Glacial Ocean Circulation and Stratification Explained by Reduced

Home , Antarctic Bottom Water, Brine rejection

Downloaded by guest on September 23, 2021 LGM storage. carbon ocean origin, in role Antarctic important of an play water may maximum age and moving radiocarbon middepth slowly observed the explains for circula- which room overturning interhemispheric makes over- the shallower tion interhemispheric for The evidence the proxy glacial. of with last consistent shoaling again circulation, and together turning the weakening goes for stratification a increased strati- inferred with The ocean salinities maximum. deep abyssal glacial increased high last strongly with a consistent enhanced to fication, associated leads The rejection Antarctica. cover brine around ice sea rate increased formation to and differences. lead temperature temperatures atmospheric atmospheric interpreted of Colder be consequence can direct climates vari- a stratifica- interglacial as and that and circulation glacial shows ocean between deep study tion the in This changes our climates. inferred to ous challenge glacial changed, major circulation of a represent ocean understanding and deep unclear ocean. still the the are which however, and by atmosphere mechanisms the The between the carbon in affecting role of by swings important partitioning dioxide an carbon circulation played atmospheric observed likely ocean the have deep which the stratification, in and rearrangements major 5–10 with of ciated 2016) order 27, June the review changes for temperature on (received high-latitude 2016 with 7, periods, and November time glacial approved between interglacial and shifts CA, dramatic Jolla, undergone La has Diego, climate Earth’s San California, of University Thiemens, H. Mark by Edited a Jansen F. Malte temperature atmospheric reduced explained by stratification and circulation ocean Glacial www.pnas.org/cgi/doi/10.1073/pnas.1610438113 in tested deep is in hypothesis changes study. This observed this circulation. the and to stratification led increased ocean then an to which amounts rate. ocean loss, strong the freezing buoyancy that into net flux hypothesis increased salt the net an resulting to to The rise led and gives Antarctica This present around 13). cooling the 12, between 6, stratification (3, differ- LGM and for circulation made an inferences in to with ences agreement lead in southward-flowing (11)—all NADW, and to AABW of NADW between shift expected separation upward clearer an a is and stratification, abyssal Antarctica the Maximum enhanced in around increase Glacial that showed loss Last recently buoyancy we and and Antarc- Specifically, stratification present (7–11). around ocean (LGM) the deep fluxes the between buoyancy in surface circulation changes and controlling in ice tica sea of between tance mixing less with originat- distinct, masses two more (6). leaving water the them appear Moreover, with depths, (3–5). masses filled Antarctica shallower water around Atlantic from to equivalent deep primarily confined ing glacial the The likely of 2). was more resur- (1, and NADW Antarctica southward of upwells around returning slowly before again then NADW the Atlantic facing the lower in into the it surface into makes formed the that basins. is AABW abyssal to the (AABW) The into up northward water spreads and back bottom Antarctica around rising Antarctic Ocean. eventually about of Southern and depth a km at southward 2–3 flowing before Atlantic North the T eateto h epyia cecs h nvriyo hcg,Ciao L60637 IL Chicago, Chicago, of University The Sciences, Geophysical the of Department utpesuishv one oadteptnilimpor- potential the toward pointed have studies Multiple ede ca oa svniae anyb w water two in formed by is mainly (NADW) ventilated water deep is Atlantic today North ocean masses. deep he | AMOC | stratification a,1 ◦ .Teeciai hfshv enasso- been have shifts climatic These C. | cooling | sea-ice 1073/pnas.1610438113/-/DCSupplemental. at online information supporting contains article This Submission. 1 Direct PNAS a is article This interest. of conflict no declares author The paper. the wrote fAB.Tepa vrunn rnprso 1. Sverdrup +17.5 of transports pathway overturning the peak representing The cell, AABW. abyssal AMOC of anticlockwise with clockwise an meridional associated the lies Atlantic Below cell is the (AMOC). of tongue part circulation salty streamfunc- as overturning stratifica- this southward overturning salinity flows that associated that the reveals NADW The of 3A) (2). reversal (Fig. abyss tion a of the to tongue in South- leading a the tion and into see southward Ocean we penetrating ern particular, middepth, at observed In water features Atlantic. salty basic present-day the the reproduce and averaged in which 2A) zonally 2C), (Fig. shows is (Fig. temperature 2 Atlantic salinity potential Fig. North of free. The sections ice contours). depth–latitude white and 1A, warmer (Fig. somewhat forms ice 28 broad − simu- sea about (∼ in point freezing from SSTs the the to varying gives tropics