Artificial intelligence (AI) and machine learning (ML) are rapidly transforming numerical weather prediction (NWP), enabling faster and more scalable forecasting systems. As part of the National Oceanic and Atmospheric Administration’s (NOAA) broader AI strategy, the Earth Prediction Innovation Center (EPIC) is advancing next-generation forecasting through the Experimental Artificial Intelligence Global and Limited-area Ensemble (EAGLE) Project.
A key regional component of this effort is the HRRRCast ensemble system, developed by NOAA’s Office of Oceanic and Atmospheric Research (OAR) Global Systems Laboratory (GSL). HRRRCast ensemble was trained with four years of HRRR analysis data. HRRRCast ensemble This ensemble system now provides, in a computationally efficient way, information on forecast uncertainties at convection allowing scales.
The latest version of the HRRRcast Ensemble system, v3.0, joins the EAGLE family of models and includes the following updates: increased resolution to 3 km, thirty new diagnostics, and significantly improved composite reflectivity and precipitation forecasts.
With support from NOAA’s National Weather Service (NWS) Environmental Modeling Center (EMC), the HRRRCast Ensemble system runs hourly out to 48 hours at 0000, 0600, 1200 and 1800 UTC cycles, and out to 18 hours for all other cycles. The HRRRCast system supports applications in severe weather prediction and other short‑range, high‑impact weather events, with a particular focus on precipitation, reflectivity, and other key meteorological variables.
Within Project EAGLE, HRRRCast serves as a regional, convection-allowing AI component that complements global AI systems and traditional NWP models. It demonstrates how AI can:
- Emulate high-resolution regional models
- Enable scalable ensemble forecasting
- Support hybrid AI-Physics approaches
HRRRCast highlights the growing role of AI in enhancing forecast speed, scalability, and accuracy. As development continues, efforts will focus on improving model fidelity and expanding integration within the Unified Forecast System (UFS) framework. Through Project EAGLE, EPIC and its partners are helping shape a future where AI and traditional modeling work together to deliver more efficient and actionable forecasts. For model output, visit GSL’s HRRRCast page.
