U.S. Dept. of Commerce / NOAA / OAR / PMEL / Publications
In classical carbon cycle model studies, emissions from fossil fuel burning are prescribed and the model computes the time evolution of atmospheric CO2 as the residual between emissions and uptake by land and ocean. Because the global carbon cycle is intimately embedded in the physical climate system, several feedback loops exist between the two systems (Friedlingstein et al., 2003). For example, increasing CO2 modifies the climate, which in turn impacts on ocean circulation and therefore on oceanic CO2 uptake. Similar effects are expected to occur on land with rising temperatures (e.g., higher soil carbon respiration rates). When a climate or carbon cycle feedback results in an increase in the atmospheric CO2 accumulation rate and thus in enhanced climate change, it is referred to as a positive feedback. A change that reduces atmospheric CO2 is a negative feedback.
The quantitative assessment of these feedbacks necessitates the use of coup led carbon cycle climate models. Three coupled models, Hadley Centre, Institute Pierre-Simon Laplace (IPSL) and Climate System Modelling Initiative (CSMI 4), have recently examined the feedbacks based on Intergovernmental Panel on Climate Change (IPCC) scenarios between 1850 and 2100 (Cox et al., 2000; Dufresne et al., 2002; Fung et al., 2005). By 2100 the results show dramatically different climate–carbon cycle sensitivities. These models simulate an enhanced increase of atmospheric CO2 as a result of the climate change impacts on the carbon cycle. However, the magnitudes of the feedbacks vary by a factor of four between the simulations. Without the feedbacks the models reach an atmospheric concentration of ~700 ppm by 2100. When the feedbacks are operating, the Hadley Centre model (Cox et al., 2000) reaches 980 ppm, leading to an average near-surface warming of +5°C, the IPSL model (Dufresne et al., 2002) attains only 780 ppm and a warming of +3°C and the CSMI 4 model reaches an atmospheric CO2 concentration of 792 ppm with a warming of +1.4°C. This different behaviour can be traced back to the land carbon cycle climate sensitivity of the Hadley Centre model being much larger than either the IPSL or CSMI 4 models as well as to the geochemical oceanic uptake being much larger in the IPSL model than in the Hadley Centre model.
These pioneering model simulations are subject to important limitations. In these models key biological processes on land and in the ocean are highly parameterized and poorly constrained (see Fung et al., 2005). Proper modelling of the coupled carbon and climate system, however, requires an improved understanding of the two primary classes of feedbacks: that of the carbon cycle and that of the carbon–climate system.
Carbon cycle feedbacks are processes that respond directly to increasing atmospheric CO2, resulting in a change of the net land–air or sea–air exchange of CO2. For example, the efficiency with which the ocean can absorb CO2 at the surface is related to how much CO2 can be converted to DIC. The measure of this is called the Revelle factor (RF) as given by Eq. 3.2 (Revelle and Suess, 1957):
(∆pCO2 / ∆DIC) / (pCO2 /DIC)
The RF of surface ocean waters varies from 8–9 in the subtropical gyres to 13–15 in the higher latitudes. Figure 3.7 shows the change in DIC concentration of the modern surface ocean in response to a uniform increase in pCO2 of 10 ppm, plotted as a function of RF. It also shows that waters with low RF (~9) are four times more efficient at taking up CO2 (∆DIC) than waters with very high RF (~15). The RF of ocean waters is controlled by the distribution of the DIC species, including the pH of the ocean.
Fig. 3.7. Plot of the change in dissolved inorganic carbon for a 10 ppm change in pCO2 as a function of Revelle factor for surface waters (<60 m) from the GLODAP bottle data-set. Inset shows a map of surface Revelle factor from the same data-set.
As the ocean takes up anthropogenic CO2, the pH of the water decreases and the RF increases. With the anthropogenic CO2 estimates of Sabine et al. (2004b), the global average RF of surface waters today appears about one unit higher than the pre-industrial values. Thus, the surface ocean today is less efficient at taking up CO2 than the preindustrial ocean providing a positive feedback. According to Fig. 3.7, the significance of this effect will vary depending on locations. Changing the RF by one in the high latitudes will have less effect than changes in the subtropics with relatively low RF. A further insight of these processes and their proper representation in ocean carbon models is important for understanding the ultimate long-term storage of anthropogenic CO2 in the ocean.
Inorganic carbon thermodynamics are reasonably well understood, but some carbon cycle feedbacks, particularly those involving biological processes, are not well understood. One example of this is the effect of anthropogenic CO2 on organisms that produce calcium carbonate (CaCO3) shells. Shallow water environments, primarily coral reefs and carbonate shelves, produce ~0.3 Pg C/year, largely as metastable aragonite and high-magnesian calcite. Open-ocean plankton produces an estimated 0.7–1.4 Pg C/year (Milliman, 1993; Lee, 2001), mostly as calcite but also some aragonite. These open-ocean calcifiers include phototrophic coccolithophorids and heterotrophic foraminifera as well as pteropods. Using Eq. 3.3, 1 mol of CaCO3 produced releases 1 mol of CO2:
Ca+2 +2HCO−3 ↔ CaCO3+CO2+H2O (3.3)
Numerous studies have suggested that the rate of calcification in a wide variety of organisms is reduced when they are exposed to elevated CO2 levels (see summary in Feely et al., 2004). As atmospheric CO2 levels increase, one might expect calcification to decrease, which would lead to a lower natural release of CO2 from the ocean, providing a negative feedback.
