3.6.3.2 Uptake of anthropogenic CO₂ by the ocean

Ocean uptake is constrained to some degree by observations of anthropogenic tracers. Three transient tracers are commonly used. First, anthropogenic CO₂ itself gives a direct benchmark for model estimates of the quantity and distribution of anthropogenic CO₂ that has penetrated the ocean since the pre-industrial era. Anthropogenic CO₂ can be inventoried by an indirect method whereby carbon concentration is compared to what would be expected from water exposed to pre-industrial air (Gruber et al., 1996). The ¹⁴CO₂ released in the early 1960s by atmospheric nuclear testing (commonly called bomb ¹⁴C) provides a second tracer; the content of bomb ¹⁴C in the ocean is used to constrain global air-sea CO₂ exchange (Wanninkhof et al., 1992), and ocean model results can be compared with its penetration depth as a benchmark for vertical transport (Broecker et al., 1995). Bomb ¹⁴C is computed by subtracting the observed ¹⁴C concentration from an estimate of its pre-industrial value (Broecker et al., 1995). Finally, CFCs also constrain the downward transport of tracers in ocean models. No natural background needs to be subtracted from CFCs. None of these three tracers provide a perfect indicator of anthropogenic CO₂ uptake: CO₂ equilibrates with the atmosphere ten times faster than ¹⁴C and ten times slower than CFCs; anthropogenic CO₂ and ¹⁴C are indirectly estimated. As part of the Ocean Carbon-Cycle Model Intercomparison Project (OCMIP), a comparison of carbon models with respect to all three anthropogenic tracers is in progress (Orr and Dutay, 1999; Orr et al., 2001).

Although regional estimates show discrepancies, modelled estimates of anthropogenic tracers agree reasonably well with observations when integrated globally. The mean value of the penetration depth of bomb ¹⁴C for all observational sites during the late 1970s is 390 ± 39 m (Broecker et al., 1995). For the same years and stations, modelled estimates range between 283 and 376 m (Orr et al., 2001). Modelled and observed CFC concentrations have been compared locally but not yet globally (England 1995; Robitaille and Weaver, 1995; Orr and Dutay, 1999). Modelled anthropogenic CO₂ inventory since 1800 is comparable to the estimate of 40 ± 9 PgC for the Atlantic Ocean (Gruber, 1998) and 20 ± 3 PgC for the Indian Ocean (Sabine et al., 1999; Orr et al., 2001). Latitude-depth profiles of anthropogenic CO₂ in the Atlantic, extracted from data and from models, are shown in Figure 3.9. Modelled CO₂ uptake for the global ocean between 1800 and 1990 ranges between 100 and 133 PgC (Figure 3.8), comparable to the preliminary data-based estimate of 107 ± 27 PgC for the global ocean, which includes the Pacific value of 45 ± 15 PgC (Feely et al., 1999a). Although in reasonable agreement with basin and global estimates of anthropogenic CO₂, modelled inventories exhibit large differences at the regional scale: models tend to underestimate the inventory of anthropogenic CO₂ between 50°S and 50°N in the Atlantic and Indian Oceans, and to overestimate it at high latitudes (Sabine et al., 1999; Orr et al., 2001). In the Southern Ocean the uptake of anthropogenic CO₂ varies by a factor of two among models (Orr et al., 2001). The difficulty for models in reproducing the spatial structure of anthropogenic tracers may be indicative of problems in ocean physics mentioned earlier, and may be responsible for the increasing range of model estimates when future CO₂ uptake is projected by the same models (Figure 3.10c).

The most recent model estimates of the ocean-atmosphere flux obtained with process-based models are -1.5 to -2.2 for 1980 to 1989 (Table 3.4), in agreement with earlier model estimates for the same period (Enting et al., 1994; Orr et al., 2001). These estimates are fully consistent with the budget based on atmospheric observations alone (Table 3.1), with estimates based on pCO₂ and ¹³C observations (Table 3.4), and with the SAR estimate of -2.0 ± 0.8 PgC/yr. Figure 3.8 shows modelled ocean CO₂ uptake for 1900 to 2000. (These results do not include natural variability and therefore appear smoother than in reality.) The oceanic regions absorbing the largest quantities of anthropogenic CO₂ according to models are those where older waters come in contact with the atmosphere, such as high latitudes and upwelling regions of the equator. In contrast, modelled sub-tropical regions rapidly saturate at atmospheric CO₂ level and do not absorb large quantities of anthropogenic CO₂ (Sarmiento et al., 1992; Orr et al., 2001).

Figure 3.9: Anthropogenic CO2 in the Atlantic Ocean (mmol/kg): comparison of data and models. The top left panel shows the sampling transect; the top right panel shows estimates of anthropogenic CO2 content along this transect using observations from several cruises between 1981 and 1989 (Gruber, 1998). Anthropogenic CO2 is not measured directly but is separated from the large background of oceanic carbon by an indirect method based on observations (Gruber et al., 1996). The remaining panels show simulations of anthropogenic CO2 content made with four ocean carbon models forced by the same atmospheric CO2 concentration history (Orr et al., 2000).

Figure 3.10: Projections of anthropogenic CO2 uptake by process-based models. Six dynamic global vegetation models were run with IS92a CO2 concentrations as given in the SAR: (a) CO2 only, and (b) with these CO2 concentrations plus simulated climate changes obtained from the Hadley Centre climate model with CO2 and sulphate aerosol forcing from IS92a (Cramer et al., 2000). Panel (b) also shows the envelope of the results from panel (a) (in grey). (c) Ten process-based ocean carbon models were run with the same CO2 concentrations, assuming a constant climate (Orr and Dutay, 1999; Orr et al., 2000). A further six models were used to estimate the climate change impact on ocean CO2 uptake as a proportional change from the CO2-only case. The resulting changes were imposed on the mean trajectory of the simulations shown in panel (c), shown by the black line in panel (d), yielding the remaining trajectories in panel (d). The range of model results in panel (d) thus represents only the climate change impact on CO2 uptake; the range does not include the range of representations of ocean physical transport, which is depicted in panel (c).

IPCC Homepage