Systematic model-data comparison for advancing global carbon cycle models

Advancing predictability of the global carbon cycle requires improved understanding of terrestrial and marine biogeochemical processes and their interactions with hydrology and climate under conditions of increasing atmospheric carbon dioxide. The increasing complexity of Earth system models and rapidly expanding volumes of observational data necessitate comprehensive, multi-faceted, and systematic evaluation of model fidelity. Model benchmarking, employing rigorous analysis methods and best-available observational data, is a powerful approach for objectively assessing and constraining model predictions, informing model development, and identifying needed measurements and field experiments. Rapid progress requires accounting for multivariate relationships that provide insights into functioning of the Earth system.