The UK Meteorological Office has partnered with London-based technology firm DeepMind to research the use of AI to improve the accuracy of short-term weather forecasts and particularly the prediction of storms and heavy rain.
The project, which was recently published in Nature, focused on the ability to forecast precipitation up to two hours ahead – ‘nowcasting’. It found that current methods, which uses radar-based wind estimates, are not handled well by supercomputer models developed to forecast weather up to two weeks in advance.
An AI-powered approach, called Deep Generative Modelling (DGM), was able to create a system that ranked first for accuracy and usefulness, by 89% of a panel of 56 professional meteorologists when compared with two competitive methods.
Researchers trained DGM to predict the development of rain and snow by analysing three years of UK radar maps. The maps detailed how much rain was falling every five minutes at 1km spatial resolution.
The DGM focused on the probability of sequences such as rainfall patterns rather than determining the shape of protein molecules, something DeepMind’s systems have proven successful at in the past.
The model produced “realistic and spatiotemporally consistent” predictions over regions up to 954 miles by 795 miles and with lead times from five to 90 minutes ahead.
The results read: “When verified quantitatively, these nowcasts are skillful without resorting to blurring. We show that generative nowcasting can provide probabilistic predictions that improve forecast value and support operational utility, and at resolutions and lead times where alternative methods struggle.”
The authors went onto say that the research was undertaken not to work towards potentially automating meteorologists, but instead improving the work of human experts.
“AI could be a powerful tool, enabling forecasters to spend less time trawling through ever-growing piles of prediction data and instead focus on better understanding the implications of their forecasts,” added Shakir Mohamed, senior researcher at DeepMind.
