Novel biogeochemical theory predicts ocean carbon reservoir response to changes in MOC strength and diapycnal mixing

Physical ocean processes strongly regulate atmospheric carbon on ocean equilibrium timescales such as those associated with glacial terminations. Here we propose a novel analytical theory for linking changes in global ocean circulation with air-sea CO2 exchange, as well as an intriguing application of this theory to deglaciation dynamics. The theory relies on a Lagrangian analysis of biogeochemical processes affecting air-sea carbon partitioning within MOC water parcels. Our theory accounts for biological uptake, remineralization and air-sea CO2 exchange. The theory predicts analytically atmospheric CO2 and the amount of carbon stored in each cell as a function of the competing roles of the time scales associated with biological uptake, remineralization, gas-exchange and surface circulation. Additionally, the strength of the overturning in each of the cells is calculated from pycnocline depth, wind strength, diapycnal mixing, buoyancy fluxes and eddies based on the analytical theory of Nikurashin and Vallis (2011, 2012). With the combined physical-biogeochemistry theoretical framework we predict how atmospheric and ocean carbon change with diapycnal mixing and the resulting MOC variations. Our theory compares well with the output from a set of 7 steady-state simulations with varying diapycnal mixings using the global ocean-only 2.8o MITgcm with a simple biogeochemical package. Our theoretical and modeling results present an interesting application and conundrum for deglaciation dynamics. Tidal models indicate that diapyncal mixing was far stronger during the Last Glacial Maximum (LGM) compared to the present. Ice melt during deglaciations significantly increases ocean volume relative to the LGM, decreasing the diapycnal mixing over most of the ocean and slowing down the MOC (Schmittner 2015). Using the proposed diapycnal diffusivity values from Schmittner (2015) (while disregarding all other possible mechanisms), our model would then predict a decreased atmospheric CO2 in present day of 40 ppm compared to the LGM. We note that this mechanism was not accounted for typically by the PMIP2 models (Otto-Bliesner, 2007) and that it acts in opposition to the net observed increase in atmospheric CO2 during deglaciation. We posit that this mechanism needs to be accounted for in future analyses of the LGM.