Downslope föhn winds over the Antarctic Peninsula and their effect

29 Pages·2014·8.75 MB·English

Atmos.Chem.Phys.,14,9481–9509,2014
www.atmos-chem-phys.net/14/9481/2014/
doi:10.5194/acp-14-9481-2014
©Author(s)2014.CCAttribution3.0License.
Downslope föhn winds over the Antarctic Peninsula and their effect
on the Larsen ice shelves
D.P.Grosvenor1,*,J.C.King2,T.W.Choularton1,andT.Lachlan-Cope2
1UniversityofManchester,CentreforAtmosphericScience,SEAES,Manchester,UK
2BritishAntarcticSurvey,Cambridge,UK
*nowat:SchoolofEarthandEnvironment,UniversityofLeeds,Leeds,UK
Correspondenceto:D.P.Grosvenor([email protected])
Received:3January2014–PublishedinAtmos.Chem.Phys.Discuss.:5March2014
Revised:10July2014–Accepted:4August2014–Published:16September2014
Abstract. Mesoscale model simulations are presented of a Thesurfaceenergybudgetofthemodelduringthemelting
westerly föhn event over the Antarctic Peninsula mountain periodsshowedthatthenetdownwellingshort-wavesurface
ridge and onto the Larsen C ice shelf, just south of the re- flux was the largest contributor to the melting energy, indi-
cently collapsed Larsen B ice shelf. Aircraft observations cating that the cloud clearing effect of föhn events is likely
showed the presence of föhn jets descending near the ice tobethemostimportantfactorforincreasedmeltingrelative
shelf surface with maximum wind speeds at 250–350m in to non-föhn days. The results also indicate that the warmth
height. Surface flux measurements suggested that melting of the föhn jets through sensible heat flux (“SH”) may not
was occurring. Simulated profiles of wind speed, tempera- becriticalincausingmeltingbeyondboundarylayerstabil-
ture and wind direction were very similar to the observa- isation effects (which may help to prevent cloud cover and
tions. However, the good match only occurred at a model suppress loss of heat by convection) and are actually can-
timecorrespondingto∼9hbeforetheaircraftobservations celled by latent heat flux (“LH”) effects (snow ablation). It
were made since the model föhn jets died down after this. was found that ground heat flux (“GRD”) was likely to be
Thiswasdespitethefactthatthemodelwasnudgedtowards an important factor when considering the changing surface
analysisforheightsgreaterthan∼1.15kmabovethesurface. energybudgetforthesouthernregionsoftheiceshelfasthe
Timing issues aside, the otherwise good comparison be- climatewarms.
tween the model and observations gave confidence that the
modelflowstructurewassimilartothatinreality.Detailsof
the model jet structure are explored and discussed and are
found to have ramifications for the placement of automatic 1 Introduction
weather station (AWS) stations on the ice shelf in order to
detect föhn flow. Cross sections of the flow are also exam- During the last 50–60years near-surface temperatures over
ined and were found to compare well to the aircraft mea- the Antarctic Peninsula (hereafter referred to as AP) re-
surements.Gravitywavebreakingabovethemountaincrest gion have increased more rapidly than anywhere else in
likely created a situation similar to hydraulic flow and al- the Southern Hemisphere, at several times the global av-
lowedföhnflowandiceshelfsurfacewarmingtooccurde- erage rate (Vaughan et al., 2003). One manned station on
spite strong upwind blocking, which in previous studies of the west side of the Peninsula (Vernadsky, formerly Fara-
this region has generally not been considered. Our results day)measuredameannear-surfacewarmingof2.94◦Cbe-
thereforesuggestthatreducedupwindblocking,duetowind tween 1951 and 2004, significant at the <1% level, com-
speedincreasesorstabilitydecreases,mightnotresultinan pared to a global average of 0.52◦C over the same period
increasedlikelihoodofföhneventsovertheAntarcticPenin- (Marshall et al., 2006). Vaughan et al. (2003) estimated the
sula,aspreviouslysuggested. mean warming trend for several of the Peninsula stations
to be 3.7±1.6◦C(century)−1 and suggested that current
PublishedbyCopernicusPublicationsonbehalfoftheEuropeanGeosciencesUnion.
9482 D.P.Grosvenoretal.:DownslopewindsandtheLarseniceshelf
o W tion of the Larsen B ice shelf in February to March 2002
120 60o when an area of 3200km2 was lost (Scambos et al., 2004).
Domain 1 W The summer warming is especially important with respect
to ice shelf disintegration since this is the season when the
vast majority of surface melting on the ice shelves occurs.
Domain 2
Crevassepropagationduetotheweightofaccumulatedmelt
Domain 3
W wateriscurrentlythoughttohavebeenthemajorfactorinthe
o
180 o84S o72S o60S o48S o36S o24S o 0 B2ar0rooM0ue2nakbdresr,het2haa0ekll0-Pu5eep)tn.,aianls.swu(l2ea0ll0(Sa6sc)aingmatbvhoeesbeervetiadakel.-n,uc2pe0o0tfh0oa,tth2ae0tr0tri4ci;beuvstahenseldvtehenes
anomaloussummerandautumnwarmingontheeastsideto
changesintheSouthernHemisphere(SH)annularmode,or
SAM, which is the principle mode of variability in the SH.
