Annals qfClaeiology 25 1997 © International G laciologica l Society

The Antarctic surface m.ass balance in a stretched grid general circulation m.odel

GERHARD KRINNER, CHRISTOPHE GENTHON

Laboratoire de Glaeiologie et Geophysique de I'Environnemenl ( C\RS, associe a l'CniversiliJoseph-Fourier), 54 rue Moliere, BP 96,38402 Saint-Martin -d'Hlms Cedex, France

ABSTRACT. The LaboraLO ire de l\lctcorologie Dyna mique (LMD ) va riable-g rid a tmospheric genera l circula tion model ( AGC~I ) was used in thi s stud y for a five-year high-resolution simulation of the Anta rctic clima te. The hori zontal resolution is about 100 km over a la rge part of the ice sheet. This study focuses on the simulated surface mass balance (precipita ti on- evaporation sublimation- melt ) and on the spati al and temporal \'a ri ability of snowfa ll in Anta rctica. The simulated annual mean surface mass ba lance fo r the whole continent is close to the observed va lue, and the model simulates well the spati al di stribution of the surface mass ba lance. The annua l cycle of snowfall ex hibits a clear minimum in summer O\'e r the high interior pl ateau as well as for Anta rctica as a whole, in ag reement with the observations. In the interior of the continent, the model p roduces a permanent li ght background snowfa ll that accounts for about 5% of the total annual precipitation. The bulk orthe snowfa ll is produccd irregula rly during period s that genera ll y las t only two or three days that a re caused by cyclones off the coast.

INTRODUCTION

One poss ible impac t of human activity on the globa l climate sys tem is cha nge in the surface mass ba lance of the pola r ice sheets, resulting in a change in the globa l sea level. \Ve need to be abl e to g ive accurate estimates of the future surface mass ba la ncc of Anta rctica because it is the largest ice sheet

on Earth. In response to a wa rmer climate, the Anta rctic ice shee t appea rs more likcly to thicken than to thin (Ohmura a nd 'vVilcl , 1995). This may be because warmer air ca n trans­port more moisture LO the continent, whereas the temper­a ture is unlikely to ri se enough to increase signifi ca ntly ice melting and runoff, although changes in cyclone frequency,

strength a nd pos ition might be expected th at may compli­cate the picture. The modifi ed surface mass bala nce of Ant­a rctica may thus pa rti a lly offset a sea-level ri se from the mclting of temperate and other pola r glaciers (including Greenland ) and the therma l expansion of the oceans. How­ever, uncert ainties remain (Houghton and others, 1995).

Atmospheric genera l circulation models (GCMs) can be used to predict the future surface m ass ba lance of the pola r ice sheets. In order for these predictions to be reli able, it is first necessa ry to be sure that the GCM produces a reason­able simulation of the current surface mass balance. Much progrcss has been madc in recent years in this respec t (e.g.

T zeng and others, 1994), due in part to increased horizontal resolution as a consequence of increased computing power (Genthon and others, 1994). H owever, even if a modcl pro­duces a good estimate ofthe current surface mass ba lance, it need not necessaril y be for the correct reasons. For example, Connolley and King (1996) report th at in the United

Kingdom Meteorological Office (UKMO) GCM, which produces a good estimate of the Anta rctic surface mass ba lance, only 30% of the water vapor precipita ted over a la rge secto r of East A nta rctica is advected by expl ici t trans-

port at the reso lved sca les. The rest is ca rried onto the con­tinent through hori zonta l diffusion, which is a numerical convenience introduced to prevent small-scale instabiliti es and which can be seen as a pa rameteri zation of sub-gricl­scale a tmospheric dyna mics. Connolley and King (1996) conclude that it is doubtful whether a predicted surface mass bala nce for future climates is credibl e.

In thi s study, we used the Laboratoire de M eteorologie D ynamique (LMD ) stretched-g rid GCM (provisiona ll y call ed LMDz, where 'z ' sta nds for 'zoom' ) with a hori zontal resolution of about 100 km over a la rge part of Anta rctica. This is quite a high reso lution, the grid spacing of most GCMs usually being only about 300- 400 km. This high re­

solution is achi eved by loca lly refining the g rid over the region of interes t at the expense of the res t of the globe. The model is globa l, so the interaction between the high-resolu­tion region and the res t of the globe is treated consistently, but computa tional cost is reasonable compa red to a model with equa l g rid spacing everywhere.