The situation, however, is not that straightforward. A decrease in carbonate precipitation in the upper ocean would also lower the RF, increasing the capacity of the ocean to thermodynamically take up CO2 from the atmosphere. A complete shutdown of surface ocean calcification would decrease surface ocean pCO2 by ~20 ppm (Wolf-Gladrow et al., 1999). On the other hand, if these organisms are primary producers, the decrease in organic matter production could result in a positive feedback. Furthermore, a decrease in CaCO3 production would affect the ratio of organic/inorganic carbon delivery to the deep sea. If processes regulating this ‘rain’ of organic and inorganic carbon to deep-sea sediments are uncoupled, a decrease in CaCO3 production would lead to increased dissolution of CaCO3 in deep-sea sediments, which would raise the ocean pH and its capacity to store CO2 (Archer and Maier-Reimer, 1994). However, if these two processes are coupled and the denser carbonate particles are necessary for transporting the organic matter into the deep ocean quickly (Armstrong et al., 2002), reducing the carbonate production could result in shallower remineralization of organic carbon, producing a positive feedback (Klaas and Archer, 2002; Ridgwell, 2003), and a diminished role of sediments in the buffering of atmospheric CO2 increases. It is also not clear how elevated CO2 selection against a certain species (e.g., calcifying organisms) will affect the overall ecosystem structure and net CO2 uptake by ocean biology in the future. Clearly, there is a need for more research on these mechanistic controls of the long-term changes in the carbonate system.
In addition to the direct impacts of elevated CO2 on the ocean carbon system, there are many possible indirect effects related to the climate changes associated with the atmospheric CO2 increase. These feedback mechanisms include: (i) reduced CO2 solubility due to the increase in sea water temperature; (ii) enhanced stabilization of the upper water masses of the water column that will lead to decreased exchange of DIC and nutrients from the ocean interior; and (iii) enhanced productivity in high-latitude regions (Table 3.5). The potential magnitude of these carbon–climate feedbacks has been examined in several modelling studies (Sarmiento and Le Quéré, 1996; Sarmiento et al., 1998; Joos et al., 1999; Matear and Hirst, 1999; Greenblatt and Sarmiento, 2004).
CO2 solubility has a strong inverse relationship with temperature. Greenblatt and Sarmiento (2004) estimate that, as the surface ocean warms over this century, ~9–14% of the CO2 that would have been stored in the ocean will be retained in the atmosphere by 2100 (a positive climate feedback). The thermodynamics of this process are well known and, consequently, the uncertainties are reasonably low.
However, there are other processes that are not fully understood. For example, increased stratification of the water column due to warming and changes in the hydrological cycle is expected to cause a decrease in the exchange of carbon and nutrients between water masses, particularly in high latitudes. The decreased carbon exchange makes it more difficult to move the anthropogenic CO2 into the ocean interior, thus decreasing the oceanic uptake efficiency and providing a positive feedback (Table 3.5). This increased stratification, however, also increases CO2 drawdown by biological activity in the Southern Ocean (negative feedback) where there is an excess of surface nutrients and the organisms are generally light-limited. The different model studies disagree on the magnitude of these two competing effects and, in some cases, do not even agree on whether the combination of these two effects will provide a positive or negative feedback (Table 3.5).
Modelling studies have suggested that the ocean will ultimately absorb up to 70–85% of the CO2 released by human activity (Le Quéré and Metzl, 2004). Including a carbon system feedback where carbonate sediments in the ocean are dissolved by the lowered pH of the waters suggests that the ocean may be the ultimate storage place for as much as 90% of the anthropogenic CO2 (Archer et al., 1997). The dissolution of carbonate sediments reverses Eq. 3.3 and increases the carbon storage capacity of the ocean. However, because of the slow mixing time to get the anthropogenic CO2 into the deep ocean, this capacity may not be realized for hundreds or thousands of years. When considering the role of carbon system and carbon–climate feedbacks, it is important to understand the timescale of these processes. With typical lifetimes ranging from weeks to months, biological processes have the potential to respond very quickly to carbon system or climate changes. Large-scale circulation changes are likely to be relevant on annual to decadal timescales, and sediment dissolution processes are presumed to be relevant on centennial to millennial timescales.As long as atmospheric CO2 continues to rise, the ocean will continue to take up CO2. Long-term feedbacks like dissolution of carbonate sediments may enhance the oceanic uptake but most indications are that the shorter-term feedbacks may reduce the rate of CO2 uptake on the decadal to centennial timescales. Although there are considerable differences in the relative magnitudes of each of the individual feedback processes described in Table 3.5, all the models showed a net decrease in the overall uptake of CO2 by the ocean over time (Greenblatt and Sarmiento, 2004). In the OCMIP-2 models, which utilized a constant biological activity and circulation, the estimated oceanic uptake of anthropogenic CO2 by 2100 ranged from 4 to 8 Pg C/year, depending on the CO2 emission scenario used in the model (Watson and Orr, 2003). This estimate has a factor of 2–4 times higher than the current value of 2 ± 0.5 Pg C/year but still lags behind the projected rate of CO2 emissions. This means that a larger fraction of the CO2 emissions will be retained by the atmosphere in the future, thus enhancing the overall climate change impact.
Return to previous section or go to next section
Return to Outstanding Publications Page
Return to PMEL Publications Page
Return to PMEL Home Page