Presenter: Neill Mackay (University of Exeter)

Description:
The Southern Ocean accounts for 40% of the uptake of anthropogenic CO₂ by the global ocean, which in turn absorbs around a quarter of all anthropogenic CO₂ emissions, mitigating climate change. Despite its importance, sampling of the Southern Ocean is sparse and is biased towards the summer months, and consequentially uncertainties in the carbon sink and its variability are largest there. Recently, autonomous platforms have begun to provide year-round coverage of the parameters needed to estimate the Southern Ocean carbon sink; however, these new observations cannot address the historical sparsity. We present a new estimate of the Southern Ocean carbon sink that attempts to address historically sparse wintertime sampling through extrapolation from summertime observations. Using observed carbonate system parameters from the Global Ocean Data Analysis Project (GLODAP), we identify winter waters in a temperature minimum layer from summertime profiles and extrapolate them to construct ‘pseudo observations’ of winter surface pCO₂ from 2004 onwards, boosting the wintertime spatiotemporal coverage by 22%. Combining the pseudo observations with direct pCO₂ data from the Surface Ocean CO2 Atlas (SOCAT), and using a Self-Organising-Map Feed-Forward-Network (SOM-FFN) gap-filling technique, we map the surface pCO2 and then estimate the Southern Ocean CO₂ sink from 1993-2018. After adjusting for surface boundary layer temperature effects, we find a strong mean sink south of 35°S of 1.29 ± 0.29 PgC yr^-1 for 2004-2018, the period covered by the pseudo observations, consistent with recent independent estimates from atmospheric data. We also show that observed strong decadal variability in the sink remains after improving the wintertime coverage.