~

INFLUENCE OF THE CLIMAP ICE SHEET ON THE CLIMATE OF A GENERALCIRCULATION MODEL; IMPLICATIOtilS FOR THE MILANKOVITCH THEORY

S. Manabe. A.J. Broccoli

Geophysical Fluid Dynamics Laboratory/NOAA, Prince-ton University, P.o. Box 308, Princeton, New Jersey08540, USA

INTRODUCTION

While the spectral analys:is of data from deep sea coressupports the suggestion that the temporal variation of theearth's orbital parameters is responsib,le for triggering thegrowth and decay of continental ice duri.ng the Quaternary (1),the physical factors responsible for maintaining a global iceage climate are less well understood. One of the important fac-tors in maintaining a lowered ;itmospheric: temperature during anice age is the presence of lllrge continental ice sheets thatreflect a large fraction of incoming solar radiation. This stu-dy investigates the influence of the 18 kyr BP (18 000 yearsbefore present) ice sheet on the earth's climate by use of amathematical model of climate constructed at the GeophysicalFluid Dynamics Laboratory of NOAA.

As the box diagram of Figure 1 indic:ates, the mathematicalmodel used for this study consists of three basic units: 1) ageneral circulation model of the atmosphere, 2) a heat and waterbalance model over the continel1lts, and 3:1 a simple model of theoceanic mixed layer. A brief description of these three unitsfollows. A more detailed description 01: this atmosphere-mixedlayer ocean model can be found in Hanabe and Stouffer (2).

The atmospheric general circulation model computes therates of change with time of the '..ertical component ofvorticity, horizontal divergence, temperature, moisture, andsurface pressure using the so-called "spectral method". Thedynamical component of this model is developed by Gordon andStern (3). Hanabe et al. (4) and Hanabl~ and Hahn (5) discuss

789

A. L. Berger et QL (ed~), Milankovitch Qnd Climate, ,Oart 2, 789- 799.e 1984 by D. Reidel Publishing Company.

790

(.) I

;=

I~

~<

:( 2.

I Z

Z

c5 10

0>- t=

: I

i=<:(

0 ~

- 1-

~

-) 10

.:=

C>

, <

:(C

t: Ll.J

I C

t:w:x:

I~

I-

L C

k:

~ (~

~--~

~

:;) ~

->

()Z

--,

-1-- loA

.

I-- c~

lIJ

Z

-:) I--

0 C

y ~U

LLJ ~

lJ-0

-~~-

z ()

Q

-::f-

C)

~

~;

::> ~

-01w

S. M

AN

AB

E A

ND

A. J. B

RO

CC

OLI

wUZ~~co

.-~wIwuz~-J~ccI-~w:I:

>-

~0-J0~0>-

:I: wU~w(/)

Qj

~='

CtI

...~ .414.J

~...u041'0~~u...4.J

;41.c4.J

~41.c4.J

041~~4.JU~~4.J~U...~~.041

.c...00c:......~~...~~...B~~00~...'0I<0IQ

INFLUENCE OF mE CLIMAP ICE SHEET791

the structure and performance of this atmospheric model indetail.

Over the continents, the assumption of zero surface heatstorage is used to determine surface temperatures from energyfluxes at the surface. Snow is allowed to accumulate on thesurface, with the change in snow depth predicted as the net con-tribution from snowfall, silJblimation, and snowmelt. A highersurface albedo is used whe:11 snow is present. Also used is awater balance model which computes ch~lQges in soil moisture fromthe rates of rainfall, evaporation, snowm,elt and runoff.Further detailr of the hydrologic coDlputations can be found inManabe (6).

The oceanic mixed laye1: model cot1lsists of a vertically iso-thermal layer of static w,iter of ut1liform depth. This modelincludes the effects of ocE,anic evaporation and the large heatcapacity of the oceanic mixE!d layer. b'ut neglects the effects ofhorizontal heat transport by ocean currents and of the heatexchange between the mixed layer and the deeper parts of theocean.

Sea ice is predicted when the mixed layer temperaturefalls below the freezing poj~nt of sea water (-2°C). and a highersurface albedo is used WherE~ sea ice is present.

EXPERIMENTAL DESIGN

In order to investigal:e the influence of continental icesheets on the climate of an ice age, two long term integrationsof the atmosphere-mixed layer ocean model described in the pre-ceding section are conductecl. The first time-integration, here-after identified as the standard expe1riment, assu~s as a boun-dary condition the modern distribution of continental ice. Thesecond time-integration assu~s the d:Lstribution of continentalice at the time of the last glacial maximum as reconstructed byCLIMAP (7), and will hereaJ:ter be identified as the ice sheetexperiment.

A sea level difference between the two experiments of 150 mis prescribed, consistent with the glacial lowering of sea levelas estimated by CLIMAP. Or't>ital paralneters in both experimentsare set at modern values, making the distribution of insolationat the top of the atmosphere with lat:itude and season the samein both experiments. This simplification is reasonable, sincethe orbital parameters at: l,~ kyr BP a:re not very different fromthe modern values. The surface albedo distribution of snow- andice-free areas is prescribed to be the same in both experiments.