observations, over forcing with chosen (SST) agreement The present-day temperature forcing. resembling conditions atmospheric surface boundary with domain, sea 1A features lated Fig. key circulation. the and reproduce state to shows ocean ability (interglacial) model’s modern the the of on focus first We 12).Results 6, 5, (3, circu- inferred elemental been most have the changes where lation The Oceans, Methods). Southern the and and resembling Atlantic configuration continental (Materials idealized an conditions uses model boundary pre- as evaporation–precipitation atmo- scribed proposed and with winds, the model, temperature, ocean–sea-ice isolate coupled spheric to a use us We allow mechanism. idealized which using simulations, changes, stratification numerical and circulation ocean and uhrcnrbtos ...dsge eerh efre eerh nlzddt,and data, analyzed research, performed research, designed M.F.J. contributions: Author mi:[email protected]. Email: ca iclto,adamshrcCO temperature, atmospheric atmospheric and circulation, between ocean loop feedback of notion positive tem- the a supports atmospheric changes transi- between circulation ocean link glacial–interglacial and direct perature of the particular, interpretation by In tions. impli- our directly important for has caused which cations are warming, or in stratification cooling changes atmospheric and observed circulation that suggest ocean proxy ocean results various and The consis- deep understanding observations. physical climates the our interglacial in both with and differences This tent glacial circulation. for ocean between model deep circulation a the in presents in fluctuations changes study observed by driven the (CO likely dioxide explain carbon to atmospheric need we interglacial climates and glacial between swings climatic understand To Significance ets h oncinbtenamshrctemperature atmospheric between connection the test We PNAS | aur ,2017 3, January 2 2 ◦ ,wihi unaemost are turn in which ), )aon nacia where Antarctica, around C) www.pnas.org/lookup/suppl/doi:10. 2 | . o.114 vol. | o 1 no. ◦ nthe in C | 45–50

EARTH, ATMOSPHERIC, AND PLANETARY SCIENCES proxy data for the LGM, the prescribed atmospheric cooling is polar amplified, ranging from 2◦C in the tropics to 6◦C around Antarctica (14, 15). All other boundary conditions are held fixed. The reduction in atmospheric temperature leads to a reduction in ocean surface temperature and an expansion of sea ice around Antarctica, as well as the appearance of sea ice in the North Atlantic (Fig. 1B). Moreover, the model suggests a cooling and salinification of AABW, leading to a strong salt stratification in the deep ocean, which replaces the reversed salinity stratification observed with present-day forcing (Fig. 2B). The strong salt stratification is consistent with pore-fluid data for the LGM (12). (To compare the absolute salinity values here to those inferred for the LGM, one needs to account for the increased bulk ocean salinity resulting from the eustatic drop in sea level, which is not included in the simulations and would add around 1 g/kg globally.) The AMOC cell weakens and shoals (Fig. 3B), again consistent with inferences for the LGM (3–5). The abyssal cell slightly strengthens and becomes somewhat more confined to the abyss, leading to an increased separation between the two overturning cells, which may have played an important role in the increased ocean carbon storage (6). Fig. 1. (A and B) Sea surface temperature (colors) and sea-ice concentration In the real ocean the abyssal overturning cell is distributed (white contours), for simulations with boundary conditions representing over multiple basins and currently overlaps in depth and density present-day conditions (A) and LGM conditions with reduced atmospheric with the AMOC cell, leading to a continuous exchange of water temperature (B). Contour interval for sea-ice concentration is 20%. between the two cells in the Southern Ocean (1, 2). Whereas this interbasin overlap between the present-day overturning cells cannot be modeled in a single-basin model, the simulated con- (SV) and −5.2 SV are roughly consistent with the observed rate traction of both the AMOC and abyssal cells is likely to be of NADW formation and the transport of AABW into the North robust (11). In a multibasin configuration, this contraction of Atlantic (1, 2). both cells is expected to remove or reduce the overlap between We now analyze the response of the model’s equilibrium solu- the two cells, thus generating a more isolated abyssal water mass. tion to a reduction in atmospheric temperature. Consistent with The rearrangement of the overturning circulation would likely

Fig. 2. (A–D) Zonal mean temperature (A and B) and salinity (C and D) for simulations with boundary conditions representing present-day forcing (A and C) and simulations with reduced atmospheric temperature resembling LGM conditions (B and D). Contour intervals are 1◦C in A and B and 0.05 g/kg in C and D. Note that the colorbar ranges have been cropped to focus on the deep ocean.

