A recently-developed AI (Artificial Intelligence) algorithm will help researchers better forecast future arctic sea ice conditions over future months.
The more accurate prediction might help underpin early-warning systems that help protect the Arctic wildlife and coastal communities from the outcomes of ice loss.
The news of the AI was published this week in Nature Communications, described y an international team of researchers conducted by The Alan Turing Institute and British Antarctic Survey (BAS).
IceNet, as it was dubbed, took up the challenge of making precise Arctic sea ice forecasts for the future season, something that puzzled researchers for many decades.
Sea ice is extremely difficult to forecast due to its intense correlation with the atmosphere and ocean.
Its sensitivity to global warming made it lose 50% of its mass over the past four decades.
That surface loss is equivalent to an area roughly 25 times greater than Great Britain.
The ever speeding up climate change has dramatic consequences for the climate, its ecosystems, and numerous communities whose activities are related to the seasonal sea ice cycle.
IceNet has an estimated accuracy of 95% in predicting whether sea ice will make its apparition two months ahead.
Tom Andersson, the project’s leader and Data Scientist at the BAS AI Lab, explained:
“The Arctic is a region on the frontline of climate change and has seen substantial warming over the last 40 years. IceNet has the potential to fill an urgent gap in forecasting sea ice for Arctic sustainability efforts and runs thousands of times faster than traditional methods.”
Here’s a short video to help you better understand the process:
In contrast to regular forecasting systems that try to model the laws of physics directly, the authors build IceNet according to DeepLearning laws and ideologies.
That approach helps IceNet “learn” how sea ice changes over thousands of years worth of climate simulation data, plus decades of observational data to figure out the evolution of Arctic sea ice.