120o A higher SAM index is associated with stronger westerly
E o E winds impacting on the Antarctic Peninsula. The SAM in-
60
dexincreasedbetween1965and2000withmorestatistically
Figure1.LocationsoftheWRF(WeatherResearch&Forecasting) significantandmuchlargerincreasesobservedintheautumn
modeldomainsusedinthisstudy.Thesquaredomainshadsidesof and summer seasons. The increase in SAM index has been
length7470,3000and840kmwithhorizontalresolutionsof30,7.5 attributedtoozoneloss(e.g.ThompsonandSolomon,2002;
and1.875kmresolutionfordomains1,2and3,respectively. Gillett and Thompson, 2003), or greenhouse gas concentra-
tion increases (Kushner et al., 2001; Cai et al., 2003). Mar-
shalletal.(2006)suggestedthatthestrongersummerwest-
temperatures are unprecedented in the context of the past erly winds associated with an increasing SAM index could
1800years for this region. It has recently been suggested, lead to a higher frequency of penetration of warm air onto
throughacombinationofinfra-redsatellitetemperaturemea- theeastsideoftheAntarcticPeninsula,leadingtoenhanced
surementsandstationdata,thatthestrongwarmingtrendex- warminginthisregion.
tends to the whole of West Antarctica, which is estimated Warmanddryairflowsdowntheleeslopesofamountain
to have exceeded 0.1◦C(decade)−1 over the past 50years are given various names around the world, the most com-
(Steig et al., 2009). In contrast, the same work and others monly known being “föhn” (when it occurs in the Alps),
(e.g. Turner et al., 2005) estimated a small and statistically “Chinook” (Rocky Mountains, North America) or “Zonda”
insignificanttrendforthelargerareaofEastAntarcticaover (ArgentineAndes).Weusetheterm“föhn”inthispaper.The
asimilarperiod. warming on the downwind/lee side relative to a position on
There is evidence that in the Antarctic Peninsula region, theupwindsideatthesamealtitudeoccursduetolatentheat
the seasonal pattern of local warming has varied with loca- releaseontheupslope(western)side(ifcombinedwithpre-
tion.ThePeninsulaconsistsofahigh,narrowmountainridge cipitationlosses)and/oradiabaticdescentofairfromupper-
that reaches over 2km in altitude and runs for a length of levelsdownwardstowardsthesurface.Bothofthesemecha-
around 1500km (see Figs. 1, 3 and 4). Its length is orien- nismswillalsotendtomakethedownwindairdrierthanthat
tated approximately from north to south and it is bounded atanequivalentaltitudeupstream.
tothesouthbytheAntarcticcontinent.Thehighmountains GiventhelikelihoodthatsuchflowsovertheAPhavein-
provide a climatic barrier between the warmer oceanic air creased in frequency over the past 50years in response to
of the west and the cold continental air of the east where a strengthening of the prevailing westerly winds, and the
annual mean temperatures are 5–10◦C colder at compara- possibility of a connection with Larsen ice shelf break-up,
blelatitudes(KingandTurner,1997).Therearefewmanned knowledgeofthedetailsoftheseflowsisimportantinorder
stations on the east side of the Peninsula, though, and they tounderstandtheconditionsinwhichtheyform,thedegree
are all close to the northern tip of the Peninsula. Whether ofwarmingtheyarelikelytoprovidetotheeastside,andthe
they reflect temperatures further south is therefore not cer- consequencesoftheflowsforiceshelfmeltrateandstabil-
tain.Thesestationshaveshownsimilarannualwarmingrates ity.However,littleisknownaboutthesedetailsinthecontext
tothoseonthewesternside(Vaughanetal.,2003;Marshall oftheAntarcticPeninsula,exceptfortheveryrecentresults
etal.,2006).However,theyrecordedamuchstrongerwarm- of Elvidge et al. (2014). In the latter some simulations of
ingtrendintheseasonsofAustralsummerandautumnthan föhnflowandcomparisonstoaircraftobservationsforthree
those on the west side (Marshall et al., 2006). The summer different types of flow regime were presented following the
trendinparticularwashighlystatisticallysignificant. OFCAP(OrographicFlowsandtheClimateoftheAntarctic
An indication of rising temperatures on the east side at Peninsula) field campaign. These results are discussed fur-
moresoutherlylatitudescamefromthedramaticdisintegra- ther in Sects. 3.6.2 and 4.3.4. Our paper will focus on the
Atmos.Chem.Phys.,14,9481–9509,2014 www.atmos-chem-phys.net/14/9481/2014/
D.P.Grosvenoretal.:DownslopewindsandtheLarseniceshelf 9483
(a) Pressure at 2.3 km for 06 Jan 06:00 UTC (hPa) (b) Pressure at 2.3 km for 07 Jan 00:00 UTC (hPa)
y (km)11225050500000000000−−101515−−11−07−−−−748−1006678426−1000−7−95−66−9−−5−90−0658880−80−−90−7805−82−−66−−−057565−547−852−8−050−−7784−−−766−−067270−64−5−4−67−0−5−06558−6−8365−5−6504−5−572−−67−4−035−0070−66−−6−6822−−56455−−64−0−−−605505286−−−5−3645055 11y (km)00 mm//1122ss5050500000000000−−110150−−7−−886−−10070060−100−8−−87 7777777777777−89256−9042223333344444−−9−58575791357913579−26−−66−960−24−5−7087−840−8−565−55−75−6−07−8−58−0−6−76−06−077860−−4752−−6−−−657−−3−5567−64560−42−45054−855−60−70−70−68−5−−−66−5260255−−6−6340−56−−46520−−−5−−3−505−545586040 1100 mm//ss 777777777777722233333444445791357913579
0 500 1000 x1 (5k0m0) 2000 2500 0 500 1000 x1 (5k0m0) 2000 2500 hPa
Figure2.Pressureandwindvectorsat2.3kma.s.l.forWRFdomain2at(a)06:00UTC,6Januaryand(b)00:00UTC,7January.
