NASA’s Prithvi Becomes First AI Geospatial Foundation Model In Orbit
Editor’s note: This article was updated on May 7, 2025 to include a link to the preprint article for this research.
A team of researchers from the University of Adelaide and the SmartSat Cooperative Research Center in South Australia have successfully downloaded and demonstrated NASA and IBM’s Prithvi Geospatial geospatial artificial intelligence (AI) foundation model aboard two orbiting platforms, making it the first geospatial foundation model to be deployed in orbit. Trained on 13 years of data, Prithvi can facilitate a wide variety of Earth observation tasks.
By uploading a compressed version of Prithvi to the South Australian government’s Kanyini satellite and Thales Alenia Space’s ISS Mounted Accessible Global Imaging Nod-e (IMAGIN-e) payload aboard the International Space Station, the researchers tested the model’s flood and cloud detection performance on two different orbital platforms and computing environments. The team shared their results in a preprint paper.
The team chose Prithvi for their research because of its strong generalizability to Earth observation tasks and its availability as an open source model.
“If Prithvi wasn’t open source, I would have to train my own foundation model,” said Dr Andrew Du, the project’s principal investigator, a postdoctoral researcher at the University of Adelaide and an AI engineer at the SmartSat Cooperative Research Center. “Having this template freely available saved a lot of time and effort.”
A base model is an AI model trained on a huge amount of unlabeled data, which allows the model to start detecting patterns in the data that humans wouldn’t notice on their own. The model can then be refined for specific applications using much smaller amounts of labeled data.
“Prithvi is the first model of its kind to be deployed in orbit, and it demonstrates exactly why we make our AI models open source,” said Kevin Murphy, chief data scientist at NASA Headquarters in Washington, whose office led the collaboration that created Prithvi. “By sharing these tools with anyone who wants to use them, we accelerate scientific and technological development in the future. »
Developed by a team of data scientists from IBM and the NASA IMPACT team within the Office of Data Science and Computing at NASA’s Marshall Space Flight Center in Huntsville, Alabama, the Prithvi geospatial model was trained on the harmonized Landsat and Sentinel-2 dataset. This dataset compiles over a decade of global geospatial data from NASA and ESA (European Space Agency) Sentinel-2 satellites. Prithvi can be adapted for tasks such as floodplain mapping, disaster monitoring, and crop yield forecasting.
Kevin Murphy
NASA Chief Data Science Officer and Acting Chief Data Officer/Chief AI Officer
Earth observation satellites collect enormous amounts of data about our planet. Processing and analyzing data in orbit before the satellite sends it back to Earth can help researchers obtain information more quickly. However, active satellites often cannot accept significant software updates due to bandwidth limitations. The AI models they carry for data analysis therefore tend to be lightweight and highly specialized.
Researchers can use the flexibility of a base model to facilitate a wide range of Earth observation tasks in a single software architecture. If they want the model to take on a new task once the satellite is in orbit, they simply need to download a small additional decoder package – using much less bandwidth than downloading an entirely new model to the satellite.
Sending Prithvi into orbit is an early demonstration of how foundation models could transform Earth observation. In addition to data analysis, the basic models could eventually help scientists interact with the instruments collecting the data.
“A large language model is also a basic type of model,” Du said. “In the future, this could allow operators to interact with satellites in natural language, asking questions about onboard data or system status and receiving responses in a conversational manner.”
The NASA team behind Prithvi continues to work on open source baseline models trained on NASA data. A heliophysics model, Surya, was released in 2025 and the team intends to create basic models for planetary science, astrophysics as well as biological and physical sciences.
The Prithvi Geospatial Foundation Model is funded by the Office of the Chief Data Science Officer in NASA’s Science Mission Directorate at NASA Headquarters in Washington. The Office of the Chief Data Science Officer advances scientific discovery through innovative applications and partnerships in data science, advanced analytics, and artificial intelligence. To learn more about NASA AI Core Models and other AI tools for science, visit:
https://science.nasa.gov/artificial-intelligence-science
By Lauren Leese
Web Content Strategist for the Office of the Chief Data Science Officer
