7.3.5.3 Sensitivity Analysis

The coupled and uncoupled model experiments can be used to separate the effects of climate change and CO₂ increase on land and ocean carbon storage (Friedlingstein et al., 2003). Table 7.4 also shows the linear sensitivity parameters diagnosed from each of the C⁴MIP models (Friedlingstein et al., 2006).

7.3.5.3.1 Increase in ocean carbon uptake with increasing atmospheric carbon dioxide

The ocean takes up CO₂ at a rate that depends on the difference between pCO₂ in the atmosphere and in the surface ocean. Model estimates of uptake differ primarily because of differences in the rate at which carbon is exported from the surface ocean to depth by the large-scale circulation (Doney et al., 2004; Section 7.3.4.1; Box 7.3) and the biological pump (Sarmiento et al., 2004). Ocean carbon cycle model intercomparisons have shown that the simulated circulation in the Southern Ocean can have a large impact on the efficiency with which CO₂, and other anthropogenic tracers such as CFCs, are drawn down (Orr et al., 2001; Dutay et al., 2002). The C⁴MIP models show ocean carbon storage increases ranging from 0.9 to 1.6 GtC ppm^–1, which is equivalent to ocean uptake increasing at between 42 and 75% of the rate of atmospheric CO₂ increase. Basic ocean carbonate chemistry suggests that the ocean-borne fraction of emissions will fall in the future, even in the absence of climate change, because of an increasing ocean buffer factor (Section 7.3.4.2).

7.3.5.3.2 Increase in land carbon uptake with increasing atmospheric carbon dioxide

In the absence of land use change and forest fires, land carbon storage depends on the balance between the input of carbon as NPP, and the loss of carbon as heterotrophic (soil) respiration (Section 7.3.3). There is an ongoing debate concerning the importance of CO₂ fertilization at the patch scale where other constraints such as N limitation may dominate; recent surveys indicate a wide range of possible responses to a CO₂ increase of around 50%, with average increases of 12 to 23% (Norby et al., 2005; see Section 7.3.3.1).

The C⁴MIP models show increases in global NPP of between 6 and 33% when CO₂ increases over the same range. These figures are not directly comparable: some C⁴MIP models include vegetation dynamics, which are likely to increase the vegetation cover as well as the NPP per unit of vegetation area, and therefore lead to higher overall sensitivity of global NPP to CO₂. The FACE experiments also typically involve an instantaneous increase in CO₂. However, most C⁴MIP models are within the range of the CO₂ sensitivities measured.

The overall response of land carbon storage to CO₂ is given by the fifth column of Table 7.4. The C⁴MIP models show time-mean land carbon storage increases ranging from 0.2 to 2.5 GtC ppm^–1, with all but two models between 1.1 and 1.6 GtC ppm^–1. This response is driven by the CO₂ fertilization of NPP in each model, with a counteracting tendency for the mean soil carbon turnover rate (i.e., the heterotrophic respiration by unit soil carbon) to increase even in the absence of climate change. This somewhat surprising effect of CO₂ is seen to varying degrees in all C⁴MIP models. It appears to arise because CO₂ fertilization of NPP acts particularly to increase vegetation carbon, and therefore litter fall and soil carbon, in productive tropical regions that have high intrinsic decomposition rates. This increases the average turnover rate of the global soil carbon pool even though local turnover rates are unchanged. In some models (e.g., model C) this acts to offset a significant fraction of the land carbon increase arising from CO₂ fertilization. Models with large responses of ocean or land carbon storage to CO₂ tend to have weaker climate-carbon cycle feedbacks because a significant fraction of any carbon released through climate change effects is reabsorbed through direct CO₂ effects (Thompson et al., 2004).

