Screenshots from the MALI and BISICLES computer models simulating what the Antarctic ice sheet would look like and how fast ice would move 200 years after the ice sheet’s floating ice shelves have disintegrated.

Screenshots from the MALI and BISICLES computer models simulating what the Antarctic ice sheet would look like and how fast ice would move 200 years after the ice sheet’s floating ice shelves have disintegrated.

Image Credit: MALI model, developed by DOE’s Los Alamos and Sandia National Laboratories and BISICLES model, developed by DOE’s Lawrence Berkeley National Laboratory, University of Bristol, and University of Swansea.

New Year’s Day, 2002. Without warning, a few large cracks in the Larsen B ice shelf appeared in satellite imagery. Within days, car-sized chunks of ice appeared in the water. Then shards of ice larger than houses collapsed into the water. By March, satellite photos showed the entire ice shelf had disappeared; more than 1,000 square miles of ice were gone.

The collapse of the Larsen B ice shelf in the West Antarctic Peninsula shattered more than ice — it shattered scientists’ prevailing view of ice sheets and their floating shelves. Scientists had previously thought ice sheets would respond to recent climate change on time scales of hundreds to thousands of years. After all, for interior regions of the Antarctic ice sheet, conditions are still very similar to those at the end of the last Ice Age.

But where ice meets ocean, change can happen fast.

Scientists who work to understand and model Earth systems, including ice sheets, had to change their approach too. They had theorized the Larsen B collapse might happen, but existing models didn’t predict the ice sheet’s sudden response. Ice shelves block rivers of ice from flowing into the ocean. With the shelf gone, the rivers of ice moved into the ocean three times faster than before. That shift dramatically increased the amount of ice moving from the land into the ocean.

In 2007, the Intergovernmental Panel on Climate Change published its regularly issued report—the world’s biggest and most comprehensive report on climate change. That report noted that then-current computer models of ice sheets weren’t good enough to accurately predict how those sheets would react to climate change and influence sea level rise, a critical gap in understanding the climate system.

The Department of Energy’s (DOE) Office of Science responded to that need.

To plan for the future, city managers, energy companies, and national security experts need to know how and when changing ice sheets will affect sea level rise and other parts of the climate. Researchers supported by DOE are working to develop new and better models that can provide these insights. In 2009, the Office of Science funded six projects under the Ice Sheet Initiative for Climate Extremes (ISICLEs). It continued that funding through the Predicting Ice Sheet and Climate Evolution at Extreme Scales (PISCEES) project in 2012 and the Probabilistic Sea-Level Projections from Ice Sheet and Earth System Models project in 2017.

“We’ve definitely been at the forefront of this effort,” said Daniel Martin, a computational scientist who focuses on ice sheet modeling at DOE’s Lawrence Berkeley National Laboratory (LBNL).

Zooming in on Ice Streams and Shelves

Mature ice sheets can cover entire continents and, in particular, currently spread over large parts of Antarctica and Greenland. Glaciers and ice sheets form when snow accumulates over many years until it compresses into ice. The accumulation of ice then becomes so thick that it flows outwards under its own weight.  However, these sheets are constantly changing, with faster moving “streams” of ice flowing into the ocean in some places and new ice forming in others.

Those changes don’t happen in isolation.

“All of these components … ocean and atmosphere and land, they are all interacting,” said Wuyin Lin, an earth systems modeler at DOE’s Brookhaven National Laboratory.

These interactions create feedback loops. Changes in the ocean and atmosphere affect the ice and vice versa. These loops can multiply outside effects, such as warming from climate change.

“It’s a domino effect,” said Sophie Nowicki, an Earth systems modeler at NASA’s Goddard Space Flight Center.

The West Antarctic Ice Sheet (WAIS), where the Larsen B shelf was located, is one of the biggest areas of concern. As the Larsen B collapse showed, the WAIS is particularly vulnerable to climate change. Large swaths of the WAIS’s continental bedrock sits as much as 1.5 miles below sea level. As a result, much of the ice doesn’t actually rest on land. Twelve percent of it hangs out into the ocean, floating in flat ice shelves on top of the water. These shelves aren’t static either. Because ice slides easily along the ocean’s surface, the shelves are continually spreading outward, becoming increasingly unstable.  

Ocean water flows through the cavities under these ice shelves, melting them from below. Liquid water on top of the ice shelves also seeps into cracks. As the water moves, it can widen the cracks and further thin the shelf. If the shelf thins and cracks enough, the entire thing can disintegrate. That collapse can open the floodgates for ice streams to flow far more quickly off the bedrock.

These changes don’t happen on a regular schedule. It may seem like nothing happens for a while. But then everything appears to happen at once.

Simulating these erratic and complex changes requires scientists to have a high level of detail — called resolution — in their models. Computer models split a physical system, like an ice sheet, up into small pieces. The higher the model resolution, the smaller the pieces and the more of them are needed.

“We break the ice sheet into little pieces and solve equations that tell us what’s going on in each little piece,” said Martin. “When you connect together all of the little pieces on a big computer, you can simulate the entire ice sheet.”

