Thu, Oct 11, 2018, 12:30 pm to 1:30 pm
In this study we examine the possibility that prior estimates of Southern Ocean CO2 fluxes based on shipboard data from the Surface Ocean CO2 Atlas missed a significant portion of the seasonal and spatial signal that is now being filled in by profiling floats from the Southern Ocean Carbon and Climate Observations and Modeling (SOCCOM) project. Previous studies suggested that the Southern Ocean absorbs approximately 1 Pg out of the 2.6 Pg of anthropogenic carbon (LeQuéré et al 2018) that dissolves into the ocean each year. A new estimate of Southern Ocean carbon dioxide (CO2) uptake based on profiling float observations has called the strength of this sink into question (Gray et al. 2018). We find that adding float estimates of pCO2 to neural network and model-based interpolations of carbon dioxide observations provides new information in the times and places where there was previously little data. Subsampling of high resolution model output at the times and locations of shipboard and float observations indicates that a bias in the shipboard-only derived products due to sampling is likely. A combined ship and float estimate provides a better constrained product by joining the accuracy of the shipboard data with the newly resolved seasonal and spatial information from float observations. The combined carbon flux estimate reduces the Southern Ocean sink by 37%, which atmospheric inversion estimates indicate must be balanced by the ocean or land south of 20°S.