Scenario and model uncertainty in ocean biogeochemical response: Isolating the role of ocean lateral mixing

Projecting the impacts of increasing atmospheric carbon concentrations on global ocean processes that vary temporally and spatially amongst Earth System Models (ESMs) represents a formidable challenge that results in large cross-model variation. ESM uncertainty currently has three known sources: uncertainty due to internal variability, uncertainty due to scenario spin-up, and uncertainty due to model structure. One source of uncertainty is the different representations of turbulent mixing parametrized with the lateral mixing coefficient, A_REDI (Redi, 1982). The lack of theoretical understanding of how the spatial pattern of eddy velocity is linked to mixing length becomes apparent in biogeochemical trends. Within this study, we analyze the differences in an ESM suite in which A_REDI is varied and CO₂ is instantaneously doubled and quadrupled. This allows us to distinguish between biogeochemically-important fields that are sensitive to the magnitude of forcing (scenario uncertainty) compared to the internal model circulation (model uncertainty). Our results show that the rate of convection under increasing CO₂ is strongly influenced by A_REDI. The northern latitudes show large reductions in chlorophyll concentration, particulate carbon, primary productivity, and particle export when the coefficient is parametrized to anything larger than 400 m²s^-1. A value of 800 m²s^-1 appears to be the “tipping point” with regard to physical fluxes, as this is where the convection in the model appears to be most unstable to hydrological perturbations. As ESMs become the basis of our knowledge for understanding the future effects of increasing atmospheric carbon, we emphasize the importance in dedicating model evaluation to understanding the physical drivers of uncertainties and various biogeochemical behavior driven by uncertainties.