Figure 3. The area of the second WRF domain (see Fig. 1) showing the topography height (coloured contours) and various landmarks,
includingtheRotheraBASresearchbase(whitedotwithredcrossinside).Theaircraftflighttrackisindicatedbyawhiteline.
simulationofadifferentföhneventovertheAntarcticPenin- served by an instrumented aircraft and was simulated using
sula, which was characterised by observations from an in- a high-resolution regional atmospheric model. In this sec-
strumentedaircraft. tionwebrieflydescribetheaircraftandthemodellingsystem
Thebreakdownofthesectionsofthepaperisasfollows: used.
Sect.2describestheaircraftdatausedandtheset-upforthe
simulation; results regarding the meteorology, structure and 2.1 Aircraftobservations
thermodynamicsofthemodelledjetsandhowtheycompare
Observations were made by an instrumented DHC6 Twin
toobservationsaredescribedinSect.3;Sect.4describesthe
Otter aircraft operated by the British Antarctic Survey. The
surfaceenergybalanceresultsandsimulatedamountofsur-
aircraft instrumentation is described by King et al. (2008).
face ice melting; and Sect. 5 provides discussions and con-
Briefly, the aircraft recorded basic meteorological variables
clusions.
(pressure, temperature, frost point temperature, wind speed
anddirection)atflightlevel.Inaddition,aremotemeasure-
2 Dataandmethods mentofsurfacetemperaturewasavailablefromadownward-
pointing infrared thermometer and upwelling and down-
Thefocusofthispaperwillbeaföhneventthatoccurredon wellinglong-andshort-waveradiativefluxesweremeasured
the east side of the Antarctic Peninsula on 6 January 2006 byaircraft-mountedpyrgeometersandsolarimeters.
whentheairflowwasfromwesttoeast.Thiseventwasob-
www.atmos-chem-phys.net/14/9481/2014/ Atmos.Chem.Phys.,14,9481–9509,2014
9484 D.P.Grosvenoretal.:DownslopewindsandtheLarseniceshelf
780000 −66 −78 −65.5−−7665 −64 −72 −64.5−63−7.50 −68−63 −66 −62.5 −6−264 22680000
2400
600 −66.5 −74 Adelaide−65 −64 −63.5 2200
−67 −66 Isla−n6d5.5 2000
500 −67.5 Rothera −64.5−62 1800
1600
y (km)400 −76I−cW−6e698i lS.k5ihneslf−68 Ale−I7s2xlaannddMeArRGU−7E0RITE −68−67.5 −−−666676.5 −64 −66 −65.5−−6605 11240000
300 BAY Larsen B 1000
1200000 −72−74−71.−5−7070 −6−968.5−71 −−6760.5 −−66−4968.5−70−68−69.5 −62−60Larsen C− 6I8c.5e −S67Wh−.55e8−El6fS8D−E−6D76A6E.5LL−56 02468 0000 0000
0 200 400 600 800 Topography
x (km) height (m)
Figure4.AsforFig.3exceptforthethirdWRFdomain(seeFig.1)andthatheretheblacktopographycontoursareevery500m.The
aircraftflighttrackisindicatedbythewhitedottedlinesurroundedbycolouredcircles,whichshowtheaircraftaltitudeonthesamecolour
scaleasthetopography.
Figure 4 shows the flight track of the aircraft with the the molecular viscosity for momentum and the friction ve-
aircraft altitude shown in colour. The aircraft took off from locity, following Zilitinkevich (1995); for the land surface
Rothera Research Station (see Fig. 4) at 19:20UTC, 6 Jan- model,thefour-layerunifiedNoahschemewasselected.As
uary and headed east. It traversed the Antarctic Peninsula describedinHinesandBromwich(2008),thelatterwasmod-
ridge at 3000m in altitude until the aircraft was ∼170km ifiedtodealwithdeepsnowpackandthedensity,heatcapac-
downwind of the ridge crest. Then, at 20:15UTC, the air- ity and heat conductivity of the snowpack are based upon
craftdescendedtowardsthesurfaceoftheLarsenCiceshelf observationsofAntarcticsnowfirn.
over a horizontal distance of ∼10km where it performed Three grid nests were used of horizontal resolution 30,
some low-level flight legs, which will be discussed later 7.5 and 1.875km for the outer, middle and inner nests, re-
(AppendixA).At22:00UTCitmadeanotherascentwithin spectively.Theinnernestis840km×840km,centredonthe
∼10km of the descent profile and returned back over the areawheretheaircraftflew.Figure1showsthenestpositions
ridgealongasimilarpath.ThereaderisalsoreferredtoKing andsizesrelativetothePeninsula.Thetwolowestresolution
etal.(2008)forfurtherinformationonthiscasestudy. nests used the Kain–Fritsch convection scheme, which pa-
rameterisesdeepandshallowconvection,whereastheinner
2.2 WRFmodellingintroduction nest did not use a convection parameterisation. There were
81 vertical levels specified and vertical resolution generally
ThemodelusedisaversionoftheWRF(WeatherResearch decreased with height. On average, the vertical resolution
& Forecasting) mesoscale model (Skamarock and Klemp, started at ∼27m near the surface and was relaxed to 240–
2008) that has been specially modified for use in polar re- 250mbythetimethemid-tropospherewasreached,andthen
gions by researchers at the Bryd Polar Research Center remainedatthisvaluethroughoutmostoftherestofthetro-
(Hines and Bromwich, 2008; Bromwich et al., 2009; Wil- posphere.