For the models to follow the details where the ice sheet begins to float and lift off the bedrock as it flows into the ocean – known as the grounding line – modelers needed at least a 500-meter (0.3-mile) resolution. That’s like zooming in to the level of five city blocks or smaller.

The higher the resolution, the more computer power a model requires. Past models could simulate the entire continent at low resolution or certain regions at high resolution, but they didn’t have the ability to do both. New models that scientists supported by DOE are developing offer a path forward.

Two Are Better than One

Work supported by DOE via the Scientific Discovery through Advanced Computing (SciDAC) program over the past nine years resulted in two new ice sheet models that tackle many of these challenges: MPAS-Albany Land Ice (MALI) and Berkeley Ice Sheet Initiative for Climate Extremes (BISICLES). Having more than one model allows DOE scientists to compare and contrast different assumptions and sources of uncertainty. If both models give the same results, it builds confidence in their answers. If not, scientists will dig to find the source of the differences.

“DOE has basically staked out a place of making accurate and efficient models that can do a really good job of trying to get the correct answer in a bunch of different ways,” said Martin.

MALI — a collaboration between DOE’s Los Alamos National Laboratory (LANL) and DOE’s Sandia National Laboratories — puts resolution only where it’s needed. It can focus high resolution on fast-moving ice streams and low resolution in the slow-moving interior. The model changes the resolution by changing the shape and size of the blocks it splits the world into.

Results have been good so far. At a test of one-kilometer (0.6-mile) resolution, MALI simulated the thickness of certain ice accurately down to the meter. Other tests have showed that it’s correctly mimicking under different circumstances how the grounding line moves.

The BISICLES model is slightly simpler than MALI, allowing scientists to run it for larger regions and over longer periods of time using the same computing resources. A collaboration between LBNL and the United Kingdom’s Universities of Bristol and Swansea, the model also places resolution only where it’s needed. Instead of changing the shape of the blocks like MALI, BISICLES splits each block into smaller and smaller ones. As the model evolves, it can automatically shift and adapt that resolution to place it where it is most needed.

Modelers are running MALI and BISICLES to test theories about how ice sheets may disintegrate. As the ice dwindles, its grounding lines and edges retreat toward the continent’s interior. Many researchers think there’s a tipping point where the ice sheet will retreat enough that even if warming stops, the ice sheet will keep on thinning and melting. Scientists have already seen the beginning of these “unstable” retreats in the West Antarctic. If all of the West Antarctic ice shelves melt and disintegrate, the follow-on changes they cause could raise the global sea level by nearly 10 feet.

Runs of the BISICLES model examining the Thwaites Glacier — one of the most vulnerable places in West Antarctica — are providing more evidence for this theory. Scientists ran a scenario where they sped warming up slightly more than the current rate. After an abrupt transition at 260 years, the ice sheet kept retreating — even after scientists “turned off” the accelerated warming. Fortunately, the less warming there was, the longer it took to reach that tipping point. In another experiment, they tested 35 different ice-shelf melting scenarios. They found that if any one of West Antarctica’s three major ice shelves collapsed, the resulting ice sheet retreat could extend all the way back to the continent’s interior. That change could result in more than six feet of sea level rise, deluging the majority of the world’s largest coastal cities, including New York City and Miami.

Looking to 2100

To take ice sheet models to the next level, the teams are working to fully integrate them into DOE’s Energy Exascale Earth System Model (E3SM).

But the very nature of ice sheets makes connecting their models to existing Earth system models both essential and challenging. Essential because modelers need to take the many relationships and feedback loops between the systems into account. Challenging because the timescales between the components don’t match up.

 “The timescales for ice sheet models are orders of magnitude larger than any other component of the climate system,” said Stephen Price, a modeler at LANL. Even though ice sheets change faster than scientists expected, they’re still glacial compared to the atmosphere. Clouds shift by the hour; ice sheets shift by the year. Those mismatches make it challenging to set an ice sheet model’s initial conditions.

If researchers connect the component models wrong, Price said, “The ice sheet model will go staggering off into some other state, which is unlikely to be representative of what any ice sheet is doing today.” It’s a delicate process.

Once the scientists finish tackling those issues, they’ll run a series of experiments. After testing the models against historical observations, they’ll look forward from 2010 to 2100. These experiments should provide the most accurate estimations of ice sheet change and sea level rise yet. 

While 2100 may seem far away, it’s not far away on the scale of an ice sheet — or even human life. Nowicki said that the time scale became real to her when she considered that her grandmother lived to be 105. These models will help us predict how actions today will affect that future.

“A 100-years gap is only a few generations,” she said. “It may seem so far away, but really it’s not. It’s something we’re dealing with every day.”


The Office of Science is the single largest supporter of basic research in the physical sciences in the United States and is working to address some of the most pressing challenges of our time. For more information, please visit the Office of Science website.

Shannon Brescher Shea
Shannon Brescher Shea ( is the social media manager and senior writer/editor in the Office of Science’s Office of Communications and Public Affairs.
more by this author