46 | www.pnas.org/cgi/doi/10.1073/pnas.1610438113 Jansen Downloaded by guest on September 23, 2021 Downloaded by guest on September 23, 2021 Jansen cir- ocean around loss deep buoyancy in surface changes Because stratification. observed and the culation to rise gives simulation to increases rate loss 2. buoyancy surface integrated buoyancy the simulation simulation surface present-day–like integrated the 4.4 about for an of Antarctica yields around which rate com- 3), loss densities by potential Fig. depth densities the in with at potential (consistent shown estimate parcel in depth km changes a can 2 on on We to referenced based fluxes ocean. fluxes salt deep buoyancy and the puting heat into of amplified sinks effect is parcel the effect water temperature a den- the surface as temperatures, on cold effect by small at counteracted a sities and has cooling cooling of a Whereas transformation by freshening. a dominated the is with noting AABW associated by to CDW increase resolved density be the can that the to contradiction and (CDW) apparent cell water The abyssal deep an AABW. circumpolar of presence upwelling the of with transformation odds at be would to loss in buoyancy appear found of lack is This Antarctica simulation. around coef- present-day–like are the loss expansion buoyancy fluxes thermal no and buoyancy virtually haline ther- ficients, surface surface the the If of using (16). dependence computed equation pressure coefficient the the expansion of particular nonlinearity mal in the and consider state to of important is it tica rywnsaesrne.A euto h agriela and load ice larger Antarctica around the from 3 of about rate from export result increases ice a peak west- the As the velocity, where stronger. export northward are farther winds extends erly ice sea goes as maximum) increases, its (near kg/m kg/m which 800 300 load, about played ice from role the up dominant in the differences with atmo- area) by the reduced, unit as is per increase temperature snow Both spheric and velocity. sea-ice transport the equatorward of as and defined mass (here zonal-mean load ice and the time- of product export rejec- the Sea-ice to brine export. proportional and enhanced is formation from sea-ice primarily with associated results tion Antarc- turn around in rate loss which buoyancy tica, increased an from result here mecha- different. the somewhat although is (9), here al. proposed et nism Ferrari in the sketched cancels that largely resemble meanders standing of contribution The 3. Fig. region.] includes in transport channel included overturning the not isopycnal in is the transport that which [Note zonal-mean (11), velocity. meanders diapycnal bolus eddy-induced standing apparent parameterized with computed the is associated and streamfunction component velocity overturning additional mean The zonal an B). ( Eulerian temperature the atmospheric of reduced sum with the conditions from LGM and (A) forcing present-day representing conditions 3. Fig. 1 h agrboac osrt rudAtrtc nteLGM the in Antarctica around rate loss buoyancy larger The Antarc- around loss buoyancy net effective the compute To h hne ntede ca iclto n stratification and circulation ocean deep the in changes The × 10 (A 4 2 and m nte“G”smlto.Teieepr eoiyalso velocity export ice The simulation. “LGM” the in 4 ×10 eiinloetrigsrafnto clr)adptnildniyrfrne o2k et baklns,frsmltoswt boundary with simulations for lines), (black depth km 2 to referenced density potential and (colors) streamfunction overturning Meridional B) · s −3 3 . m 4 · s ×10 −3 2 .I h LGM the In Methods). and (Materials nte“rsn”smlto oabout to simulation “present” the in 7 gst bu 14 about to kg/s ×10 7 kg/s. n G iuain Tbe1 atcolumn). last present 1, the (Table between simulations doubles LGM almost and m) the (300 below thermocline mixing to upper input energy implied stratified the unchanged more diffusivities, assumed vertical which the simulations, mix our In kinetic to (22). column required turbulent water be more LGM would as the dissipation mixing, during energy vertical stratification suppressed ocean have other deep may the increased On 21). the (20, mixing hand, vertical enhanced turn in caused which have ocean, may deep the in dissipation energy tidal increased dis- simulations LGM amplifies and rate present above. loss cussed the buoyancy between increased differences The the and 1). export Table “LGM ice (experiment in Antarctica windS” in around increase loss Hemi- moderate buoyancy Southern associated a the a shows of incorporating westerlies strengthening simulation sphere 20% A and over 19). shift westerlies (18, southward surface Ocean the Southern slight of a the strengthening indicate and 1). instead shift Table models poleward in climate windN” and “LGM observations most (experiment Proxy solution at the of on shift impact equatorward an for about the Observational allows in LGM. however, differences and sur- evidence, present for Hemisphere the mechanism Southern between potential circulation the ocean a of as latitude westerlies the face in shift torward following. the in discussed briefly in are summarized and are 1 simulations Table these of temper- turbulent results atmospheric The of vertical change. structure the ature spatial the and as sensitivity well stress of as diffusivity, wind number the a modifications consider varying additional we experiments, to conditions, results boundary the the of in robustness the test To Experiments Sensitivity (11). Nadeau and Jansen The of shift results upward state). the an with to of consistent leads NADW, equation then of LGM the somewhat the is in in stratification stratification nonlinearity increased and the loss by buoyancy complicated in compar- changes quantitative of a (although ison simulations the approxi- between LGM stratification relationship and ocean depend This present deep (11). the to in rate increase loss expected the buoyancy explains basin, the is the on stratification in linearly mately diffusion ocean vertical deep by balanced the be to has Antarctica h oso hlo hl esdrn h G a ieyldto led likely has LGM the during seas shelf shallow of loss The equa- an proposed (17) al. et Toggweiler paper, seminal a In 3 ◦ 1,1) hc ntr shr on ohv negligible have to found here is turn in which 19), (18, PNAS | aur ,2017 3, January | o.114 vol. | o 1 no. | 3 47 ◦

EARTH, ATMOSPHERIC, AND PLANETARY SCIENCES Table 1. Summary of results from sensitivity experiments R 2 ˙ Experiment BdA Ndeep ΨNADW ΨAABW zNADW zΨ=0 zAABWreturn emix Present 0.4 × 104 0.6 × 10−4 17.5 5.2 1,450 2,050 2,370 0.53 × 10−6 LGM 2.1 × 104 2.3 × 10−4 9.6 6.2 1,070 1,590 2,480 1.02 × 10−6 LGM windS 3.4 × 104 3.8 × 10−4 9.5 7.2 1,000 1,500 2,530 1.42 × 10−6 LGM windN 1.7 × 104 2.0 × 10−4 9.6 6.1 1,040 1,550 2,430 0.95 × 10−6 LGM κ F02D 50% 2.3 × 104 3.9 × 10−4 6.4 4.8 810 1,250 2,490 0.73 × 10−6 LGM κ + 50% 2.2 × 104 1.8 × 10−4 12.2 8.6 1,210 1,710 2,250 1.35 × 10−6 LGM dTSH 1.6 × 104 2.1 × 10−4 10.9 6.2 1,120 1,660 2,520 0.98 × 10−6 LGM dTconst 1.8 × 104 2.2 × 10−4 8.7 6.3 1,080 1,620 2,500 0.96 × 10−6 Present seas 0.3 × 104 0.4 × 10−4 15.2 5.0 1,550 2,220 2,530 0.46 × 10−6 LGM seas 1.6 × 104 1.8 × 10−4 9.9 6.1 1,190 1,750 2,410 0.85 × 10−6 Warm 0 0.1 × 10−4 19.9 — 2270 — — 0.35 × 10−6