son et al., 2011; Hines et al., 2011) through improvements Themodelwasinitialisedwithandreceivedlateralbound-
intherepresentationofthepolarsurface;theWRFparame- ary information from ECMWF operational analysis data,
terisationoptionsthatarenowlistedwereselectedaccording whichfortheperiodinquestionwasavailableat0.5◦×0.5◦
to these studies and the reader is referred there for further horizontal resolution with 61 vertical levels. The simula-
details and for justifications for these choices: the rapid ra- tion was started at 00:00UTC, 5 January 2006 and ran un-
diative transfer model (RRTM) was selected for long-wave til 00:00UTC, 8 January 2006. It was decided to perform
radiation (“LW”) and the Goddard scheme for short-wave nudging on all model nests so that the model fields of hor-
radiation (“SW”); the Mellor–Yamada–Janjic´ Turbulent Ki- izontal wind, temperature and vapour mixing ratio are con-
neticEnergy(TKE)schemewasusedfortheboundarylayer stantlybeingmovedtowardstheabovementionedECMWF
optioninconjunctionwiththeJanjic´ Etaschemeforthesur- analysis fields. This was done since otherwise it was found
facelayer(Janjic´,2002),whichisbasedonMonin–Obukhov that the fields drifted away from the analysis, most likely
similarity theory, but with moisture and thermal roughness as a result of the combination of rapidly changing analysis
lengthsthatscalewiththoseformomentumasafunctionof
Atmos.Chem.Phys.,14,9481–9509,2014 www.atmos-chem-phys.net/14/9481/2014/
D.P.Grosvenoretal.:DownslopewindsandtheLarseniceshelf 9485
fields,thelargedomainsizesandthefairlylongtimeperiod and 3000m. This period is associated with föhn flow that
of the simulation. The nudging was only applied above the willbedescribedinmoredetaillater.
10th vertical model level, which corresponds to a height of Figure 2b shows the situation at 00:00UTC, 7 January.
∼1.15km above the terrain. The relaxation timescale was Thesplittingofthelow-pressuresystemoverSAmericacan
t =55.6min.Iftherearenootherforcings,thenamodel beseeninthisfigure.Asthissystemmovedtotheeastside
relax
variable,q(t),wouldchangeduetothenudgingaccordingto of S America it started to affect the low-pressure system to
thefollowing: theSEoftheAPcausingittowidenandmoveslightlyeast-
wards.Atthistimethewindsassociatedwiththelattersys-
q(t)=q(0)+(1−e−t/trelax)(qtarget−q(0)). (1) tem at the western side of the AP are more southerly and
no longer impact the AP in a direction perpendicular to the
Here,t isthetimeinminutessincethestartoftherelaxation ridge. The overall change in direction is around 45◦. In ad-
andq isthetargetanalysisvalue.
target dition,thewindstherealsoweakenafter00:00UTC,7Jan-
uary. After 15:00UTC, 6 January, the föhn flow started to
die down and the changes in the wind direction and speed
3 Thethermodynamicsandmeteorologyofthe
justdescribedarelikelythemainreasonsforthis.
föhnflow
Further details about the properties of the upstream flow
3.1 Thesynopticsituation (windspeed,Froudenumber,stabilityprofiles,etc.),itsevo-
lution and its relationship to the föhn flow are described in
Thegeneralsynopticsituationduringtheperiodofthesimu- Sect.3.6.
lationwasdominatedbycircumpolarflowaroundAntarctica,
which carried a succession of low-pressure systems around 3.2 Aircraftobservationsoftheföhnjet
the pole. At the start of the simulation (05:00UTC, 5 Jan-
uary), two such systems were located to the west and east The flight track of the aircraft was described in Sect. 2.1;
of the southern tip of S America with surface low-pressure we now discuss the observations that were made during the
centresatapproximately52.5◦Sinlatitudeandatlongitudes flight. During the initial ascent (close to Rothera) the mea-
of 100 and 40◦W, respectively. As the systems progressed sured wind direction between ∼1700 and 3000m varied
eastwards,theedgeofthewesternmostsystemstartedtoim- from ∼225 to 250◦ and the wind speed was between 8.5
pactontothewestcoastofSAmericabyaround12:00UTC, and12ms−1 (notshown).Thus,theanalysiswinds(Fig.2)
5 January and by 06:00UTC, 6 January part of the system were in a similar direction to, but were a little weaker than
waslocatedovertheeasterncoastofSAmerica.Thesurface those measured. Wind profile data was not available below
pressurefieldofthesystemhadsplitintotwoalmostequally 1700montheascentduetoinstrumentmalfunction.Asthe
sized low-pressure systems by 09:00UTC, 7 January on ei- aircraftcrossedthePeninsula,headingfromwesttoeast,the
thersideofSAmerica.Bytheendofthesimulationthebulk winddirectionremainedwesterlytosouthwesterlyrevealing
of the system was on the east side and had travelled south- thatcrossridgewindsprevailedatthistime.
