Google DeepMind open-sources SatWeave and PARDIG weather datasets

Google DeepMind announced on November 13, 2026, that it is open-sourcing two datasets, SatWeave and PARDIG, that power its WeatherNext 3 weather forecasting model, making them freely available on Google Cloud Storage and Google Earth Engine, as detailed in its blog post on opening the datasets behind its AI weather models.

What's new in SatWeave and PARDIG

• SatWeave combines imagery from five geostationary satellites: GOES-East, GOES-West, Meteosat 0, Meteosat Indian Ocean, and Himawari.
• SatWeave provides 17 visible, water-vapor, and infrared channels, updated hourly, with more than a decade of historical archive.
• PARDIG stands for Precipitation AI Reanalysis: Densely Inferred GPM, with roughly 10-kilometer spatial resolution, 15-minute temporal intervals, and an average latency of 5 hours.
• PARDIG's historical archive spans 1999 to 2026.
• PARDIG improves hourly precipitation detection skill over IMERG-Final by more than 30% on light-to-moderate precipitation.
• Both datasets underpin Google DeepMind's WeatherNext 3 forecasting model and are available on Google Cloud Storage, with PARDIG also available in Google Earth Engine.

Why it matters for AI builders

The release gives developers and researchers free access to the same satellite imagery and precipitation reanalysis data that power Google's WeatherNext 3 model, letting them build their own forecasting or climate tools on the same inputs Google uses. Google lists potential applications including solar power forecasting, wildfire monitoring, flood prediction, landslide modeling, and drought assessment.

Availability

Both SatWeave and PARDIG are available now on Google Cloud Storage, and PARDIG is also available in Google Earth Engine, the same sources that feed Google DeepMind's WeatherNext 3 forecasting model.

Sources

• Opening the datasets behind our AI weather models (Google DeepMind, November 13, 2026, official)

Written with AI assistance, strictly from the sources listed above. Before it went live, automated checks verified its dates, sources and every language edition.