Google Unveils New AI Weather Model With Faster, More Accurate Forecasts

Google DeepMind has launched WeatherNext 2, an AI-driven weather forecasting system promising to generate global predictions eight times faster than traditional methods. This innovative model produces hundreds of weather scenarios in under a minute, enabling timely updates crucial for responding to extreme weather conditions associated with climate change. WeatherNext 2 is already operational within Google's Search, Gemini, Pixel Weather, and Maps, with plans for broader integration. It surpasses the previous version, WeatherNext Gen, in both speed and resolution—providing forecasts at one-hour intervals rather than six and achieving a sixfold increase in data resolution. The advances stem from a new approach called Functional Generative Networks, which allows the model to better understand and predict how different weather variables interact. This enhanced system significantly improves accuracy on measures such as temperature, humidity, and cyclone tracking, benefiting decision-making in sectors reliant on accurate weather data, from agriculture to energy management.

Source 🔗