Google DeepMind AI improves tropical cyclone forecasting

Google DeepMind and Google Research have introduced WeatherNext Cyclones (WN-C), an artificial intelligence system designed to forecast tropical cyclones with greater lead time than current operational models. According to a study published in the journal Nature, the AI can predict the track, intensity, and wind radii of storms approximately one day further in advance than existing leading models while maintaining comparable accuracy.

The system was trained on nearly 20 terabytes of global atmospheric data and the International Best Track Archive for Climate Stewardship (IBTrACS) database, which includes historical records for nearly 5,000 storms. By analyzing these patterns, WN-C can generate large ensembles of possible global weather and cyclone scenarios extending up to 15 days into the future.

Meteorologists note that increasing warning times is critical for disaster preparedness, as tropical cyclones have caused over 700,000 deaths and $1.4 trillion in economic damage globally over the last 50 years. The technology aims to bridge the gap between tracking large-scale atmospheric currents and the localized phenomena that drive storm intensity.