7.3.5.3.3 Transient climate sensitivity to carbon dioxide

The strength of the climate-carbon cycle feedback loop depends on both the sensitivity of the carbon cycle to climate, and the sensitivity of climate to CO₂. The equilibrium climate sensitivity to a doubling of atmospheric CO₂ concentration remains a critical uncertainty in projections of future climate change, but also has a significant bearing on future CO₂ concentrations, with higher climate sensitivities leading to larger climate-carbon cycle feedbacks (Andreae et al., 2005). The fourth column of Table 7.4 shows the transient global climate sensitivity (i.e., the global climate warming that results when the transient simulation passes doubled atmospheric CO₂) for each of the C⁴MIP models. All but two models (models E and I) have transient climate sensitivities in the range 1.9°C to 2.7°C. However, differences in carbon cycle responses are likely to occur because of potentially large differences in regional climate change, especially where this affects water availability on the land.

7.3.5.3.4 Dependence of ocean carbon uptake on climate.

Climate change can reduce ocean uptake through reductions in CO₂ solubility, suppression of vertical mixing by thermal stratification and decreases in surface salinity. On longer time scales (>70 years) the ocean carbon sink may also be affected by climate-driven changes in large-scale circulation (e.g., a slowing down of the thermohaline circulation). The last column of Table 7.4 shows the sensitivity of ocean carbon storage to climate change as diagnosed from the C⁴MIP models. All models indicate a reduction in the ocean carbon sink by climate change of between –14 and –60 GtC °C^–1, implying a positive climate-CO₂ feedback.

7.3.5.3.5 Dependence of land carbon storage on climate.

The major land-atmosphere fluxes of CO₂ are strongly climate dependent. Heterotrophic respiration and NPP are both very sensitive to water availability and ambient temperatures. Changes in water availability depend critically on uncertain regional aspects of climate change projections and are therefore likely to remain a dominant source of uncertainty (see Chapter 11). The overall sensitivity of land carbon storage to climate (Table 7.4, seventh column) is negative in all models, implying a positive climate-CO₂ feedback, but the range is large: –19 to –175 GtC °C^–1. These values are determined by the combined effects of climate change on NPP and the soil carbon turnover (or decomposition) rate, as shown in Table 7.5.

The C⁴MIP models utilise different representations of soil carbon turnover, ranging from single-pool models (model A) to nine-pool models (model E). However, most soil models assume a similar acceleration of decay with temperature, approximately equivalent to a doubling of the specific respiration rate for every 10°C warming. This temperature sensitivity is broadly consistent with a long history of lab and field measurements of soil efflux (Raich and Schlesinger, 1992), although there is an ongoing difficulty in separating root and soil respiration. Note, however, that the expected dependence on temperature was not found at the whole ecosystem level for decadal time scales, in forest soils (Giardina and Ryan, 2000; Melillo et al., 2002), grasslands (Luo et al., 2001) or boreal forests (Dunn et al., 2007). These apparent discrepancies may reflect the rapid depletion of labile pools of organic matter, with strong temperature responses likely so long as litter inputs are maintained (Knorr et al., 2005). Nevertheless, the temperature sensitivity of the slow carbon pools is still poorly known.

Table 7.5 shows that all C⁴MIP models simulate an overall increase in soil carbon turnover rate as the climate warms, ranging from 2 to 10% per °C. The use of a single soil carbon pool in the Hadley model (A) cannot completely account for the relatively large sensitivity of soil respiration to temperature in this model (Jones et al., 2005), as evidenced by the lower effective sensitivity diagnosed from the UVic model (model G), which uses the same soil-vegetation component. It seems more likely that differences in soil moisture simulations are playing the key part in determining the effective sensitivity of soil turnover rate to climate. Table 7.5 also shows the effective sensitivities of NPP to climate, ranging from a significant reduction of 6% per °C to smaller climate-change driven increases of 2% per °C under climate change. This variation may reflect different time scales for boreal forest response to warming (leading to a positive impact on global NPP), as well as different regional patterns of climate change (Fung et al., 2005). The models with the largest negative responses of NPP to climate (models A, B and C) also show the tendency for tropical regions to dry under climate change, in some cases significantly (Cox et al., 2004).