Greenhouse Gas Stabilization Experiments
U. Mikolajewicz, M. Gröger, E. Maier-Reimer, G. Schurgers, M. Vizcaíno & A. Winguth*
Max-Planck-Institute for Meteorology, Bundesstr. 53, 20146 Hamburg, Germany, firstname.lastname@example.org
*) Center for Climatic Research, Madison, USA
MPI Earth System Model for paleo studies including atmosphere (ECHAM), ocean (LSG), ocean biogeochemistry (HAMOCC), terrestrial biosphere (LPJ), and ice sheets (SICOPOLIS).
Experiments: In order to study the sensitivity of the feedbacks between the carbon cycle and the climate, a control experiment and seven atmospheric sensitivity experiments with prescribed atmospheric CO2 concentrations and with different stabilization levels (as used by IPCC ) were conducted. The prescribed CO2 concentration increases by 1% per year until it stabilizes after 70 years for 2xCO2, 105 years for 3xCO2, and 140 years for 4xCO2, respectively.
2. Climate change
Relatively moderate climate sensitivity to changes in the atmospheric CO2 concentration [IPCC, 2001] is indicated by an increase of global mean surface temperature of 2.1 K, 3.2 K, and 4.2 K for the 2xCO2, 3xCO2, and 4xCO2 experiment, respectively (Figure 1a). A drastic change is predicted for the Arctic Ocean, with a substantial decline in sea ice thickness due to a warming of the atmosphere (Figure 2). Reduced formation of NADW and a decline in the strength of the thermohaline circulation (Figure 1b) are particularly obvious in the 4xCO2 simulation. In the experiments with 2xCO2 and 3xCO2, this reduction in the thermohaline circulation is modest and the strength of the overturning reaches almost the values of the control run after a century, but the North Atlantic overturning cell is shallower than in the control run (not shown). In this scenario ocean convection is intensified in the northeastern Pacific (intermediate depths) by reduced freshwater input from ECHAM3 and Southern Ocean convection is limited to mid-depth with increased ventilation over the last 100 years of integration. Overall, the deep Pacific circulation in the 3xCO2 run below 2.5 km is substantially reduced and thus prone to low oxygen or anoxic conditions.
The reduction in the North Atlantic thermohaline circulation in all the CO2 perturbation scenarios is related to an increased stratification, resulting from a positive freshwater anomaly and surface warming (Figure 2). This freshwater perturbation is caused by a strengthening of the atmospheric moisture transports into the North Atlantic and Arctic of up to 0.2 Sv in the case of the 4xCO2 experiment (Figure 1c). The changes in the surface buoyancy fluxes result in a weakening of the meridional overturning circulation involving all the well-known feedbacks shown in other melt-water experiments, such as reduced tropical salt transport by a more sluggish Gulf Stream, and additional melt-water input from the Greenland Ice Sheet (Figures 1c and 1d). Comparison of three further experiments, 2x, 3x, and 4xCO2_NOICE (without ice sheet feedback), clearly reveals that the changes in atmospheric surface buoyancy fluxes are sufficient to bring the system across the threshold for a collapsed thermohaline circulation and that the meltwater input from the Greenland Ice Sheet, which is substantially (roughly a factor of 10) smaller than the changes in atmospheric freshwater input, is of minor importance. The melt water input from the Greenland Ice Sheet is sensitive to changes in air temperatures, with the largest changes predicted for the 3XCO2 experiment which simulates the warmest surface temperature over Greenland, with an overall contribution to global mean sea level rise of about 1.0 m after 1000 years (Figure 1d).
Fig. 1: (a) Change in global mean surface temperature, (b) strength of the North Atlantic thermohaline circulation, (c) freshwater transport into the North Atlantic (filled circle: contributions from the Greenland ice sheet), (d) global sea level change by melt water input from Greenland, and (e) Antarctica (black: control run; blue: 2xCO2; green: 3xCO2; red: 4xCO2; thick: with ice sheet feedback; thin: without ice sheet feedback).(high resolution):
The 2xCO2 and 3xCO2 experiment exhibit a rather similar two-dimensional pattern of ice thickness changes in the Greenland Ice Sheet: reduced thickness in the margins due to intensified melting along the shore by the warming and increased thickness in the interior of North Greenland due to higher snowfall rates. The decrease in Greenland ice volume in the 3xCO2 experiment is almost balanced by the increase in volume of the Antarctic Ice Sheet (Figure 1e), which sees enhanced snowfall. In the experiment with 4xCO2, the pattern clearly differs from the 2x and 3xCO2 experiments. Cooling over the North Atlantic in the 4xCO2 experiment leads to a reduction in snowfall and moderate melting rates.
