Cambridge scientists have made a major breakthrough in weather forecasting after developing a new AI prediction model that is tens of times better than current systems.
The new model, called Aardvark Weather, replaces the supercomputers and human experts used by forecasting agencies with a single artificial intelligence model that can run on a standard desktop computer.
This turns a multi-stage process that takes hours to generate a forecast into a prediction model that takes just seconds.
“Aardvark reimagines current weather prediction methods, offering the potential to make weather forecasts faster, cheaper, more flexible and more accurate than ever before,” said Richard Turner, a professor of machine learning in the Department of Engineering at the University of Cambridge.
Tests of the Aardvark model revealed that it is able to outperform the United States national GFS forecasting system using just 10 per cent of the input data, leading researchers to say it could offer a “revolution in forecasting”.
The researchers noted that its simple design and ability to run on standard computers means it has the potential to be used to create bespoke forecasts for a huge range of industries – from predicting wind speeds for offshore European wind farms, to rainfall and temperature predictions for farmers in developing countries.
“Aardvark’s breakthrough is not just about speed, it’s about access,” said Dr Scott Hosking, Director of Science and Innovation for Environment and Sustainability at the Alan Turing Institute.
“By shifting weather prediction from supercomputers to desktop computers, we can democratise forecasting, making these powerful technologies available to developing nations and data-sparse regions around the world."
Anna Allen from the University of Cambridge, who led the research, added: “These results are just the beginning of what Aardvark can achieve. This end-to-end learning approach can be easily applied to other weather forecasting problems, for example hurricanes, wildfires, and tornadoes.
“Beyond weather, its applications extend to broader Earth system forecasting, including air quality, ocean dynamics, and sea ice prediction.”
The new AI weather model was detailed in a study, titled ‘End-to-end data-driven weather prediction’, published in the journal Nature.