Wc compa re the spatial pattern of the simulated surface mass ba lance (in water equivalent (w.e.) a 1) to the obser­\'a ti ons, and, in order to evaluate the credi bility of the simul ation, we examine the annual cycle of the modelecl precipita ti on o\'C r the whole continent and over the high plateau of East Antarctica. Furthermore, the high frequency

variability of the simulated precipitation in Antarctica is addressed, and a typical synoptic situation leading to signif­icant snowfa ll is bri efl y di scussed.

THE MODEL

The model is a new version oC the Lr../ID grid point GCM. The physical pa rameteri zations a re e sentiall y those of the LMD-GCM Cycle 5 (Polcher a nd others, 1991; H arzall a h

73 Downloaded from https://www.cambridge.org/core. 25 Dec 2021 at 04:19:24, subject to the Cambridge Core terms of use.

Krinner and Cen/hon: An/arctic Sll1jace mass balance

a nd Sadourl1Y, 1995). The mos t important differences between the LMD-GCM Cycle 5 and the GCM used in thi s study is the abilit y to stretch the g rid hori zontally, a nd the different pa rameter ization of the horizonta l diffusion o /" small-scale waves that may be subj ect to future improve­men t. There is no hori zontal diffusion o[ water \·apor in the model, so all moisture transport is by the reso lved wind s. Thu ·, the problem identifi ed by Connoll ey and K ing (1996) in the UK~IO GCM does not a ffect this model. Some of the parameterizations a rc further ada pted 10 1' pola r climate and high resolution, and a re described by Krinner a nd others (1997).

The primitive eq uations arc integrated on a g rid of 64 equall y spaced points in longitude and 72 points in latitude, which a rc concentrated oyer Antarctica. This yields a meri­

dional resolution of about 100 km oyer Antarctica, a nd approx imately 700 km at the North Pole. Due to the conver­gence ot the meridi ans towards the poles, the zona l reso lu­tion becomes high in the polar regions e\-en with a low number o[ gridpoinrs in longitude. Figure I shows the eITec­tive zona l a nd meridiona l resolution as a fun ction of lati­tude. Note that, a lthough the nominal zonal resolution becomes infinite a t the Pole because the meridi a ns con­\·erge, the effective zona l model reso lution does not, because ot th e act ion of a numerical zona l wave filter close to the pole. This filtering is necessary because an infinitely fine g rid requires a vanishing integra tion time-step as a conse­quence of the Courant- Friedrich- Levy stability criterion. The zonallilter thus limits the effective zonal resolution to

about 100 km polewa rds of 82 S. The simu lation presented here covers the five years from

September 1986 to August 199 1. The model was spun-up for three months from a synoptic snapshot. AMIP (Atmo­

spheric Model Intercomparison Proj ect) (Gates, 1992) boundary conditions for the sea-surface temperatures were

74

.--..

~ 800+---~~---L----~----~---L----+ --C 700

.Q 600 ...... :::J 0 500

~ 400 ••••••• _ 300 •• '

§ 200! / ......

:0 100 ~ .;:: ID 0+-----,----,-----,-----.----,-----+ E ·90 -60 ·30 0 30 60 90

a latitude .--.. 800+-----~----~----~----L-----L---__+

~ 700

C 600

.Q 500 ...... :::J 0 400 Cl) 300 ~

- 200 ctS C 100 o

// ................... .

N 0+-----,----.-----,-----,----,-----+ ·90 ·60 -30

b o

latitude 30 60 90

Fig. 1. The model resolution in km as afunction rif latitude, ( a) meridional resolulionJ ( b) zonal resolution.

used. Sea-ice fraction was obta ined from SSMjl satellite data (NSIDC, 1992). This is the same simula tion as that pre­sen ted in Krinner a nd others (1997).

RESU LTS

The a nnua l Inean s u rface Inass b a l ance

Figure 2 shows the annua l mean surface mass balance as simu lated by the model, as well as observat ions by G iovinetto and Bentl ey (1985). The model captures well the shape and the ex tent of the large area with low acc umul ation below 5 cm a \ a lthough this region extends too far towards the coast on the Lambe rt G lac ier and the eastern edge of the Ross Ice Shelf. Low acc umul ation on the ice shelves is a fea­ture both of the model and the observations. H owever, the model underestima tes the mass balance in these areas. Due to the relatively high horizonta l resolution, the strong accu­lllulation grad ient near the coast is well captured in the silllul ati ons, and the model even succeeds in reproducing the higher accumulation values at the foo t o /" the Trans­antarctic Mountains on the Ross Ice Shelf.

o·

180·

o·

180·

Fig. 2. The annual mean Sll1jace mass balance in CIII a I w.e., (a) as simulated b} the model, ( b) after Ciovilletto and Benl!C)1 (J985), interpolated onlo the model grid.

