WeatherNext 2 advances cyclone track, intensity and wind forecasts

Google DeepMind and Google Research introduced WeatherNext 2 on 6 August. Google says the model can produce forecasts up to eight times faster, at hourly resolution, and improve on the previous WeatherNext model across 99.9% of evaluated variables and lead times.

A companion announcement reports state-of-the-art accuracy for tropical cyclone track, intensity and wind structure, with the potential for an extra day of warning. The system generates many possible scenarios, which is important because operational decisions need to account for uncertainty rather than one deterministic path.

This is a forecasting aid, not a replacement for weather agencies. Businesses exposed to severe weather should use official warnings and local expertise, then test whether probabilistic forecasts improve their own continuity, logistics or emergency-response decisions.