Annals cifGlaciology 25 199 7
© International Gl aciological Society
Evaluating ll1.oist physics for Antarctic ll1.esoscale
sill1.ulations
K E ITH
i Polar
M. HINES, i DAVID
H . BROMWIC H,2
R. I.
C U LLATH E R
i
IV1eteorology GroujJ, By rd Polar Research Cenle1; T he Ohio State University, Colum bus, OH 43210, US.A.
2Atmospheric Sciences Program, T he Ohio Stale Universil)l, Columblls, OH 43210, US. A.
ABSTRACT. The perform ance of a n explicit clo ud physics p arameteri zati on is exa mined with simul a tions of hig h so uthern la titude winter cl im a te using a ve rsio n of th e
Penn sylvani a Sta te U nive rsity/Na ti ona l Cent er for Atm os ph eri c R esea rch M esosca le
M odel, version 4. Th e results r eveal that th ere a re three moist physics regimes in the ve rtical over the eleva ted interior oCAnta rctica: th e ve ry co ld upper tro posphere, the rel ati vely warm middle tropos phere a nd th e co ld bo und a r y laye r. D efi ciencies for th ese
layers include excessive clo ud ice in the upper troposphere, excessive clo ud ice in the inve rsion laye r nea r the ice surface, overly wa rm temperatures in th e lower trop osphere,
overl y cold temperatures in the upper troposph ere a nd excess ive downwa rd longwave r adi ati on at th e Ea rth's surface. Three sensiti vity experiments we re performed to investigate
possibl e improvements in the cloud para meteri zation. The res ults indicate tha t a reducti on of the numerous cloud condensati on nuclei, whil e reducing some errors, appea rs to be
insuffieient to improve the simul ation. A reduction in the excessive cloud ice in the upper
troposphere significa ntl y improves the simulati on of upper-tropospheric temperature.
INTRODUCTION
Cl ouds play a key rol e in th e regulati on of global climate
(e.g. R a ma na tha n a nd others, 1989). Rel ativel y little is
know n, however, about the clo ud cove r and elo ud physics
in high so uthern la titudes (Stone, 1993). Ex isting studi es of
the regional hydrologic cycle reveal uni q ue features, including the slow ice accumulati on over interior Anta rctica by
"clear sky" precipitation without the presence of visible
clouds (Bromwich, 1988). Th e proper treatment of cl oud
and cloud-radiati ve p ro perties is often a serio us obsta cl e for
numcrical modeling studies of tropical a nd mid-l atitude
regio ns, which a re much beLLer observed than the pola r
regions. Not surpri singly, m a ny advanced globa l clim ate
models poorly simul ate th e hydrologic cycle of the Anta rctic
region (Chen a nd others, 1995). For example, th e Nati ona l
Center for Atmospheric Research (NC AR ) Community
Clim ate M odel ve rsion 2 (CCM 2) improperly simul ates
highly persistent, deep clo uds over the interi or of th e Antarctic ice sheet (T zeng a nd others, 1994).
C loud physics para meteri zati ons for hig h-resoluti on
mesoscale model s, often designed for warmer tropical a nd
mid-l atitude cl im a tes, can a lso perform poorl y in the polar
regions. The explicit moisture physics param eterization
(H sie a nd Anthes, 1984) in the Pennsylvani a Sta te Uni versity (PSU )/NCAR M esoscale M odel ve rsion 4 (MM4) has
difficult y representing well the physics of Arctic stratus
(Pinto a nd Curry, 1997). Furthermore, Hin es a nd oth ers
(1997) found improper simul ati on of Antarctic winter
clo ud s by MM4. Their model ed zonall y a\'eraged prec ipitation minus evap ora tion (PME ) and evaporati on during
June 1988 a re displ ayed in Fig ure l. Simul ated evapora0
tion/sublimation is relatively sma ll south of 70 S. Fig ure I
also shows the PME derived by Bromwi ch a nd oth ers
I
I
I
I
I
-5c::
I
I
o
E
§
......
I
I - - - ....J
2
,'PME (ECMWF)
I
I
I
I
I
- / //
Evaporati on (W ETJUN)
·75
·70
·65
·60
Latitude
Fig. 1. Zonalry averaged jmcipitation- evapomtion (cm
month ) dllTing J une 1988JoT WETJUN (thick line) and
derivedfiom Ihe ECi\1W F analyses (dashed line) and zon ally averaged evaporation (tizillline, cm month- I) Jor rVE TJ UN
(1995) frolTl the weather a na lyses produced by the European
Centre for Medium-Range vVeathcr Forecasts (EC MvVF).