Downloaded from https://www.cambridge.org/core. 25 Dec 2021 at 04:19:24, subject to the Cambridge Core terms of use.

The zona l acc umu la tion g radient ove r the Anta rct ic

Peninsul a is not sufIi cient ly strong in the m odel. Because the Penins ul a is relatively fa r north, th e zona l resoluti on is o nl y a bo ut 250 km. Thi s means that th e m ounta in ranges in thi s a rea a rc no t resolved , leading to a n underestima ti on of

the m-ographic forcing of precip ita ti on. H owcver, th e a\ 'er­

age acc umul a t ion over the A ntarc tic Peninsula is acc ura tely

r eprod uced. As a whole, the ge neral p a ttern of th e surface m ass

bala nce is well sim u la ted by th e m odel. Including the ice

sh eh-es, the overall continenta l mean surface m ass bala nce

[o r the fi ve yea rs is 16.2 cm w.e. a I with the \'a lues fo r the

ind iv idua l yea rs ranging from 16.0 to 16.+ cm without a ny

\ 'isible trend over the live years of th e simul a ti on. T he \ 'a lue o r 16.2 cm a I can thus be rega rded as represe nt a ti ,'C o r th e m odel climate. T his com pa res ravo rabl y with th e mos t recent esti mati ons of the surface m ass ba la nce (e.g.

15.4 cm a I by Frolich (1992) fo r the whole ice shee t). T his es­

tima ti on is based on th e value of 14.3 cm a I given by Gi O\' i­

netto a nd Bentley (1985) fo r the Anta rctic excluding the Anta rctic Peninsu la . The un ce rta int y o r these va lues is a bo ut 10%.

The annual cycle of t he su rface m ass bala nce

Fig ure 3 displays the fi ve-year mea n monthl y surface m ass ba lance far th e whole continent (Fig. 3a), a nd fo r th e g ri d­point s a bO\'e 3000 m a ltitude, w hich thus be long to th e East

Antarct ic p la tea u (Fig. 3 b ), In bo th cases, there is a clea r

minimum in summer, perhaps earli er on the Anta rctic pla­

teau tha n on the rest o f the continent.

-.s:: 1: 25.-L--L __ L--L __ L--L __ L--L~L--L~ __ -LT

E 20 r----l--l-" I

~ "r iT+-1'l C / \ 'I :.8 10 / '" CO -l " " 'I,-

:::J 5 -E :::J ~ O~r-.--'--.--.--.--r-.--.--.--.--r~ CO 2 3 4 5 6 7 8 9 10 11 12

-.s:: +-' c o E -

a month

E 5 E -c o

:;::::; CO :::J

E (?"'O-::J ~ 0 ~-'---r----'-'---r----'--;--r----'--;----....L CO 2 3 4 5 6 7 8 9 10 11 12

b month

Fig. 3. (a) Simulated month ly meall swfilce mass balance ill mm mOll th lfor the whole Antarctic ice sheet. (b) Simulated monthly mean suiface mass balance ill //1 //1 montl! 1 for the gridpoints above 3000 Ill . T he diamonds show jJrecijJitation observationsfor the Vostok stationp'om Dolgina and Petrova (1977). The enor bars denote the standard devia tioll between the individualJive jiears rif the simuLatioll.

/{rinner and Gent/IOn: Antarctic S1l1face mass balance

For the continent as a whole, the snow[a ll m aximum

typicall y occurs in April or M ay, with a broad m ax imum rrom Februa r y to O ctober. In rac t, deyeloping the m onthly tim,e seri es or precipitat ion o\'(' r the G" e yea rs (Fig. +a ) shows there is often a seconda r y m aximum betwee n Jul y

and Septcm ber, which is not \ 'isibl e in the live-yea r m ean

time series (Fig. 3a ), because its ti m ing is \·a r iable.

