Aardvark Weather, a groundbreaking AI system, delivers accurate forecasts with a fraction of the computing power.

A major breakthrough in AI-driven weather prediction could transform how forecasts are made, significantly reducing costs and computing power while increasing speed and accuracy.

Researchers have developed Aardvark Weather, an AI-based forecasting system capable of producing accurate predictions in a fraction of the time required by traditional supercomputer models. Unlike existing forecasting systems, which require massive computing power and teams of experts, Aardvark can be run by a single researcher using a desktop computer.

The system, developed by the University of Cambridge, the Alan Turing Institute, Microsoft Research, and the European Centre for Medium-Range Weather Forecasts (ECMWF), was unveiled in Nature on Thursday.

Richard Turner, professor of machine learning at Cambridge, described Aardvark as a complete departure from traditional forecasting.

“The writing’s on the wall that this is going to transform things,” he said. “It’s going to be the new way of doing forecasting.”

Faster, Smarter, and More Accessible Forecasting

Aardvark uses AI models trained on raw weather data from satellites, weather balloons, ships, planes, and weather stations worldwide. The system can generate forecasts tens of times faster while using thousands of times less computing power than current supercomputer-based methods.

Turner explained that Aardvark could provide bespoke, industry-specific forecasts, such as predicting temperatures for African agriculture or wind speeds for European renewable energy companies.

This approach eliminates the need for years of research and large development teams, making high-quality weather predictions more accessible to policymakers, businesses, and even developing nations.

Dr Scott Hosking, director of science and innovation at the Alan Turing Institute, called the breakthrough a “democratisation of forecasting.” He said it could assist with emergency planning, climate research, and disaster response.

Dr Anna Allen, lead author from the University of Cambridge, highlighted its potential for predicting natural disasters, including hurricanes, wildfires, and tornadoes, as well as monitoring air quality, ocean changes, and sea ice shifts.

How AI is Outperforming Traditional Forecasting

Aardvark builds on previous AI research by Huawei, Google, and Microsoft, which showed that an AI model could replace the numerical solver, a key component in weather prediction.

The ECMWF has already started integrating this AI-driven approach into its forecasting systems.

Despite using only 10% of the data required by traditional systems, Aardvark has already outperformed the US national GFS forecasting system in certain areas and is proving competitive with the US Weather Service forecasts.

With its ability to provide eight-day forecasts compared to the current five-day limit, Aardvark is set to reshape the future of meteorology, making forecasting faster, more accurate, and more accessible than ever before.