£1 million to create tropical storm forecast methods using artificial intelligence has been awarded to researchers at the National Centre for Atmospheric Science, University of Leeds, and the UK Centre for Ecology and Hydrology.

The new research project, known as Nowcasting with Artificial Intelligence for African Rainfall Intelligence for African Rainfall (NAIAR), aims to use artificial intelligence (AI) to create nowcasting for tropical storms across Africa – a process which can predict weather conditions 0 to 6 hours ahead.

There is an urgent need for information about storms, and the related risks to lives and livelihoods, across Africa. Tropical storms are unpredictable and can change rapidly. For this reason, these kinds of storms can be particularly hazardous and challenging to predict accurately. 

Nowcasting informs people about what is currently happening, and provides information about how a storm is likely to move and develop over the next few hours. 

The lack of weather radar in most African countries means that nowcasting is almost completely absent. NAIAR will deliver information on storm hazards within minutes to users across Africa. 

Other projects led by the National Centre for Atmospheric Science, such as GCRF African SWIFT, have shown that satellite technology can be used instead of weather radar for nowcasting storms.

Existing satellite and AI nowcasting methods are limited in their ability to forecast realistic storms. NAIAR aims to combine machine-learning (a form of AI), theoretical fluid dynamics, operational prediction, and meteorology to innovate satellite and AI nowcasting of tropical storms, using data from the Meteosat Third Generation satellite.

NAIAR was awarded £1 million from UK Research and Innovation, through the Natural Environment Research Council.