Primary Convener:
Yang Song, University of Arizona

Convener:
Umakant Mishra, Sandia National Laboratories
Pamela Weisenhorn, Argonne National Laboratory
Dipankar Dwivedi, Lawrence Berkeley National Laboratory
Vanessa Garayburu-Caruso, Pacific Northwest National Laboratory

Early Career Convener:
Allison Myers-Pigg, Pacific Northwest National Laboratory

Chair:
Avni Malhotra, Pacific Northwest National Laboratory
Forrest Hoffman, Oak Ridge National Laboratory
Vanessa Garayburu-Caruso, Pacific Northwest National Laboratory
Umakant Mishra, Sandia National Laboratories

Earth systems, especially biogeochemical processes, are complex and heterogeneous, complicating predictive understanding across scales. Traditional approaches struggle to create physical parameterizations for these processes, creating critical knowledge gaps. Artificial intelligence and machine learning (AI/ML) can overcome these limitations by enhancing the integration of models and experiments (ModEx), expediting model-experiment cycles, optimizing experimental design, and improving model calibration. The unique properties of biogeochemical data, like multiple scales, limited samples, and spatial autocorrelation, challenge AI/ML, requiring systematic evaluation of model structure, interpretability, robustness, and uncertainty.

This session explores how AI/ML facilitates coupling between models and experiments in multiscale biogeochemical cycles while advancing prediction through hybrid data-driven and process-based approaches. We welcome contributions demonstrating innovative AI/ML applications that bridge observations and models across terrestrial, aquatic, and coastal environments. We encourage work highlighting systematic evaluation approaches and interdisciplinary contributions that overcome heterogeneity challenges and accelerate predictive understanding of biogeochemical cycles across the Earth system.

Index Terms
0412 Biogeochemical kinetics and reaction modeling
0430 Computational methods and data processing
1894 Instruments and techniques: modeling
1942 Machine learning

Cross-Listed:
IN - Informatics
H - Hydrology
GC - Global Environmental Change

Neighborhoods:
3. Earth Covering