wardsslightly,lyingjusttotheNEofthetipoftheAntarctic When the aircraft descended towards the Larsen C ice
Peninsula(centredat57◦S,50◦W). shelf a strong low-level wind jet was observed. This is
Figure 2a shows the WRF pressure field at a height of showninFig.5alabelledas“Aircraftdescent”.Windspeed
2.3kmat06:00UTConthemorningoftheaircraftflighton peakedat15ms−1 250mabovethesurface,whilethewind
6January.ThisisverysimilartotheECMWFanalysispres- direction changed quite sharply from being approximately
sure field at the same height. This height is just above the southwesterly to westerly (245–265◦) at 800–3000m to-
maximum height of the ridge and therefore the wind at this wardsasoutherlydirectionattheheightofthejetmaximum
levelwilllikelybeimportantindeterminingthecrossPenin- (Fig.5b).Below,thejetwinddirectionsharplychangedback
sulaflow.Thewesternlow-pressuresystemcanbeseentothe tobecomealmostwesterlyagainclosetothesurface.
north-westoftheAP.Ahighpressureridgetothewestofthe Warmairtemperatures(Fig.5c)wereobservedataround
PeninsulaextendedeastandnorthbeyondthePeninsulatip. thesameheightasthejetwindspeedmaximumwithamax-
It separated a large low-pressure system centred south-east imumof4.6◦Cat283mabovethesurface.Thepresenceof
ofthePeninsulafromtwolow-pressuresystemsoffthewest thiswarmaircausedastrongtemperatureinversionabovethe
andeastcoastsofthesoutherntipofSAmericathatwerede- icesurface.Thesurfaceitselfremainedcloseto0◦C,ascon-
scribedabove.Atthelocationoftheflight(seeFigs.3and4), firmed by the surface infrared aircraft measurements (King
the western branch of the clockwise circulation of the low- et al., 2008). King et al. (2008) also showed that the down-
pressuresystemtothesouth-eastisimpactingontothewest wind air had a considerably higher potential and equivalent
side of the Peninsula with flow that is almost perpendicular potential temperature and was drier than that at equivalent
across the ridge. The analysis winds over the ridge are ap- altitudes on the upwind side. This indicates either adiabatic
proximatelysouthwesterly(240–245◦)andhavespeedsthat warming due to the descent of dry air that originated from
vary between 5 and 10ms−1 between the ridge top height above the mountain, or diabatic warming of air that came
www.atmos-chem-phys.net/14/9481/2014/ Atmos.Chem.Phys.,14,9481–9509,2014
9486 D.P.Grosvenoretal.:DownslopewindsandtheLarseniceshelf
Figure5.Profilesfromtheaircraftobservationsandfromdomain3oftheWRFmodel.Aircraftprofilesweretakenduringthedescentdown
totheiceshelfbetween20:14and20:24UTC,6Januaryandduringtheascentawayfromtheiceshelfbeforethejourneybacktobasefrom
22:00to22:09UTC.Modelprofilesarefor12:00UTC,6Januaryfromvariouslocations(labeledwithlettersinthelegend).SeeFig.8bfor
amapofthelocationsoftheseprofiles.Theprofilesare(a)windspeed,(b)winddirectionand(c)temperature.
frombelowthemountainontheupwindsideandexperienced oftheascenthadrotatedtowardsamoresoutherlydirection
latentheatwarmingduetoiceorliquidformationanddrying (Fig.5b).
byprecipitationloss. Thus, the observational evidence suggests that a cross
Figure 7 shows MODIS images over the peninsula ridge ridgeflowgeneratedaföhneventthatproducedstrongwind
from13:00UTC,6January.Figure7bshowsthatmostofthe jets and temperatures higher than 0◦C above the ice shelf
LarsenCiceshelfwasrelativelycloudfreesincetheicesur- surface to the east of the mountain barrier. Such tempera-
face shows up as red, whereas cloud shows as white. There tures could promote melting of the ice surface; the issue of
is cloud upwind; however, Fig. 7a demonstrates that this is ice melting, including results on the amount of melting at
quitethin.Alinearbandofthickercloudcanbeseenorien- differentlocationsontheiceshelfaspredictedbytheWRF
tatedalongtheridgecrestthatisassociatedwiththemoun- model and the likely contributions from different processes
tainwave,althoughthereisagapinthiscloudjustnorthof arediscussedlaterinSect.4.
AdelaideIslandandRothera.Theseobservationssuggestthat
latent heating through condensation followed by precipita- 3.3 Descriptionofthesimulatedföhnjets
tionremovalisnotabigcontributortothedownwindwarm-
Figures6and8showplanviewsofthehorizontalwindfields
inginthiscase.
