Analogous simulations have been performed with several ocean carbon models (Figure 3.10c,d). To compute the impact of increasing CO₂ alone (no climate change), OCMIP models were forced to follow the atmospheric CO₂ concentration derived from the IS92a scenario as in the DGVM experiment (Figure 3.10a,b) (Orr and Dutay, 1999). All models agreed in projecting that the annual ocean-atmosphere flux of CO₂ continues to become larger, reaching -6.7 to -4.5 PgC/yr by 2100 (Figure 3.10c). Since surface conditions (temperature, wind speed, alkalinity) were prescribed, the range in model estimates stems only from different representations of physical transport processes.
Several atmosphere-ocean models were used to project the effect of climate change (Maier-Reimer et al., 1996; Sarmiento et al., 1998; Matear and Hirst, 1999; Joos et al., 1999b; Bopp et al., 2001). These models include most processes previously discussed, including all processes associated with carbonate chemistry and gas exchange, physical and biological transport of CO₂, and changes in temperature, salinity, wind speed, and ice cover. They account for simple changes in biological productivity, but not for changes in external nutrient supply, species composition, pH, or Redfield ratios, all of which could be involved in more complex biological feedbacks. Coupled models estimate the impact of climate change as a departure, reported in per cent, from a �control� experiment modelling the effect of increasing atmospheric CO₂ alone.

In climate change simulations, warming of surface waters and increased stratification of the upper ocean produced an overall positive feedback that reduced the accumulated ocean uptake of CO₂ by 6 to 25% between 1990 and the middle of the 21st century, as compared with the CO₂-only case. In the first part of the simulation, the climate-mediated feedback is mainly due to the temperature effect on CO₂ solubility (Sarmiento and Le Quéré, 1996; Matear and Hirst, 1999). Towards the mid-century, the impact of circulation changes becomes significant in most models, with the net effect of further reducing ocean CO₂ uptake. To investigate the effect of climate change on the IS92a scenario, the average of the OCMIP CO₂-only projections (mean of results in Figure 3.10c) was used as a baseline and the reduction in atmosphere-ocean CO₂ flux caused by climate change (in per cent since the beginning of the simulation) was applied to this curve (Figure 3.10d). The range in model results (Figure 3.10d) must be attributed to uncertainties related to climate change feedback, and not to uncertainties in the modelling of physical transport as shown in Figure 3.10c.

The range of model estimates of the climate change impact is dependent on the choice of scenario for atmospheric CO₂ and on assumptions concerning marine biology (Joos et al., 1999b). At high CO₂ concentrations, marine biology can have a greater impact on atmospheric CO₂ than at low concentrations because the buffering capacity of the ocean is reduced (see Box 3.3) (Sarmiento and Le Quéré, 1996). Although the impact of changes in marine biology is highly uncertain and many key processes discussed in Section 3.2.3.1 are not included in current models, sensitivity studies can provide approximate upper and lower bounds for the potential impact of marine biology on future ocean CO₂ uptake. A sensitivity study of two extreme scenarios for nutrient supply to marine biology gave a range of 8 to 25% for the reduction of CO₂ uptake by mid-century (Sarmiento et al., 1998). This range is comparable to other uncertainties, including those stemming from physical transport (Figure 3.10c).

IPCC Homepage