Scientists have discovered that uncertainty about the landscape beneath Antarctica’s kilometre-thick ice sheet can influence future sea-level forecasts just as much as the difference between high and low CO₂ emissions. The findings suggest that, in Antarctica, better forecasts may depend as much on better maps as on better models.
Antarctica contains enough ice to reshape coastlines around the world. This is why scientists devote enormous effort to predicting how quickly the continent’s ice sheet will melt as the climate warms.
Most of the uncertainty is usually assumed to stem from one source: future greenhouse-gas emissions.
However, a new study suggests that another uncertainty may be just as important – and it lies hidden beneath kilometres of ice. In the simulations, the effect amounted to more than 40 centimetres of sea-level rise by 2150 and around 1 metre by 2300.
“Many people believe that the greatest uncertainty stems from how much greenhouse gas we emit. But our results show that something as fundamental as the shape of the bedrock beneath the ice can be just as crucial,” says co-author Helene Seroussi, a glaciologist and Associate Professor at the Thayer School of Engineering, Dartmouth College, Hanover, NH, USA.
The team used advanced ice-sheet models to test what happens when the maps of Antarctica’s hidden landscape are altered within their known margins of error.
“We have spent decades improving climate models, but if we do not know exactly what the ice is resting on, it is like trying to predict a car’s route without knowing the road.”
The findings suggest that better forecasts may depend as much on better maps as on better models.
The landscape beneath the ice
To most people, Antarctica looks like one enormous, white mass of ice. But for glaciologists, what lies beneath the ice is at least as important as the ice itself.
Not the climate above the ice. Not the sea around the ice. But the landscape beneath the ice itself.
Beneath the ice sheet, which is up to several kilometres thick, lies a complex landscape of mountains, valleys, ridges and deep basins. In some places, the bedrock lies hundreds of metres above sea level. In others, gigantic valleys extend hundreds of kilometres inland far below sea level.
The shape of this hidden landscape helps to determine how the ice moves.
“The ice does not flow across a flat surface. It flows through a very complex landscape, and small changes in the topography can have major consequences for how the ice responds to climate change.”
As glaciers move towards the sea, the bedrock acts as a kind of steering system. Ridges can slow the ice’s movement, whereas deep channels can funnel ice flows and expose them to warm seawater. Of particular importance is the grounding line – the boundary between ice resting on bedrock and ice floating on the ocean.
“If you do not know the exact topography around the grounding line, predicting future developments becomes much harder.”
The hidden continent is still being mapped
Over the past decades, aircraft carrying ground-penetrating radar have criss-crossed Antarctica. The radar signals pass through the ice and bounce back from the bedrock below, enabling researchers to map the hidden landscape.
Yet large gaps remain. The aircraft only measure directly beneath their flight paths, and in some regions hundreds of kilometres separate one survey line from the next.
In some of the best-studied areas, the uncertainty is only a few metres. In remote regions, the bedrock may still be wrong by several hundred metres.
“We have made enormous progress. But Antarctica is larger than Europe and covered by kilometres of ice. There are still large areas where we lack direct measurements.”
Despite this, most international ice-sheet models rely on the same underlying maps of Antarctica’s bedrock.
“Many studies have looked at uncertainties in climate, ocean conditions and ice physics,” says Helene Seroussi. “But far fewer have asked how much uncertainty in the bedrock itself might affect the forecasts.”
To answer this question, Seroussi and colleagues designed a simple experiment.
What if the maps are wrong?
No one knows exactly what the landscape beneath Antarctica looks like. But researchers do know how uncertain their maps are. Modern maps contain not only a best estimate of the bedrock’s height but also an estimate of the uncertainty in each area.
Instead of creating yet another map, they asked a different question: what if the existing uncertainties turned out to be right?
“We asked a relatively simple question: if the bedrock lies as high or as low as the uncertainties allow, how much does that change our forecasts?”
Using BedMachine Antarctica – one of the most widely used maps of the landscape beneath the Antarctic Ice Sheet – the researchers created three versions of the continent: one based on the best current estimate of the bedrock, one in which the bedrock was systematically raised within the reported uncertainties and one in which it was lowered.
“We did not change the climate. We did not change the physics of the model. The only thing we changed was the bedrock on which the ice rests.”
Three Antarcticas, one experiment
For each version, the researchers calibrated the model to present-day Antarctica and simulated future ice-sheet evolution through 2300.
They used the ISSM ice-sheet model, which is also used in international sea-level projections. To place the effect of bedrock uncertainty in context, they ran the simulations under both low- and high-emissions scenarios, enabling them to compare geological uncertainty directly with one of the most widely discussed uncertainties in climate science.
“We wanted to be able to say whether this uncertainty is small or large compared with the uncertainties climate scientists usually deal with.”
The team also tested the effect at three different scales: across Antarctica as a whole, within the Aurora Basin and Bellingshausen regions and for individual glacier systems. If the signal appeared across all three scales, the researchers could be more confident that it reflected a genuine feature of the system rather than an artefact of a particular model setup.
“If topography really does matter, we expected the effect to become clearer the closer we got to the individual glaciers.”
Aurora contains some of Antarctica’s largest topographic uncertainties, whereas Bellingshausen is comparatively well mapped.