-.s:: C 30 ,......~~~u......~~.......J..~~~ ........ ~~~ ......... ~~ ..........

o E 25 -E 20 E -C 15 o

:;::::; CO 10

::J E 5 ::J ~ O~~~~~~~~~~~~~~~~~~~ CO 12 24 36 48 60

a month

Fig..J. T he lIIol1th£)' mean values if the aCClImlllationJor the individual 60 months qf the Jive:venr simulation in mm 1II0nt/1 I T he time series begins in September 1986 and end., in August 1.991. (a) Jar the 1.l'!lOle rif An/arctira, (b) Jor the gridjJoinls abol'e 3000 //1.

Above 3000111 , when de\'e1oping the same fi\ 'C-yea r

monthly time series (Fig. +b), thcre is strong semi-a nnu a l \'a ria tio n \'isible fo r m ost of th e years, but aga in the timing of th e indiv idua l peaks is \ 'ari a ble, so th a t in th e fi\ 'C-yea r mean time seri es (Fig. 3 b), thi s sign a l becomes a lmost insig­

nilicant a nd only th e a nnua l cycle with a broad m aximum

in winter and a rela ti vely sha rp minimum in summer m ay

be regarded as sig nili ca nl. The m onth ly m ean accumul ati on can be compa red to

prec ipita ti o n obsen 'a ti ons Crom \'Clrious Antarctic sta ti ons p ubli shed by D olg in a nd Pc trov (1977) and di sc ussed by Bro mwich (1988). Fo r the high interio r sta ti ons (Vostok a nd

South Polr ), the a nnual cycle seem s to b e reaso na bly simu l­

a ted by the m odel, exce pt tha t the m ax imulll is proba bly a

littl e ea rly in the m odel a nd the summer minimum is too sha rp. There is no sign of semi-a nnua l " a ri a ti on in th e obse r­, ·ati ons.

However, the obsen 'ati ons in centra l A nta rctica, where

the precipita ti on ra tes are low, a re subj ect to sig nilicanl un­

certa inty (Bro lllwich, 1988). The obser vcd values from th e

E as t Anta rctic coas ta l sta ti ons sho\\' th e same seasona l trend, a nd Brol11 wich concludes th a t thi s kind or cycle a m aximum in winter a nd a m inimum in summer prob-

75 Downloaded from https://www.cambridge.org/core. 25 Dec 2021 at 04:19:24, subject to the Cambridge Core terms of use.

Krinner and Centhon: Antarctic surface mass baLance

ably cha racteri zes thc cnti rc East Antarctic ice shcct. H ow­ever, Budd and others (1995), using the net atmospheric moisture convergence from the Gamma-R ay Astronomy at the South Pole (GASP) a nalyses, derived an annual cycle of the surface mass ba lance for the highest parts of Antarctica (above 2500 m ) with littl e systemati c vari ati on dur ing the year, and maybe with a slight max imum in summer. On the other hand, as in thi s study, they fo und rel atively high vari ability from yea r-to-yea r, and frequently the highest acc umulation occ urred a round the equinoxes.

For the whole Antarctic ice sheet, the simulatcd annual cyclc of thc acc umulation rate compares favorably to the a nnua l cycle from the GASP (Budd and others, 1995) and the Nationa l Meteo rological Center (NMC) (Ya maza ki, 1992) ana lyses, al though it should be noted that the NMC analyses show a stronger semi-annual cycle than either the LMDz model or the GASP ana lyses. Note however that the NMC analyses we re shown to be severely defi cient in somc respects (Bromwich and others, 1995). lVI OI"eover, European Centre for M edium R ange Weather Forecasting (ECMWF) analyses (Bromwich and others, 1995) suggest a peak inJuly, a nd a broad maximum from April to September, which is again simila r to the cycle simul ated by LMDz.

Cyclonic activity in the Southern O cean m ay be or con­siderabl e importa nce to precipitati on occ urrence in Antarc­tica. The continent is surrounded by a low pressure belt, which is commonly called the circumpolar trough (CPT).

The depth and position of the CPT a re linked to the strong latitudina l insolation gradients (van Loon, 1967). In the LMDz G CM , as in m any other G ClVls (T zeng and others, 1993), the latitude of the e PT is genera lly underestim.ated by several degrees (Krinner and others, (997) conl.pa red to

observa tions (Xu and others, 1990). But these obse rvations

a re drawn from va rious sources prior to 1981, so that, when comparing them to the LMDz simul ations, one has to bea r in mind that the differing time periods may be at leas t partly responsible for the differences. Note that the simul­ated latitude of the CPT compares well wi th the ECMWF

ana lyses from 1985- 91 (Krinner and others, 1997). The mean

intensity of the ePT is also simulated well by the LMDz GC M , although it fails to reproduce the observed intensity or the semi-annual cycle. Note, however, that these cycles a re subj ect to g reat inter-annual variability in the m odel, so that the details of the cycles may not be statistically robust (Krinner and others, 1997).

