Foundation models and downstream applications for Earth Observation and Weather and Climate.
IBM and NASA ❤️ Open Source AI
IBM and NASA have teamed up to create a family of AI foundation models for Earth called Prithvi. These geospatial foundation models make AI accessible for open science users, startups, and enterprises by simplifying model training and deployment.
Three foundation models have been released to date: Prithvi-EO-1.0 and Prithvi-EO-2.0 which uses earth observation data from NASA’s Harmonized Landsat and Sentinel-2 (HLS), and Prithvi-WxC-1.0 which uses weather and climate data from NASA’s MERRA-2.
Technical papers for the Prithvi-EO models can be found here and the Prithvi-WxC model here. Each foundation model has been finetuned for several tasks using IBM TerraTorch – model cards, checkpoints and code to get you started with the finetuned models can be found in the links above.
IBM has also developed enterprise-ready finetuned versions of the Prithvi models, which are named Granite. Please visit the IBM Granite Geospatial Hugging Face for more information.
The Prithvi models would not have been possible to develop without exceptional collaborators. This includes Forschungszentrum Jülich super computing centre who co-developed Prithvi-EO-2.0 and Oak Ridge National Lab who co-developed Prithvi-WxC-1.0.
By embracing the principles of open AI and open science, IBM and NASA are actively contributing to the global mission of promoting knowledge sharing and accelerating innovations. We hope that the AI community find our efforts useful and that our models help fuel progress in addressing critical environmental challenges.
If you have any questions, please visit our discussions page.