The initial condition for both time integrations is a dry,isothermal atmosphere at rE:st coupled with an isothermal mixed

S. MANA BE AND A. J. BROCCOLI792

layer ocean. In both cases, the model is time integrated for 20seasonal cycles. A quiisi-equilibrium model climate is achievedafter 15 model yea1:s, and the :;ubsequent five-year period isuse.d for ana lys is in e~ch experiment.

In Figure 2, the g;eographical distribution of February sur-face air temperature olf the model atmosphere is compared withthe surface distribution compile,d by Crutcher and Meserve (8)and Taljaad et al. (9). This comparison indicates that, despitesome exceptions, the IIlodel succe,eds in reproducing the generalcharacteristics of the observed distribution of surface air tem-perature. The success of the model in simulating the geographi-cal distribution of climate and Lts seasonal variation encoura-ged the authors to colr1duct the present study by use of thismodel.

'nIERMAL RESPONSE

Sea Surface Temperature

To illustrate the effect of continental ice on the distri-bution of sea surface temperature (SST), Figures 3 and 4 areconstructed. These fiE;ures show the geographical distributionsof the SST difference of the model mixed layer ocean between theice age and standard experiments for February and August, res-pectively. These can be comparE!d to the distributions of SSTdifference between 18 kyr BP and the present as determined byCLIMAP, which are added to the lower half of each figure.

In the Northern HE~misphere in both winter and summer, SSTsfrom the ice sheet experiment are significantly lower than thecorresponding temperattlres in thE! standard experiment. The SSTdifferences are most pronounced in the mid-latitudes of theNorth Atlantic and the North Pacific, with the ice sheet-inducedcooling over the Atlant:ic generally larger than the correspond-ing cooling over the Pacific. This is in good qualitativeagreement with the difference between the modern and 18 kyr BPdistributions of SST as obtained by CLIMAP.

In contrast, the SS! differences in the Southern Hemisphereof the model are very I;mall in both seasons, while the S5! dif-ferences as estimated by CLIMAP are of significant magnitude.Since most of the changes in the distribution of continental icebetween the two experiments are located in the Northern Hemi-sphere, this result suggests that the presence of an ice sheetin one hemisphere has relatively little influence on the distri-bution of sea surface temperature in the other hemisphere.

INFLUENCE OF mE CLIMAP ICE SHEET 793

~N

60

30

0

30

60

90s0 GOE 120 180 120 6OW 0

90N

60

30

0

30

60

90S

Figure 2 Geographical distributions of monthly mean surfaceair temperature (degrees Kelvin) in February. Top: comput-ed distributi.on from the standard experiment. Bottom: ob-served disitribution (8,9). The computed surface air tempe-rature represents the tem];>erature of the model atmosphere atthe lowest finite difference level located at about 70 mabove the earth's surface.

794S. MANABE AND A. J. BROCCOLI

SEA SURFACE TEMPERATIJRE DIFFERENCE (ICE SHEET EXP.-STANDARD EXP.

FEaRUARY

60'

JO':k).

fO

~;;

30':M)'

60',600

Figure 3 Geographical distribution of February sea surface tem-perature differen<:es (degrees Kelvin). Top: differencebetween ice sheet and standard experiments. Bottom: diffe-rence between 18 kYI: BP and ipresent (as reconstructed byCLIMAP).

---9O.S' :.. .~- oj-' ., , .'9O"S

O' ~. 60" 90. 120. 150.e 1.,. I50OW 120' 900 600 ~ o'

SEA SURfACE TE~\PERATURE DiffERENCE (ICE AGE-MODERN)08SERveo

INFLUENCE OF THE CLIMAP ICE SHEET

795

C~D

AUGUST

~

)O'

600 900 1200 IsooE ISO"", 1200 900 600

SEA SU~FACE TEMPE~ATU~E DIFFE~ENCE (ICE AGE-MODERN)~. O'

bO'

60'

O' ~. 9()' 1200 I~'E 1~. 150"W 120. 90' 60' :k)° 0-

S. MANABE AND A. I. BR<Xx:oLl796

Hemispheric Heat Balance

To evaluate the lack. of a Southern Hemisphere response tothe presence of widespread continental ice in the Northern Hemi-sphere, the hemispheric heat budgets of the model atmosphere areobtained from the standard and ice sheet experiments. These areillustrated by the box diagrams in Figure 5, as is the differen-ce in heat budget between the two experiments.

For the Northern Hemisphere, the net downward flux of solarradiation at the top _~f the model atmosphere in the ice sheetexperiment is 5.6 W m less than the corresponding flux in thestandard experiment. This is primarily due to the reflection ofinsolation by the l;irge area of continental ice. The differencein incoming solar radiftion is essentially counterbalanced by adifference of 5.8 W m in outgoing terrestrial radiation at thetop of the model atmosphere. Although the rate of interhemi-spheric heat exchan;ge is also different between the tw~ experi-ments, the magnitude of the difference is only 0.4 W m- , and ismuch smaller than the differences in the net incoming solar ra-diation and net outgoing terrestrial radiation.

