Scientists have built an AI tool that forecasts Arctic sea ice conditions, which could help protect local wildlife and people from changes caused by global warming. The deep learning system, named IceNet, was developed by a research team led by British Antarctic Survey (BAS) and The Alan Turing Institute. The model was trained on climate simulations and observational data to forecast the next six months of sea ice concentration maps. In tests, the system was almost 95% accurate in predicting whether sea ice will be present two months ahead. Study lead author Tom Andersson said IceNet outperforms the leading physics-based models — and runs thousands of times faster:
AI takes on global warming
The Arctic has warmed at around two to three times the rate of the global average. As a result, the size of sea ice in the summer is half of what it was just 40 years ago. The decline is having a huge impact on local people and wildlife. IceNet could help them adapt to future changes by providing early warnings about the timing and locations of sea ice loss. Per the study paper: The researchers are now investigating whether adding ice thickness to IceNet’s inputs improves its accuracy. They will also implement an online version of the tool that operates on a daily temporal resolution, which could enhance performance at short lead times. The system will not prevent global warming, but it could at least mitigate some of the damage. Greetings Humanoids! Did you know we have a newsletter all about AI? You can subscribe to it right here.
