Formulating approximate, closed-form parameterizations for biogeochemical processes—such as global element cycles—remains challenging due to process complexity and environmental heterogeneity. Advances in artificial intelligence (AI) offer new opportunities by combining data-driven methods with process-based models, but key gaps persist. Biogeochemical data are often multi-scale, sparse, biased, and spatially autocorrelated, complicating model evaluation and increasing uncertainty. Robust integration of AI and biogeochemical modeling requires systematic approaches to assess model structure, improve interpretability, and quantify data uncertainty.

This session brings together advances at the AI–biogeochemistry interface to support next-generation ecosystem and Earth system models. We invite contributions on: (1) hybrid AI–process modeling; (2) AI-assisted parameter optimization; (3) AI-derived data products for model evaluation; and (4) large language model–enabled biogeochemical data integration, mining, and management.

Our goal is to foster an interdisciplinary community advancing predictive understanding of biogeochemical cycles across ecosystems and the Earth system.

• GC - Global Environmental Change
• H - Hydrology
• IN - Informatics
• MR - Mineral and Rock Physics