In short: The Aardvark season, an AI-based system, promises to increase weather forecast by rapidly transporting dozens of times using less computing power compared to current methods. The system has been developed by researchers at the University of Cambridge, with the support of Allen Turing Institute, Microsoft Research and European Center for Medium Range Weather Forecasts.
The speed and efficiency of modern forecasting systems is important, as traditional methods depend on powerful supercomputers and broad teams of experts, often requiring several hours to produce forecasting.
Recent innovations of technical giants like Huawei, Google, and Microsoft have displayed that AI can significantly improve the specific aspects of the forecast process, including numerical solvers, which are important in weather forecasts because they simulate how the atmospheric status develops over time. These companies have gained rapidly and more accurate predictions by integrating AI into these solvers.
As an example, Google is developing AI models for weather forecasting and currently marketing two models for its enterprise cloud customers. Developed by Google Deepmind, models use historical weather data 10 to 15 days in advance to predict future conditions.
Aardvark represents a significant advancement by changing traditional forecast processes with a single, streamlined machine-learning model. Using a standard desktop computer, it can process data from various sources, including satellites and weather stations, to generate global and local forecasts in minutes.
Richard Turner, Professor of the Engineering Department of Cambridge, said, “Aardvark re -explains the current weather methods, providing the weather forecasts rapid, cheap, more flexible and more accurate than ever.” “Aardvark is thousands of times faster than all previous weather forecasting methods.”
Despite working with only one fraction of data used by existing systems, Aardvark crosses the US National GFS forecast system in several major matrix and is competitive with the forecast of the National Meteorological Service, which usually involves several models and expert analysis.
Anna Alan, the first writer of Cambridge’s Computer Science and Technology Department, said, “These results are the only beginning of what Aardvark can achieve.” He said the end-to-end approach can easily be applied to other weather forecasting problems, such as storms, forest fire and tornado. It can also be used for the forecast of a comprehensive Earth system, including air quality, ocean dynamics, and sea ice prediction.
One of the most interesting aspects of Aardvark is its flexibility and simple design. Because it learns directly from data, it can be quickly adapted to produce Bispoke forecasts for specific industries or locations, whether it predicts temperature to support African agriculture or air conditions for European renewable energy firms. This is rapidly opposite with traditional systems, which require years of work by large teams to customize.
This capacity has the ability to change the weather prediction in developing countries, where expertise and access to computational resources are limited. Dr. from Allen Turing Institute. Scott Hosing said, “By transferring the weather prediction to desktop computers from supercomputers, we can democratization of the forecast, allowing these powerful technologies to be made available to developing countries and data-sparrow areas around the world.”
Aardvark is expected to play an important role in expanding the scope of weather forecast. Turner mentioned that the model may eventually make an accurate prediction of eight-day forecasts, which could cross the capabilities of the current model for three days. It keeps it as a transformational force in meteorological, with the adaptability and efficiency of Aardvark, Aardvark.
The next stages for Aardvark include developing a new team within the Alan Turing Institute that will deploy technology in the global south and integrate it in a comprehensive environmental forecasting initiative.
