Google DeepMind Unveils AI that Predicts Cyclones

Google DeepMind has introduced Weather Lab, an interactive platform showcasing its new AI-powered cyclone prediction model.

The tool uses a deep neural network trained on decades of historical data and nearly 5,000 real cyclone events to forecast a cyclone’s track, intensity, size, and structure.

Weather Lab provides real-time and historical cyclone predictions, comparing outputs from traditional and AI models.

"This model can predict a cyclone’s formation, track, intensity, size and shape — generating 50 possible scenarios, up to 15 days ahead," Google said in a blog post.

In one test, the AI accurately predicted Cyclone Alfred’s weakening to a tropical storm and its eventual landfall near Brisbane, Australia.

The system generates multiple simulations, enhancing the accuracy of forecasts.

"Internal testing shows that our model's predictions for cyclone track and intensity are as accurate as, and often more accurate than, current physics-based methods," it adds.

Developed in partnership with the U.S. National Hurricane Center (NHC), the AI model aims to improve early warnings, allowing officials to better prepare, allocate resources, and mitigate damage.

The project marks a step forward in using AI to address real-world climate challenges, offering a more informed response to natural disasters like tropical cyclones.

Last month, Google DeepMind unveiled SignGemma, its most advanced model for translating sign language into spoken text.

Set to join the Gemma family later this year, SignGemma supports multiple sign languages but is optimised for American Sign Language (ASL) and English.