Between 65 ° S a nd 70° S, the PME is about 3 clTllll onth I
for both the model and ana lyses. From 70° S to 80° S,
modcl ed PME is roug hl y 0.2 to 0.5 cm month I greater tha n
the clerived PME. Th e relative error, however, is high onl y
for the mountainous inland region betwee n about 78 ° S a nd
90 0 S. An acc urate depiction of the ice acc ulllulati on over
Anta rctica is critical for studi es of th e cha nge in g loba l sea
level. Th e goal of this paper is to stud y th e deficiencies in the
pa ra meteri zati on of the hydrologic cycl e a t high southern
latitudes as a step towa rd s improved para meteri zations th at
Downloaded from https:/www.cambridge.org/core. IP address: 88.99.165.207, on 16 Jun 2017 at 19:22:24, subject to the Cambridge Core terms of use, available at
282
https:/www.cambridge.org/core/terms.
https://doi.org/10.1017/S0260305500014166
Hines and olhers: Evaluating Anlarctic mesoscale simulations
ca n m ore acc urately depict th e regiona l clima te a nd globa l
wa ter bala nce.
MODEL DESCRIPTION
Fo r the simula tions desc ribed in thi s p a per, \Ve use th e hig hreso lutio n mesoscaiC' m odel used by Hin cs a nd o th ers (1997).
T hi s m od el is a m odifi ed ve rsio n o f th e widely used MM4
(A l1lh es a nd o thers, 1987), a three-d imensio na l (3-D ), hydrosta ti c, primiti\"C-equ a ti on m odel. The key lon gwave radi atio n a nd turbule nt bo unda r y-l aye r pa ra me teri za ti o ns we re
m odifi ed to silTlLll ate better the co ld, pe rsistent ka ta ba tic
wind s th a t d ra in orr th e high ice terra in (Hines a nd o thers,
1995). L ongwave ra dia ti o n is co mputed by th e el1ieient g raybod y me th od o f Cerni a nd Pa ri sh (1984) fo r dr y clo ud-free
simul a ti ons. A rad ia tio n sch eme simil a r to tha t o f CC ~Il is
used for m oist sim ulati o ns. H o ri zo nta l resoluti o n is 100 km
in a square do m a in, 7900 km wide o n eac h side, ce ntered a t
the South Pole. Th c grid ex tends to 4 1.5 S at th e corners.
T he vc rtica l disc reti za ti on consists of 15 er levels with a
100 hPa top fo r dry runs a nd 16 er leve ls with a 10 h Pa top
fo r m oist run s. Ini t ia l co nditi o ns a nd ti me-\·a r ying bo undary conditi o ns a re interpola ted fr om E C ~rWF a na lvses fo r
Jun e 1988. Thi s win ter mo nth has been prev io usly studi ed by
Hin es a nd othe rs (1995, 1997).
M oist p hysics arc simul a ted with the ex plicit m oisture
sc heme (H sie a nd A nthes, 1984) m odifi ed for treatment o f
ice a ncl snow a t tempe ratures < 273.1 5 K (Dudhi a, 1989).
A simil a r m o ist physics sch eme is em pl oyed by Ga ll ce
(1995) in hi s simul a tion of m esocyclones O\"C r t he R oss Sea.
The exp li cit m oisture sc heme includes prog nostic equ a tio ns
[o r the local m ass frac ti o n, in m ass of wa ter substa nce per
mass of moist a tmos phere, o f wa ter \·apo r, clo ud a ncl precipita ti o n.
In the sc hem e of H sie a nd Anthes (1984), p rocesses th a t
co nvert vapor to clo ud i nclude initi a lization o f c lo ud pa rticles a nd depos ition o n to cl o ud pa rticles. C lo ud m oisture
becomes p recipitatio n by a utoco nve rsio n a nd acc re tion.