These results indicate that, in the ice sheet experiment,the effective reflection of incoming solar radiation reduces thesurface and ftmospheric temperatures in the Northern Hemisphereof the model and, accordingly, the outgoing terrestrial radia-tion at the top of the atmosphere (10). The relatively lowsurface temperature in the Northern Hemisphere induces a smallincrease in the heal: supplied from the warmer atmosphere in theSouthern Hemisphere. However, the radiative compensation in theNorthern Hemisphere i:; much more effective than the thermaladjustment through the interhemispheric heat exchange in themodel atmosphere.

CONCLUSIONS

The results of t:his study suggest that the effects ofincreased continental ice extent alone are insufficient toexplain the glacial climate of the Southern Hemisphere, althoughvariations in the eartli1's orbital parameters may be responsiblefor including the large fluctuations in the extent of NorthernHemisphere ice sheets during the Quaternary. Thus it is neces-sary to look for mechanisms other than the interhemisphericexchange of heat in the atmosphere in order to explain the lowtemperature of the Soul:hern Hemisphere during the 18 kyr BP iceage. This is consistent with the results of Suarez and Held(11) using a simple energy balance model to study the

astronomical theory of the ice ages.

INFLUENCE OF THE CLIMAP ICE SHEET191

N. HEM. EQ. S. HEM.

DiffERENCE

Figure 5 Box diagrams representing the primary energy fluxesin the atmosphere-ocean-.cryosphere system of the model. Thesurface fluxes shown at the bottom of the Northern Hemi-sphere box represent the h'~at energy involved in the growthand decay of sea ice, land ice, and permanent snowcover, andlong-term changes in heat storage in the oceanic mixed

layer. Top: standard expe:riment. Center: ice sheet expe-riment. Bottom: differen(:e between ice sheet and standard

experiments.

798 S. MANABE AND A. J. BROCCOLI

A~ng the potential mechanisms for the cooling of the Sou-thern Hemisphere duri:l\g glacial times is the cross-equatorialheat transport by the ocean circulation. For example. if theintensity of the interhemispheri,= thermohaline circulation chan-ges from a glacial to an interglacial period. as is suggested byRooth (12). it is posElib1e that the interhemispheric ocean heattransport does also. resulting in a change of Southern Hemi-sphere temperature.

Other processEs which can cause an almost simultaneouschange of temperature in both hemispheres are fluctuations inthe concentration of carbon dioxide of the loading of aerosolsin the atmosphere. IndE~ed. the results from the recent analysisof ice cores from the J~ntarctic and Greenland ice sheets suggestthat the atmospheric concentration of CO2 during the last gla-cial maximum was about: 200 ppm by vo1u~ and is significantlyless than the curreclt concentration of 340 ppm (13.14).Broecker (15) proposes a mechanism by which a reduction of atmo-spheric carbon dioxidE~ might result from the lowering of sealevel associated with the growth of continental ice. If suchreduction did occur, it could account for much of the SouthernHemisphere cooling indicated in the CLIMAP results.

REFERENCES

1. Hays, J.D., Imbrie, J., and Shackleton, N.J. : 1976, Science194, pp. 1121-1132.

2. Manabe, S., and StoLJffer, R.J. : 1980, J. Geophys. Res. 85,pp. 5529-5554.

3. Gordon, C.T., and Stern, W.F. : 1982, Mon. Wea. Rev. 110,pp. 625-644.

4. Manabe, S., Hahn, D.G., and Holloway, J.L. : 1979, GARPPubl. Ser. 2:!, World Meteorological Organization,Geneva.

5. Manabe, S., and Hahn, D.G. : 1981, Mon. Wea. Rev. 109, pp.2260-2286.

6. Manabe, S. : 1969, Mon. Wea. Rev. 97, pp. 739-774.7. CLIMAP Project MembE!rs : 1981, The Geological Society of

America Map and Chart Series, MC-36.8. Crutcher, H.L., and Meserve, J.M. : 1970, NAVAIR 50-IC-52,

U.S. Naval Weather Service, Washington, D.C.9. Taljaad, J.J., van Loon, H., Crutcher, H.L., and Jenne, R.L.

: 1969, NAVAIR 50-IC-55, U.S. Naval Weather Service,Washington, D.C.

10. Bowman, K. : 1982, J. Geophys. Res. 87, pp. 9667-9674.11. Suarez, M.J., and Held, I.M. : 1979, J. Geophys. Res. 87,

pp. 9667-9674.12. Rooth, C. : 1982, Prog. Oceanog. 11, pp. 131-149.

INFLUENCE OF mE CLIMAP ICE SHEET 799

13. Berner, W., Oeschger, H., and Stouffer, B. : 1980, Radiocar-bon 22, pp. 227-235.

14. Delmas, R.J., Ascencio, J.M., and Legrand, M. : 1980, Nature284, pp. 155-157.

15. Broecker, W. : 1982, Prog. Oceanog. II, pp. 151-197.

lStrickly speaking, the ice sheet-induced reduction of sur-face temperature results not only from t:t1e large surface reflec-tivity but also from the high elev,altion of the ice sheetsurface.