North Atlantic anthropogenic carbon: methods, trends, budgets, variabilities, and uncertainties

Abstract

Since the advent of the industrial revolution, atmospheric CO2 has increased from 275 ppm to over 400 ppm, enhancing the associated Greenhouse effect and being suggested as the main cause of recent climate change. The global ocean sequesters around a third of the CO2 emitted by human activity, mitigating climate impacts, with the highest anthropogenic CO2 (Cant) storage per unit area occurring in the North Atlantic. However, ocean Cant cannot be measured directly, but it is calculated with published uncertainties that range between ±10 % and ±20 %. Here, we assess five methods used to estimate Cant, named ∆C*, ΦCT0, TrOCA, TTD, and eMLR, by using the outputs of four climate models (CCSM, CM2Mc, OCCAM, and GFDL-ESM2M) between 1860 and 2100, the most recent observation database (e.g. GLODAPv2) between 1980 and 2013, and the repeated time series collected along the 24.5◦N Atlantic transect between 1992 and 2016. We focus on the North Atlantic upper 1000 m, where the Mode waters store the largest Cant amount. In this layer, the TTD and ∆C* estimates confine the probable range of Cant concentrations, therefore we focus on these two methods. For both, we quantify a total (analytical precisions + methodological assumptions) uncertainty of ±34 %, which is higher than previously suggested. However, the Cant uncertainties depend on timeframes and regions: between 1992 and 2010, observations enable us to reliably decrease these uncertainties to ±13 % (TTD) and ±14 % (∆C*) in the upper 1000 m of the subtropical North Atlantic (20-30◦N). Here, we estimate with a quasi Monte Carlo approach that the Mode waters Cant pool increases by 0.5 (TTD) and 0.8 (∆C*) ± 0.2 μmol kg−1 yr−1, thus the estimates diverge over time. We associate the divergence to unsteady CO2 disequilibrium between the atmosphere and ocean (0.3 (∆C*) and 0.5 (TTD) ± 0.3 μmol kg−1 yr−1), and biogeochemical changes, as suggested by the increasing (0.3 ± 0.1 μmol kg−1 yr−1) dissolved inorganic carbon from remineralised soft tissue: these alterations are unequally captured by the TTD and ∆C* techniques. Changes in ocean biogeochemistry are further explored using the output of a CM2Mc pre-industrial ‘control’ simulation over two millennia. Here, the statistically significant drivers of the enhancement in remineralised soft-tissue carbon are increasing mean residence time (R2 = 0.86) and acidification (R2 = 0.68). Feedback mechanisms have the potential to shift the oceanic carbon cycle towards new equilibria, significantly influencing the future North Atlantic carbon uptake.