Precipit a tio n ca n be lost (gain ed ) to (fr om ) vapo r by evaporatio n (condensatio n ) or sublim ati o n. R a in a ncl snow fa ll
o ut o f th e a tm os phere at a calcul a ted termin a l vel ocit y. At
tem p era tu res < 273. 15 K , th e co ndensa ti o n nuclei co ncentra tio n, n e, (a num ber de nsity ) impac ts ice pa rticle co ncentra ti on. Duclhi a (1989) o bta ins n e by th e equ a tio n:
'ne
=
n e(0 ) cxp [0. 6(2 73.1 5 -
T )]I P
(2)
w here !lfo is 10 12 kg, qe is e1 0 ud m ass fr ac ti o n, a n cl fi t is the
timestep. Autoconve rsio n, P HC' is given by the equ a ti o n:
=
M A X [qe - qcril) / fi t , O]
RESULTS
Th e impac t of th e m o ist-ph ys ics pac kage ca n be evalua ted
by co m paring simil a r mesosca le simul a ti o ns with a nd witho ut th e hyd ro logic cyc le o f the a tm osph ere. Hines a nd
oth ers (1997) co nducted such a n a na lys is fo r hi g h south ern
la titudes. Th eir simul a ti o ns of June 1988 with clo ucl- fr ee
conditi o ns a nd with m o ist physics a re hereby refe rred to as
DRYj U ~ a nd W E1JU N , res pec ti\"ely. Compared to DRYJ U N , W E1]UN was fo un d to have a d eep er circ umpo la r
troug h surro un d ing A nt a rctica a nd sig nifica ntl y increased
geopo te nti a l heig hts a t 500 hPa o\'er A nt a rctica. The m oist
phys ics parameterizati on a ppears to h ave improved th e
m oclel cl ima tology over the Southe rn O cea n a ncl cleoTa d ed
"
it over th e interi o r of Anta rctica.
Th e impac t o f the m o ist-ph ys ics para me teri zati o n is
dem o nstra ted by Fi g ure 2, w hich di spl ays \"C rt·ica l pro fil es
of tempera tu re a\·e raged fo rJun e 1988. These p ro fil es a t th e
South Po le a re represe n tative of co nd iti ons ove r the interi o r
o f Anta rcti ca. Th e pro fi le fo r the E C MWF a nalyses has a
coa rse ve rtica l resoluti o n co nsisting o f seven sta ncla rcll e\·el s.
Consequentl y, the strong A nta rctic inve rsio n is no t well
represe nted in th e a n alyses. Th e clo ud- fr ee DRYJ U ~ represents the tempera ture pro fil e reaso na bl y well th ro ug ho ut
th e tro pos phere. A st rong inve rsio n is seen for both DRYJ UN a nd WETJ U i\'. With th e m oist physics in WE1J U :J,
however, th e tempera ture a t the bo und a r y laye r lOp is a bo ut
5 K wa rmer th a n tha t o f DRYJU N. L ongwa\"C ra dia ti on
fr om thick clo uds in \v ETJ U ~ is directl y rel a tcd to th e
increased tempera ture in the lowe r troposphere.
l00r-----~
__,---~----,---~----~----
200
(1)
wh ere nc(O) is 0.01 m :' , T is temperature (K ) a nd p is a ir
d ensity. By thi s equ a ti o n, ne inc reases by a bo ut 20 o rd ers o f
mag nitude as th e temperature fa ll s from 273 K to 193 K.
Th e pa ra me teri zed nuclei co ncentra ti o n appears to be
excessive a t realistic tempe ratures for t he intensely cold a ir
in th e uppe r trop os phere a nd near the E as t Anta rctic plateau surface. The ini tializati o n a ncl a utoconversio n of ice
pa rticl es arc bo th linea rl y related to th e nuclei co ncentrati on. \\' he n th e a tmos phere is supersatura ted with res pect
to ice, initia li zati on, Pn is give n by th e equ a tion:
Pnc
icles. Th e m oist ra di a tio n schem e is impac ted by th e clo ud
m ass fr ac tio n, but not directl y by th e c lo ud pa rticle size
d etermin ed by t he ex plicit m o isture sc hem e. Thro ug h Equ ation (3) th e sm a ll ice pa rticle size is hi ghl y unfavo ra bl e 10
a utoco nve rsio n of clo ud to precipita tio n a t yer y low tempera tures.
(3)
w here qcri1 (kg/kg ) is 9.4 x 10 12 kg ne. Thus, a t colcl Anta rctic tempera tures, para m eteri zed clo ud s fo r the ex plicit
m oisture schem e will consist of numerous, ye t sm a ll ice p a rt-
Fig. 2. Verlical lemperatllre ( K ) pnifiLes al the So uth PoleJor
J une 1988.from ECMT I·F analyses ( dashed line ), DRl ~
]e' ·( thin lil1e ) and TI 'E T]C Y" ( thick line ).
