Google DeepMind's WeatherNext 3 Delivers Hourly, 5-Kilometer Global Forecasts

Google DeepMind and Google Research unveil WeatherNext 3, an AI weather model that ingests live satellite data to produce hourly forecasts at up to 5-kilometer resolution.

Google DeepMind and Google Research have introduced WeatherNext 3, which they describe as their most advanced and accurate global weather AI model to date. Unlike earlier versions trained on supercomputer-driven physics simulations that carry a six-hour lag, the new model ingests a live mosaic of geostationary satellite data to generate a fresh forecast every hour.

WeatherNext 3 resolves surface temperature and moisture at 5-kilometer resolution — roughly five times sharper than WeatherNext 2's 25-kilometer, six-hour grid — and adds variables built for renewable energy, including 100-meter wind speeds for turbines and cloud cover for solar farms.

The company says the model delivers up to 50% more accurate precipitation forecasts a day or more in advance, with the biggest gains in regions historically underserved by high-resolution forecasting. It is now powering weather features in Google Search, the Gemini app, Google Maps, and Google Earth Engine.

Sources: Google | Gizmodo