GraphCast is an AI model developed by Google DeepMind, designed for fast and accurate medium-term
weather forecasts. Instead of being based on traditional physical equations,
GraphCast uses machine learning and graph neural networks (GNNs). The system was trained using decades of
historical weather data from the ERA5 reanalysis archive of the European Center for Medium-Range Weather Forecasts
(ECMWF) to learn cause-and-effect relationships in weather evolution.
The model is initialized with the ECMWF analysis data and can generate forecasts for up to ten days in advance
. GraphCast works with a high resolution of 0.25 degrees. The datasets are updated twice daily with
near real-time data (00Z and 12Z initializations).
Copyright statement
Forecast maps derived from the AI Weather Prediction (AIWP) Model Reforecast Archive produced by CIRA and NOAA-GSL.
Citation: Radford et al, BAMS 106, E68-E76 (2025), DOI:10.1175/BAMS-D-24-0057.1.
Data are released as open data with no use restrictions; provided “as is” with no warranties.