In DRYJUN, th e rela ti vel y co ld 211.4 K surface tempera ture a t thc South Po le is m a inta ined la rge ly by ra di a ti\"C
heat loss as o utgoing long \,·a\"e ra di a ti o n, 111.3 \\. m ~,
exceecl s incoming 10ng wa\"C ra dia ti on by 50.2 \V m 2 By
co ntra t, a t the So uth Pole in \ VETJ U N, the rela ti\·ely wa rm
227.2 K surface tempera tu rc is influencecl by th e ba la nce a t
Downloaded from https:/www.cambridge.org/core. IP address: 88.99.165.207, on 16 Jun 2017 at 19:22:24, subject to the Cambridge Core terms of use, available at
https:/www.cambridge.org/core/terms. https://doi.org/10.1017/S0260305500014166
283
Uines ami'others: Evaluating Antarctic mesoscale simulations
the surface between outgoing radiation, 1495Win " and
incoming radiation, 149.5 W m ~. Heat loss into the colder
ice sheet maintains the inversion in the lower atmosphere
forWETJUN. The temperature difference between simulations is reversed in the upper troposphere, where WEXJTJN
is up to 16K colder than DRYJUN and up to 12 K colder
than the E C M W F analyses. Outgoing longwave radiation
from the tops of relatively dense clouds certainly contributed
to the low temperatures in the upper troposphere in WETJUN.
Evidence for surprisingly thick, persistent clouds over
the high interior is displayed in Figure 3, showing the average over longitude and time of vertically integrated watervapor and cloud mass in centimeters of liquid water equivalent depth f o r W E T J U N . North of 75 S, vertically integrated water vapor is slightly larger (by about 0.1)5 cm
than that of the ECMWFanalyses. Over interior Antarctica, water vapor is excessive in the model results, as
simulated water-vapor mass fraction is as much as double
that of the analyses for latitudes south of 75 S. The obvious
reason for the excess vapor is the increased saturation vapor
pressure with relatively warm temperature in the lower and
middle troposphere. Figure :J also shows that, over the
Southern Ocean, clouds mass is relatively small compared
to water-vapor mass. Near the South Pole, however, cloud
mass is the same order of magnitude as water-vapor mass.
The surprising and clearly spurious resull of WF-TJUN is
that the maximum cloud mass in a vertical column is not
located over the relatively warm northern sections of the
grid. Rather, the maximum is located over the elevated terrain of interior Antarctica.
Fig. .7. Liquid water equivalent depth (an) oftonally averaged wafer iw/wr and timid mass in a vertical column during
June 1988JOT ll'ETJl'.Y (water vapor, thick fine: chad
mass, thin line) ami the ECMWFanalyses (water vapor,
dashed line J.
'w"i
10
r
10
z
10
-a
10
kg kg'air
Fig. /. Vertical projUesofthe massfraction (kgkg ') of water
vapor /thiu line), cloud (thick tine) ami precipitation
(dashed line) at the Smith
MeforWETJUM
longwave radiative cooling at the top of the thick cloud layer
in the upper troposphere contributes to the cold upper troposphere seen in Figure 2. T h e moisture stored in falling ice
precipitation is at least an order of magnitude smaller than
that stored in cloud ice within the troposphere. The difference is two orders of magnitude at the model's lowest level.
Figure 1 also shows that between 300 and 200 hPa,
atmospheric moisture stored in clouds is one or two orders
of magnitude larger than that stored in water vapor. T h e
modeled cloud density within this regime appears to be
greatly excessive, although the actual (loud density is not
well known for this region. Stone (1993) estimated wintertime cloud-ice concentration to be M x 1(1 ' to fi x 10 " k g m l
from radiative measurements at the South Pole. Simplv In
dividing Stone's estimate by the atmospheric density, which
is close to I k g m at the top of ihe inversion, we can compare his numbers with the cloud-mass fraction displayed in
Figure t.This indicates that cloud ice fraction is an order of
magnitude too large in the model results. Furthermore.
Stone's clouds are concentrated in the relatively warm
middle troposphere,The cloud bases were located neat the
inversion top. and cloud tops were located between 300 to
280hPa. Furthermore, Bromwich (1988) notes thai clearsky precipitation tends to form in the relatively warm air
near the top of the inversion. T h e model's maxima in cloud
concentration in the lower and upper troposphere are apparently spuriouN. li is quite possible, however, that persistent, very thin cirrus occurs in the Antarctic upper
troposphere. T h e time variation of the modeled cloud concentration in the upper troposphere indicates that a cloud
layer over Antarctica appears in the upper troposphere
early in the simulation I Fig. 5a! and persists for the rest of
the month. The appearance of the cloud layer coincides
closely with falling temperatures near the upper troposphere i Fig. (>;, indicating a link through outgoing longwave
radiation.