Figure 5 shows the five-year time seri es of the simu lated depth or the C PT. T here is a clear co rrclation between thi s

76

? 0 A ! ~ I1 V"b )l 1\ 19 cq, I I Q -AQ ? L ~ ? \ i 'tm",

Of 0\/\' V" Q I Cl..) 9 ' i-n. Q I u-

' Q ' : ~Q ! \f ~, ! ~,/ Q f g \ J d\: 0 I f CD ~ QI ' ! \ 6 o 66 q;

If b o

12 24 36 48 60

month Fig. 5. The month61 mean values of the swj ace pressure if the ePT for the individuaL 60 months of the five -:Far simulation in hPa.

time series a nd the live-year time seri es for the acc umul a­tion over the whole of Anta rctica (Fig. 4a ), although this correlation does not hold for a ll details (i. e. there a re differ­ences in the relative amplitudes of the minima and maxi­ma). This is due to influences from other pertinent va ri ables such as the latitudinal position of the CPT and the sea-ice extent.

The accumulation over the Anta rctic plateau above 3000 m seems to be more strongly influenced by the latitudi­na l position of the CPT (not shown), a lthough the correla­tion is not clear since both signals are complicated and, again, other influences such as sea-ice extent and the in ten­sity of the trough a rc important.

T h e s hort-terlll va riability of precipitation

Fig ure 6a di splays the 1990 time seri es of the simulated da ily prec ipitati on a t the griclpoint that contains Dome C (123.0° E, 74.5° S, 3280 m a.s. 1.) in East An tarctica. The fre­quency of precipitation events is g reatest during spring and autumn, and least in winter and summer. This is consistent

1.3 1.2 1.1

~ 1.0 E 0.9 .§. 0.8 5 0.7 ~ 0.6 .§" 0.5 ~ 0.4

0.3 0.2 0.1 0.0 4LI .... ...a.I~I¥--..., ....... II...tf-..... +W.~ .... ~I4i~

o 30 60 90 120 150 180 210 240 270 300 330 360 a day

680 +-~--~--~~--~~~~--~~--~~~-+

.. a.

670

:S 660 ~ :J

"' !3 650 a. ~ 640 "t:

~ 630

620

b

-20

-30

U -40 l!.-a> 5 "50 ~ Q)

It -60 ~

-70

-80

C

0

0

30 60 90 120 150 180 210 240 270 300 330 360 day

30 60 90 120 150 180 210 240 270 300 330 360 day

Fig. 6. Daily vaLues for 1990 for Lite gridbox rejJresenting Dome C. The LMDz modeL has a 360-day year, (a) precipi­talion in mm rI, ( b) surface pressure in hPa, (c) surface temperature in 0 C.

Downloaded from https://www.cambridge.org/core. 25 Dec 2021 at 04:19:24, subject to the Cambridge Core terms of use.

\I,ith the five-year ti me seri es of the m onthly acc umulati on above 3000 m presented in Figure 4 b,

T he m odel simul ates very few days with no so lid precipi­ta ti on, (H erearter, "snowfa ll " will be used in the sense or "so lid prec ipita ti on" that comprises prec ipita ti on of' snO\\'

a nd ice crysta ls.) R a ther, there is il1\'ari abl y a t leas t a light snowfa ll bac kg round of' abo ut 0.005 mm d I (about 0,2 cm a \ which thus acco unts fo r about 5% orthe annua l accu m ula ti on, T hi s backgrou nd precipita ti on might corres­pond to the light ice-crys ta l precipita ti on oft en reported . e\'en from clear skies that occurs d uring m ost of the year in

the hig her parts of Anta rctica (Schwerdtfege r, 1984; Bro m­wich, 1988). or course, clouds are d iagnosed in the model when precipitat ion is simulated , but because oC the \Try low temperatures these clo uds ca n be optically su thin th at the process can be ident ifi ed as "clear sky" precipit ation,