Each row indicates a different numerical simulation: Present denotes the simulation with boundary conditions resembling present-day conditions. LGM denotes the simulation with reduced atmospheric temperature, resembling LGM conditions. LGM windS denotes a simulation with atmospheric temperatures as in LGM, but also including a 3◦ southward shift and 20% strengthening of the Southern Hemisphere (SH) westerlies. In LGM windN SH westerlies instead are shifted 3◦ northward (Fig. S3B). LGM κ-50% and LGM κ + 50% denote simulations with boundary conditions analogous to LGM but with diapycnal diffusivities reduced or enhanced by 50%, respectively (Fig. S5). LGM dTSH denotes a simulation where atmospheric temperatures have been reduced only in the Southern Hemisphere, whereas LGM dTconst denotes a simulation where temperatures have been reduced homogeneously over the whole domain by 5◦C relative to Present. Present seas and LGM seas denote simulations analogous to Present and LGM, but with seasonally varying air temperature forcing (Fig. S1A). Warm denotes a global warming simulation, with similar but opposite signed changes in the atmospheric temperature as in the LGM simulation (Fig. S3A). Columns show various diagnostics: R BdA (in m4s−3) is the integrated buoyancy loss rate around Antarctica (Materials and 2 ◦ −2 Methods). Ndeep is the mean stratiﬁcation (referenced to 2 km depth) averaged over the deep ocean basin below 1,500 m and north of 40 S (in s ). ΨNADW indicates the maximum AMOC overturning transport and ΨAABW is the maximum overturning transport in the abyssal cell (both in Sv). zNADW denotes the median depth of NADW, computed as the depth (in meters) at which the basin-averaged overturning streamfunction is reduced to 50% of its maximum value. zΨ=0 is the depth at which the basin-averaged overturning streamfunction changes sign; and zAABWreturn is the depth where the streamfunction reaches 50% of its minimum value, thus denoting the median depth of the southward return ﬂow of AABW. e˙ mix is the globally averaged energy input per unit area by vertical mixing below 300 m (in units m3·s−3).

To test the sensitivity of our results to the rate of vertical conditions, mixed layer dynamics, and sea-ice thermodynamics mixing, we consider two additional LGM simulations in which make this model arguably less suited to represent the seasonal the vertical diffusivity is reduced or increased by 50% (“LGM κ cycle, the simulations do exhibit strong seasonality in sea-ice −50%” and “LGM κ + 50%”, respectively). Due to the stronger cover around Antarctica (Fig. S1B). Whereas the addition of a stratification, the implied energy input to mixing below 300 m seasonal cycle causes some minor changes in the deep ocean is still slightly larger than for present-day conditions in LGM circulation and stratification in both the present and LGM sim- κ −50%, and the energy input is almost tripled in LGM κ + ulations, the main results regarding differences in the stratifi- 50%. Reduced vertical mixing weakens and shoals the AMOC cation and circulation between the present and LGM remain cell during the LGM and increases the abyssal stratification— unchanged by the inclusion of a seasonal cycle (Fig. S1C). thus amplifying the effects of atmospheric cooling. Enhanced To address the potential implications of our results for long- vertical mixing instead strengthens and deepens the AMOC cell term future climate change, we finally examine a “global warm- and reduces the abyssal stratification—thus counteracting the ing” simulation (experiment “Warm” in Table 1), which shows effects of atmospheric cooling. However, even the simulation essentially reversed results from the cooling experiments: Above- with enhanced mixing maintains a stronger stratification and freezing temperatures around Antarctica lead to ice-free condi- weaker and shallower AMOC cell compared with the simulation tions and net buoyancy gain. As a result AABW formation is shut representing present-day conditions, indicating that the effect of down, leaving the entire deep ocean filled with nearly unstrati- atmospheric cooling remains dominant. fied water of North Atlantic origin (Fig. S2). The effect of these Significant uncertainty also exists in the spatial pattern of potential rearrangements on ocean carbon storage represents an atmospheric temperature change between the present and LGM important topic for future research. (14, 15). We consider two sensitivity experiments, which represent extreme cases: one where atmospheric cooling is restricted Discussion and Conclusions to the Southern Hemisphere (experiment “LGM dTSH” in The results of this study suggest that atmospheric cooling Table 1) and one where a globally constant cooling is applied alone—via a modification of the buoyancy loss rate around (experiment “LGM dTconst”). Both experiments show roughly Antarctica—leads to a strong increase in the deep ocean strat- similar results as the LGM reference case with symmetric polar ification, a shoaling of NADW, and an increased separation amplified cooling, as long as the reduction in atmospheric tem- between NADW and the underlying abyssal overturning cell. perature at high southern latitudes remains about the same. This The simulated response of the deep ocean circulation and strat- result confirms our interpretation that circulation and stratifica- ification to atmospheric temperature change is consistent with tion changes are controlled primarily by differences in the sur- differences between the present and LGM inferred from paleo- face boundary conditions around Antarctica. proxy observations (3, 6, 12, 13). A series of sensitivity experi- The simulations here are highly idealized and, among other ments suggests that the dominant control over circulation and things, do not include a seasonal cycle, which may affect even stratification changes is exerted by the atmospheric temperature the mean sea-ice growth and export rate around Antarctica. To around Antarctica. Temperature changes in other regions, as analyze the effect of seasonality, the present and LGM simula- well as differences in the atmospheric wind stress, instead play tions were repeated using seasonally varying air temperatures, only a relatively minor role. but leaving the annual mean temperatures unchanged (experi- The results here are broadly consistent with some complex ments “Present seas” and “LGM seas” in Table 1 and Fig. S1). coupled climate model simulations (7, 8), as well as with global Even though the idealized representations of surface boundary ocean-only simulations under LGM boundary conditions (10).