figure 4 displays the vertical distribution of atmospheric
moisture at the South Pole for WETJUN. Based upon
extremes in the vapor and temperature profiles, it is conveThe high cloud concent ration seen within the boundarynient to separate the troposphere into three regimes: the
layer regime forWETJUN is also of questionable validity, as
upper troposphere, the middle troposphere and the boundStone 1993) determined that South Pole clouds normally
ary layer. Water-vapor concentration has minima at the
have their base near the top of the inversion. Equation [3]
relatively cold surface and upper troposphere, and a maxindicates that a sufficiently dense cloud is required for autoimum lor the relatively warm layer above the boundary
conversion to precipitation to occur, especially at very low
layer. Cloud-ice concentration, by contrast, is relatively high
temperatures. on
Observations,
suggest
that toprecipiDownloaded from https:/www.cambridge.org/core. IP address: 88.99.165.207,
16 Jun 2017however,
at 19:22:24,
subject
the
at the surface and in the upper troposphere. It appears that
tation can form without a visible cloud present.
Cambridge Core terms of use, available at https:/www.cambridge.org/core/terms.
https://doi.org/10.1017/S0260305500014166
284
Hinesand others: Eva/ua/ing Antarc/ic mesoscalesimu/atiolls
0.00012
a
0.00010
( Nl':;'
+...... _............ __ . .+.... _.......... _. . +/
"'I -'.7~;
0.00008 1-........................;.....__ ._..............c_....._...... _._ .......
,'l
,'l
...
.~ 0.00006 f-
--..........._... t-.. _. . ·__ ..... _;.. .............._. · . ·i~.. ·-....·-..........·+-·........· . . . . . "f ...}_.._-_... __...
i
""be
,
if
..>(
be
0.00004.I-... --................+- .. -............. . + -.... -...--.... -.......~-.- .....--..·_·_·+_·--..·J·· ..·/
..>(
,
,A~ !
'Ii-..·--
/ /' Ii
~ ~.j
0.00002
0.00000
0.00020
b
0.00015
...
. ~ 0.00010
""be
..>(
be
..>(
0.00005
O.OOO4·r--- - - , - -- - - , - - - -,--- --,-- ---,- - - - ,
C
0.0003'1-..·_ ............ __.. +··_....·__ ·_/
0.0001
4
H ence, precipitation is likely to form more easily th a n sugges ted by Equations (1)-(3),
These co nsidera ti ons inspired th ree se nsiti\'it y experiments that conT th e period from 0000 h UTC I Jun e 1988
to 0000 h lJTC on 7 Jun e 1988, Since the errors in cloud
mass fraction a rc so la rge, this 6 day period is long en o ug h
to establi sh th e de\'eloplllent of th e errors, To test the sens itivit y o f' the Antarctic simul a ti o ns to reduced clo ud nuclei
co ncen tra ti on . two simul a ti o ns were performed, In experi ment :\"1 , \\'hi ch assumes a pristine el1\'iro nmen t, th e nu c lei
conce ntra ti on is Illultipli ed by 0,01. In ex pc rim ent 1'\ 2,
which a lso ass umes the number o f' efTccti\'C ice-cloud nuclei
is limited , nu clei conce ntra tio n is multipli ed by 0,01 for
temperatu res a bO\'C 233.15 K a nd set a t 2,65 x 10 6 m :1
diyided by densit y [or temperatures at or < 233 .15 K. Thus
in N2. th e nucl ei conce11lration is eq ua l to or less than that of
:\11 for the same temperature la rger (sm a ll er) th an 233, 15 K.
Expcriment S3 is d esig n ed to redu ce th e excessi\'c cloud
co nce ntra ti o n compared to the \\'a ter-vapor concentratio n.