onthefourthverticalmodellevelfortheinnerdomainofthe
Ontheascentbeforethereturnbacktobaseat22:00UTC
simulationatvarioustimes.Theactualheightrepresentedby
thewindjetwasagainobserved(Fig.5a,labelledas“Aircraft
ascent2”) but with a lower maximum speed of 12.4ms−1 thismodellevelvarieswithpositioninthedomaindepending
ontheterrainheightandpressureleveldistortion.However,
andtheheightofthismaximumhadrisenfrom250to345m
overtheiceshelfthemodellevelheightisapproximatelyuni-
abovetheiceshelfsurface.Thewinddirectionattheheight
format293mabovethesurface.Thisheightisclosetothat
of the maximum was southerly, as was the case on the de-
atwhichthemaximumwindspeedwasobservedduringthe
scent. However, above here (between 600 and 2000m) the
initialaircraftdescentovertheiceshelf(250mabovetheice
wind was closer to westerly on the descent, but by the time
shelf),andalsoduringthefinalascent(350mabovetheice
Atmos.Chem.Phys.,14,9481–9509,2014 www.atmos-chem-phys.net/14/9481/2014/
D.P.Grosvenoretal.:DownslopewindsandtheLarseniceshelf 9487
05 Jan 00:00 UTC Islandontheupwindsideoftheflow.Inaddition,thewinds
800 −−7865.5−65−76 −64.5−−7264 −63−.750 −68−63 −66 −62−−.65624 1155 mm//ss 1178 itmenmsiefydiinatge.lByyab0o3v:0e0thUeTrCid,g6eJaonfutahrey,Pjeentsinasgualianwstearreteadlstooainp--
700 −66 16 pear at several locations over the Larsen C ice shelf at the
15
600 −67.5−67−66.5−74 −−6665.5−65 −64.5−64 −63.5 111234 ejeatsstehranvfeofootromfetdh,etrhidegeed.gAets0o9f:0w0hUicThCha(Fveigp.r8oag),rethssreedemeaasitn-
500 11 wardsbyaround100km.Thesearemarkedasjets1,2and3
y (km)400−76−68.5 −68 −72 −70 −67−6.85−67−66.5−66 −64 −62 −−6665.5−−6605 8910 ainndthwesielljbeetsreaftetrhriesdhteoigahstsurecahcfhreodmunpotwo1o8n.mTsh−e1w.indspeeds
1230000000−7−2−7−−4−77672−1907...5155−70−70−68 −6−966 −−−666489..55−−6270−68−60 −B69A−5−8A6−86.W75.5−5S6−6−7−6866.5−54 1234567 lo(tohsfeweAteh-tflewFteeivPogree.tmlnh8siiobnoss)sut.tutihTlmnaeoherariltstythhtnehweorenijrnesthdjtoeswsutsstsaht,traaedjrrrettneptstdreoo1ntgdodreaevomnsefdsoltiovho2peen,atnLooloaofrrmnttshhgeeewnrthgjaCeeertdseitscoacesgwatesuhshtshiieledesldefrt
0 200 400 600 800 m s-1 by 12:00UTC, 6 January (Fig. 8b). By this time the jets
x (km)
reachedalmostasfareastasthelocationswheretheaircraft
Figure 6. Domain 3 horizontal wind speed and wind vectors on observedthestrongjetonthedescentandascentat20:23and
the fourth vertical model level at the beginning of the simulation 22:01UTC(labeledAandB,respectively,inFig.8).
at00:00UTC,5January.Thisthereforerepresentstheinitialwind
fieldasinterpolatedfromtheECWMFanalysis.Markedonhereis 3.3.1 TheinfluenceoftheCorioliseffect
theflightpathoftheaircraft(whitedottedline).Themarkedloca-
tion, A, is where the maximum wind speed was measured during GiventhehighlatitudeofthislocationafairlystrongCorio-
theaircraft’sdescentontotheiceshelf.LocationBistheposition
liseffectisexpectedthatwouldturnwindstotheleft.Since
of the maximum wind speed during the ascent from the ice shelf
the modelled jets were fairly strong it seems feasible that
beforeflyingbackoverthepeninsula.Alsomarkedisthelocation
their movement northwards could have been due to this ef-
oftheautomaticweatherstation(AWS).
fect. To examine this possibility, the surface pressure field
andthewindvectorsatthefourthmodellevelat15:00UTC
are shown in Fig. 9a. At this time there is a small low-
shelf;seeFig.5).Theformerwasobservedat20:23UTCat pressuresystemcentredneartheeasternedgeoftheiceshelf
thelocationmarkedaspointAinFigs.6and8andthelatter (x=578, y=216km) that has a fairly weak cyclonic cir-
at22:01UTConly10kmfromthedescentmaxima(labeled culation associated with it. It is clear that at the locations
pointB). on the northern part of the ice shelf, where the jet speed
The initial conditions for the WRF run (Fig. 6), which is the greatest, the jets are turning northwards (to the left
were taken from the European Centre for Medium-Range in the sense of the jet direction) across the direction of the
WeatherForecasts(ECMWF)analysisat00:00UTC,5Jan- isobars, suggesting that the influence of the Coriolis effect
uary, show moderate winds on this model level of up to is dominating there. Using the pressure gradient between
10.9ms−1 in the form of a fairly wide jet that covers ap- x=469km, y=159km and the centre of the low-pressure
proximatelythesamelatituderangeasthegapinthetopog- system,thepressuregradientaccelerationiscalculatedtobe
raphybetweenthehighterrainofAlexanderIslandandAde- 6×10−4ms−2,whereastheCoriolisaccelerationcalculated
laideIsland.ThejetstartsattheeasternfootofthePeninsula using the wind speed of the southernmost jet (14ms−1, jet
mountains and continues past the edge of the Larsen C ice 3)atx=450,y=230kmis1.9×10−3ms−2,whichisover
shelf and beyond the edge of domain 3. This suggests that threetimeslargerthanthepressuregradientacceleration.