48 | www.pnas.org/cgi/doi/10.1073/pnas.1610438113 Jansen Downloaded by guest on September 23, 2021 Downloaded by guest on September 23, 2021 c)(3,i yrsai osieqcngrto.Teielzddomain (MIT- idealized The model configuration. 70 Boussinesq circulation from hydrostatic extends general a in (MIT) (33), Technology gcm) of Institute sachusetts Configuration. Model Numerical Methods understanding and our Materials to central changes. climatic be future will and loop past of cru- feedback this a of CO play sion atmospheric to temperature, global and likely between circulation, loop are feedback ocean positive they a here, in role argued cial tem- as by driven swings, directly perature themselves are CO changes circulation carbon atmospheric ocean ocean thus modulating in and and role circulation storage ocean key the a in played changes have glacial– that are stratification likely mechanisms of is exact it understanding debated, the still Although our stratification swings. and for climate circulation interglacial implications ocean important the in has differences of driver is but them, study. this between more of scope differences A the understand 30). is beyond simulations (26, better climate sheets to LGM ice needed comprehensive the boundary of in the analysis changes detailed in and/or as differences (29), such other formation conditions, to insufficient and due cover 28), effects sea-ice (27, compensating Antarctic solutions in equilibrated between increase fully differences include and Pos- disagreement 26). transient this (25, stratifica- for simulations and reasons LGM widely circulation sible and ocean and preindustrial deep between different the tion show in other however, changes Many diverging models, (10). climate conditions boundary coupled LGM under ocean (CESM)1.1.2 simulation a Model in System stratification Earth ocean deep Community high decoupled the explain domi- to have argued the Antarctica been around be also fluxes is to freshwater likely in export and Changes driver. sufficient here sea-ice nant we is 1) enhanced alone p. cooling that 7, that (ref. suggest argue westerlies,” increased (7) by “ultimately al. caused et Antarctica CO Shin around reduced whereas export ultimately have sea-ice changes and enhanced these (7) to here, attributed discussed strongly results been 24). a the 23, with show (7, NADW Consistent CCSM3 shallower Cli- and and stratification Community abyssal (CCSM)1.4 increased the Model using System simulations mate climate LGM Coupled Jansen The levels. 29 of total a with ocean, 69 deep 1 between a the by in bounded m is 200 ocean to surface the near 1 is resolution horizontal nadto otesnil etflx nielzdrdaiersoigi pre- is restoring radiative 35). idealized (34, an flux, formulas as heat bulk scribed sensible standard atmo- the using the to described addition between con- are In exchange are ocean heat fields the sensible and forcing and sphere atmospheric Momentum all time. temperatures, in stant air in cycle ana- seasonal idealized in of form shown the functions in lytical prescribed are winds surface and evaporation), Southern the representing channel, reentrant a to Ocean. rise give conditions ary ture, etrn fteieoensraetmeaue(6 ( (36) temperature surface linear ice/ocean simple a the with of performed restoring were simulations additional conditions, global ary the close to adjustments latent flux and artificial moisture budget. salt for require formula would bulk transfer layer- a heat boundary of and use about radiation Moreover, assumptions humidity. long-wave additional specific and short- avoid downward sta- to layer in chosen boundary changes and was speed idealization wind by This modified bility. is that rate temperature—albeit restoring boundary atmospheric a thermal heat prescribed with the the latent to simplifies restoring by effect a in primarily to which conditions balanced here, is ignored are which Both heating, loss. radiative net a riences W·m h oiatrl famshrctmeauecag sthe as change temperature atmospheric of role dominant The topei eprtrs e rswtrflx(.. precipitation– (i.e., flux freshwater net temperatures, Atmospheric ots h estvt fterslst h eal ftetemlbound- thermal the of details the to results the of sensitivity the test To −2 T a ·K stepecie topei eprtr,and temperature, atmospheric prescribed the is −4 F ◦ rad n 48 and S steSea–otmn osat nraiyteoenexpe- ocean the reality In constant. Stefan–Boltzmann the is = ◦ o65 to S σ (T ◦ s 4 ,wee bv mdph oal eidcbound- periodic zonally depth, km 3 above where, S, ◦ − ◦ ◦ ,cvr 72 covers N, ti fln nalsds hc sitrutdonly interrupted is which sides, all on land of strip × xeti h estvt iuain with simulations sensitivity the in Except S3. Fig. T a 4 1 ,where ), ◦ n h etclrslto agsfo 0m 20 from ranges resolution vertical the and h ueia iuain s h Mas- the use simulations numerical The ◦ 2 T nlniue n s4k ep The deep. km 4 is and longitude, in s ocnrtos() However, (8). concentrations 2 steoeniesraetempera- surface ocean/ice the is ocnrtos(1 2.If 32). (31, concentrations 2 etrcomprehen- better A . i.S4 Fig. σ = .Tesurface The ). 