In thi s case, the nuclei co nce ntra ti o n is identi ca l to th a t of
1'\2 for th e sall1e tempe ra ture, Furthe rm ore. th e critical
cloud co ncentra ti on [or aULOeo l1\'Crs ion, qcril . is reduced to
th e sa wra ti o n-nlpo r concc ntra tion when it exceed s th a t
quantit y (sce Equation (3)),
Fig ure 5 displays th e clo ud co nce nt ra ti o n for WETJ :\",
:\11 , i\' 2 a nd S3 a t three m ode llC\'C ls: +, 12 a ndl 6 a t ab o ut 230
hP,!' 6+0 hPa a nd 23 III a bo\"(' g r ound le\'C I, resp ec tiyely. [o r
the South Po le. In th e upp er tropos ph e re (Fig. Sa ), the cloud
is co nside ra bl y thinner in S3, Thi s leads to a m o re reasonable tempera ture at thi s level (Fig. 6), The nuel ei co ncentrati on ill NI and ~ 2 is probably st ill too la rge, despite th e
red uc ti o n. Co nsequ entl y. the cloud s a re on ly slig htl y thinn er in the upper tropos phere a nd th e temperatures a ppea r
to be too cold, rn th e middle trop osp here, \\'here the clo ud
concen tra ti o n in \\' ETJ U 1\ ap pears to be more rcaso nab le
than a t hig hcr a nd lower le\'Cls, th e reduced nuclei co nccntra ti o n in NI , ~ 2 and S3 appea rs to imprQ\'C the res ults (Fig,
5 b ), Atll10dellcvel 12, n ear the to p o f'th e Ant arc tic sur[acc
inve rsio n, th e a\'erage cl o ud th ic kn ess is reduc ed in th e sensitivit y experim ents, and th e sy no pti c \'ar iabilit y can clea r ly
be see n. The clo ud co nce ntra ti o n o f':\I1 is nea r'" identica l to
th a t of N2 in Fig ure 5 b, Peri od ica ll y, this b 'e! b ecomes
June 1988
Fig. 5. Time el'olation cif cloud mass ,[raction ( kg kg ') al
modellel'e/s. a) -J ( ajJjJrorimale0' 230 hPa): b) 12 ( ajJ/JTOXil7lale[)' 6-J0 hPa ): and c) 16 ( ajJjJ1'O,\ill1ale[J' 23 m above the
s1l1jace ) JOT ' I'ET]L~ \ -( Ihick !lne ), Xl ( shal'l dashed line ),
V"2 ( long dashed fine) and S3 ( Ihin line).
210r------,------,------,------,-----~------,
205r .. - .. ·········· .... ; ·····.. ···-.. ··.. ·· .. ·.. ··i · .. ~~ .. · .. ;···· .. ·· .. ·-·· .. ··· .. ··......t ..·....· ~~ · ..
v
Th e probl ems a n sing from Equations (1)-(3) ca n b e
d em o nstra ted. According to Equati o n (I), fo r a temperature
o f 220 K a nd a density of 1 kg m :\ the nuclei co ncentrati o n
is 7,07 lO" kg I, This la rge number sugges ts th a t cloud ice
particl es individu a ll y will b e very sm a ll , which is un[a\'orable fo r au toeo nvers io n to precipitati o n, Initi a ti o n o f clo ud
part icles, acco rd i ng to Equa tion (2), \Vi 11 t hen proceed
rapid ly fo r sup ersat ura ti o n a nd a ny rea li sti c c lo ud-m ix in g
ra ti o. Th e crit ical cloud m ass fr ac ti o n, qnil , for a utoCO ll\'ersion is 6,65, which is obviously unreali st ic. f leteher (1962),
howe\ 'C r, no tes that the nucl ei co nce ntrati o n ca n \'ar y
severa l orders o f m ag nitude [or a g i\'e n tempe r at ure, The
Antarctic a tm ospher e has a low aerosol eo nte lll that wou ld
appea r to fayor less numero us, but large r cloud particles.
:><:
ID
200
r .......... · ......·_·+·_...... _........ ·_....+ ....·_......·........ _...,.._......_..............+.
.. .. i···.. ···············.. ··· ..i
\
Q)
b'o
Q)
o
195r.. ··....· ···· ....··_·..;.. ·.. ··.... _······- · j .. ·_··...... ······ ......··;····..·····_.. ····· ........·1 .. ·......_·..,
June 1988
Fig. 6, Time el'o/ltlion ciflelll/Jera/ure ( It") a/lllode//el'e/-Jjor
J I ET}L: \ . ( thick line ), ' \ 1 ( .rhor/ da.rherl/ine), ,\ "2 ( long
da shed line) and S3 ( thinlille).
Downloaded from https:/www.cambridge.org/core. IP address: 88.99.165.207, on 16 Jun 2017 at 19:22:24, subject to the Cambridge Core terms of use, available at
https:/www.cambridge.org/core/terms. https://doi.org/10.1017/S0260305500014166
285
J-Jines and others: Evaluating Antarctic mesoscale simulations
clo ud free. At level 16, the red uced nuclei concentrati on in
N I a nd N2 reduces th e cloud concent ra ti on by abo ut 40 %
from that ofWETJ UN (Fig. 5c ). The easier autoco nve rsion
of cloud to precipitation in S3 has a much more significant
effect at level 16, th a n the reduced nuclei co ncentrati on, as
the cloud thi ckness is abo ut a qu a rter of th at of N I and N2.
