Google DeepMind AI improves cyclone warning times by 30 hours
TL;DR. Google DeepMind and Google Research developed WeatherNext Cyclones, an AI system that significantly enhances tropical cyclone forecasting accuracy. - The AI system outperformed existing models in tracking storm paths, intensity, and wind extent. - It provides over 30 hours of additional warning time for communities compared to traditional forecasts. - WN-C learned from decades of global weather data and nearly 5,000 historical cyclones.
- Google DeepMind and Google Research created WeatherNext Cyclones (WN-C), an AI model for tropical cyclone prediction.
- WN-C combines global atmospheric patterns with individual storm physics for improved forecasting.
- The AI offers 30 additional hours of warning time for storm location compared to European models.
- Its intensity forecasts for peak wind speeds are sharper than top US hurricane models.
- The model was trained on global weather analyses and historical cyclone data.