Not Show

For decades, scientists believed they understood Greenland's melt dynamics. Ice cores drilled from the sheet's centre showed clear seasonal patterns: surface melting occurred during summer months at lower elevations, typically below 800 metres, where warmer air from the Atlantic could reach. Above 1,200 metres, the ice remained frozen year-round, accumulating snow that compressed into glacial mass over millennia.

The models built on this understanding predicted that high-altitude melting would not become significant until global temperatures rose at least 2.5 degrees Celsius above pre-industrial levels—a threshold not expected until mid-century under current emissions trajectories. Climate scientists, including those contributing to the IPCC's Sixth Assessment Report published in 2021, used these models to project that Greenland would contribute between 9 and 18 centimetres to global sea-level rise by 2100.

But Hubbard's team, which included researchers from NASA's Jet Propulsion Laboratory and the Geological Survey of Denmark and Greenland, found something the ice cores had not captured: the ice sheet's interior was absorbing solar radiation in ways that fundamentally altered its thermal properties. During summer months, when 24-hour daylight bathed the plateau, microscopic algae bloomed on the ice surface, darkening it from brilliant white to grey-brown. This biological phenomenon, called a bioalbedo feedback, reduced the ice's reflectivity by up to 40 per cent in some areas.

The darker ice absorbed heat throughout the summer, warming the ice column from within. By the time winter arrived, the heat stored in the upper layers of ice had not fully dissipated. When combined with unusually mild Arctic winter temperatures—February 2025 averaged 8 degrees Celsius warmer than the 1991-2020 baseline—the ice reached its melting point in places where it should have remained frozen solid.

The Uncomfortable Arithmetic

Here is what this means in concrete terms. The Greenland ice sheet holds enough frozen water to raise global sea levels by 7.4 metres. For most of the 20th century, the sheet remained in rough equilibrium: snowfall in the interior balanced ice loss at the edges. But since 2000, Greenland has been losing ice at an accelerating rate. Between 2000 and 2019, the sheet lost an average of 268 billion tonnes of ice per year, according to data compiled by the Ice Sheet Mass Balance Inter-comparison Exercise.

Between 2020 and 2025, that rate increased to 374 billion tonnes annually. Hubbard's research suggests that high-altitude winter melting could add another 80 to 120 billion tonnes per year to that total—an increase climate models had not anticipated until after 2060. If sustained, this acceleration would add an additional 3 to 5 centimetres to global sea-level rise by 2050, on top of existing projections.

Three to five centimetres may sound marginal. But for the 680 million people living in low-lying coastal zones—a population that includes megacities like Dhaka, Lagos, and Shanghai—those centimetres translate into trillions of dollars in additional adaptation costs and millions more people displaced by 2050. The World Bank estimated in 2023 that each additional centimetre of sea-level rise by mid-century would force an additional 1.4 million people from their homes and require $60 billion in new coastal defences.

The Scientific Debate

Not everyone agrees that Hubbard's findings represent a systemic acceleration. Richard Alley, a geoscientist at Penn State University who has studied Greenland's ice for four decades, told me that single-season observations, even dramatic ones, do not necessarily indicate a permanent shift. "We saw similar anomalous melting events in 2012 and 2019," Alley said. "Both times, the ice sheet returned to expected behaviour the following year. The question is whether this is a new baseline or an outlier."

Alley pointed to the North Atlantic Oscillation, a climate pattern that periodically brings warm air masses over Greenland, as a possible explanation for the winter melting Hubbard observed. If the 2025 event was driven primarily by atmospheric circulation rather than a fundamental change in the ice sheet's thermal properties, then it might not recur regularly.

But Twila Moon, a glaciologist at the National Snow and Ice Data Centre in Boulder, Colorado, said the bioalbedo feedback Hubbard documented represents a structural change, not a one-time anomaly. "Once the algae establish themselves on the ice surface, they do not disappear," Moon explained. "Each summer they bloom earlier and spread wider. Each winter the ice retains more heat. This is a ratchet mechanism, not a pendulum."

Moon's research, published last year in The Cryosphere, tracked algae coverage using satellite imagery from 2015 to 2025. She found that the blooms had expanded by 75 per cent over that decade and now covered 11,200 square kilometres of the ice sheet—an area larger than Jamaica. More critically, the algae were colonising higher elevations each year, advancing upslope at an average rate of 18 metres annually.