5.67 × 10 −8 etflxi optdas computed is flux heat ihavral dydfuiiyfruae olwn ibc ta.(43), al. et Visbeck following formulated parameterized parameterizations, diffusivity are (42) eddy with Redi fluxes and variable eddy (41) zero- a (GM) Mesoscale a with McWilliams cover. and use conduc- snow Gent heat and the and different using (40) ice the al. (39). of for al. et et account tivities Zhang Losch but and on approximation, (38) based Hibler heat–capacity and are Zhang introduced thermodynamics by rheology modified Sea-ice viscous-plastic and the (37) on Hibler based by is which package, sea-ice condi- boundary thermal main the the of that details ). S4 suggest the (Fig. results to tions in The sensitive that not article. to are this identical conclusions in is discussed however, simulations, simulations LGM the and present difference the temperature air between The unchanged). approximately temperature 2.5 1 air by by latitudes reduced, highest equator the slightly the at at temperature was temperature gradient by air temperature the implied present-day–like decreasing air realistic average meridional a on obtain the still is simulation To what conditions. boundary than original restoring the effective faster somewhat e.Te“oa uynyls aeaon nacia scmue by computed 55 of is south gridpoint Antarctica” every include around to defined rate loss buoyancy integrating “total chan- circumpolar The the in nel. northward farther negative but Antarctica, around ambient an dbar. thermal for 2,000 the computed of depth, are pressure at coefficients parcel a contraction of haline density and the expansion on fluxes surface of effect the in and salinity, acceleration, melt), itational and formation sea-ice coefficient, of expansion effects the include (which where of velocity piston a with drag, bottom linear a mm·s by 1 represented bound- bottom is the in layer momentum ary of Dissipation unstable. statically is diffusivity cation convective a background with a ment diffusiv- to in reduces vertical shown but a is abyss, by the 2×10 represented in of is enhanced value following strongly mixing slopes, is Diapycnal isopycnal which (44). steep ity, of al. presence et the Gerdes in tapered is eterization between ue ae ntentha n rswtrflxsas fluxes freshwater and heat net the on based puted Rates. Loss Buoyancy of Computation entire the for acceleration sensitiv- tracer all use circulation, study negligible and integration. this the state in to mean discussed Due the experiments found. on ity was acceleration ice) tracer sea of of strat- effect seasonality and thermal circulation the the ocean on deep in (or the cycle ification on seasonal effect significant a no includes Again, which in forcing. explicitly simulation, virtually tested seas are accel- also present paper was tracer the acceleration this the tracer in to by discussed sensitivity affected circulation The be unaffected. and state to mean appear prop- the variability Whereas eration, acceleration. internal and tracer the acceleration without of tracer y erties 1,000 with another obtained for from initialized solution integrated impact was the equilibrium simulation test LGM statistical To the equilibration. stepping step, the the time time up tracer tracer accelerated speed accelerated the to of an used been of has use steady-state which the a (47), of question lack the into Nevertheless, puts simula- 1–3. Figs. LGM solution in and out be present averaged may the com- is but and small between study, tions is obtained this with variability differences of the the associated of scope with magnitude be the pared The beyond to work. is follow-up appears Atlantic, in which North addressed the variability, in this ice of sea nature AMOC. the exact in simula- variability the The centennial to cold Whereas decadal the simulation. some solution, exhibits equilibrium each simulation steady of LGM a y reaches forcing 500 present-day last with the tion over taken are ages viscosi- biharmonic and respectively. Laplacian coefficients using viscosity suppressed with ties, is noise grid-scale and e-c yaisadtemdnmc r ecie sn h MITgcm the using described are thermodynamics and