The rapid growth of geophysical data (from remote sensing, ground observations, and physically based models) offers new opportunities to advance Earth and space sciences through data-driven approaches and closer collaboration between domain experts and AI researchers. This session explores an emerging frontier where challenges in geosciences drive the development of novel AI methodologies with broader applicability.

We seek contributions highlighting innovative developments in AI for geosciences, including approaches that incorporate physical constraints, improve interpretability, and uncover new processes or relationships in complex systems. Topics may address a wide range of data regimes, from large-scale datasets to data-sparse environments critical for predicting high-impact events, as well as methods for robust learning, uncertainty quantification, and generalization.

Contributions are invited across the geosciences, including atmospheric and ocean sciences, hydrology, climate and weather modeling, space weather, planetary processes, solid Earth, and natural hazards. Funding agency perspectives will also be included.