Satellites can now track sea ice thickness year-round

Melting Arctic sea ice photographed from the airborne sea-ice survey IceBird
Melting Arctic sea ice photographed from the airborne sea-ice survey IceBird. Esther Horvath/Alfred Wegener Institute

Scientists have devised a way to monitor Arctic sea ice throughout the summer – a time of year for which information on ice thickness was previously lacking, according to a new study.

The novel technique could help researchers improve forecasts for Arctic shipping and projections of how much warmer the planet will be in coming decades.

Sea ice is a key component of northern coastal environments and communities. Northerners travel on top of it, and shipping companies need to know where it lies in order to navigate safely. Sea ice also plays a role in regulating climate. Sea ice is melting as the climate warms, exposing dark ocean water that absorbs heat, leading to even more warming and melting.

While the area covered by sea ice is an integral part of this feedback loop, the thickness of that ice is important, too. “The thickness of the ice is what kind-of controls where it is and where it’s not,” said Jack Landy, associate professor of earth observation at UiT The Arctic University of Norway, who led the new work. Whereas thick sea ice tends to stick around, Landy said, a thinner layer is more likely to disappear during the summer months.



Scientists have used satellites to measure sea ice thickness since 1993. As the BBC has reported, they typically draw on radar data to gauge the difference in elevation between floes and leads – floes are sheets of floating ice, leads are the cracks between them that expose open water. Researchers use those measurements to calculate sea ice thickness.

But they can only do so in the winter, from October to April. During the summer months, meltwater ponds that form on top of floes create complications because radar systems have trouble distinguishing the ponds from the open ocean. The presence of those ponds also skews estimates of sea ice elevation, throwing off thickness calculations.

“They’re beautiful things, but they’re a real pain,” Landy said.

Scientists used artificial intelligence to measure summer ice thickness based on satellite data gathered by the European Space Agency’s CryoSat-2. Video: ESA/Planetary Visions

In the new work, described in the journal Nature earlier this month, Landy and his colleagues used artificial intelligence to overcome these issues, creating a year-round record of sea ice thickness from 2011 to 2020.



Previous records of summer sea ice were limited to measurements gathered by buoys distributed throughout the Arctic or planes and ships travelling through the region, as reported in Scientific American, which provide only snapshots of certain locations or times.

Landy said the new record, by contrast, offers an Arctic-wide picture of ice conditions every few weeks for the summer season.

Predicting conditions into the future

To gauge sea ice thickness during the summer, the researchers turned to machine learning, a type of artificial intelligence. They fed satellite images and radar data into an algorithm, training it to distinguish meltwater ponds from the ocean. They also simulated radar responses from different melt conditions, using the information to correct the estimated elevation of pond-covered ice from satellite data.

The researchers say satellite-derived measurements are fairly accurate and align with measurements obtained from ships, buoys and planes.

In addition, the extent of the ice-covered area in September correlates with sea ice volume – a measure calculated using ice thickness – several months earlier, the researchers found.

That suggests the new measurements might extend forecast lead times. If researchers can collect sea ice thickness measurements in May or June, Landy said, they should be able to produce more accurate sea ice forecasts for late summer and early fall, which is when a lot of Arctic shipping occurs.

In the longer term, researchers hope year-round sea ice records can help to calibrate climate models, providing tighter projections for both changes in sea ice and future warming.

The new study presents “an important contribution in being able to measure and monitor the changes in the polar ice pack,” said Trevor Bell, a geography professor at Memorial University and founding director of SmartICE, a program that combines traditional knowledge and technology tools to provide northern communities with real-time sea ice measurements, who was not involved in the work.



Improved forecasts might help northerners know when to expect more ships travelling through their areas, Bell said. The technique’s resolution is too low to inform safe ice travel, however. Each thickness measurement averages an area of 80 km by 80 km. “One data point for that sort of spatial scale is not really useful for communities,” he said, although advances in remote sensing might bring finer-grained assessments in the future.  

According to Landy, more work is also needed to reduce uncertainty in the measurements and the time it takes to turn satellite data into ice thickness information that can be used in forecasts.

Currently, it takes about six months, he said. “I’d like to get that shortened right down to less than a month, if possible.”

This article is produced under a Creative Commons CC BY-ND 4.0 licence through the Wilfrid Laurier University Climate Change Journalism Fellowship.