the ECMWF analysis has some ability to resolve the föhn From15:00UTConwards,thewindspeedsofmostofthe
flow in this region and that the föhn may have been active modelledjetsstartedtoreduceinintensity.Thesurfacepres-
before00:00UTC,5January.However,theresolutionofthe sure field evolved such that the low-pressure system moves
ECMWFanalysismodelislikelytobetoocoarsetoresolve east and by 21:00UTC, 6 January its centre was located
alotofthedetailsofthetopographyandtheflow. beyond the edge of the ice shelf (Fig. 9b). The associated
As the high-resolution WRF model began to spin up and pressuregradienthasincreasedslightlyandthewindsinthe
evolve it started to resolve this single large jet into smaller southern half of the ice shelf are now stronger and directed
moreintensejetsatvariouslocationsalongtheeasternfootof fromthesouthacrossmostofthathalfoftheiceshelf.The
the mountains (not shown). However, these jets were short- remnants of the jets continue to move northwards, which is
livedandby21:00UTC,5Januarythewindswererelatively probably due to the influence of the southerly wind driven
calmovertheiceshelf.Atthistime,though,low-levelwinds by the pressure gradient, since at this point the Coriolis ef-
that were directed towards the Peninsula were starting to fect is likely secondary over most regions. For example, at
build up around the base of the northern part of Alexander the location of jet 3 used above (x=450, y=230km), the
www.atmos-chem-phys.net/14/9481/2014/ Atmos.Chem.Phys.,14,9481–9509,2014
9488 D.P.Grosvenoretal.:DownslopewindsandtheLarseniceshelf
(a) 1-4-3 (visible) image (b) 3-6-7 image
Figure 7. MODIS images over the Antarctic Peninsula region from 13:00UTC, 6 January. (a) shows the visible image (bands 1, 4 and
3 used for red (R), green (G) and blue (B), respectively). (b) shows a false colour image using, respectively, bands 3, 6 and 7 for RGB.
In (b) ice covered land shows up as red, whereas cloud shows up as white. The image is orientated approximately with north at the top
andsouthatthebottom.Theoutlineoftheiceshelf,theicecoveredlandandseaicetotheeastoftheiceshelfcanbediscernedin(a)
–seeFig.4toaididentification.(b)demonstratesthatmostoftheLarsenCiceshelfwasrelativelycloudfree.(a)showsthatthecloud
upwind(west)oftheridgeisquitethin,whereasmuchthickercloudispresentalongtheridgecrest(exceptinthecentralportionofthe
ridgejustnorthofAdelaideIsland).Imagesweretakenfromhttp://lance-modis.eosdis.nasa.gov/cgi-bin/imagery/single.cgi?image=crefl1_
143.A2006006130000-2006006130459.1km.jpg
wind speed has dropped to 2.8ms−1 giving a Coriolis ac- those observed at this time throughout the boundary layer.
celerationof4×10−4ms−2.Thepressuregradientaccelera- Theheightsofthewindspeedmaximainthejetprofilesdo
tion calculated between x=516, y=219km and x=638, agreewellwiththeobservedheight,though.
y=212km is now 6×10−4ms−2 and so is around 50% Betteragreementisobtainediftheobservedprofiles(mea-
largerthantheCoriolisacceleration. 12 surements between 20:00 and 22:00UTC) are compared
with modelled profiles at 12:00UTC. At the model time of
3.4 Modelcomparisontotheobservations 12:00UTC,themodelledföhnjetshavenotyetstartedtodie
down and jet 1 has just reached near to the regions where
3.4.1 Windspeed the real jets were observed. Profiles at this time are shown
inFig.5forvariouslocations,whicharemarkedinFig.8b.
Themodeloutputtimeof21:00UTC,6Januaryistheclos-
Location C is near the centre of the combination of mod-
est available time to that of the aircraft observations of the
elled jets 1 and 2, which at this time is ∼55km away from
strongwindjetsat20:23and22:01UTC.Atsimilarheights,
wheretheobservationsweremade.LocationDisatthecen-
themaximumwindspeedsofthesimulatedflowintheregion
tre of jet 3, which is further from the observation region at
of the maximum observed jet speed are around 9.3ms−1
thistime(∼100kmaway).Below∼1400m,thewindpro-
(Fig. 8d). The simulated jets extend further east than where
files at both locations are very similar to those observed on
the observations were made showing that they penetrate at the aircraft descent with maxima of ∼14ms−1 located at
leastasfaracrosstheiceshelfastherealföhnflow.However,
the same height as the observed maximum. However, since
the jet intensities are weaker than the observed jet winds,
at 12:00UTC the modelled jets do not reach as far east as
whichhadmaximaof12.4and15ms−1 forthedescentand
thelocationwheretheaircraftobservationsweretaken,this
ascent,respectively.Profilesthroughthecentres(locationsof
suggestssomespatialoffsettothejetlocationscomparedto
themaximumwindspeed)ofthesimulatedjets(notshown)
reality;modelprofilesAandBtakenattheaircraftlocation
confirm that the modelled jet wind speeds are lower than
Atmos.Chem.Phys.,14,9481–9509,2014 www.atmos-chem-phys.net/14/9481/2014/
D.P.Grosvenoretal.:DownslopewindsandtheLarseniceshelf 9489
Figure 8. As for Fig. 6 except in close up view over the ice shelf and at different times on 6 January: 09:00UTC (a), 12:00UTC (b),
15:00UTC(c)and21:00UTC(d).AlsomarkedarethelocationsofvariousotherpointswherethemodelprofilesinFigs.5and10have
beentaken.Theblackstraightlinein(a)isthelineoverwhichthecrosssectionsinFig15weretaken,andthelinein(c)isthatforthecross
sectioninFig13.
