Sign up to save tools and stay up to date with the latest in AI
bg
bg
1

New AI Weather Forecasting Model Improves Accuracy Up To 20%

Mar 03, 2025 - forbes.com
The European Centre for Medium-Range Weather Forecasts (ECMWF) has launched the Artificial Intelligence Forecasting System (AIFS), which it claims is the first fully operational open AI model for weather forecasting. AIFS reportedly outperforms traditional physics-based models by up to 20% in some measures and can predict hurricane tracks 12 hours further ahead than existing models. The system was refined over 18 months with input from Member States and users. AIFS is designed to run alongside existing models, using the same data sources but leveraging historical data to evaluate real-time conditions, unlike the physics-only approach of the Integrated Forecasting System (IFS).

AIFS is also noted for its energy efficiency, using approximately 1,000 times less computing energy than traditional simulations, which could reduce supercomputing emissions and costs. ECMWF Director-General Florence Rabier described AIFS as a "milestone" that will make high-quality forecasts freely available, potentially improving disaster preparedness and renewable energy planning. The system is seen as complementary to existing models, providing a range of forecasting products to meet diverse user needs. ECMWF emphasizes collaboration with 35 nations to advance weather science and improve global predictions.

Key takeaways:

  • The European Centre for Medium-Range Weather Forecasts has launched the Artificial Intelligence Forecasting System (AIFS), claiming it outperforms existing models by up to 20% in some measures.
  • AIFS is the first fully operational open machine learning forecasting model, predicting hurricane tracks 12 hours further ahead than other models.
  • ECMWF claims AIFS uses approximately 1,000 times less computing energy than physics-based simulations, making it more efficient and faster.
  • AIFS runs alongside existing models, using historical data to evaluate real-time conditions, and is seen as complementary to traditional physics-based models.
View Full Article

Comments (0)

Be the first to comment!