MANY MINDEDWeatherNext 3 Launch Enables Precision Agriculture Planning
Google DeepMind launched WeatherNext 3 on September 3, 2026, an AI weather forecasting model delivering hourly forecasts with five times higher spatial resolution than its predecessor. The model provides 5-kilometer resolution for surface variables like temperature and moisture, 10-kilometers for other surface data, and 25-kilometers for atmospheric conditions such as wind speed. By using real-time geostationary satellite data instead of traditional numerical weather prediction models, WeatherNext 3 reduces prediction lag for rapidly changing weather events. It was evaluated by Brightband as the most accurate global weather model at launch and is integrated into Google Search, Maps, Gemini, and Google Cloud.
This precision enables farmers to anticipate seasonal shifts and extreme weather with greater accuracy than previous models. The 5-kilometer resolution—critical for localized agricultural decisions—allows for targeted irrigation and crop management, while the atmospheric resolution supports better disaster preparedness. By reducing the time between weather events and forecasts, the model helps minimize crop damage and economic losses from sudden weather disruptions.
For the SUSTENANCE sector, WeatherNext 3 moves abundance by making agricultural planning more responsive to weather conditions. This directly supports food security in regions vulnerable to climate shocks by reducing the risk of crop failure and food waste. The model also advances GOODS and SECURITY through improved supply chain resilience and early warning systems for natural disasters.
What to watch: Brightband’s independent evaluation is current at launch but may require retesting as the model scales. The model’s real-world impact on food waste reduction remains unquantified in the source material.
Source: Google DeepMind