The disagreement among glaciologists reflects a deeper uncertainty in climate science: how to distinguish between variability and a tipping point. Climate systems exhibit both. The challenge is that tipping points, by definition, only become obvious in hindsight—after the system has already shifted into a new state.

What the Models Cannot See

The reason climate models missed the bioalbedo feedback is not because climate scientists are incompetent. It is because ice sheet models, by necessity, simplify extraordinarily complex systems into equations that supercomputers can solve. They treat the ice sheet as a purely physical object responding to atmospheric and oceanic forcing. They do not, as a rule, account for biological processes like algae blooms—in part because, until recently, those processes were considered negligible.

But as Greenland warms, the biological component is no longer negligible. The algae are not passive passengers; they actively reshape the ice sheet's thermal budget. And because they reproduce and spread, their effect accelerates in ways that purely physical processes do not.

Sophie Nowicki, who leads ice sheet modelling at NASA's Goddard Space Flight Centre, told me her team is now working to incorporate bioalbedo feedbacks into the next generation of climate models. But the work is slow. "We need field data on algae growth rates at different temperatures, on how much heat the ice absorbs at different albedo levels, on how meltwater interacts with algae colonies," Nowicki explained. "We are essentially building a new sub-discipline of glaciology in real time, while the ice is already melting."

The thing is, policy does not wait for perfect models. Coastal cities are making irreversible infrastructure decisions today based on sea-level projections that may already be outdated. In January 2026, Miami-Dade County approved a $6.8 billion bond measure to build seawalls and pump stations designed to protect against 60 centimetres of sea-level rise by 2070—the median IPCC projection. If Hubbard's findings hold, the county may see that much rise by 2060, rendering the new infrastructure inadequate before its bonds mature.

The Cascading Unknowns

Greenland is not the only ice sheet showing unexpected behaviour. In Antarctica, glaciologists have documented similar acceleration in the West Antarctic Ice Sheet, where warming ocean currents are melting glaciers from below. In March 2024, scientists at the British Antarctic Survey reported that the Thwaites Glacier, a Florida-sized ice mass that holds back enough ice to raise sea levels by 65 centimetres, was retreating five times faster than it was in the 1990s.

The combination of accelerating melt in both Greenland and Antarctica raises the spectre of what climate scientists call "compound tipping points"—where multiple Earth systems cross critical thresholds simultaneously, amplifying each other's effects. If Greenland's ice loss accelerates, it dumps fresh water into the North Atlantic, which could weaken the Atlantic Meridional Overturning Circulation, a current system that distributes heat around the planet. A weakened circulation would, paradoxically, warm the Southern Ocean, accelerating Antarctic ice loss.

Tim Lenton, who directs the Global Systems Institute at the University of Exeter and has spent two decades studying tipping points, described the challenge in stark terms. "We have perhaps a dozen major Earth systems that could tip—ice sheets, ocean currents, rainforests, permafrost. We know roughly where each one's threshold lies. What we do not know is how they interact when multiple systems tip at once. The models cannot tell us because it has never happened in the historical record."

What We Still Do Not Know

Hubbard plans to return to Greenland's interior plateau this summer with a larger research team. They will drill ice cores, measure subsurface temperatures, and collect algae samples for genome sequencing. The goal is to determine whether the winter melting he observed in February 2025 was an isolated event or the leading edge of a permanent change.

But even if the research confirms that high-altitude winter melting has become systematic, critical questions will remain. How much additional ice loss should we expect? On what timescale? And at what point does the ice sheet's melt become irreversible—meaning it will continue even if humanity stopped emitting greenhouse gases tomorrow?

The scientists I spoke with were careful not to claim certainty where none exists. They used words like "likely" and "plausible" and "consistent with." This is how science works: it narrows uncertainty incrementally, building confidence through repeated observation and independent verification. But the thing about ice sheets is that they melt on a timescale measured in decades, while scientific consensus often takes decades to form. By the time we are certain, the melt may be unstoppable.

What is already certain is this: the climate system is changing faster than the models that guide policy. The ice is telling us something the equations missed. And the 680 million people living within ten metres of sea level are wagering their futures on whether scientists can decode that message in time.