For generations, Arctic residents rely on seasonal sea ice, which increases and recedes throughout the year. Polar bears and marine mammals rely on it as a hunting place and resting place; indigenous people fish from openings on the ice known as ice lakes and travel from one place to another using the famous route through the ice. But according to a report in May 2021, since 1971, the Arctic air and water have warmed three times faster than the rest of the planet. Arctic Council report, And this warming is causing ice to expand and contract in unpredictable ways.
Some scientists and research companies are now deploying artificial intelligence-driven tools to more accurately and timely predict which parts of the Arctic Ocean will be covered by ice, and when. Artificial intelligence algorithms complement existing models that use physics to understand what is happening on the surface of the ocean, a dynamic area where cold underwater ocean currents meet strong winds to form floating ice rafts.This information is becoming more and more valuable to Tribe members of the ArcticCommercial fishermen in places such as, Alaska, and global shipping companies interested in taking shortcuts in open waters.
Polarctic CEO Leslie Canavera (Leslie Canavera), a scientific consulting company headquartered in Lorton, Virginia, has developed artificial intelligence-based predictive models. He expressed the uncertainty about the speed of climate change. As the existing sea ice models become increasingly inaccurate. That’s because they are based on rapidly changing environmental processes.
“We don’t have a good understanding of climate change and what is happening on Earth. [Arctic] System,” said Canavera, who is a member of the Yup’ik tribe and grew up in Alaska. “We have statistical models, but then you will see more averages. Then you have artificial intelligence, which can see the trends in the system and learn. “
Existing physics-based models capture hundreds of years of scientific records on ice conditions, current weather conditions, the speed and location of polar jets, cloud cover, and ocean temperature. These models use this data to estimate future ice cover. But processing these numbers requires a lot of computing power, and it takes hours or days to generate forecasts using traditional programs.
Although artificial intelligence also requires complex data and a large amount of initial computing power, according to Data Thomas Anderson, once the algorithm is trained on the appropriate amount and type of data, it can detect weather conditions faster than physical-based models. model. Scientists from the British Antarctic Survey have developed an artificial intelligence ice prediction called IceNet. “As we found in our model IceNet, the speed of artificial intelligence methods can be increased thousands of times,” Anderson said. “And they will learn automatically. Artificial intelligence is not smart. It will not replace physics-based models. I think these two sources of information are being used in the future.”
Anderson and his colleagues published their new sea ice prediction model in the journal in August Nature Communications. IceNet uses a form called artificial intelligence Deep learning (It is also used to automatically detect credit card fraud, operate self-driving cars, and run personal digital assistants.) According to Arctic simulations, train yourself to provide six-month forecasts for every 25 km square in the area between 1850 and 2100. Climate and actual observational data recorded from 1979 to 2011.Once the model is trained and given the current weather and ocean conditions, IceNet beats the leading physics-based model in seasonal prediction of the presence or absence of sea ice. In each grid, especially in summer, when ice When retreat occurs once a year, according to nature Learn.