How AI is helping researchers understand Arctic ice thickness

Sea ice melts on the Franklin Strait along the Northwest Passage in the Canadian Arctic Archipelago, Saturday, July 22, 2017. Because of climate change, more sea ice is being lost each summer than is being replenished in winters. Less sea ice coverage also means that less sunlight will be reflected off the surface of the ocean in a process known as the albedo effect. The oceans will absorb more heat, further fueling global warming. (AP Photo/David Goldman)

Artificial Intelligence (AI) has become the answer to understanding ice thickness in the Arctic year-round, according to researchers.

Before using AI, researchers would use satellites to measure the thickness of ice from October to March, however, as the temperatures got warmer, satellites weren't able to distinguish between the ocean and the melting ice.

Now researchers from around the world, including those from the University of Manitoba are using AI, as it can take the data from the satellites and help interpret the ice thickness year round.

"Thickness is a really important property of ice, whether it's here in Manitoba, and we're talking about river ice or lake ice, or whether we're in the Arctic Ocean and we're talking about the thickness of sea ice," said David Babb, a research associate at the U of M.

He said the previous data from satellites was helpful, but left gaps in research.

"This new study fills that gap and is able to discern between melting sea ice and open water, and then, in turn, estimate sea ice thickness throughout the annual cycle across the entire Arctic Ocean."

Babb said during the summer there are around four million square kilometres of sea ice in the Arctic Ocean that is spread out.

He said understanding the thickness during the summer months can help ships that travel through the Arctic.

"So this is when you have the greatest risk posed by sea ice towards ships because ships are present. So if we're able to monitor sea ice thickness in the Arctic Ocean, we're able to provide better information to the ships that can then adjust routes and adjust their plants to avoid areas of potentially thicker ice or thinner ice."

He said this new research can help predict ice levels months in advance and this information would be beneficial to the shipping industry as it would give them more time to forecast their routes.

Babb noted using AI has really helped researchers understand the Arctic and know more than ever before because of how large the area is and the amount of information that can be collected.

"I think this is where AI can really help kind of streamline some of that interpretation and really sort of grasp the volume of data that is available from some of the satellites and then really highlight the processes that the satellites are observing."

The first dataset showing the thickness of sea ice for a whole year in the Arctic has been published in the journal Nature.