E\ 'en though mean prec ip itation has a defin ite mini­

mu m in summer (sce the sec ti on en titled: T he a nnua l cyc le of the surface mass bala nce ). the la rges t single snow fa ll e\'C'n t occurs on day 30 of t he year. \\'arme r a ir can conta in more mo isture, so that summer prec ipita ti o n e\ 'ents. a lthough ra rer than in winte r. can depos it a compa ratively large quantity or water and contribute sig nifica ntl y to the

a nnua l surface m ass ba lance. It is interesting to note that,

du ri ng a nother sim ulated yea r, a simila rly importa nt single su mmer precipita ti on cvent was a lso simul ated,

Both cvents a rc charac terized by a particul a r synoptic situa ti on: a st rong cyclone off thc coast downslopc of D ome C, and a rela ti \·e1 y high-pressure a rea to the east of it (Fig. 7),

T he two systems together ad vec t wa rm a nd moist a ir into the inter ior where prec ipita tion then occ urs. This a ir mass can even be tracedLO Vostok, where rela ti \'ely high sno\l,fa ll is simula ted onc day la ter tha n a t Dome C. The synoptic situat ion di scussed abo\'e is simil a r to the situati on leading

to intrusions of warm moist air into Antarc tica a nd to sub­

sequent prec ipita tion in the interi or as di scussed by Brom­wich (1988).

o·

180'

Fig, 1 The sill1u{aledsea-{e l'e{ jmssure (ill hPa) 011 da.,)' 30 qf 1990 minus Ihe mean sea-La'e! presslIreJor Decell1bl'1'.]allll ­m)' and Februm),.for 1989 and 1990. Areas belo~t' -20 hPn are bl(lck, areas above +20 hPa (Ire gr~)I . . Vote the strollg pressure gradient off the coast facing Dome C

it i'illll er alld Gm/hall: AIl/arc/ic SlI1Jace mass balance

In most or the cases of high prec ipita ti on in the interi or, a simila r d ipole situatio n offth e coast ca n be obsen 'Cd, Note howeve r th at the low pressure does not illlrude into the inter ior or Anta rctica, a lt houg h its a llli cyclonie counterpa rt to the east could help "push" it towa rd the interi or. This is

rea li stic, as it is well known tha t cyclones ra rely penetra te into East Anta rctica because of the blocking effec t of the icc sheet (M echoso, 1980),

In ract, the surface pressure a t D ome C is particul a rl y high a nd railing d uring the (T ent. a tta ining it s ma ximum

one day befo re the precipita ti on (Fig, 6b). This is in agree­

ment with the di sc ussion by Bromwich (1988). At leas t in summer, p rec ipita ti o n in the interior occ urs m ore often in assoc ia ti o n \\'ith high th a n \I,ith 10\\' surrace press ure, a ltho ug h the co rrela tio n is not strong. In winter, there a re sometimes snowfa ll episodes \I'hen the surrace pressure O\'Cr the \\'hole intrr ior of Anta rctica d rops by more tha n to hPa

in one or two days, but that thi s is due to the intrusion of a cyclone is do ubtful. On the other hand . a nd not surpri singly, there is a clea r co rrelati on bet\\'Ccn prec ipita ti on a nd temp­nat ure (Fig. Gc " pointing LO the ocea nic ori gin or the a ir m asses th at bring moisture into the continent. :\iote that Fig­

LIlTS 6a a nd 6c reflec t the increased cyclonic acti\ 'ity in win­

ter a nd the rcl a ti\ 'C ly calm weather in summer.

SUMMARY AND CONCLUSIONS

T he Li\ID stretched-g rid GCi\I with a loca l horizonta l resolution or 100 km produces a n acc ura te simulati on or the surraee mass bala nce o f'th e Antareti c ice shee t, a lthoug h the shelf a reas a rc too dry in the m odel a nd, due to a rel a ti\'el y low zona l resolution o\ 'er the Anta rctic Peninsul a,

the strong zona l \'a ri a tions of the mass ba la nce in thi s region

a re not well captured. The model reproduces a m aximum sno\l' acc umulati on a t th e foo t of the Tra nsanta rctic \Ioun­ta ins, but it underestimates the mass ba la nce o\ 'C r the Amery Ice Shelf' a t the to ng ue or the L a mbert Gl acier. The tota l continenta l surface m ass ba la nce is 16.2 cm a I, which

compa res fa\'o rably with estimates for the actua l ba la nce.

