IPCC emission scenarios

U. Mikolajewicz, M. Gröger, E. Maier-Reimer, G. Schurgers, M. Vizca&íno & A. Winguth*
Max-Planck-Institute for Meteorology, Bundesstr. 53, 20146 Hamburg, Germany, mikolajewicz@dkrz.de
*) Center for Climatic Research, Madison, USA

1. Introduction

Model: MPI Earth System Model for paleo studies including atmosphere (ECHAM), ocean (LSG), ocean biogeochemistry (HAMOCC), terrestrial biosphere (LPJ), and ice sheets (SICOPOLIS).

Experiments: The model was forced with greenhouse gas emissions following the IPCC Scenarios B1, A1B, and A2 (Fig. 1) to the year 2100.

Fig. 1: Prescribend CO2 emission following the IPCC scenario B1 (green), A1B (red), and A2 (blue) .(high resolution):

For each scenario an ensemble of at least 3 simulations has been performed. After the year 2100 an exponential decline of of anthropogenic emissions was assumed. Individual simulations were integrated up to year 8000.


2. Atlantic Overturning Circulation

Fig. 2: Top: Strength of the North Atlantic overturning stream function at 30N at 1500 m depth in Sv. Below: Same as top but at 2500 m depth. .(high resolution):

The changes in the meridional overturning circulation of the North Atlantic (NAMOC, see Fig. d and e) reveal a strong sensitivity on the scenario. In the low emission scenario B1, the NAMOC is almost unchanged during the entire experiment (maximum reduction approximately 2 to 3 Sv). In the high emission scenario A2, the NAMOC in year 2100 is reduced to values between 18 to 20 Sv compared to 22 Sv in the long-term mean of the control run. In the year 2250, however, the deep water formation in the North Atlantic collapses completely, and in year 3000 the NADW cell has vanished in all A2 experiments. The experiments forced with SRES emission scenario A1B behave similar to the A2 experiments until the year 2100.
In the following years, however, the individual ensemble members behave quite differently. The model response ranges from experiments with a relatively fast collapse to experiments with a moderate reduction of the NAMOC. Here, the model is obviously close to a bifurcation point, the threshold for the NAMOC seems to be close to 10 Sv. Whereas two experiments make the transition during the strong warming phase between the years 2150 and 2250, one experiment crosses the threshold between the years 2600 to 2750, a period where the rate of change of surface temperatures is already rather small. The weakening of the NAMOC is always associated with a much shallower NAMOC, indicated by the much stronger reduction at 2500 m than at 1500 m. Whereas the NAMOC at 1500 m in the A1B_2 simulation has almost the same strength after the year 3000 as the control run, the NAMOC at 2500 m (see Fig. 11e) shows substantially reduced values. The reduction and collapse of the deep convection in the North Atlantic delays the global mean warming. This is visible as a relative plateau in the temperature time series of e.g. the A2 experiments between the years 2200 and 2350 (Figure 2). Mode transitions in the A1B experiments occur between the 2x and 3xCO2 levels, but the A1B-on simulations experience, for some time, atmospheric concentrations slightly higher than three times the preindustrial control value of 280 ppm. Winguth et al. (2005) report a collapse of the NAMOC for a 4xCO2 experiment with the present ESM, whereas the 2xCO2 and 3xCO2 experiments resulted in a weaker, but noncollapsed NAMOC.

2. Temperature change
For the near surface air temperature, the B1 simulations show a distinct warming pattern (see Fig. 2a). Over the ocean, the warming is roughly 1 K; over most of the land, the warming is between 1 and 2 K. Exceptions are the desert areas of North Africa and southwest Asia and the high northern latitudes with a warming between 2 and 3 K. In the northwest Pacific and close to the deep water formation sites of the Southern Ocean, strong warming is simulated due to reduced sea ice coverage. In the southern Indian Ocean, the southeast Pacific, and directly at the southeast coast of Greenland, areas of slight warming or even cooling are simulated. In the North Atlantic, these are the consequences of reduced deep convection.
For scenario A1B temperature anomalies are shown in Figs. 2b and c. The average of experiments A1B_2 and A1B_5 (henceforth also called A1B-on), which both have a noncollapsed NADW cell, show essentially the same spatial pattern as in the B1 experiments, but with stronger amplitude.

Fig. 3: Average Temperature change for the emission scenarios.Displayed are ensemble averages over the years 2801 to 3000. Units are K. Reference value is the climate of the control run. (high resolution)

The change in near surface air temperature in the two simulations with a fast collapse of the NADW cell is shown in Fig. 2c. The strong cooling southeast of Greenland is absent in the A1B-on simulations. Strong warming is simulated in the northwest Pacific (more than 8 K) and in the Ross Sea (>12 K). These two regions are associated with enhanced convection. In the North Pacific, a strong, relatively shallow overturning cell has developed with enhanced poleward oceanic heat transport. In the high emission A2 scenario, the NADW cell collapses in all simulations. The surface warming pattern (Fig. 2d) is very similar to the the A1B-off simulations. In contrast to the collapsed A1B simulations, Europe shows a warming everywhere due to the higher CO2 concentration, but the warming over northwest Europe is relatively small. The warming over the oceans lies typically between 3 and 5 K.

3. Changes in atmospheric moisture transports
The change in atmospheric moisture transports between A1B_2 (an experiment with non collapsed NAMOC) and the control run is displayed in Fig. 13, together with the climate of the control run. To first order the changes in moisture transport in this simulation reflect just an enhancement of the transport pattern of the control run. Exceptions are the subtropics and the Indian Ocean. The poleward atmospheric moisture transport is enhanced in all oceans. As a consequence, the Arctic and the Atlantic north of 60 N receive 0.1 Sv more freshwater from atmosphere, rivers, and Greenland; the Southern Ocean (south of 45 S) receives almost 0.2 Sv more freshwater. The atmospheric moisture transport across North America increases by 50%, adding 0.16 Sv additional freshwater input to the Atlantic.

(high resolution):
Fig. 4:Vertically integrated atmospheric moisture transports in kg/m/s. (a) Climate of the control run. The crosses indicate the watersheds used for the calculation of interbasin transports in table 2. (b) Difference between A1B_2 (mean over the years 2801 to 3000) and control run. (c) Difference between A1B_1 and A1B_2 (mean over the years 2801 to 3000). (d) Difference in precipitation minus evaporation between A1B_1 and A1B_2 (mean over the years 2801 to 3000) in mm/month. Only every second vector in each direction is shown.

In the tropics, however, the Atlantic export across America towards the Pacific is strongly enhanced. Together with an additional net export across Africa, Asia, and by transports towards the Southern Ocean, this almost completely cancels out the increased import across North America. The Atlantic north of 30N and the Arctic receive together 0.07 Sv more freshwater input from anomalous convergence of atmospheric moisture transports. These changes in interbasin moisture transports explain the weakening of the NADW formation in the North Atlantic seen in almost all experiments. In the North Pacific, the atmospheric moisture transports are divergent, associated with stronger evaporation in this region. As a consequence of these changes in atmospheric moisture transport, the northwest Atlantic becomes fresher (between 0.25 and more than 0.5 0) and the North Pacific becomes saltier (more than 0.5 0 in the northwest Pacific, regionally more than 1 0) (not shown). The saltier surface waters reduce the vertical stability in the northwest Pacific quite substantially. In the experiments with collapsed NADW cell (A1B-off), the convection in the northwest Pacific gradually deepens and induces a shallow meridional overturning cell. The associated ocean heat transport leads to an additional surface warming in the northwest Pacific (see Fig. 12), further enhancing evaporation. In the North Atlantic, the colder surface temperatures reduce evaporation. The resulting anomaly between a simulation with convection in the North Pacific and a simulation with convection in the North Atlantic is shown in Fig. 13c and d. In the tropical Atlantic, the ITCZ is shifted towards the southeast. An additional moisture export from the North Pacific into the Arctic and across North America into the Atlantic is obvious.

4. Ice sheets
The warming influences the evolution of the Greenland ice sheet. Until the year 2200, the changes are rather small. In the A1B-on experiments, a considerably reduced volume of the Greenland ice sheet is simulated for the year 3000. This change corresponds to approximately 0.9 m of global mean sea level. In the A1B-off simulations, this reduction in Greenland ice sheet volume is considerably smaller with approx. 0.2 m sea level equivalent (SLE). In the A2 experiments, the volume is reduced by 0.6 m SLE. In the following 1000 years, the melting of Greenland is quite substantial, reducing the volume of the Greenland ice sheet to approximately two thirds of its original value. In the A1B-on experiments, the volume is reduced by 1.7 m SLE; in the A1B-off experiments, by only 0.3 m SLE. In the B1 experiments, the change in volume of the Greenland ice sheet is rather small, with a reduction of 0.2 m SLE in the year 4000.

At the end of this millennium, a strong reduction in ice thickness near the coasts is simulated in the A1B-on experiments. This is caused by enhanced melting. The thickness in the interior of northern Greenland is increased due to enhanced accumulation. Ice thickness at the southeastern tip of Greenland is growing slightly for elevations above 2000 m due to stronger snow fall. The moderate warming, due to weakened poleward Atlantic heat transport, is too small to cause a substantial change in melting. On the western and eastern flanks of the ice sheet, a few grid points become ice free. In the A2 simulations the pattern of the changes in ice thickness is relatively similar to the A1B-on experiments. Except for southern Greenland, ice thickness decreases at almost all margins. In the east and central part of Greenland, reduced snow fall in the interior can explain the lowered thickness. This effect also plays a role near the margin of the ice sheet. In southern Greenland, increased precipitation rates over the whole area and low melting rates at the margins lead to larger ice thickness in elevated areas and at several locations the ice sheet is expanding. In the year 4000, the reduction in ice sheet thickness and extent is drastic. An exception is southern Greenland, where the thickness is almost identical to the thickness 1000 years earlier.

(high resolution):
Fig. 5: Changes in ice thickness relative to the control run. Top row averaged for the years 2801-3000, bottom row averaged over the years 3600 to 4000. Results are shown for one A1B simulation with a noncollapsed NAMOC (A1B_2), one with a noncollapsed NAMOC (A1B_1) and for one A2 simulation (A2_1). Units are m. Changes in the ice mask relative to the control run are indicated by encircling in pink (ice sheet retreat) and light blue (ice sheet advance). Isolines of surface topography are plotted in black. Displayed are the contours for 0 (land/sea mask) and 2000 m.

The effect of the changes in ice sheets on the deep water formation must be expected to be relatively small. The additional freshwater input into the North Atlantic is, in all experiments, dominated by the signal from the changes in atmospheric moisture transports. The effect of the melting of the Greenland ice sheet is typically smaller by an order of magnitude. The Antarctic ice sheet is growing in all greenhouse simulations. The accumulation in the interior is dominating over enhanced melting at the coasts. It must be noted that the model does not include an adequate treatment of ice shelves. A potential disintegration of the West Antarctic Ice Sheet due to a grounding line retreat caused by enhanced basal melting on the ice shelves due to warmer ocean temperatures (Warner and Budd, 1998), cannot be simulated. The control run is showing a slight drift of 0.2 m SLE in 1000 years. After the year 2200, the increase in volume is almost linear in all experiments. In the A2 experiments, the volume has increased by 3.2 m SLE. In the A1B experiments, the increase lies between 2 and 2.2 m SLE and in the B1 experiment at 0.7 m SLE.

2. Carbon Cycle
During the first centuries of all greenhouse experiments, when the prescribed anthropogenic emissions are strong, both ocean and terrestrial biospheres take up anthropogenic carbon at comparable rates. Initially, the terrestrial uptake is even slightly larger. After the peak in atmospheric CO2 concentrations has been passed, the amount of carbon stored in the terrestrial biosphere levels off and even starts to decrease slowly. Until the year 4000, the carbon content in the ocean continues to increase in all greenhouse experiments. At the year 4000, the increase in oceanic carbon content is 2540 Pg C (A1B), 3930 Pg C (A2) and 1395 Pg C (B1). The corresponding numbers for the terrestrial biosphere are 710 Pg C (A1B), 1025 Pg C (A2) and 360 Pg C (B1). The difference between the individual realisations of the A1B simulations are relatively small, indicating the minor importance of the MOC. In the year 4000, the fraction of emitted carbon that remains in the atmosphere increases substantially with increasing emissions: Results range from 11% for the B1_1 simulation up to 25% for the A2_1 experiment. However, especially for the high emission scenario A2, equilibrium is not yet reached in the year 4000 and the ocean still takes up approx. 0.63 Pg C yr-1 with the long time constants of the deep circulation. The atmospheric and terrestrial carbon pools are shrinking by 0.35 and 0.27 Pg C yr-1 to compensate for the reduced carbon content in the surface ocean.

(high resolution):
Fig. 6: Changes in carbon storage (kg/m2)

The change in vertically integrated carbon for both land and ocean at the end of this millennium is shown in Fig. 19. Carbon content increases almost everywhere, except for some land areas in the mid-latitudes of the Northern Hemisphere. Here, warmer temperatures enhance respiration. The reduced soil carbon locally leads to a decrease of the total carbon content. The simulated large increase in carbon content in all scenarios for the regions north of 60N is caused by a poleward shift of the boreal forest due to warmer temperatures. Forest expansion can be seen as well for the African and Indian monsoon regions, Australia, and dryer areas in North and South America. Only some regions around the Mediterranean show a decrease in forest, because precipitation decreases here. The increase of carbon content in the terrestrial biosphere is mainly caused by the increased CO2 concentration.
The entire water column is exposed to higher carbon content. Three main mechanisms are contributing to the carbon enrichment in the deep ocean: (1) Physical transport (advection, convection, mixing) of surface waters that are highly enriched with anthropogenic carbon (including changes of the solubility pump); (2) accumulation of particulate organic matter and CaCO3 shells sinking down from the surface ocean, and (3) dissolution of marine sediments due to the more acidic water. In the simulations without collapse of the NADW formation, the first process dominates in the North Atlantic. In the simulations with a collapsed NAMOC cell, the deep waters of the North Atlantic transform from a low nutrient/high oxygen water mass into an oxygen depleted water mass enriched with remineralized carbon and nutrients. In the A1B experiments, the difference between the simulations with and without collapse of NADW formation is comparably small. The amount of calciumcarbonate in the sediment is reduced due to dissolution caused by more acidic conditions in the overlying water masses.

Matthias Gröger (matthias.groeger@zmaw.de)
Last modified: May 18 2006.