Google DeepMind Just Rolled Out Its Most Accurate AI Global Weather Model (gizmodo.com)

Google DeepMind has announced the launch of WeatherNext 3, its most accurate global AI weather model to date, which promises to enhance hyper-local weather predictions significantly. Set to reshape forecasting from 2026 onwards, WeatherNext 3 utilizes real-time satellite data and traditional weather analysis to generate high-resolution forecasts every hour. By achieving a resolution of up to 5 kilometers for temperature and moisture, and 25 kilometers for wind speed, the model enables more precise predictions in areas with rapidly changing weather conditions, such as coastlines and mountainous regions. The significance of WeatherNext 3 lies in its potential to overcome limitations faced by traditional numerical weather prediction models, which often lag by several hours. This enhanced capability is particularly vital for regions like Latin America, Africa, and Asia-Pacific, where high-resolution forecasting has been restricted due to infrastructure costs. In addition to improving general weather predictions, WeatherNext 3 also addresses the energy sector, providing forecasts for wind speeds and solar radiation important for renewable energy operations. Already topping Operational WeatherBench’s leaderboard for accuracy, this advancement marks a substantial leap forward in the AI/ML community's ongoing efforts to integrate machine learning with meteorological science.