dynamics Sea-ice B l iuain r nertdt qiiru o tlat400y n aver- and y, 4,000 least at for equilibrium to integrated are simulations All α sgnrlypstv dntn ca uynyls)wti strip a within loss) buoyancy ocean (denoting positive generally is Q = −1 stentha u and flux heat net the is 1admxn length mixing and 0.01 K osi oiotlbudr odtosaeapida h walls, the at applied are conditions boundary horizontal No-slip . min B ovcini aaeeie sn ifsv adjust- diffusive a using parameterized is Convection S5. Fig. C p = −5 vrteetr eino uynyls rudAntarctica, around loss buoyancy of region entire the over steha aaiyo ewtr eas eaeinterested are we Because seawater. of capacity heat the is 0 m 200 m 2 ·s B −1 β 2 ρ ·s 0 = stehln otato coefficient, contraction haline the is PNAS −1 ntetemcie(5 6.Tedfuiiyprofile diffusivity The 46). (45, thermocline the in = gαQρ F and ,3 kg·m 1,035 = ν 2 | 5W/(m 25 = F 0 −1 K l aur ,2017 3, January = max κ 2 h e rswtrflxoto h ocean the of out flux freshwater net the C conv × 6 m h dydfuiiyi capped is diffusivity eddy The km. 160 p −1 uynyls rmteoeni com- is ocean the from loss Buoyancy = 10 = + ,0 m 2,000 4 −3 2 K)(T m 0m 10 ◦ gβ wihkestegoa mean global the keeps (which C 2 sarfrnedensity, reference a is ·s F s ◦ −1 2 S where S − ·s 2 ρ ·s 0 −1 −1 and | T −1 ◦ a , o.114 vol. hl nraigthe increasing while C .Ti mut oa to amounts This ). hnvrtestratifi- the whenever n h Mparam- GM the and , ν 4 B = > α 2 × bcuethe (because 0 stethermal the is | g 10 stegrav- the is o 1 no. 13 m S 4 sthe is ·s | −1 [1] 49 ,

EARTH, ATMOSPHERIC, AND PLANETARY SCIENCES areas of buoyancy loss and gain are well separated, the results are not sen- uration ﬁles and model output data are available from the author sitive to the exact choice of this latitude). upon request. Funding for this work was provided by the National Sci- ence Foundation under Award 1536454, and computational resources ACKNOWLEDGMENTS. I thank Alice Marzocchi, K. Thomas, and two anony- were provided by the Research Computing Center at the University of mous reviewers for their very valuable comments. The MITgcm conﬁg- Chicago.

1. Lumpkin R, Speer K (2007) Global ocean meridional overturning. J Phys Oceanogr 25. Weber SL, et al. (2007) The modern and glacial overturning circulation in the Atlantic 37:2550–2562. ocean in PMIP coupled model simulations. Clim Past 3(1):51–64. 2. Talley LD (2013) Closure of the global overturning circulation through the Indian, 26. Muglia J, Schmittner A (2015) Glacial Atlantic overturning increased by wind stress in Pacific and Southern Oceans: Schematics and transports. Oceanography 26:80–97. climate models. Geophys Res Lett 42(22):9862–9868. 3. Curry WB, Oppo DW (2005) Glacial water mass geometry and the distribution of δ13C 27. Stouffer R, Manabe S (2003) Equilibrium response of thermohaline circulation to large P of CO2 in the western Atlantic Ocean. Paleoceanography 20(1):PA1017. changes in atmospheric CO2 concentration. Clim Dynam 20(7-8):759–773. 4. Lynch-Stieglitz J, et al. (2007) Atlantic meridional overturning circulation during the 28. Zhang X, Lohmann G, Knorr G, Xu X (2013) Different ocean states and transient char- last glacial maximum. Science 316:66–69. acteristics in Last Glacial Maximum simulations and implications for deglaciation. Clim 5. Burke A, et al. (2015) The glacial mid-depth radiocarbon bulge and its implications Past 9:2319–2333. for the overturning circulation. Paleoceanography 30(7):1021–1039. 29. Roche DM, Crosta X, Renssen H (2012) Evaluating Southern Ocean response for the 6. Lund DC, Adkins JF, Ferrari R (2011) Abyssal Atlantic circulation during the Last Last Glacial Maximum and pre-industrial climates: PMIP-2 models and data evidence. Glacial Maximum: Constraining the ratio between transport and vertical mixing. Pale- Quat Sci Rev 56:99–106. oceanography 26:PA1213. 30. Klockmann M, Mikolajewicz U, Marotzke J (2016) The effect of greenhouse gas con- 7. Shin SI, Liu Z, Otto-Bliesner BL, Kutzbach JE, Vavrus SJ (2003) Southern Ocean sea- centrations and ice sheets on the glacial AMOC in a coupled climate model. Clim Past, ice control of the glacial North Atlantic thermohaline circulation. Geophys Res Lett 1829–1846. 30(2):1096. 31. Brovkin V, Ganopolski A, Archer D, Rahmstorf S (2007) Lowering of glacial atmo- 8. Liu Z, Shin SI, Webb RS, Lewis W, Otto-Bliesner BL (2005) Atmospheric CO2 forcing on spheric CO2 in response to changes in oceanic circulation and marine biogeochem- glacial thermohaline circulation and climate. Geophys Res Lett 32(2). istry. Paleoceanography 22:PA4202. 