Overview

• Aurora, a foundation model developed by Microsoft and partners, delivers 20–25% greater accuracy in five-day cyclone forecasts compared to seven leading forecasting centers.
• The AI system correctly predicted Typhoon Doksuri's track four days in advance, outperforming traditional models that missed the storm’s path.
• Unlike physics-based models requiring supercomputers, Aurora uses machine learning to analyze historical data, enabling high-resolution forecasts on standard desktops.
• Aurora's flexibility allows fine-tuning for diverse Earth system predictions, including air pollution, ocean dynamics, and extreme weather events.
• Experts predict major weather centers will integrate AI alongside traditional methods within 2–3 years, with AI-driven forecasts becoming primary by 2027–28.