Google DeepMind Open-Sources WeatherNext 2 AI Model for Cyclone Forecasting
Google DeepMind and Google Research have open-sourced WeatherNext 2 (WN2), a weather forecasting AI model published in Nature that predicts cyclones with unprecedented accuracy. The model provides forecasters with an extra day of lead time compared to existing systems. This model shifts weather forecasting from resource-intensive supercomputers to modern GPU hardware like the NVIDIA H100, making advanced meteorological predictions more accessible. The extra day of warning for cyclones can significantly improve disaster preparedness and save lives. WeatherNext 2 is a global, medium-range atmospheric and cyclone forecasting model that can generate forecasts 8x faster with up to 1-hour resolution. The open-source code is available on GitHub in the google-deepmind/weathernext repository.
## BACKGROUND
Traditional weather forecasting relies on Numerical Weather Prediction (NWP) models, which require massive supercomputers to simulate physical atmospheric processes. Recently, deep learning models trained on historical weather data have emerged as a faster, highly accurate alternative that can run on standard GPU accelerators.