Carbon Cycle Variability Due to the Atlantic Meridional Overturning Circulation
Abstract:The North Atlantic is the most intense region of CO2 uptake by the world oceans. Though characterization of the mean sink is robust across methodologies , a detailed understanding of variability remains lacking, seriously complicating interpretation of observations [2,3]. We investigate the causes of decadal scale variability in the North Atlantic carbon cycle using a regional numerical simulation driven by realistic climate for 1948-2013 and preindustrial atmospheric pCO2. Modeled decadal-timescale variability in air-sea CO2 fluxes and surface ocean pCO2 are dominantly controlled by basin-averaged sea surface temperature (SST). This SST signal is composed of two parts: the Atlantic Multidecadal Oscillation (AMO), associated with the model AMOC, and a positive trend. AMO dominates long-term pCO2 variability, with positive AMO leading to pCO2 declines in the subpolar gyre and pCO2 increases in the subtropical gyre. Decomposition of pCO2 into chemical (pCO2-chem) and temperature (pCO2-SST) drivers is instructive. Maximum positive AMO causes subpolar pCO2-SST to increase by ~10 uatm, but also for pCO2-chem to decline by ~20 uatm. Reduced subpolar pCO2-chem is due to reduced supply of dissolved inorganic carbon (DIC) by winter deep mixing and to enhanced DIC horizontal divergence. On net, positive AMO substantially depresses subpolar North Atlantic pCO2.
AMO had maximum negative amplitude in the 1980s and maximum positive amplitude in the mid-2000s, which coincides with the observed record of surface ocean pCO2 . This model suggests that the changing sign of AMO drove trends in the natural component of surface ocean pCO2 of approximately -7 uatm / decade in the subpolar gyre since the 1980s. This trend is significant in comparison to observed changes in surface ocean pCO2 , and thus impacts our understanding of the changing ocean carbon sink in this critical region.
 Schuster et at 2013
 McKinley et al. 2011, McKinley and Fay 2013
 Metzl et al. 2010