Anthropogenic CO2 estimates in the Southern Ocean: Storage partitioning in the different water masses

The role of the Southern Ocean (SO) remains a key issue in our understanding of the global carbon cycle and for predicting future climate change. A number of recent studies suggest that 30 to 40% of ocean uptake of anthropogenic carbon (CANT) occurs in the SO, accompanied by highly efficient transport of CANT by intermediate-depth waters out of that region. In contrast, storage of CANT in deep and bottom layers is still an open question. Significant discrepancies can be found between results from several indirect techniques and ocean models. Even though reference methodologies state that CANT concentrations in deep and bottom layers of the SO are negligible, recent results from tracer-based methods and ocean models as well as accurate measurements of Ar-39, CCl4 and CFCs along the continental slope and in the Antarctic deep and bottom waters contradict this conclusion. The role of the SO in the uptake, storage and transport of CANT has proved to be really important for the global ocean and there is a need for agreement between the different techniques. A CO2-data-based (back-calculation) method, the C-T(O) method, was developed with the aim of obtaining more accurate CANT concentration and inventory estimates in the SO region (south of 45 S). Data from the GLODAP (Global Ocean Data Analysis Project) and CARINA databases were used. The C-T(0.)). method tries to reduce at least two of the main caveats attributed to the back-calculation methods: the need for a better definition of water mass mixing and, most importantly, the unsteady state of the air-sea CO2 disequilibrium (Delta C-dis) term. Water mass mixing was computed on the basis of results from an extended Optimum Multi-Parametric (eOMP) analysis applied to the main water masses of the SO. Recently published parameterizations were used to obtain more reliable values of ACd,, and also of preformed alkalinity. The variability of the ACcus term (8Cdis) was approximated using results from an ocean carbon cycle model. Results from the C7. method are compared with those from the AC method, the TrOCA method, and two different tracer-based approaches, the transit-time distribution (TTD) and Green's function (GF) methods. We find that the TTD, GF and CT methods give very similar estimates for the SO's inventory (with reference to the year 1994) of 30 2, 22 2,29 3 PgC, respectively. Importantly, Antarctic Bottom Water shows CANT concentrations of 9 +/- 1, 3 +/- 0.3, 6 +/- 1 mu mol kg(-1), contributing 6-12% of the SO's inventory. The AC* and TrOCA methods seem to underestimate and overestimate, respectively, both the total C-ANT inventory and C-ANT concentrations in deep and bottom layers. Results from the O. method suggest that deep and bottom layers of the water column in the SO contain, in general, low concentrations of C-ANT compared with subsurface and intermediate layers but higher than those recorded in the global databases. It is important to note that, as deep and bottom layers in the SO fill two of the most voluminous water masses of the global ocean, even these relatively low values of C-ANT can be of considerable importance when computing the inventories in the water column, mostly in the SO but also in outer regions where bottom waters spread. (C) 2013 Elsevier Ltd. All rights reserved.