C uri ously, th e reduced clo ud thickness in N I, N2 a nd S3 for
the middl e a nd lower troposphere did not have a significant
impact on th e temperature fi eld for those laye rs. Th e cha nge
in nuclei co ncentrati on with temperature apparentl y does
not have a great effec t for temperatures bel ow 233. 15 K ,
since the r e~ ults for N I a nd N2 a re simil a r. These experiments help us unde rsta nd the deficiencies in th e moist-ph ysics
para meteri zati on, but more obse rvati ona l a nd modeling
wo rk is req uired before acc urate and physicall y realistic
parameterizations for the high so uthern latitude hydrology
can be implemented in numerica l models.
DISCUSSION AND CONCLUSION
Simulati ons of Anta rctic meteorology with a mod ifi ed version of the PSU/NCA R MM4 have demonstrated th a t the
moist-phys ics pa ra meteri zation is inadequ ate for the unique
winter climate of Anta rctica. Inadequ acies in simul ated
Anta rctic cloud s lead directly to errors in th e longwave radi ati ve calcu la tions. T he lack of cloud a nd clo ud physics
studi es in this ve ry cold a nd low-ae rosol environment limits
our abilit y to formul ate a nd ve rify para meterizati ons.
Nevertheless, we can begin to improve th e clo ud pa ra meterization after first identifying model defi ciencies. ro r the
simul ati on of June 1988 with the explicit moisture physics,
wc find that ove r the high interi or of Anta rctica th ere a rc
th ree different moist- p hysics regimes in the vc rtica l: the
ve r y cold upper troposphere, th e rela tively warm middle
troposphere a nd th e co ld bound ary laye r. D eficiencies for
these laye rs include the following: (1) there is excessive clo ud
ice in the upper trop osphere; (2) excess ive clo ud ice i. a lso
fo und in the inve rsio n near the ice surface; (3) temperatures
a rc too wa rm in th e lower troposphere; (4) temperatures are
LOO cold in th e upper troposph ere; a nd (5) downwa rd longwa" e radiatio n is excessive at the Earth's surface. Findings
(1) a nd (4) a rc directl y rela ted , as a re findings (2), (3) a nd
(5). Findings (1) and (3) m ay also be related through downwa rd longwave radiati on from high-level clo uds. It appears
tha t th e explicit clo ud pa ra meteri zation retains too much
wa ter substa nce in ice cloud s at ve ry low temperatures.
T he excessive ice clo uds p robabl y arise from th e co nve rsion term s between vapor, cloud a nd precipitation pa nicles,
as th e m eridional m oisture flu x towa rd s higher so uthern
latitudes is reasonablc (Hin es a nd oth ers, 1997). Th e la rge
variati on of para lTleteri zed condensati on nuclei with tempel-alUre indicates th at low temper ature clo ud s co nsist of
hi ghl y numerous and sma ll ice pa rticl es. The pa rameteri zed
autoconve rsion process, by which cloud pa rticles obtain suffi cient size to become fa lling precipitation, is not favo red at
low temperatures. Yet, observati ons of slow, but hydrologica ll y significant "clear-sky" precipitati on over the high
interior of A ntarctica suggests th at precipitati on pa rticles
ca n form under rea li stic conditi ons with out visible clo uds.
If present during episodes of clear-sky prec ipita tio n, clo ud s
must have ve ry low water content.
SUMMARY
Three se nsitivity ex periments we re performed with th e goal
of improving the moist-physics simul ations by reducing
cloud co ndensati on nuclei a nd increas ing autoconversion
of clo ud to precipi ta li on. A red uction of clo ud co ndensati o n
nuclei, whil e red ucing some of the excess cloud ice, appears
to be insufficient to so lve the difficulti es. A reducti on in the
excessive clo ud co ncentrati on in the upper troposphere
does, howe\'er, sig nifica ntly improvc th e simul ation of
upper tropospheric temperature. Th e res ults of this stud y
emphas ize the need for testing a nd development of physical
pa ramcterizations adapted espec ia lly for high southern la titudes in par ticul ar, a nd for th e pola r regions in ge neral.
ACKNOWLEDGEMENTS
This resea rch was suppo rted 111 pa rt by NASA grant
NAGW-27 18 to D. H . Bro mwich. Th e numerica l model was
prO\'ided byT. Pa rish at the Universit y of Wyo ming, and th e
radiation code fo r moist physics was provided by R. L eung
at the Pac ific Nor thwest Laboratory. Th e numerical simul ations were pet-form ed on th e C R AY Y-MP at th e Ohi o
Superco mputer Cenler, whi ch is support ed by th e Sta te of
Ohio a nd on th e C RAY Y-MP at NC AR, which is suppOl-ted by the Nati ona l Science Fo und ati on. This pap er i
co ntributi on 1026 of the BYI'd Polar R esearch Center.
