WeatherNext AI: 1-Day Advance in Cyclone Forecasts

Google DeepMind's WeatherNext AI model has significantly improved cyclone forecasting. It offers an additional day of predictive accuracy for storm track, intensity, and wind structure. This advance is like a decade of meteorological progress, and it could save lives by providing earlier warnings for devastating tropical storms. WeatherNext outperformed traditional models in a study, delivering three-day forecasts as accurate as previous two-day predictions. During the 2025 hurricane season, it helped the National Hurricane Center anticipate Hurricane Melissa's rapid intensification and landfall in Jamaica. This early warning allowed for critical preparations, mitigating potential loss of life and damage. The model combines large-scale atmospheric models with high-resolution data, training on 20 terabytes of global atmospheric data and nearly 5,000 historical cyclones. It can produce a 15-day forecast in under a minute, generating up to 1,000 scenarios for each cyclone. WeatherNext is now open source, enabling the global research community to build on its capabilities. This extra day of warning could significantly reduce deaths and economic losses from tropical cyclones.