A new suite of operational, artificial intelligence (AI)-driven global weather prediction models is launching at the National Oceanic and Atmospheric Association (NOAA), supposedly advancing speed, efficiency, and accuracy in forecasts.
“NOAA’s strategic application of AI is a significant leap forward in American weather model innovation. These AI models reflect a new paradigm for NOAA in providing improved accuracy for large-scale weather and tropical tracks, and faster delivery of forecast products to meteorologists and the public at a lower cost through drastically reduced computational expenses.” – Neil Jacobs, Ph.D., NOAA administrator.
According to the agency, the suite includes three distinct applications. AIGFS (Artificial Intelligence Global Forecast System) provides faster global forecasts while using about 99.7% less computing power than the traditional GFS, and it improves large-scale and tropical cyclone track forecasts.
AIGEFS (Artificial Intelligence Global Ensemble Forecast System) is an AI-based ensemble model designed to provide multiple forecast scenarios, performing similarly to the typical GEFS while using 9% of the computing resources and extending the useful forecast skill by 18-24 hours. Finally, the HGEFS (Hybrid-GEFS) is a first-of-its-kind hybrid “grand ensemble”, combining AI-based and physics-based ensemble models, consistently outperforming either both approaches alone in forecast uncertainty.
The model suite was developed through Project EAGLE, building on Google DeepMind’s GraphCast Model to improve the use of NOAA’s own data. This is a huge shift toward AI-enhanced weather predictions, potentially creating faster, more accurate, and lower-cost forecasts to better protect lives and property.
