Although a new study of the mass of the Greenland Ice Sheet shows that previous research underestimated its ice loss by about 20 percent, which could lead to unexpected increases in sea level rise, it also held good news about the technological advancements used to make such measurements.

The study, published last month in the journal Nature, effectively captures ice loss at the edges of glaciers where they meet the sea. The researchers manually recorded changes at the edges of the ice and also trained algorithms to track melting along the border where the glacier meets the sea. 

Previous research focused primarily on capturing melting within a fixed ice mask, an established perimeter, of the ice sheet, which often missed the notable impacts of calving ice breaking off from the edges of the ice sheet. The new research indicates that 1,000 gigatons of ice went unaccounted for in previous estimates. 

Since the lowest edges of the ice sheet sit submerged in the water, the previously unnoticed ice loss didn’t directly affect sea level rise. However, the thinning of the ice sheet could accelerate further melting and allow ice currently sitting on land to slide into the water to raise sea levels.

The researchers studied 207 glaciers using 236,328 satellite observations gathered from 1985 to 2022, some manually collected and some assembled by AI, to observe melting happening along the perimeter of the ice. This approach addresses the limitations of previous Greenland ice loss measurement methods, which faced challenges in capturing changes in the edges of the ice due to the fixed perimeters studied.  

Faster access to larger data sets and AI has not only changed the way that research can be conducted, but also the types of questions that can be asked and answered. Researchers can now look at melting processes on a much larger scale, rather than just in one place, said Chad Greene, the lead author of the study and a glaciologist and remote sensing specialist at NASA’s Jet Propulsion Laboratory. 

“Twenty years ago, the field was data starved. We were just so hungry for any bit of extra data that we could get,” Greene said. “And now that paradigm has really flipped on its head, where suddenly we have so much data…and processing power and so it’s just a matter of writing algorithms to sift through and find what are the most coherent, most meaningful signals.”

Estimates of Greenland ice mass loss were previously acquired by three methods: calculations of the elevation of the glaciers using satellite altimetry, which shows how thick they are, studies of their surface velocity and measuring the gravitational pull of ice sheets via The Gravity Recovery and Climate Experiment (GRACE), a set of satellites. Each method presents challenges in accurately measuring ice mass loss. 

Satellite altimetry uses satellites to determine the elevation of the Earth’s surface. The method uses laser shots to measure the precise elevation of the ice surface, but struggles to include the edges of the ice sheet. To mitigate this, scientists define a fixed perimeter of the ice sheet that excludes the troublesome edges. Changes in the ice sheet beyond that border are not recorded.

To study the velocity of the ice’s surface, researchers compare images of a glacier taken at different times to see how it has moved. But beyond a fixed “flux gate,” another perimeter established by researchers, changes are not captured. Ice flux describes the movement of ice at the interior of the ice sheet to the edges.

When ice sheets lose mass, they lose a little bit of their gravitational pull, but GRACE satellites struggle to measure those changes in the narrow deep fjords Greenland’s ice drains out through, Greene explained. Ice in these fjords is often already below sea level, so when it is lost, it is immediately replaced by seawater and the gravitational pull doesn’t change.

When Marco Tedesco, a Lamont professor at the Lamont-Doherty Earth Observatory of Columbia University and adjunct scientist at the NASA Goddard Institute for Space Studies, started his research around 2001, he had to wait for weeks or months to receive the data sent to him on a CD. But with the standardization of model data and the availability of satellites like GRACE, which launched in March 2002, researchers could widen their scope of observation.

Early GRACE observations showed that between 2002 and 2023, Greenland lost about 270 gigatons of ice per year, which would cause a global sea level rise of 0.03 inches per year. 

New tools found that more ice had melted than previously thought, Greene and his team found in their recent study. Although the loss has minimal direct impact on sea rise, they say, an addition of more than 1,000 gigatons of freshwater to the North Atlantic Ocean can change the buoyancy of the water, which could strengthen Greenland’s coastal currents and alter the course of future interactions between ice and the ocean. Melting icebergs release large amounts of freshwater into fjords, which can influence ocean heat flux.

Sea level rise and ocean expansion are not uniform worldwide, and as Greenland’s ice melts, the redistribution of sea level rise will also be affected as the gravitational pull from the shrinking masses of ice declines and the locations where Greenland releases water change.

By 2100, according to a study published in Nature, Greenland will shed ice at a faster rate than any period in the last 12,000 years, and, according to another study, the Northeast Greenland Ice Stream will contribute to sea level rise six times as much as previous models predicted. 

“Mathematically speaking, we’re chasing something that is not only changing, but is also accelerating and almost escaping from our skills to observe things,” Tedesco says. 

But machine learning and artificial intelligence will become big players in helping the new generation of scientists access data and answer questions in a less time-consuming way, he said. 

“With the intersection of computer science [and data availability], there will be even better understanding of the processes and the tools which allow us to do better simulations, using better resources and having answers on a much shorter timescale than before,” he said.

“A key concern is avoiding a tipping point where the thinning of the ice sheet is so severe that even if we return to pre-industrial climate conditions, the ice sheet’s thickness will be too small to recover.”

A thinning of the ice sheet can accelerate the melting, said Xavier Fettweis, an ice mass balance expert at the University of Liège in Belgium, by a process known as melt-elevation feedback that occurs when the ice sheet enters a cycle of declining surface height and the loss of ice mass from its surface mass increases. As ice melts and the glacier’s surface drops, surface temperatures rise, promoting further melting of ice. 

The loss of 100 meters of ice can add an additional warming of one degree Celsius locally. The ice sheet is projected to thin by approximately 200-300 meters by the year 2100 in areas closest to the edges.

“A key concern is avoiding a tipping point where the thinning of the ice sheet is so severe that even if we return to pre-industrial climate conditions, the ice sheet’s thickness will be too small to recover,” Fettweis says.

Rates of ice mass loss in the 21st century are projected to be significantly higher than those of the pre-industrial era, but vary widely in response to different scenarios of the concentration of greenhouse gasses in the atmosphere, a 2020 study from the Nature journal found. In a low emission scenario, the predicted mass loss in the twenty-first century is 8,800 billion metric tons, but under a high emission scenario, as much as 35,900 billion metric tons could be lost in that 100-year period.

The only way to save the ice would be the cooling and formation of new ice, Tedesco says. But he believes this will not happen, because even if CO2 is no longer pumped into the atmosphere, it would take at least 50-100 years for the atmosphere to phase out the carbon dioxide that’s already there. That is why processes like carbon sequestration will become increasingly important.

Even with the advancements in understanding from 20 years ago, “there’s always more that we need to know and there’s always new things that we need to do to improve our understanding of Greenland,” Tedesco said. “It’s not settled. It will never be settled.”

But new technologies and research methods can help resolve some of the uncertainties, he said.

“The more we learn about the interaction among the different components, the better we can [develop] these estimates,” Tedesco says. “We need more observations of the ocean interaction with the glaciers, and we need to understand better where the water is going.