NOAA High Resolution Rapid Refresh (HRRR)

sam_annual_hr_ir_tropopause_difference_99th_percentile

sam_clim_hr_ir_tropopause_difference_99th_percentile

sam_seas_hr_1995_2021_cloud_top_height_99th_percentile

South America Climatology Hourly IR Anvil Hour Count

The nested-Experimental Artificial Intelligence (AI) Global and Limited-area Ensemble (nested-EAGLE) is part of the NOAA EAGLE product suite that is designed to bridge the skill of short and medium range weather forecasts (see arXiv for details). It is based on the Anemoi framework developed by the European Centre for Medium-Range Weather Forecasts (ECMWF) and partner agencies, and is trained on the combination a 0.25° Global Forecast System (GFS) data and 6-km High-Resolution Rapid Refresh (HRRR) data over CONUS that produces 10-day forecasts for a range of atmospheric variables. This work was a joint effort between NOAA Research Laboratories and the Earth Prediction Innovation Center (EPIC), and the National Weather Service (NWS). Project EAGLE provides NOAA, the weather enterprise (including government, industry, and academia), and the broader community members with the ability to rapidly test, develop, and demonstrate AI models for global and limited-area ensemble forecasting. For more information, please visit the EPIC website, EAGLE Overview page or the EAGLE repository on GitHub. The nested-EAGLE data are available as netCDF files. It's offered on a free, unsupported basis. We do not recommend using it for any critical operational purposes. Domain Information The nested-EAGLE global data is at a 0.25-degree resolution. To account for the nested grid, the high-resolution nest over the CONUS domain is upsampled by conservatively regridding it from the original 6-km resolution to the 0.25-degree grid. CONUS data at the original 6-km resolution is distributed in a separate collection (nested-eagle-conus). Technical Specifications Archive length October 1st, 2024 - Present Update frequency This model generates forecasts in near-real-time and is initialized 4 times daily for 00, 06, 12, and 18 Z cycles. Forecasts are typically available 6 hours after initialization time. Forecast length Ten days (240 hours) with predictions every 6 hours Levels 2D: surface, 10 meter, and 80 meter 3D: 100, 150, 200, 250, 300, 400, 500, 600, 700, 850, 925, and 1000 hPa Specs The data are stored as netCDF files, one netCDF file per forecast. See the example Jupyter notebooks for guidance on how to load the data. License Information These data are in the worldwide public domain and are licensed under CC BY 4.0.How to Cite The Earth Prediction Innovation Center (EPIC) and the NOAA AI for Numerical Weather Prediction (AI4NWP) Working Group. (2026). The nested-EAGLE (Experimental AI Global and Limited-area Ensemble forecast system) v1.1.0 (Version v1.1.0). Zenodo. https://doi.org/10.5281/zenodo.19672026

sam_monthly_ir_tropopause_difference_99th_percentile