The uncertainty turned out to be enormous
“We knew it would have an impact, but I was not expecting it to be so large,” says Helene Seroussi. “The impact is a lot bigger than what I was anticipating.”
In some simulations, small changes in the topography beneath the ice sent glaciers down very different paths over the coming centuries.
Differences in the seabed topography altered Antarctica’s contribution to sea level by more than 40 centimetres by 2150. By 2300, the difference had grown to around one metre.
One result kept appearing, regardless of which simulation the researchers ran: uncertainty in the bedrock was often as important as the difference between a low- and a high-emissions scenario.
Until around 2250, the effect of bedrock uncertainty was greater than the effect of the climate scenario itself in several cases.
“It was surprising to see how long this uncertainty dominated the signal. We expected the climate scenarios to take over much earlier.”
Sometimes the bedrock mattered more than the climate
The effect was not evenly distributed across the continent. Large parts of West Antarctica, including the Amundsen Sea, Ross Basin and Filchner-Ronne region, proved particularly sensitive.
That was not what Seroussi expected. Some of these glaciers, including Thwaites and Pine Island, are among the best surveyed in Antarctica.
“I was expecting the poorly mapped regions to be affected,” she says. “I was surprised that some of the places where we already have many measurements were still so sensitive to these uncertainties.”
Small changes in the topography could alter both the timing and extent of glacier retreat.
“We did not just see differences in how much ice was lost. We also saw differences in when specific areas began to retreat.”
A few hundred metres can change everything
The reason is that glaciers often behave like threshold systems. A glacier can remain relatively stable until its grounding line reaches a critical ridge or drops into a deeper basin. Once retreat begins on a bed that slopes downward inland, the ice becomes thicker, flows faster and can trigger a self-reinforcing retreat.
If those features are mapped incorrectly, the model can produce very different outcomes.
“In some places, a difference of just a few hundred metres in the bottom topography can determine whether a grounding line stabilises or continues to retreat.”
The importance of bedrock uncertainty increased as the models became more detailed. In the Aurora region, uncertainty increased by around 14% in regional simulations and by around 30% in the highest-resolution glacier simulations compared with continental-scale models.
Rather than reducing uncertainty, the finer models often revealed uncertainties that coarser models had concealed.
“When we increase the resolution, the models become more sensitive to the information we feed them. If the topography is uncertain, that uncertainty also becomes more visible.”
Different parts of Antarctica play by different rules
The Aurora Basin in East Antarctica proved especially sensitive to uncertainties in the bedrock, whereas the Bellingshausen region was influenced more strongly by the climate scenario itself. Together, the two regions showed that Antarctica was not responding as a single system.
The contrast suggests that different parts of Antarctica are governed by different mechanisms and sensitivities – in some places better maps matter most, whereas in others climate uncertainty remains dominant.
“Antarctica is not a single ice sheet that behaves the same everywhere. Each region has its own geometry, its own dynamics and its own sensitivities.”
Taken together, the continental, regional and glacier-scale simulations suggest that a significant share of the uncertainty in future sea-level projections may stem not only from future climate but from incomplete knowledge of the landscape beneath the ice.
The next breakthrough may come from better maps
For decades, researchers have focused on improving climate models, understanding ice physics and collecting more precise measurements of temperature, ocean currents and precipitation.
But the results point in a different direction: better forecasts may begin with better maps of the continent hidden beneath the ice.
“If we want better forecasts, the next major improvement may not come from better climate models. It may come from better maps of Antarctica.”
The researchers are not suggesting that climate change is less important than previously thought.
Climate change remains the fundamental driver of future ice loss. But the study suggests that uncertainty about Antarctica’s geology may limit how precisely scientists can predict the ice sheet’s response.
“This does not change our understanding that the climate is getting warmer. It changes our understanding of how accurately we can predict the ice’s response.”
One uncertainty scientists can actually reduce
The findings also raise a practical question: where should future investments be made? Unlike many uncertainties in climate science, this is one researchers can actually reduce.
“We cannot predict future emissions,” says Seroussi. “But bedrock topography is something we can go out and measure. We have the technology. We know where the biggest gaps are.”
Some of the places that mattered most in the simulations are also among the places scientists know least about. These are precisely the places where new measurements could have the biggest impact on future sea-level forecasts.
Some of the largest uncertainties lie in remote parts of East Antarctica, where direct measurements remain sparse.
“We actually know quite precisely where we are lacking information. This allows us to target future measurement campaigns at the areas where new data will make the biggest difference.”
We are still discovering the continent we are trying to predict
Meanwhile, new generations of radar instruments, satellites and autonomous systems are enabling researchers to map the continent more easily than ever before.
“We are at a point at which observations and models can increasingly work hand in hand. The models can tell us where data is missing, and the observations can help us to reduce the uncertainties.”
Whether the researchers looked across Antarctica, within individual regions or at specific glaciers, the same message emerged: models are only as good as the data that underpin them.
“There is often a great deal of focus on the models, but models are only as good as the data on which they are based.”
“Antarctica is one of the most inaccessible places on Earth,” says Seroussi. “In many ways, we are still in the process of discovering the continent whose future we are trying to predict.”