REFERENCES
Anlhes, R. 1\., E. -Y. Hsie a nd Y. - H. !':.uo. 1987. Desai/Jlioll o/Ihe Pellll Slale/
. \'G~ R lIIesoscate model "fI'siOIl 4 (,If. \'4). Bo ulder, CO, Na ti ona l Centr r for
Atmospheric Resea rch. (i\CA R Techn ical i\ote T N-282 + STR.)
Bromwich, O. H. 1988. Snowfa ll in high southern latitudes. Ret,. Gen/)/9's.,
26 (1 ). 1'.9 168.
Bromwirh, D. H .. F. r-1. Robas ky, R . I. C ull ather a nd r-l. L. ,·an \\loe rt. 1995.
Th e a tmospheric hydrologic cycle O\'er the South ern Ocean a nd Antarctica fro III operat io na l nu merica l a na lyses. /\Jon. f Ifa/Itn Rn '., 123 (12),
3518- 3538.
Cerni , T. 1\. a nd T. A. Pa ris h. 198.J.. A rad iati vc m odel of the stable nocturna l bou ndary layer wi th application to [he polar night. J Clima/e AP/JI.
,1Ie/eoro/. , 23 (1 1), 1563 1572.
Chen, B. , D. H. Bromw ich, K. r-1. Hines a nd 'C . Pan. 1995. Simu lations or
the 1979 88 po la r clim ates by globa l clim ate model s. A 1111. (;Iarioi., 21,
83 90.
Dudhia, .J. 1989. Ku merica l study oreom'cetio n obsen 'ed d uring t he winter
monsoon experiment using a mesosca le two-di mensiona l model. J
A/lIlos. Sei., 46 (20), 3077 3107.
Fletcher, N. H. 1962. Thepl!ysicsq/raillciouds. Cambridge, Ca mbridge Uni ,'CI'sit )' Press.
Ca llee, H. 1995. Simul ation of' the mcsocyclonic act i"ity in th e Ross Sea,
A ntarc ti ca. M Oll. 11 ealher Rpu.. 123 (7).2031 2069.
Hin es, K . ~I " D. H. Bromwich a nd T. R. Pa rish. 1995. A mesoscaic modeling study of the atmosph eric circul ati on of high southern latitucies. JIolI.
Il ea/her Re1'.233.15
Hi nes, K. :,,1., D. H. Bromwich a nd Z. Liu. 1997. Combi ned CC r-1 and
mesosea lc m odel sim ula ti ons of An ta rcti c cl im ate. J Ceo/Jlrys. Res.,
102 (D 12), 13,747- 13)60.
I-Tsie. E. -'I'. a nd R . A. An thes. 1984. Sim ulations of frontogenesis in a moist
atJllosphcre using a llcrnati \'e pa ra mc ter izal io ns of concle ns:u io ll and
prec ipita ti on. J ;] /lI1os. Sri., 41, :l701 2716.
Pi nto, J. 0. a ndJ A. C urry. 1997. Role of radi ali" e tra nsfer in th e modelcd
mesoscalc de"elopmelll of sum me rtime Arctic stratus. J CeopI9's. Res.,
102 (01 2).13,861 13,872.
R a ma narh a n, v. , B. R . Barkstrom a nd E. F. Harri son. 1989. C lim ate a nd th e
Eart h's rad iati on budget. Ph),s. 7od(I)', 42 (5), 22 32.
Stone, R . S. 1993. Pro perties of austral wi nter clouds c1er iwd from rad iometr ic pro fi les at the SOllth Pole. J G'eoph),s. Res., 98 (D 7), 12,961 12.971.
T zeng, R. -Y., D. H. BroJ11wich, T. R. Pa rish and B. C hen . 199.J.. Ne AR
CC r-f 2 simul ati on of the modern Anta rcti c cl im ate. J Ceoph)'S. Res.,
99 ( DII ), 23,131-23,1.J.8.
Downloaded from https:/www.cambridge.org/core. IP address: 88.99.165.207, on 16 Jun 2017 at 19:22:24, subject to the Cambridge Core terms of use, available at
286
https:/www.cambridge.org/core/terms.
https://doi.org/10.1017/S0260305500014166