Monday, September 7

Google and DeepMind have released WeatherNext 3, an AI weather model that skips traditional physics simulations and learns directly from real-time satellite data. This model produces hourly forecasts on a five-kilometer grid, five times sharper than its predecessor, capturing local weather events and terrain details with greater precision.

The model uses live geostationary satellite data to generate a fresh forecast every hour at up to five-kilometer resolution. It also trains on individual weather station data to capture coastlines, valleys, and mountains, and produces dense gridded fields for atmosphere and precipitation. The model is now live across Google Search, Maps, and Gemini, with up to 50 percent more accurate precipitation forecasts.

WeatherNext 3 is a significant improvement over previous AI weather models, which relied on numerical weather prediction (NWP) data and carried a six-hour delay. The new model's ability to learn directly from live satellite data and produce hourly updates means earlier and more accurate predictions, particularly for fast-developing storms or precipitation. This will be especially beneficial for regions where predictions have been less reliable, such as Latin America, Africa, and the Asia-Pacific.