Figure9.Surfacepressure(colourcontours;hPa)withwindvectorsforthefourthmodellevelat15:00UTC(a)and21:00UTC(b)6January.
show much lower wind speeds than those measured by the therealjetreducedinintensity.Inthemodelthejetwasdy-
aircraft(Fig.5a). ingdowninintensityafter12:00UTC,6January,whichindi-
Thewindspeedsobservedinthejetduringtheaircraftas- catesthatasimilarreductioninjetintensityoccurred,except
centat22:01UTCwereweakerthanthoseduringthedescent atanearliertimethaninreality.
at 20:23UTC up to an altitude of ∼400m, suggesting that
www.atmos-chem-phys.net/14/9481/2014/ Atmos.Chem.Phys.,14,9481–9509,2014
9490 D.P.Grosvenoretal.:DownslopewindsandtheLarseniceshelf
Figure10.AsforFig.5exceptforat15:00UTC,6Januaryandforwindspeed(a)anddirection(b)only.SeeFig.8cforamapofthe
locationsoftheseprofiles.
(a) 3.4.2 Winddirection
Wind speed (m s−1) for AWS data
15
Larsen AWS
WRF at 10m
Figure 5b shows the wind direction at the same model time
(12:00UTC)andlocationsasintheprevioussection.There
−1m s)10 is generally a very reasonable match between the modelled
Wind speed ( 5 astwenrdeveeondbtswheeirnvgderdoduiwrneidnctdaiondndircethhceatinohgneeispgrhfortofiomlfestwhaeetsjateeltlrlwaylittniotdusdsoepuset.heTderhmleyaobxbei---
mumat∼250–350m.Thetwomodelprofilesatthecentres
ofthestrongjets(locationsCandD)exhibitasimilarrota-
050− J a n 0 6 1 2 1 8 0 6 − J a n 0 6 1 2 1 8 0 7 − J a n 0 6 1 2 1 8 0 8 − Jan tion in wind direction over the same height range, although
Time (UTC) the wind direction only reaches 215–220◦ at the jet maxi-
mum height compared to the observed ∼190◦. The model
(b)
400 Wind direction (degrees) for AWS data profilesatCandDhaveamoresoutherlydirectionthanthe
Larsen AWS
A and B profiles that are outside of the jets, which is likely
350 WRF at 10m
duetotheCorioliseffect(seeSect.3.3.1).
Wind direction (degrees)122350500000 tctbiiliooonBnnseeeydodtv1woee5iart:t0hhrtl0ehiteheUare,inTritconChrrfliat,shfujteeeisrtonnb3cliespkheaoearrlvfstyattohtudifeorunntseheoledtuootctihocatehetrhielosayhnvCeweo(lFfiraniioagdsnlo.sidus8dtcrhepi)feav.frseeAlsncyestsbdmcyivoreetemhnrcye---
100
pressuregradient(seeFig.9)thathaveincreasedinstrength
50 comparedto12:00UTC.Aprofilethroughthejetcentre(lo-
050− J a n 0 6 1 2 1 8 0 6 − J a n 0 6 1 2 1 8 0 7 − J a n 0 6 1 2 1 8 0 8 − Jan cationE),just20.7kmawayfromthelocationB,showsthat
Time (UTC) thewinddirectionsatthejetmaximumheightandthrough-
out the heights sampled by the aircraft are very similar to
(Fci)g ure11.TimeseriesfromtheLarsenCAWSalongwiththemod-
elledvaluesattheAWSloTecmapteiroatnurei n(oCd) ofomr AaWiSn da3ta;(a)10mwindspeed thoseobserved,asistheheightofthemaximumwindspeed
12
(S)and(b)10mwinddirection(φ).AlsomarkedaredLaertsaeinl sAWoSfair- (Fig.10).
10 WRF at 2m
craftobservationsmadeduringtheA–L1lemga st-11 5maltitude.ForS Thissuggeststhattheobservedjetmayhavelookedsome-
8
themeanvaluesareshownbythebluecrossandthefilledsquares what similar to the simulated jet 3, which emanated from
denC)ote6±1σ.Forφthesquaresdenotethefullrangeofthewinddi- halfway down the Larsen C Ice shelf (68.5◦W) and experi-
rTaebhcoeotTemperature (vioceni024rtchdleuersAinsWghoSthwelotlhceaegtioaobnnsd.etTrhvheeedcrsvoaasmlsuesehiswowmhseantrhkteehdemfaiodirrpcotrhainefttlowaftaetsrheddierrasenccegtlneyt. eainntgctheoidnslchyoen∼igsih1dt0e,.rt5ahbmoluesg−nho1,r,itsahlwsthoaomrudegswhphrtaohtgelrohewessiegiorhntth.oaTfnhtheoebwsmienradvxeisdmpeubemed-
towardstheiceshelfjustbeforethefinalascentfromtheregion.
−2 is very similar to that observed. The likelihood that the ob-
servedjetwasgenerallystrongerthanthesimulatedonepro-
−4
videsmoreevidencethattheobservedjetdidnotstartatthe
0−56− J a n 0 6 1 2 1 8 0 6 − J a n 0 6 1 2 1 8 0 7 − J a n 0 6 1 2 1 8 0 8 − Jan
Time (UTC)
Atmos.Chem.Phys.,14,9481–9509,2014 www.atmos-chem-phys.net/14/9481/2014/