NASA-IBM Unveil Open-Sourced Lunar Foundation Model with Enhanced Ice Detection

NASA and IBM have released the NASA-IBM Lunar Foundation Model as open source, giving scientists a powerful tool for studying the Moon. The model is built on TerraMind, an Earth-observation model developed by IBM with the European Space Agency, chosen for its ability to handle mixed data types and resolutions.

The model was trained on a unified dataset of over 2 million image tiles from nine instruments across four missions, including imagery from NASA's Lunar Reconnaissance Orbiter and gravity-field maps from the GRAIL mission. This dataset is a significant improvement over previous hand-sifted maps and narrow, task-specific models.

In initial benchmarks, the model cut root-mean-square error in flagging likely ice deposits by up to 22% against a SwinV2-B baseline. The model also performed well in detecting volcanic features and craters at various scales, outperforming other models in some cases.