For generations, a Arctic lands count the ice that grows in a year. Polar bears and aquatic mammals depend on them as a hunting ground and resting place; Indian Indians fish on the ice floes called polynyas, and they use well-known means of crossing the ice. But air and water in the Arctic have been three times hotter than the rest of the world since 1971, according to May 2021. Arctic Council report, and the heat is causing the glaciers to overflow and to rub in unexpected ways.
Some scientists and research companies are now launching artificial intelligence tools to provide accurate and timely forecasts for the Arctic ice sheet, and when. AI algorithms complement existing forms that use physics to understand what is happening on the surface of the oceans, a changing environment in which cold underwater currents are exposed to storms to form floating glaciers. This is becoming more and more common members of nations in the Arctic, commercial fishermen in places like Alaska, and international shipping companies who want to follow shortcuts on open waters.
Leslie Canavera, CEO of Polarctic, a science-based technology company based in Lorton, Virginia that has developed AI predictive models, says the unpredictable speed of climate change means that existing forms of marine ice are inaccurate. It’s because it comes from natural phenomena that move fast.
“We do not fully understand climate change and what is happening in the country [Arctic] system, ”says Canavera, a member of the Yup’ik tribe who grew up in Alaska. “We have statistical numbers, but you’re looking for more. Then you have a creative mindset, where you can see what’s going on in the system and learn.”
Existing physics models reflect centuries of scientific knowledge of the weather, the current climate, the speed and location of the polar jet stream, the number of clouds, and the temperature of the oceans. The models use the same to predict the future spread of ice. But it takes a lot of computing power to add numbers, and hours or days to make predictions using common software.
Although AI also requires complex data and a lot of basic computer power, once the information is taught on the right number and type of data, it can detect weather events faster than physical models, according to Thomas Anderson, data. a scientist at the British Antarctic Survey who developed an AI ice prediction called IceNet. “AI solutions can run thousands of times faster, as we have found in our brand, IceNet,” Anderson says. And they learn on their own. AI is not smart. It does not replace physics. I think the future uses all sources of information. ”
Anderson and his colleagues published their new model of sea ice in August magazine Nature Communications. IceNet uses a type of AI called in-depth study (also used to detect credit card fraud, self-driving, and computer-assisted driving) training to provide six-month forecasts in every 25-square-kilometer grid across the region, based on Arctic weather forecasts between the 1850s and into the 1850s. 2100 is a precise monitoring data recorded from 1979 to 2011. When the model was trained and equipped with state-of-the-art weather and ocean capabilities, IceNet won the leading physics-based model by making weather forecasts for the presence or absence of sea ice. in a large group of people, especially for the summer season, when the ice passes year after year, according to Nature learning.