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

Google DeepMind has introduced a new AI-powered weather forecasting system named WeatherNext 2, capable of generating global weather predictions eight times faster than traditional models and producing results in under a minute. This system is designed to enhance preparedness for severe weather events in a climate of increasing natural disasters. The model works by generating hundreds of potential weather scenarios from a single data point and operates on specialized Google Tensor Processing Units. WeatherNext 2 has been integrated into several Google products like Search, Gemini, and Maps, with plans for wider deployment. The new approach, called Functional Generative Networks (FGN), has improved accuracy for critical metrics like wind and cyclone tracking. The model enhances the precision of forecasts in various weather conditions and demonstrates significant performance improvements over its predecessor, achieving substantial gains in predicting temperature and wind patterns.

Source 🔗