Rapid advances in machine learning are transforming many areas of biogeosciences. Beyond traditional successes of machine learning in making predictions, novel combinations of data-driven and process-based approaches are generating new insights and accelerating scientific discoveries about terrestrial, aquatic, and marine ecosystems. Efficiency, interpretability, mechanistic understanding, and uncertainty quantification are among the key benefits of these new synergies.

In this session, we invite contributions that leverage emerging machine learning, artificial intelligence, and data science approaches to deepen our understanding and build generalized representations of ecosystem processes across all scales. Example areas include but are not restricted to applications of physics-informed machine learning, explainable machine learning, causal inference, information theory, unsupervised machine learning, Bayesian approaches, and uncertainty quantification. The focus of this session is on applications that provide novel understanding. Method development, reviews, syntheses, perspectives, and theoretical analyses are also welcome.