The seasona l cycle of prec ipita ti on shO\\,s a clear mini­mum in summer a nd a la rge maximum in winter, wh ich is in keeping with obs(T\'a ti ons. There is a large iJ1ler-a nnua l \'ariability o rthe seconda ry fCa tures orthe a nnua l cycle, but in most or the fi\ 'e years of the simulati on, a semi-a nnua l

\'a ri a ti on or the acc umulati on is \'isiblc. This may be linked

to the semi-annua l va ri a ti on orthe intensit y of the circuJ11-po la r trough,

The short-t erm va ri a bilit y or snO\\,ra ll in the nl.ocle l sho\\'s the clea r influence or th e cyclones off the East Anta rc­tic coast, a lthough, a t leas t in summer, there is no sig n of

intrusio n orlhese cyclones into the interi or of the continent.

Prec ipita ti on ra rely d rops to ze ro in the model. Rather, a n a lmos t perm a nent background snowfall acco unts for abo ut 5% or the tota l a nnua l mea n preeipita ti on, This light, stead y prec ipita ti on might be rela ted to the occ urrence of ice crys ta ls ill the lower a tmosphere e\ 'en during cloudless

days,

O \'e ra ll . the model produces a rea li stic simulati on of the p resent-day prec ipita tion in A11l a rcti ca. Further ex peri­ments \I'ith the model a re pla nned , including a n assessment or the dyna mic aspec ts or cyc lones, H O\l'e\'Cr, wc a re sufTi­ciently confident to contemplate its use for the estimation of

77 Downloaded from https://www.cambridge.org/core. 25 Dec 2021 at 04:19:24, subject to the Cambridge Core terms of use.

Kril1l1eT and Gentlwn: An/arctic S1l1jclce mass balance

Anlarctic mass balance in other climates (e.g. ice ages and CO2 doubling experiments).

ACKNOWLEDGEMENTS

This research was supported by the Commission of the Eur­opean Communi li es, Directora le General XII, under con­tract ENV4-CT95-0I24 (Climate and Sea Level Change), and by the French Programme National d'Etude de la Dyna­mique du Climat. Computer time on a Cray C90 was pro­vided by the Institut du DcYcloppement et des Ressources en lnform alique Sci enlifique (CNRS) and by the Cenlre Grenoblois de Calcul Vectoriel (CEA). \Vc a rc graleful to the Laboratoire de Meteorologic D ynamique ( E~SI

CNRS) for providing the LMDz GCM, and P. Le Van and L. Li for lheir help. Comments and suggestions by D. Brom­wich and two anonymous rev iewers helped improve this work.

REFERENCES

BrOlmrich. D. H . 1988. Sno\\'fal l in hig h southe rn lat itudes. Ra. Geoplll'5 .. 26 (1), 1+9 168.

Brol1l\\·ich. D. 1-1. , F. ~1. Robasky, R. 1. Cu ll a t her a nd ~ I. L. , 'an \ \ 'Den. 1995. The atmosphe ri c hydrologic cycle ovcr the South,Tn Ocean and Ant­

a rctica fi'om o pcrational numerical analyses . . \1011. li en/ha Rn '., 123 (121, 3518 3538.

Buckl . \\'. E. P. A. Rcid and I..J Hinty. 1995. Antarctic moisture flu x a nd

nct accul1lu lation fi 'o m g loba l atmospheric a na lyses. rflll/. Glacial" 21. HI) 156.

Connolley. \I'. :-'1. andJ. C. King. 1996. ,I modcling and obscn'ati onal st ud), of East Anta rctic su rface mass balance . .1 Ceopl!l's. R"., 101 (D I). 1335 13++.

Dolgin. I. lI1. a nd L. S, PctrO\'a, eds. 1977. Sj)favotcilllik jlo klill7alll Alllarkl;d; !lalldbook q/ Alllarc/;" climale . rol. 2. Le ningrad. H ydrological Institute.

1I n Russ ian.]

Fro lich. R . r,l. 1992. The su rface mass balance of the Antarctic Peninsula ice

sheet. III i\lorri s. E. 11,1., ed. The cOII/ribll/ioll q/ AII/arc/ic Pellillsula ice los(a IfI'ei rise. Re/Jar/ Jar Ihe COII/missioll q/ Ihe Ellro/)f(1II COIII 1111111 ili,'" Projl'(l EPOC-CT90 -0015. Cambridge. Briti, h Antarctic Su n TY, 3 9. r Ire and

78

Climate Special R.eport I.) Gates. \ I'. L. 1992. A1I 11 P: th c Atmospheric M odc l I ntcrcomparison Project.

B ui! . • 1111 . . 1/e/eorol. Soc.. 73 (12 ).1962- 1970..

