Artificial intelligence can improve the accuracy of short-term weather forecasts, according to new research by the UK Meteorological Office and Alphabet-owned AI firm DeepMind.

The organisations’ research was aimed at predicting short term weather fluctuations - within the next two hours - otherwise known as “nowcasting”.

The research, published in science magazine Nature, claimed that current supercomputer-based weather prediction models are less effective at predicting short term weather fluctuations.

DeepMind’s tool - dubbed “deep generative modelling” or DGM - was evaluated alongside two existing rain prediction tools in a blind test involving over 50 Met Office meteorologists.

The vast majority - 89 per cent of the panel - ranked the tool first for accuracy and usefulness.
Deepmind’s model was based on radar imagery collected by the Met Office between 2016 and 2018.

The Met Office sends beams into the atmosphere and evaluates how long the beams take to reflect to estimate moisture in the atmosphere, which it then uses to predict rain fall.

Deepmind, which was founded in 2010 in London, gained public recognition in 2016 when their AI system AlphaGo beat Go world champion Lee Sedol in a five-game tournament.

“Improving the accuracy of short-term forecasting is an incredibly important endeavour,” said Niall Robinson, head of partnerships and product innovation at the Met Office. “Extreme weather has catastrophic consequences including loss of life and, as the effects of climate change suggest, these types of events are set to become more common.”

He added: “This research demonstrates the potential AI may offer as a powerful tool for improving our short-term forecasts and our understanding of how our weather patterns are evolving."

Share Story: