Antarctic ice-sheet melting provides negative feedbacks on future climate warming

Antarctic ice-sheet melting provides negative feedbacks

on future climate warming

D. Swingedouw,1 T. Fichefet,1 P. Huybrechts,2 H. Goosse,1 E. Driesschaert,1

and M.-F. Loutre1

Received 21 April 2008; revised 9 July 2008; accepted 16 July 2008; published 10 September 2008.

[1] We show by using a three-dimensional climate model,which includes a comprehensive representation of polar icesheets, that on centennial to millennial time scales AntarcticIce Sheet (AIS) can melt and moderate warming in theSouthern Hemisphere, by up to 10�C regionally, in a 4 �CO2 scenario. This behaviour stems from the formation of acold halocline in the Southern Ocean, which limits sea-icecover retreat under global warming and increases surfacealbedo, reducing local surface warming. Furthermore, weshow that AIS melting, by decreasing Antarctic BottomWater formation, restrains the weakening of the Atlanticmeridional overturning circulation, which is a newillustration of the effect of the bi-polar oceanic seesaw.Consequently, it appears that AIS melting strongly interactswith climate and ocean circulation globally. It is thereforenecessary to account for this coupling in future climate andsea-level rise scenarios. Citation: Swingedouw, D., T. Fichefet,

P. Huybrechts, H. Goosse, E. Driesschaert, and M.-F. Loutre

(2008), Antarctic ice-sheet melting provides negative feedbacks

on future climate warming, Geophys. Res. Lett., 35, L17705,

doi:10.1029/2008GL034410.

1. Introduction

[2] Current anthropogenic greenhouse gas emissions arelikely to affect climate for millennia, notably due to the largethermal inertia of the oceans and the long memory of the icesheets [Meehl et al., 2007; Hasselmann et al., 2003].Archives of the past suggest noticeable Antarctic Ice-Sheet(AIS) melting contributions to sea-level changes during thelast deglaciation [Clark et al., 2002; Philippon et al., 2006]and glaciation [Kanfoush et al., 2000; Rohling et al., 2004],illustrating the possibility of massive freshwater input intothe Southern Ocean, which could have influenced theclimate [Weaver et al., 2003]. Recent observations reportan accelerated melting of the West Antarctic Ice Sheet[Rignot and Thomas, 2002; Cook et al., 2005; Velicognaand Wahr, 2006; Shepherd and Wingham, 2007]. This icemelting may partly explain the freshening of the Ross Seaobserved during the past four decades [Jacobs et al., 2002].Freshening also appears in the Antarctic Bottom Water(AABW) [Rintoul, 2007] and could limit this deep-waterformation in the future and affect climate. While none of thecoupled climate models participating to the IPCC Fourth

Assessment Report [Meehl et al., 2007] take into account theice sheets melting for projections going up to the year 2100,it is necessary to evaluate the potential effect of this meltingfor longer projections.[3] Potential irreversible changes both in the ice sheets

and ocean could actually lead to dangerous effects forthe environment, society and economy [Rahmstorf andGanopolski, 1999; Oppenheimer and Alley, 2004]. It istherefore urgent to account correctly for ice-sheet-climateinteractions in climate projections. Ice-sheet retreat canregionally enhance climate warming through changes intopography and albedo. Furthermore, ice-sheet meltingreleases freshwater into the ocean that can modify theocean circulation and sea ice cover [Weaver et al., 2003;Fichefet et al., 2003; Swingedouw et al., 2006], and thus theclimate. The Greenland and Antarctic ice sheets are ratherdifferent from each other since the total melting of theformer would represent around 7 m of sea-level rise, whilethe latter would correspond to about 61 m [Huybrechts,2002]. Moreover, contrary to the Greenland Ice Sheet(GIS), the AIS has massive ice shelves, bordering the Rossand Weddell Seas, where the bulk of AABW is formed. Theimpact of GIS melting on climate and ocean circulation hasbeen evaluated in several studies [Fichefet et al., 2003;Ridley et al., 2005; Swingedouw et al., 2006; Driesschaertet al., 2007], contrary to its southern counterpart, the AIS.In this study, we quantify the interactions of future AISmelting with climate, using the climate model LOVECLIM.

2. Experimental Design

[4] To capture the respective roles of the AIS and GISimpact under global warming, we performed 5 differentexperiments (Table 1) using LOVECLIM, a three-dimensionalEarth system model of intermediate complexity (EMIC)that includes representations of the polar ice sheets (seemethods section in the auxiliary materials).1 The firstexperiment is a control simulation (CTRL) under pre-industrial conditions that satisfactorily reproduces the climatemean state [Driesschaert et al., 2007]. In the other simula-tions, the atmospheric CO2 concentration is increased by 1%per year (compounded) until it reaches four times its initialvalue, where it remains unchanged for 3000 years. These areidealized experiments (called scenarios hereafter) designed tocapture the relevant ice-sheet-climate interactions in a warm-ing world at the millennial timescale. The first scenario(iAiG) has fully interactive ice sheets over Antarctica andGreenland, while in the second one (fAfG), climate compo-

1Auxiliary materials are available in the HTML. doi:10.1029/2008GL034410.

GEOPHYSICAL RESEARCH LETTERS, VOL. 35, L17705, doi:10.1029/2008GL034410, 2008ClickHere

for

FullArticle

1Institut d’Astronomie et de Geophysique Georges Lemaıtre, Uni-versite Catholique de Louvain, Louvain-la-Neuve, Belgium.

2Department of Geography, Vrije Universiteit Brussel, Brussels,Belgium.

Copyright 2008 by the American Geophysical Union.0094-8276/08/2008GL034410$05.00

L17705 1 of 4

nents are forced with a fixed ice-sheet configuration. In thisexperiment, we still force the ice sheets ‘‘off line’’ with thesimulated warming, but without the potential feedback ofmelting on climate. The ice sheets in this experiment aretherefore only ‘‘one-way’’ coupled. Two complementaryexperiments have been conducted to isolate the individualrole of the AIS and GIS. Experiment iAfG (fAiG) hasinteractive (fixed) AIS and fixed (interactive) GIS.

3. Results

[5] The AIS begins to loose mass after a few centuriesin iAfG and iAiG. This is in contrast with previous studies[Meehl et al., 2007; Mikolajewicz et al., 2007] and isrelated to a large warming over the AIS in this model,which leads to a larger increase in ablation than accumu-lation for the grounded AIS (see Figure S1 and Text S1 inthe auxiliary material). The melting of the AIS reduces theincrease in surface air temperature by 10% (0.3�C) on aglobal average after 500 years and beyond in iAfG andiAiG compared to fAfG and fAiG (Figure 1a). The relativecooling between iAiG and fAfG occurs mostly in thesouthern high latitudes (Figure 1b) and reaches 10�C inthe Weddell Sea sector (Figure 1c). This is associated with asmaller decrease in sea-ice cover in the Southern Ocean iniAiG compared to fAfG (Figure 1d). A slightly largerwarming appears north of 60�N in iAiG compared to fAfG,mostly after 2000 years. At that time, 70% of the GIS hasmelted (Figure S2), which explains this larger warming northof 60�N when GIS is interactive, and is due to a reduction inelevation and albedo over Greenland [Driesschaert et al.,2007]. In the Northern Hemisphere, the annual mean sea-iceextent decreases approximately at the same rate in thedifferent scenarios and evolves from 15 � 1012 km2 to 6 �1012 km2 after 3000 years. The annual mean sea-ice extentin the Southern Hemisphere decreases from 10 � 1012 km2

to 3� 1012 km2 in iAiG and to 0.9� 1012 km2 in fAfG after3000 years. Contrary to the melting of the GIS, the climaticimpact of AIS melting is therefore mainly due to interactionswith the ocean and sea ice. After 3000 years, there is anadditional freshwater input into the Southern Ocean of up to0.14 Sv in iAiG as compared to fAfG. This freshwaterdecreases the surface density of the Ross and WeddellSeas leading to the formation of a shallow halocline.

Consequently, the weakening of the deep convection andhence the reduction in vertical heat exchange in the oceanenhance the sea-ice extent, which cools the climatethrough the higher sea-ice albedo [Stouffer et al., 2007].

Table 1. Description of the 3000-Year Numerical Experiments Performed With LOVECLIM

Name Description

CTRL Control simulation with a constant forcing correspondingto pre-industrial conditions, notably with the CO2

concentration in the atmosphere set to 277.6 ppm.fAfG Scenario simulation in which the CO2 concentration increases

from the pre-industrial level by 1% per year and is maintainedconstant after 140 years of integration when it reaches a valueequal to four times the pre-industrial level (4 � CO2 scenario).The climate components experience constant Antarctic and Greenlandice-sheet areas and elevations, fixed at their preindustrial estimate.The potential melting of the ice sheets due to warming is howevercalculated ‘‘off line’’, but the corresponding freshwater fluxesare not released to the ocean.

iAiG Same as fAfG but with fully interactive Antarctic andGreenland ice sheets. Freshwater fluxes associatedwith melting are released to the ocean. Ice-sheet areaand elevation are free to evolve and to influence the climate.

fAiG Same as fAfG but with fully interactive Greenland ice sheet.iAfG Same as fAfG but with fully interactive Antarctic ice sheet.

Figure 1. Time series of the annual mean surface airtemperature (SAT in �C): (a) globally averaged from CTRL(black), iAiG (red), fAfG (green), iAfG (blue) and fAiG(purple dotted line) and (b) zonally averaged: differencebetween iAiG and fAfG. A 10-year running mean has beenapplied to all time series. (c) SAT difference between iAiGand fAfG averaged over years 2900 to 3000 expressed in �Cand (d) same difference but for sea-ice concentration foreach grid (ratio between 0 and 1), which is an index of sea-ice cover.

L17705 SWINGEDOUW ET AL.: AIS MELTING PROVIDES NEGATIVE FEEDBACKS L17705

2 of 4

[6] Furthermore, the freshwater input associated with AISmelting influences the ocean circulation in the scenarios.Without AIS melting, the annual mean AABW export at30�S (which is an index of the strength of the AABW cell)weakens during the first 300 years and then recovers (inagreement with studies from Bi et al. [2001] and Bates et al.[2005]), and is even enhanced compared to CTRL after 1000years (Figure 2a). This is caused by changes in the sea-icefreshwater forcing related to the retreat of the sea-ice cover(Figure S3). Indeed, the net annual mean sea-ice melting inthe Weddell and Ross Seas is lower in fAfG compared toCTRL. This increases the surface salinity and density, andcounteracts the density loss stemming from the temperatureincrease, leading to an increase in AABW formation in theseseas in fAfG compared to CTRL after 3000 years.[7] The AABW export is 35% smaller in iAiG than in

fAfG, due to a decrease in surface density around Antarcticaand a reduction in AABW formation, associated with AISmelting. Interestingly, the AIS melting also affects the NorthAtlantic Deep Water (NADW) export (which is an index ofthe strength of the NADW cell). At 30�S, this exportdiminishes in all the scenarios (Figure 2b), but recovers after1000 years in iAfG contrary to fAfG, illustrating the stabi-lizing effect of AIS melting on the NADW cell weakening.When GIS melting is accounted for, the NADW cell furtherweakens. This melting notably leads to a peak difference of3.3 Sv (23% of NADW export at 30�S in CTRL) in fAiG

compared to fAfG after 2000 years. The AIS melting oncemore reduces the NADW cell weakening by 1.2 Sv in iAiGcompared to fAiG. This stabilization effect of the AISmelting on the NADW cell can be explained by the so-calledbi-polar ocean seesaw [Stocker et al., 1992; Seidov et al.,2001; Brix and Gerdes, 2003], which emphasizes that areduction in AABW density allows the NADW to penetratedeeper and further south in the Atlantic, enhancing theassociated cell (see Text S1).[8] Another important impact of ice-sheet melting con-

cerns the sea-level rise. Here, we evaluate how interactionsbetween climate and ice-sheet melting can feed back on thismelting and influence sea-level rise in the various scenarios(Table S1). According to its relative warming effect, the GISmelting yields a positive feedback: in line with earlier findingusing LOVECLIM [Driesschaert et al., 2007], the whole icesheet has melted in fAiG and iAiG after 3000 years, while60% remains in fAfG and iAfG. This positive feedback is dueto the reduction in albedo and altitude of the ice sheet, whichaccelerates the melting. On the contrary, according to itsrelative cooling effect, the AIS melting produces a negativefeedback, quantified by the comparison of the Antarcticcontributions to global sea-level rise in iAiG (3.2 m) and infAfG (10.0 m, calculated but not released to the ocean) after3000 years. Moreover, the AIS melting tends to increase theoceanic heat content (Figure 3) and leads to a larger thermalexpansion in iAiG compared to fAfG. This effect increasesthe sea-level rise by 1.4 m in iAiG compared to fAfG andcorresponds to a warming at depth, while the surface,particularly in the Southern Ocean, experiences cooling. Thisis due to the capping of the ocean surface by freshwatercoming from the AIS melting, which inhibits the verticalmixing of heat in high latitudes and warms the ocean interior.On the whole, after 3000 years, the sea-level rise is 13.8 m iniAiG, or 0.8 m less than the 14.6 m calculated in fAfG,illustrating the compensation, in terms of sea-level rise,between the GIS positive feedback and the AIS negativefeedback.

4. Conclusions

[9] A number of factors should however be borne inmind when interpreting our results. The model used is an

Figure 2. Time series of the annual mean value of (a) theminimum of the oceanic global meridional overturningstreamfunction at 30�S (in Sv, 1 Sv = 106 m3/s), representingthe export of Antarctic and Circumpolar Deep Water(AABW and CDW) at 30�S, and (b) the maximum of theAtlantic meridional overturning streamfunction at 30�S,representing the export of North Atlantic Deep Water(NADW) at 30�S. CTRL is in black, iAiG in red, fAfG ingreen, iAfG in blue and fAiG in purple dotted line. A 21-yearrunning mean has been applied to all time series.

Figure 3. Latitude-depth distribution of the annuallyaveraged temperature difference (in �C), years 2900 to3000, of iAiG minus fAfG in the global ocean. Blue (red)shading indicates values where the water is colder (warmer)in iAiG than in fAfG. The contour interval is 0.2�C.

L17705 SWINGEDOUW ET AL.: AIS MELTING PROVIDES NEGATIVE FEEDBACKS L17705

3 of 4

EMIC and has therefore a rather coarse resolution. Thiscould affect deep water formation and the interactionbetween the ocean and the ice-shelves [Nicholls, 1997]but this is presently unavoidable to simulate the long-termevolution of climate. Nonetheless, LOVECLIM has reachedsufficient realism concerning ice-sheet-climate interactionsto correctly capture the underlying mechanisms we haveillustrated here. The present study should not be seen as aforecast but gives insight on the potential feedbacks betweenclimate and ice sheets melting for a given warming scenario.Regarding the ice-sheet model, some of the potentially fastprocesses (basal lubrication from penetrating surface meltwater, ice-flow acceleration induced by ice-shelf disintegra-tion) by which warming may contribute to the ice-sheet massloss are not fully represented [Alley et al., 2005] so that afaster decay could potentially happen. Note that ice sheetmelting might also be more rapid if processes responsible forthe widespread glacier acceleration currently observed inAntarctica [e.g., Rignot et al., 2008] were taken into accountin the model. We therefore argue that ongoing efforts in ice-sheet modelling should continue and that AIS models shouldbe incorporated interactively in current ocean-atmospheregeneral circulation models for centennial and millennialprojections of the climate system.

[10] Acknowledgments. We thank Chris Konig-Beatty, GillesRamstein and Susan Solomon for comments on an earlier version of themanuscript. We gratefully acknowledge the constructive comments fromtwo anonymous reviewers. This work was supported by the Marie CurieResearch Training Network NICE from the EU FP6 programme and by theASTER project of the Belgian Federal Science Policy Office Programme onScience for a Sustainable Development. The authors wish to acknowledgeuse of the Ferret program for analysis and graphics in this paper and the helpof Patrick Brockmann for the use of this program.

ReferencesAlley, R. B., et al. (2005), Ice sheet and sea-level changes, Science, 310,456–460.

Bates, M. L., M. H. England, and W. P. Sijp (2005), On the multi-century Southern Hemisphere response to changes in atmosphericCO2-concentration in a global climate model, Meteorol. Atmos. Phys.,89, 17–36.

Bi, D., W. F. Budd, A. C. Hirst, and X. Wu (2001), Collapse and reorga-nisation of the Southern Ocean overturning under global warming in acoupled model, Geophys. Res. Lett., 28, 3927–3930.

Brix, H., and R. Gerdes (2003), North Atlantic Deep Water and AntarcticBottom Water: Their interaction and influence on the variability of theglobal ocean circulation, J. Geophys. Res., 108(C2), 3022, doi:10.1029/2002JC001335.

Clark, P. U., et al. (2002), Sea-level fingerprinting as a direct test for thesource of global meltwater pulse IA, Science, 295, 2438–2441.

Cook, A. J., et al. (2005), Retreating glacier fronts on the AntarcticPeninsula over the past half-century, Science, 308, 541–544.

Driesschaert, E., T. Fichefet, H. Goosse, P. Huybrechts, I. Janssens,A. Mouchet, G. Munhoven, V. Brovkin, and S. L. Weber (2007),Modeling the influence of Greenland ice sheet melting on the Atlanticmeridional overturning circulation during the next millennia, Geophys.Res. Lett., 34, L10707, doi:10.1029/2007GL029516.

Fichefet, T., C. Poncin, H. Goosse, P. Huybrechts, I. Janssens, and H.Le Treut (2003), Implications of changes in freshwater flux from theGreenland ice sheet for the climate of the 21st century, Geophys. Res.Lett., 30(17), 1911, doi:10.1029/2003GL017826.

Hasselmann, K., et al. (2003), The challenge of long-term climate change,Science, 302, 1923–1925.

Huybrechts, P. (2002), Sea-level changes at the LGM from ice-dynamicreconstructions of the Greenland and Antarctic ice sheets during theglacial cycles, Quat. Sci. Rev., 21, 203–231.

Jacobs, S. S., C. F. Giulivi, and P. A. Mele (2002), Freshening of the RossSea during the late 20th century, Science, 297, 386–389.

Kanfoush, S. L., et al. (2000), Millennial-scale instability of the Antarcticice sheet during the last glaciation, Science, 288, 1815–1818.

Meehl, G. A. et al. (2007), Global climate projections, in Climate Change2007: The Physical Science Basis. Contribution of Working Group I tothe Fourth Assessment Report of the Intergovernmental Panel on ClimateChange, edited by S. Solomon, et al., pp. 747–845, Cambridge Univ.Press, Cambridge, U. K.

Mikolajewicz, U., E. Groger, E. Maier-Reimer, G. Schurgers, M. Vizcaino,and A. M. E. Winguth (2007), Long-term effects of anthropogenic CO2

emissions simulated with a complex earth system model, Clim. Dyn., 28,599–631.

Nicholls, K. W. (1997), Predicted reduction in basal melt rates of anAntarctic ice shelf in a warmer climate, Nature, 388, 460–462.

Oppenheimer, M., and R. B. Alley (2004), The West Antarctic ice sheetand long term climate policy—An editorial comment, Clim. Change, 64,1–10.

Philippon, G., G. Ramstein, S. Charbit, M. Kageyama, C. Ritz, andC. Dumas (2006), Evolution of the Antarctic ice sheet throughout the lastdeglaciation: A study with a new coupled climate—North and southhemisphere ice sheet model, Earth Planet. Sci. Lett., 248, 750–758.

Rahmstorf, S., and A. Ganopolski (1999), Long-term global warmingscenarios computed with an efficient coupled climate model, Clim.Change, 43, 353–367.

Ridley, J. K., et al. (2005), Elimination of the Greenland ice sheet in a highCO2 climate, J. Clim., 18, 3409–3427.

Rignot, E., and R. H. Thomas (2002), Mass balance of the polar ice sheets,Science, 297, 1502–1506.

Rignot, E., et al. (2008), Recent Antarctic ice mass loss from radarinterferometry and regional climate modelling, Nat. Geosci., 1, 106–110.

Rintoul, S. R. (2007), Rapid freshening of Antarctic Bottom Water formedin the Indian and Pacific oceans, Geophys. Res. Lett., 34, L06606,doi:10.1029/2006GL028550.

Rohling, E. J., et al. (2004), Similar meltwater contributions to glacial sealevel changes from Antarctic and northern ice sheets, Nature, 430,1016–1021.

Seidov, D., E. Barron, and B. J. Haupt (2001), Meltwater and the globalocean conveyor: Northern versus southern connections, Global Planet.Change, 30, 257–270.

Shepherd, A., and D. Wingham (2007), Recent sea-level contributions ofthe Antarctic and Greenland ice sheets, Science, 315, 1529–1532.

Stocker, T. F., D. G. Wright, and W. S. Broecker (1992), Influence ofhigh-latitude surface forcing on the global thermohaline circulation,Paleoceanography, 7, 529–541.

Stouffer, R. J., D. Seidov, and B. J. Haupt (2007), Climate response toexternal sources of freshwater: North Atlantic versus the Southern Ocean,J. Clim., 20, 436–448.

Swingedouw, D., P. Braconnot, and O. Marti (2006), Sensitivity of theAtlantic Meridional Overturning Circulation to the melting from northernglaciers in climate change experiments, Geophys. Res. Lett., 33, L07711,doi:10.1029/2006GL025765.

Velicogna, I., and J. Wahr (2006), Measurements of time-variable gravityshow mass loss in Antarctica, Science, 311, 1754–1756.

Weaver, A. J., et al. (2003), Meltwater pulse 1A from Antarctica as a triggerof the Bølling-Allerød warm interval, Science, 299, 1709–1713.

�����������������������E. Driesschaert, T. Fichefet, H. Goosse, M.-F. Loutre, and D.

Swingedouw, Institut d’Astronomie et de Geophysique Georges Lemaıtre,Universite Catholique de Louvain, Chemin du Cyclotron 2, B-1348Louvain-la-Neuve, Belgium. ([email protected])P. Huybrechts, Department of Geography, Vrije Universiteit Brussel,

Pleinlaan 2, B-1050 Brussels, Belgium.

L17705 SWINGEDOUW ET AL.: AIS MELTING PROVIDES NEGATIVE FEEDBACKS L17705

4 of 4

SWINGEDOUW ET AL.: SUPPLEMENTARY MATERIALS X - 1

1. Methods: Description of the climate model LOVECLIM

LOVECLIM consists of five components representing the atmosphere (ECBilt), the

ocean and sea ice (CLIO), the terrestrial biosphere (VECODE), the oceanic carbon cy-

cle (LOCH) and the Greenland and Antarctic ice sheets (AGISM). ECBilt is a quasi-

geostrophic atmospheric model with 3 levels and a T21 horizontal resolution (Opsteegh et

al. 1998), which contains a full hydrological cycle and explicitly computes synoptic vari-

ability associated with weather patterns. Cloud cover is prescribed according to present-

day climatology, which is a limitation of the present study. CLIO is a primitive-equation,

free-surface ocean general circulation model coupled to a thermodynamic-dynamic sea-

ice model (Goosse and Fichefet 1999). Its horizontal resolution is 3◦× 3◦, and there are

20 levels in the ocean. VECODE is a reduced-form model of vegetation dynamics and

of the terrestrial carbon cycle (Brovkin et al. 2002). It simulates the dynamics of two

plant functional types (trees and grassland) at the same resolution as that of ECBilt.

ECBilt-CLIO-VECODE has been utilized in a large number of climate studies (please

refer to http://www.knmi.nl/onderzk/CKO/ecbilt-papers.html for a full list of references). LOCH

is a comprehensive model of the oceanic carbon cycle (Mouchet and Francois 1996). It

takes into account both the solubility and biological pumps, and runs on the same grid

as the one of CLIO. This model was not activated in the present study, and we pre-

D R A F T August 4, 2008, 11:29am D R A F T

X - 2 SWINGEDOUW ET AL.: SUPPLEMENTARY MATERIALS

scribe the evolution of the atmospheric CO2 concentration. Finally, AGISM is composed

of two three-dimensional thermomechanical models of the ice-sheet flow, coupled to a

visco-elastic bedrock model and a model of the mass balance at the ice-atmosphere and

ice-ocean interfaces (Huybrechts 2002). For both ice sheets, calculations are made on a 10

km × 10 km resolution grid with 31 sigma levels. Given the long time-scales investigated

here, the model is among the most complex climate models that can be applied to study

this type of questions at present. Note that the mask of the ice shelves is fixed under

present-day configuration, and the land-sea mask is not modified during the integration

for the ocean, but can change for the ice-sheet models.

The atmospheric variables needed as input for AGISM are surface temperature and pre-

cipitation. Because the details of the Greenland and Antarctica surface climates are not

well captured on the ECBilt coarse grid, these boundary conditions consist of present-day

observations as represented on the much finer AGISM grid onto which climate change

anomalies from ECBilt are superimposed (Driesschaert et al. 2007). Monthly temper-

ature differences and annual precipitation ratios, computed against a reference climate

corresponding to the period 1970-2000 AD, are interpolated from the ECBilt grid onto

the AGISM grid and added to and multiplied by the observed surface temperatures and

precipitation rates, respectively. The oceanic heat flux at the base of Antarctic ice shelves

is also calculated in perturbation mode using the parameterization proposed by Beckmann

and Goosse (2003). After performing mass balance and ice dynamics computations, AG-

ISM transmits the calculated changes in land fraction covered by ice and in orography to

ECBilt and VECODE. In addition, AGISM provides CLIO with the geographical distri-

D R A F T August 4, 2008, 11:29am D R A F T

SWINGEDOUW ET AL.: SUPPLEMENTARY MATERIALS X - 3

bution of the annual mean surface freshwater flux resulting from ice sheet runoff, iceberg

calving, runoff from ice-free land and basal ice melting from both below the grounded ice

sheet and its surrounding ice shelves. All of these sources of fresh water are added to the

surface layer of coastal oceanic grid boxes. The Greenland (Antarctic) ice-sheet module

was first integrated over the last two (four) glacial cycles up to 1500 AD with forcing

from ice core data to derive initial conditions for coupling with the other components of

LOVECLIM. The control experiment (CTRL) of 3000-year duration was then conducted

with LOVECLIM under forcing conditions corresponding to 1500 AD. The same initial

conditions are used for all the scenario simulations performed in this study.

The model version used here is LOVECLIM1.1. Three main improvements have been

incorporated in this version compared to LOVECLIM1.0 (Goosse et al. 2007). First, the

land-surface scheme has been modified (see http://www.astr.ucl.ac.be/ASTER/doc/E AR SDCS01A v2.pdf)

in order to take into account the impact of the changes in vegetation on the evaporation

(transpiration) and on the bucket depth (i.e. the maximum water that can be hold in

the soil). Second, the emissivity, which was the same for all the surface types in LOVE-

CLIM1.0, is now different for land, ocean and sea ice. Third, in order to reduce the

artificial vertical diffusion in the ocean caused by numerical noise, the Coriolis term is

now treated in a fully implicit way in the equation of motion for the ocean, while a

semi-implicit scheme was used in LOVECLIM1.0.

2. Supplementary discussion 1: Mass balance of the Antarctic ice sheet

Under global warming conditions, the mass balance of the grounded AIS depends on the

rate of change in accumulation over the ice sheet and ice loss around its perimeter, through

D R A F T August 4, 2008, 11:29am D R A F T

X - 4 SWINGEDOUW ET AL.: SUPPLEMENTARY MATERIALS

surface runoff and ice discharge across the grounding line into the surrounding ice shelves.

Few studies have analysed the long-term mass balance of the AIS in millennial projections.

Huybrechts and de Wolde (1999) find a negative sea-level rise contribution from the AIS

of -0.3 m after 1000 years in a 4×CO2 experiment similar to the one performed in the

present study, but with the forcing derived from a two-dimensional climate model. An

annual mean temperature rise over Antarctica of 5.5◦C is simulated in the Huybrechts

and de Wolde (1999) study after 1000 years. In an 8×CO2 experiment, Huybrechts and

de Wolde (1999) simulate an 8.5◦C warming over Antarctica and a positive sea-level

rise contribution of 0.8 m after 1000 years. More recently, Mikolajewicz et al. (2007),

using another ice-sheet model coupled to a state-of-the-art climate model, find negative

contribution in terms of sea-level rise for the AIS in different projections using emission

scenarios going from B1 up to A2. In their model the increase in accumulation over the

grounded AIS is always larger than the increase in ablation for the grounded AIS.

In the present study, the simulated warming over Antarctica after 3000 years is rather

large in the projections as compared to CTRL. The annual mean temperature rise over

Antarctica reaches values of 9◦C in iAiG and 12◦C in fAfG for a spatial average over

Antarctica. This is due to a large polar amplification in this model that leads to an

important warming over the AIS in fAfG (Supplementary Figure 1.b). This polar ampli-

fication put the model used here in the higher range of polar amplification as simulated

in climate models (Meehl et al. 2007, Masson-Delmotte et al. 2006). Nonetheless, since

this model is on the lower range for climate sensitivity (Meehl et al. 2007), the simulated

warming over the AIS is not unrealistic, given the several centuries necessary to reach

D R A F T August 4, 2008, 11:29am D R A F T

SWINGEDOUW ET AL.: SUPPLEMENTARY MATERIALS X - 5

such a warming, due to the thermal inertia of the Southern Ocean. This large warming

over the AIS leads to a rather rapid decay of the West AIS in fAfG and also a substan-

tial retreat in some coastal parts of the East AIS, notably in the Ninnis-Mertz glacier

basins in Victoria Land (Supplementary Figure 4). The processes causing this decay are

both the development of a peripheral surface ablation zone, as in Greenland today, as

the demise of the surrounding ice shelves from both large increases in surface and bottom

melting. Consequently, the grounded AIS already looses a substantial fraction of its mass

after 1000 years in iAiG and fAfG (Supplementary Figure 2), due to a larger increase of

ablation (sum of surface runoff from grounded ice, basal melting below grounded ice, flux

across grounding line) over accumulation (Supplementary Figure 1.a). The AIS melting

corresponds here, after 1000 years, to a sea-level rise of 0.5 m in iAiG and 1.5 m in fAfG.

This result differs from Huybrechts and de Wolde (1999) for a 4×CO2 experiment, but is

coherent with the AIS response to a larger warming as found in the 8×CO2 experiment,

in which the warming over Antarctica in Huybrechts and de Wolde (1999) is similar to our

4×CO2 experiment. The differences in mass balance response of the AIS compared to the

Huybrechts and de Wolde (1999) and Mikolajewicz et al. (2007) are therefore due to the

different climate model used here that exhibits a large polar amplification and warming

over Antarctica.

3. Supplementary discussion 2: Issues concerning the bi-polar ocean seesaw

The effect of the bi-polar ocean seesaw has been illustrated in numerical simulations

(Stocker et al. 1992, Seidov et al. 2001) and it has been shown that changing surface

buoyancy forcing in key deep-water formation areas can disturb the balance between

D R A F T August 4, 2008, 11:29am D R A F T

X - 6 SWINGEDOUW ET AL.: SUPPLEMENTARY MATERIALS

NADW and AABW cells on millennial time-scales. Thus a decrease in AABW formation

reduces the AABW cell and enhances the NADW cell. The exact oceanic mechanism that

yields this interaction however remains unclear. Furthermore, recent simulations (Stouffer

et al. 2007, Seidov et al. 2005) show that, on centennial time-scales, an additional 1 Sv

input of freshwater into the ocean, south of 60◦S, has nearly no impact on the NADW cell.

This result questions the validity of the bi-polar ocean seesaw since AABW formation is

strongly reduced in those experiments. Two explanations arise to account for this issue:

(i) the Seidov et al. (2005) and Stouffer et al. (2007) experiments use transient simulations

and the bi-polar ocean seesaw effect applies on longer time-scales, due to adjustment in

the ocean interior that necessitates thousands of years; (ii) the experimental design of

the numerical simulations from Stocker et al. (1992) and Seidov et al. (2001) on the one

side, and Seidov et al. (2005) and Stouffer et al. (2007) on the other side, are different

since the first-named impose surface buoyancy forcing anomalies in some key regions of

the Southern Ocean, using an ocean-only model, while the last-named, using an ocean-

atmosphere coupled model, put freshwater anomalies in the ocean south of 60◦S, which

can spread through the intense currents of the Southern Ocean. Furthermore, observations

over the last decades suggest that variability in the NADW circulation is hardly influenced

by AABW (Koltermann et al. 1999).

In the present study, we have shown that even with an experimental design where fresh-

water is released into the ocean, using a coupled climate model, the bi-polar ocean seesaw

applies in LOVECLIM. We propose to analyze in further depth the exact mechanisms of

D R A F T August 4, 2008, 11:29am D R A F T

SWINGEDOUW ET AL.: SUPPLEMENTARY MATERIALS X - 7

the bi-polar ocean seesaw in a future study, since it is not the focus of the present one, in

order to try piecing some of the puzzle together (see Swingedouw et al. 2008).

References

Beckmann, A., and H. Goosse (2003), A parameterization of ice shelf-ocean interaction

for climate models, Ocean Modell., 5, 157–170.

Brovkin, V., et al. (2002), Carbon cycle, vegetation, and climate dynamics in the Holocene:

Experiments with the CLIMBER-2 model. Global Biogeochem. Cycles, 16, doi:10.1029/

2001GB001662.

Goosse, H., and T. Fichefet (1999), Importance of ice-ocean interactions for the global

ocean circulation: A model study, J. Geophys. Res., 104, 337–355.

Goosse, H., E. Driesschaert, T. Fichefet, and M.-F. Loutre (2007), Information on the

early Holocene climate constrains the summer sea ice projections for the 21st century.

In press in Climate of the Past Discussion.

Huybrechts, P., and J. de Wolde (1999), The dynamic response of the Greenland and

Antarctic ice sheets to multiple-century climatic warming. J. Climate, 12, 2169–2188.

Huybrechts, P. (2002), Sea-level changes at the LGM from ice-dynamic reconstructions of

the Greenland and Antarctic ice sheets during the glacial cycles. Quat. Sci. Rev., 21,

203–231.

Koltermann, K. P.,et al. (1999), Decadal changes in the thermohaline circulation of the

North Atlantic, Deep-Sea Res. II, 46, 109–138.

Masson-Delmotte, V., et al. (2006), Past and future polar amplification of climate change:

climate model intercomparisons and ice-core constraints, Clim. Dyn., 27, 437–440.

D R A F T August 4, 2008, 11:29am D R A F T

X - 8 SWINGEDOUW ET AL.: SUPPLEMENTARY MATERIALS

Mouchet, A., and L. M. Francois (1996), Sensitivity of a global oceanic carbon cycle model

to the circulation and to the fate of organic matter: Preliminary results. Phys. Chem.

Earth, 21, 511–516.

Opsteegh, J. D., et al. (1998), ECBILT: A dynamic alternative to mixed boundary con-

ditions in ocean models. Tellus A, 50, 348–367.

Seidov, D., R. J. Stouffer, and B. J. Haupt (2005), Is there a simple bi-polar ocean seesaw?

Global Planetary Change, 49, 19–27.

Swingedouw, D., T. Fichefet, H. Goosse, M.-F. Loutre (2008), Impact of transient freshwa-

ter releases in the Southern Ocean on the AMOC and climate. Clim. Dyn., submitted.

D R A F T August 4, 2008, 11:29am D R A F T

!! Please write \lefthead{<AUTHOR NAME(s)>} in file !!: !! Please write \righthead{<(Shortened) Article Title>} in file !!X - 1

Table 1. Sea-Level Rise for the Different Experiments in Comparison With CTRL After 3000

Yearsa

Antarctica Greenland ThermalExpansion

Total

iAiG 3.2 8.0 2.6 13.8fAfG 10.0 3.4 1.2 14.6fAiG 9.8 7.9 1.5 19.2iAfG 3.2 3.6 2.3 9.1

a (in m). Sea-level rise is decomposed into the contribution from Antarctic and Greenland

ice sheets melting and thermal expansion. The figures in italic stand for the fact that they have

been calculated, but the associated melting has not been released to the ocean and has therefore

no impact on ocean circulation.

D R A F T August 4, 2008, 11:17am D R A F T