The feedbacks between climate and carbon cycle are most obviously documented by Figures 3 and 4a. CO2 uptake of the oceans by climate change [Sabine et al., 2004] is significantly reduced because of an increased equatorial gas exchange from the ocean to the atmosphere by increased upwelling.1 Moreover, a decrease in solubility due to increased stratification and warming in high latitudes reduces the storage of carbon by the oceans. The marine CO2 storage is also affected by the change in the biological pump, which is remarkably reduced by a decline of the carbon export (16% for experiment 3xCO2) from the euphotic layer into the abyssal layers (Figure 3b). The reduced export in the Pacific is somewhat contrary to an increased upwelling in the global warming scenarios. Increased upwelling (e.g., as observed during La Nina events) [e.g., Feely et al., 1999] is commonly associated with an increased nutrient supply from subsurface water. However, in experiment 3xCO2 the surface ocean is generally more depleted in nutrients than in experiment CONTROL (where a change in the atmospheric CO2 concentration does not have an effect on climate change) (Figure 4b). This change of the vertical nutrient gradient is highly related to a change in the high-latitude ventilation of subsurface water masses [Sarmiento et al., 2004], especially in the Pacific Ocean. A stagnant deep sea is prone to nutrient accumulation, since the loss by mixing is much slower than the gain due to the particulate organic carbon flux from the surface. Thus, surface and intermediate water masses are low in nutrients, showing similarities to glacial circulation reconstructions [Boyle, 1988]. Accordingly, negative nutrient anomalies are most pronounced in areas with intense upwelling, i.e., the east Pacific divergence zone and the North Pacific. With relatively low nutrient concentrations in equatorial surface waters, export of carbon into the twilight zone is substantially reduced (Figure 3b), exerting a positive feedback on these negative nutrient anomalies.
Fig. 2: Change in surface temperature [K] for 3xCO2 minus control (100-year mean for model years 900 1000).(high resolution):
 The decrease of the export of carbon into the deepsea in the 4xCO2 experiment relative to the control run also affects the rain ratio of CaCO3 to organic carbon with a change up to +0.02 or 20% in the eastern equatorial Pacific. Anthropogenic-induced changes in pH are predicted to be the largest in the Cenozoic (the last 66 million years) [Zeebe, 1999]. Our simulations show maximum changes of the pH value at the surface ocean related to the high atmospheric CO2 concentration and a more stagnant thermohaline circulation [Caldeira and Wickett, 2003; Heinze, 2004]. The pH value in the Southern Ocean and Antarctic Intermediate Water (Figure 4c) decreases up to 0.7 after 1000 years integration for the 4xCO2 experiment, which could enhance carbonate dissolution and changes in calcification of calcareous organisms such as corals, foraminifera, pteropods, or coccolithophores [e.g., Riebesell et al., 2000].  The sensitivity of carbon uptake on land by global change is mainly determined by the CO2 fertilization of plants, which has a positive effect on carbon storage (Figure 4a), and an increase in soil temperature with overall global warming of the landmasses, which has a negative
effect. Increase in temperatures causes a poleward shift of the boreal forest cover, with high soil carbon storage in the Northern Hemisphere high latitudes (Figure 4a). Due to increased soil temperatures, respiration of soil carbon is enhanced in the temperate latitudes, which causes a decrease in carbon storage. Changes in soil moisture have a minor influence on carbon storage. The overall changes associated with climate change reduce the carbon uptake of the land biosphere up to 43% (Figure 3a).  The total sensitivity of carbon uptake by the ocean and by the land with respect to global warming is about the same in the initial phase of the CO2 perturbation. After that, the terrestrial biosphere gets saturated quickly and the total uptake of carbon by ocean and land is primarily determined by the oceans (Figure 4a; 69% over the last 100 years of integration for 4xCO2). Overall, total carbon uptake of the ocean and the land is reduced by 24 28%, if changes in the atmospheric CO2 concentration have an effect on climate change (Figure 3a).
Fig. 3: (a) Cumulative carbon uptake by the ocean (solid) and by land (dashed). NOFEED denotes an experiment without climate change due to radiative forcing. (b) Export of carbon from the surface layer into the abyssal layers (black: control run; legend see also Figure 3a).
Summary and Conclusions
Fig. 4: Differences between the 3xCO2 and the control run (a) of cumulative uptake of carbon [kg C m 2] in the water column and on the land surface, (b) in a vertical section of phosphate concentration [mmol L 1] in the central Pacific, and (c) in a vertical section of pH value in the western Atlantic (100-year mean for model years 900 1000).
In summary, our simulations with the Earth system model confirm previous results that there is a strong feedback between the carbon cycle and the climate. In contrast to previous model simulations, we consider interactions between climate and continental ice sheets. The following main conclusions can be drawn from this study:  1. Changes in freshwater input from ice sheets by global warming have a secondary impact on variations in the thermohaline circulation and carbon cycle for the first 1000 years. However, uncertainties remain in ice sheet modeling, for example with regard to the role of ice shelves (not modeled in this approach but relevant for the evolution of the Antarctic Ice Sheet) and basal temperate ice layers potentially causing ice sheet instabilities.  2. Increased equatorial upwelling related to a decrease in the pole-to-equator temperature gradient and to an increase in moisture transport and a reduced biological pump enhance the gas exchange of carbon from the oceans to the atmosphere, lowering the CO2 storage by the oceans. On land, CO2 fertilization of plants and carbon emissions by increase in soil temperature reduce the anthropogenic carbon uptake, if changes in the atmospheric CO2 concentration have an effect on climate change. The importance of feedback of the land biosphere on climate is particular significant in the northward migration of forest over Asia and North America. The reduction of carbon storage due to climate change is in agreement with other studies [Cox et al., 2000; Dufresne et al., 2002; Govindasamy et al., 2005]. However, the reduction due to climate change does not exceed the uptake due to fertilization, as was reported by Cox et al.  due to changes in the Amazon region. Uncertainties in the feedbacks between the climate and the marine carbon cycle remain and a more detailed representation of the important processes [IPPC, 2001] needs to be subject of future investigations.