Genth on, G. J. J o uzcl a ndl\L D cqu c. 199+. Accumulation a t the surface of'

po la r ice sheets: obse n 'a tio n a nd modelling for global climate change. III D esbois. :--1. and F Dcsa lmand, eds. Global/Hen/liLa/ioll alld dill/a/e challge. Be rlin. etc" Spri ngcr-Verlag. 53- 75. t :--lATO ,I SI Seri es I: Global Em'ir­onmellta l Change 26.}

GiO\'inetto. ~r. B. and C. R. Be ntl ey. 11)85. Surface balance in ice drainagc

systcms of Antarcti ca. AI/hucl. J CS" 20 (+).6- 13.

Harzalla h, A. and R. Sadou rny. 1995. Intern a l \'(TS US SST-forccd a ltl1O­

spheric \'ariability as simulated by a n atmosp heric ge nera l circu lat ion

modeL J Climale, 8 (3 ), '[7+-+95.

Houghton. JT . G.J.Jcnkins a ndJJ Ephraums. f(ls. 1990. Ciill/a/l'Chnllge: Ihe IPC(;sriflllijit aSJeSJlllelll. Cambri dge. ctc.. Ca mbridge Un i"crsity Press.

I\. ri nner. G. C. GClllhon. L. Li and P. Le \ 'an. 1997. Studies of' the Antarctic

climatc with a stretched grid GCH J Ceoplo's. Rr"".102 (DI2I, 13,731 - 13.7+5.

Loo n. H. \·an . 1967. Thc h a ll~ )'ear ly osc il lat ions in middle and hig h sOlllhcrn

latitudes a nd the cor('lcss \\' in ter. J A /IlIOS. Sci., 24(5). +72- +86.

~kchoso. G R . 1980. The atmosphe ric circu lation aro und Antarct ica: line­

ar stability and finite anlplituclc int eracti ons w ith mig rating cycl ones . .7 .·llmo,. Sci., 37, 122+ 12+8.

:--I"tiol1<11 Sno\\ a nd Ice Data Centcr I :"S I D C \. 1992. DJ ISPSS.\ I I brighllle.,." Lelll/lera/ure rllld sea ier I'OlIcetllmlioll ,~rids./or Ihe /JOlar regiolls 011 CD -RO. II. l :rers'gllide. Boulder. CO. \Tational Sno\\' and Ire Data Cente r. (Special

R eport 1.) Ohmura, A. and 111. \I'ild. 1995. A possible cha nge in mass balance or

Greenland and Antarctic ice sheets in the coming century. GrOlllallds Geologiske l·lIrlmogeise. Opell File 9.'>/\ .19- 77.

Po lche r, J (llId 10 olhers. 1991. Le C)de 5 dlllllodele de (irrlllalioll gellfmle dll L\JD. Pa ri ,. LHD/ EC'lS. (Tec hnica l Report. LMD Intcrnal 1'\OIe 170.)

Sch\\'erdtlcgcr. \\ '. 198+. I I ra/her alld dimn/e q/ /he AII/ardi£'. Amsterd am.

Else\' ier. ( De,'elopments in Atmospheric Science IS.I Tzeng. R . -Y . D. H. Brom\Vich a ndT. R . Pari sh . 1993. Prescnt-day Antarct ic

clima to logy of' the ;\ICAR communit y climate m odel \'Crsion I. J Climale. 6 21. 205 226.

T zcng. R .-Y.. D. H . Brolll\\'ich, T. R . Parish and R Chen. 1 99~ . l'\CJIR CC!\ 12 ximul at ion or the moderll .. \Illarclic cl i III ale. ] (;('ol)/~J'.f. RI'S., 99 (DII ).23.131- n l+8.

Xu , J. -S .. H . , 'on Storch a nd H . \'a n Loon. 1990.. The performance or four

spectral GCI\1s in th e Southern Hemisphere: the J anuary a nd Jul y

climatology and the semiannual \\'<H'CS. J Clima/e. 3 (1),53 70..

1~1 mazaki, K. 1992. 1I Ioisture budge t in the Anta rctic atmosphere. IJrocmlillg.r q/ Ihe, \ '1 PN '~) ' III/lorilllll 011 Polar '\IeLeorol~~) ' alld (;Iaciolog)' 6. 36 ~s

Downloaded from https://www.cambridge.org/core. 25 Dec 2021 at 04:19:24, subject to the Cambridge Core terms of use.