Google's AI Weather Model Gets Sharper, Faster

Google's WeatherNext 3 marks a step forward in AI-driven weather forecasting by integrating live satellite data, a departure from traditional models that rely on supercomputers and physics simulations. The company highlights that this approach enables faster predictions and higher resolution, addressing time-lag issues and improving accuracy for fast-moving weather events. The model's ability to provide hourly updates at up to 5-kilometer resolution is a notable improvement over its predecessor's 6-hour cycle and 25-kilometer grid.

The push for higher accuracy in underserved regions, where ground-based rain gauges are sparse, suggests a focus on global coverage. Additionally, the model's applications extend to renewable energy, with wind speed predictions at turbine height, reflecting Google's interest in supporting clean energy as its data center energy needs grow. While AI models are gaining traction, the article notes they still work alongside physics-based models, with agencies like the US National Hurricane Center using them to enhance forecasts, implying a complementary rather than replacement role for now.