9. Ferrari R, et al. (2014) Antarctic sea ice control on ocean circulation in present and 32. Watson AJ, Vallis GK, Nikurashin M (2015) Southern Ocean buoyancy forcing of ocean glacial climates. Proc Natl Acad Sci USA 111(24):8753–8758. ventilation and glacial atmospheric CO2. Nat Geosci 8(11):861–864. 10. Sun S, Eisenman I, Stewart AL (2016) The influence of southern ocean surface buoy- 33. Marshall J, Hill C, Perelman L, Adcroft A (1997) Hydrostatic, quasi-hydrostatic, and ancy forcing on glacial-interglacial changes in the global deep ocean stratification. nonhydrostatic ocean modeling. J Geophys Res 102:5753–5766. Geophys Res Lett 43(15):8124–8132. 34. Large WG, Pond S (1981) Open ocean momentum flux measurements in moderate to 11. Jansen M, Nadeau LP (2016) The effect of Southern Ocean surface buoyancy loss on strong winds. J Phys Oceanogr 11(3):324–336. the deep ocean circulation and stratification. J Phys Oceanogr 46(11):3455–3470. 35. Large WG, Pond S (1982) Sensible and latent heat flux measurements over the ocean. 12. Adkins JF, McIntyre K, Schrag DP (2002) The salinity, temperature and δ18O content J Phys Oceanogr 12(5):464–482. of the glacial deep ocean. Science 298:1769–1773. 36. Haney RL (1971) Surface thermal boundary condition for ocean circulation models. 13. Roberts J, et al. (2016) Evolution of South Atlantic density and chemical stratification J Phys Oceanogr 1:241–248. across the last deglaciation. Proc Natl Acad Sci USA 113(3):514–519. 37. Hibler IIIWD (1979) A dynamic thermodynamic sea ice model. J Phys Oceanogr 9(4): 14. Schmittner A, et al. (2011) Climate sensitivity estimated from temperature reconstruc- 815–846. tions of the Last Glacial Maximum. Science 334(6061):1385–1388. 38. Zhang J, Hibler WD (1997) On an efficient numerical method for modeling sea ice 15. Annan JD, Hargreaves JC (2013) A new global reconstruction of temperature changes dynamics. J Geophys Res Oceans 102(C4):8691–8702. at the Last Glacial Maximum. Clim Past 9(1):367–376. 39. Losch M, Menemenlis D, Campin JM, Heimbach P, Hill C (2010) On the formulation of 16. Killworth PD (1977) Mixing of the Weddell Sea continental slope. Deep Sea Res sea-ice models. Part 1: Effects of different solver implementations and parameteriza- 24(5):427–448. tions. Ocean Model 33(1):129–144. 17. Toggweiler J, Russell JL, Carson S (2006) Midlatitude westerlies, atmospheric CO2, and 40. Zhang J, Hibler IIIWD, Steele M, Rothrock DA (1998) Arctic ice-ocean modeling with climate change during the ice ages. Paleoceanography 21(2):PA2005. and without climate restoring. J Phys Oceanogr 28(2):191–217. 18. Kohfeld KE, et al. (2013) Southern Hemisphere westerly wind changes during the Last 41. Gent PR, McWilliams JC (1990) Isopycnal mixing in ocean circulation models. J Phys Glacial Maximum: Paleo-data synthesis. Quat Sci Rev 68:76–95. Oceanogr 20:150–155. 19. Sime LC, et al. (2013) Southern Hemisphere westerly wind changes during the Last 42. Redi MH (1982) Oceanic isopycnal mixing by coordinate rotation. J Phys Oceanogr Glacial Maximum: Model-data comparison. Quat Sci Rev 64:104–120. 12:1154–1158. 20. Egbert GD, Ray RD, Bills BG (2004) Numerical modeling of the global semidiurnal tide 43. Visbeck M, Marshall J, Haine T, Spall M (1997) Specification of eddy transfer coeffi- in the present day and in the last glacial maximum. J Geophys Res Oceans 109:C03003. cients in coarse-resolution ocean circulation models. J Phys Oceanogr 27:381–402. 21. Schmittner A, Green JAM, Wilmes SB (2015) Glacial ocean overturning inten- 44. Gerdes R, Koberle¨ C, Willebrand J (1991) The influence of numerical advection sified by tidal mixing in a global circulation model. Geophys Res Lett 42(10): schemes on the results of ocean general circulation models. Clim Dynam 5(4): 4014–4022. 211–226. 22. Wunsch C, Ferrari R (2004) Vertical mixing, energy, and the general circulation of the 45. Nikurashin M, Ferrari R (2013) Overturning circulation driven by breaking internal oceans. Annu Rev Fluid Mech 36:281–314. waves in the deep ocean. Geophys Res Lett 40:3133–3137. 23. Otto-Bliesner BL, et al. (2006) Last glacial maximum and Holocene climate in CCSM3. 46. Waterhouse AF, et al. (2014) Global patterns of diapycnal mixing from measurements J Clim 19(11):2526–2544. of the turbulent dissipation rate. J Phys Oceanogr 44(7):1854–1872. 24. Brandefelt J, Otto-Bliesner BL (2009) Equilibration and variability in a Last Glacial 47. Bryan K (1984) Accelerating the convergence to equilibrium of ocean-climate models. Maximum climate simulation with CCSM3. Geophys Res Lett 36:L19712. J Phys Oceanogr 14(4):666–673.

50 | www.pnas.org/cgi/doi/10.1073/pnas.1610438113 Jansen Downloaded by guest on September 23, 2021