Artificial intelligence is being brought into lunar science as NASA and IBM Research, along with academic institutions, launch the NASA-IBM Lunar Foundation Model to help researchers analyse the Moon’s surface faster and more efficiently.
The model is among the 1st open-source AI foundation models developed specifically for lunar science. It is trained mainly on data from NASA’s Lunar Reconnaissance Orbiter (LRO) and is publicly available on Hugging Face, with its complete codebase available on GitHub for testing and experimentation.
The model can help scientists process large volumes of lunar data, study the Moon’s geological history, map its rugged surface and support planning for future lunar research.
“NASA has spent decades building an extraordinary scientific record of the Moon, but collecting data is only part of the job,” said Kevin Murphy, chief science data officer and acting chief data and AI officer at NASA Headquarters. “We also have to make data easier for scientists to explore and use. The NASA-IBM Lunar Foundation Model shows what’s possible when we bring AI to NASA’s petabytes of scientific data. That’s a real opportunity we see with AI: turning large-scale data into new discoveries.”
Unlike traditional AI models that are trained from scratch for individual tasks, foundation models are pre-trained on large, unlabeled datasets. Their broad knowledge allows them to be quickly adapted to different scientific applications with smaller amounts of labelled data.
Data collected by LRO over the past 17 years provided the foundation for the model. The mission covers most of the Moon in detail and has produced a dataset larger than those from all other NASA planetary missions combined.
The model was trained on around 2 million image tiles, including more than 1 million high-resolution camera images at 1-meter resolution and nearly 964,000 multispectral images at 100-meter resolution. It also uses high-resolution imagery and terrain data from NASA’s GRAIL, Lunar Prospector and JAXA’s Selenological and Engineering Explorer missions.
Researchers can adapt the model to tasks such as crater mapping, identifying young volcanic features and estimating the location and stability of potential polar ice using relatively small amounts of labelled data.
For polar research, the model can help estimate where ice deposits may remain stable on or below the surface. Permanently shadowed regions can remain cold enough to preserve ice for billions of years, making them important for understanding lunar history and identifying potentially useful resources for future exploration.
The model can also help identify irregular mare patches, unusual volcanic features that appear relatively young. Mapping these structures could provide new insights into the Moon’s thermal evolution.
Crater mapping is another key application. Since craters are created by impacts, counting and measuring them helps scientists estimate the age of lunar surfaces and reconstruct aspects of solar system history.
Across evaluated tasks, the NASA-IBM model matched or exceeded several strong baseline models. It delivered comparable results for crater mapping and irregular mare patch segmentation, while showing a clear advantage in estimating polar ice stability.
The model forms part of NASA’s broader AI for science strategy and its ongoing collaboration with IBM. Other models from the partnership include the Prithvi Models, which use Earth observation data for applications such as disaster monitoring, flood mapping, crop yield prediction and hurricane prediction, and the Surya Model, which uses solar observations to predict space weather events such as solar flares.
NASA teams from Marshall Space Flight Center, the Science Mission Directorate, Goddard Space Flight Center and Ames Research Center worked with researchers from industry and academia to develop the lunar model.
The project also includes machine learning-ready pre-training datasets and benchmark collections and is integrated into the open-source TerraTorch toolkit. A technical paper has also been released to support reproducible research and allow scientists worldwide to build, compare and improve AI models for lunar exploration.
The science team included experts from the Universities Space Research Association, SETI Institute, University of Maryland, Baltimore County, Howard University, NASA’s Planetary Science Division, NASA Ames and NASA Goddard.
Also read: Viksit Workforce for a Viksit Bharat
Do Follow: The Mainstream LinkedIn | The Mainstream Facebook | The Mainstream Youtube | The Mainstream Twitter
About us:
The Mainstream is a premier platform delivering the latest updates and informed perspectives across the technology business and cyber landscape. Built on research-driven, thought leadership and original intellectual property, The Mainstream also curates summits & conferences that convene decision makers to explore how technology reshapes industries and leadership. With a growing presence in India and globally across the Middle East, Africa, ASEAN, the USA, the UK and Australia, The Mainstream carries a vision to bring the latest happenings and insights to 8.2 billion people and to place technology at the centre of conversation for leaders navigating the future.