Ocean carbon uptake and storage is inferred from changes in the inventory of anthropogenic carbon. Cant cannot be measured directly but is calculated from observations of ocean properties (Appendix 3.A discusses the sampling on which the ocean carbon inventory is based). Two independent data-based methods to calculate anthropogenic carbon inventories exist: the ΔC* method (Sabine et al., 2004), and the transit time distribution (TTD) method (Waugh et al., 2006). The Green’s function approach that applies the maximum entropy de-convolution methodology (Khatiwala et al., 2009) is related to the latter. These approaches use different tracer data, for instance, the TTD method is based mostly on chlorofluorcarbon measurements. Changes due to variability in ocean productivity (Chavez et al., 2011) are not considered.
Estimates of the global inventory of Cant (including marginal seas) calculated using these methods have a mean value of 118 PgC and a range of 93 to 137 PgC in 1994 and a mean of 160 PgC and range of 134 to 186 PgC in 2010 (Sabine et al., 2004; Waugh et al., 2006; Khatiwala et al., 2009, 2013). When combined with model results (Mikaloff-Fletcher et al., 2006; Doney et al., 2009; Gerber et al., 2009; Graven et al., 2012), Khatiwala et al. (2013) arrive at a “best” estimate of the global ocean inventory (including marginal seas) of anthropogenic carbon from 1750 to 2010 of 155 PgC with an uncertainty of ±20% (Figure 3.16). While the estimates of total inventory agree within their uncertainty, the different methods result in significant differences in the inferred spatial distribution of Cant, particularly at high latitudes.
The Cant inventory “best” estimate of 155 PgC (Khatiwala et al., 2013; Figure 3.16) corresponds to an uptake rate of 2.3 (range of 1.7 to 2.9) PgC yr –1 from 2000 to 2010, in close agreement with an independent estimate of 2.5 (range of 1.8 to 3.2) PgC yr –1 based on atmospheric O₂/N₂ measurements obtained for the same period (Ishidoya et al., 2012). The O₂/N₂ method resulted in 2.2 ± 0.6 PgC/yr for the time period 1990 to 2000 and 2.5 ± 0.6 for the period from 2000 to 2010 (Keeling and Manning, 2014). These estimates are also consistent with an independent estimate of 1.9 ± 0.4 PgC/yr for the period between 1970 and 1990 based on depth-integrated d 13 C changes (Quay et al., 2003) and with estimates inferred from ∆pCO₂.
The storage rate of anthropogenic CO₂ is assessed by calculating the change in Cant concentrations between two time periods. Regional observations of the storage rate are in general agreement with that expected from the increase in atmospheric CO₂ concentrations and with the tracer-based estimates. However, there are significant spatial and temporal variations in the degree to which the inventory of Cant tracks changes in the atmosphere (Figure 3.17). The North Atlantic, in particular, is an area with high variability in circulation and deep water formation, influencing the Cant inventory. As a result of the decline in Labrador Sea Water (LSW) formation since 1997 (Rhein et al., 2011), the Cant increase between 1997 and 2003 was smaller in the subpolar North Atlantic than expected from the atmospheric increase, in contrast to the subtropical and equatorial Atlantic (Steinfeldt et al., 2009). Perez et al. (2010) also noted the dependence of the Cant storage rate in the North Atlantic on the NAO, with high Cant storage rate during phases of high NAO (i.e., high LSW formation rates) and low storage during phases of low NAO (low formation). Wanninkhof et al. (2010) found a smaller inventory increase in the North Atlantic compared to the South Atlantic between 1989 and 2005.
Ocean observations are insufficient to assess whether there has been a change in the rate of total (anthropogenic plus natural) carbon uptake by the global ocean. Evidence from regional ocean studies (often covering relatively short time periods), atmospheric observations and models is equivocal, with some studies suggesting the ocean uptake rate of total CO₂ may have declined (Le Quéré et al., 2007; Schuster and Watson, 2007; McKinley et al., 2011) while others conclude that there is little evidence for a decline (Knorr, 2009; Gloor et al., 2010; Sarmiento et al., 2010). A study based on atmospheric CO₂ observations and emission inventories concluded that global carbon uptake by land and oceans doubled from 1960 to 2010, implying that it is unlikely that on a global scale both land and ocean sinks decreased (Ballantyne et al., 2012).
In summary, the high agreement between multiple lines of independent evidence for increases in the ocean inventory of Cant underpins the conclusion that it is virtually certain that the ocean is sequestering anthropogenic carbon dioxide and very likely that the oceanic Cant inventory increased from 1994 to 2010. Oceanic carbon uptake rates calculated using different data sets and methods agree within their uncertainties and very likely range between 1.0 and 3.2 PgC/yr.