JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 104,NO. DI6, PAGES 19,027-19,038,AUGUST 27, 1999
Arctic winter
climate and its interannual
variations
simulated by a regional climate model Annette Rinke and Klaus Dethloff Alfred WegenerInstitutefor Polar andMarine Research,Potsdam,Germany
Jens H. Christensen DanishMeteorologicalInstitute,Copenhagen, Denmark
Abstract. The meanArctic Januaryclimatologyandits interannualvariationhavebeenexamined by simulations with a regionalclimatemodelof theArcticatmosphere. To thisend,an ensembleof monthlongsimulations (Januaryof the 11 years1985-1995)hasbeeninvestigated in whichthe ensembleis largeenoughto represent a broadrangeof climaticconditions. The modelproduces crediblesimulationsof the meteorological patterns.Only smalldeviationsoccurbetweensimulationsandEuropeanCentrefor Medium-RangeWeatherForecasts(ECMWF) analyses;thatis, over mostof themodelareathesedifferencesin sealevel pressure,850 hPatemperature,and500 hPa heightarebelow3 hPa, 1 K, and5 m, respectively. Largerdeviations(up to 5 hPain the sealevel pressure and20 m in the500 hPaheight)arefoundoverpartsof theArcticOcean,whichseemsto be relatedto the crudeseaice representation at the lowermodelboundaryanddeficienciesin the planetaryboundarylayerparameterization. It is shownthatthedynamicalaspects of theinterannual variabilitycanbe adequatelycapturedby themodelsimulations; thatis, themaximumof themodel biasof thedynamicalvariablesis significantlysmallerthantheinterannual variabilitythroughout theentiredomain.To gainmoreinsightinto the spatialandtemporalstructures of the model's variability,an empiricalorthogonal function(EOF) analysishasbeenappliedto determinethemost significant structures in thefluctuations of themonthlymeandynamicalfields.EOF 1 of the500 hPa heightfield describes a regimewith well-pronounced polarvortexandcorresponds to the"Arctic Oscillation",whereasEOFs 2 and3 showwavestructures. A pronounced interannualvariabilityis noticedin the time seriesof the amplitudesof theEOFs.
1. Introduction
regional interannualvariability with the RCM method. In the
Studiesof generalcirculationmodel(GCM) performance in the Arctic describethe discrepancies amongdifferentmodelsandbetween modelsand observations[i.e., Walshand Crane, 1992; Bromwich et al., 1994; Chen et al., 1995; Tao et al., 1996; Kattsov et al., 1999]. The alternative concept of limited area climate
modelinghasbeen sucessfullyappliedfor simulatingthe Arctic climate at high spatial resolutionwith the developmentof the Arctic regionalclimate systemmodel ARCSyM [Lynchet al., 1995, 1997] and the regionalclimatemodel (RCM) of the Arctic atmosphereHIRHAM [Dethloff et al., 1996; Rinke et al., 1997]. BecausemultiyearRCM simulationsneed a large computational effort,up to nowhigh-resolution RCM simulations havebeendone only over selectedshortperiods.The perpetualJanuaryapproach usedcommonlyin GCMs in the pasthasbeenusedwith RCMs in the works by McGregor and Walsh [1993] and Hostetler et al. [1994]. Data analysesshow that the circulation of the Arctic atmosphere undergoeslarge fluctuationsaboutits monthly and annual means [Walsh and Chapman, 1990; Power and Mysak, 1992; Serrezeet al., 1993]. A necessary prerequisiteto correctly simulateclimatic changesis that modelsare able to realistically representthisnaturalvariability.Luethiet al. [1996] providedwith their studyin midlatitudespreliminaryinformationon simulating Copyfight1999by theAmericanGeophysical Union. Papernumber1999JD900296. 0148-0227/99/1999JD900296509.00
presentstudya firstassessment of a RCM's skillin simulating the meanArctic climatologyand its interannualvariabilityis made. Forthisreasonanensemble of monthlong simulations (January of the 11years1985-1995)hasbeeninvestigated in whichtheensemble is large enoughto representa broad range of climatic conditions.
A shortmodeldescription is givenin section 2. Theinvestigated monthlongsimulations of the Januaryensemble1985-1995with theirspecificsynopticconditions are presented in section3. The meanclimatologyof theregionalclimatesimulations of theeleven member ensemble forJanuary is described in section 4. Thesehigh horizontalresolutionmodelresultsare discussed by comparing themwith observational data.The abilityof the RCM to simulate theinterannual variabilityin theArcticis investigated in section5.
2. Model Description The employedregionalatmospheric climatemodelis called HIRHAM [Christensen et al., 1996]andis appliedon the whole Arcticregionnorthof 65øN.Themodelusesforthehorizontal grid a rotatedlatitude/longitude gridwith a rotatednorthpoleat (0øN, 0øE).The simulationsare performedat a horizontalresolutionof 0.5øin rotatedlatitudeandlongitude and19verticallayerswiththe modeltopat 10hPa.TheRCM HIRHAM usesthephysical parameterization packageof the GCM European Center/Hamburg 4 (ECHAM4) [Roeckner et al., 1996]. Physicalparameterizations include radiation,cumulusconvection,stratiformclouds,land
19,027
19,028
RINKE ET AL.: ARCTIC WINTER CLIMATE
January1986
January1985
January1987
January1988
•'•"L•,•,, .•i.,•', i. -_, _..,....
' .i ,i
....
!
January1989
....
,
....
,
....
,
....
,
....
,
....
January1990
ß',.......
....
,
.........
; .... ,,' '---• ............
,
....
,
....
,
....
!
....
,
....
q-',;........
,
....
!
....
!
....
,
....
•
....
'h,;,.......
,
....
,
....
,
Figure1. Modelsimulations of monthly meanpressure distribution fortheselected 11January simulations. Solid linesshowsealevelpressure inhectopascals (contour intei•,als: 3 hPa)anddashed linesthe500hPaheight inmeters (contourintervals: 50 m).
surfaceprocesses, hydrology,seasurface-,andseaice processes, zone.At thelowerboundary the modelis forcedby ECMWFplanetary boundary layer,andgravitywavedrag.A semi-implicit analyzedsea surfacetemperatures (SST) and sea ice fraction, updateddaily. TheseECMWF data are basedon a blend of the time-stepping schemewith a 5 min timestepis used. At the lateralboundariesthe modelis forcedby European NationalCentersfor Environmental Prediction(NCEP) SST at CentreforMedium-Range WeatherForecasts (ECMWF)analyses, 0.5øx0.5 øforopenwaterandtheiceboundary fromSpecial Sensor updatedevery6 hours.The informationfromthelateralboundaries MicrowaveImager(SSM/I). The qualityof theSSM/I ice bounis transferredto the model by a relaxationprocedure[Davies, daryis discussed by Nomura[1995].Seaicethickness is crudely 1976]for theprognostic fieldsusinga 10 gridpointwideboundary represented in themodelas2 m for eachgridboxwithseaice.
RINKE ET AL.: ARCTIC WINTER CLIMATE
19,029
January! 992
January1991
,
January!993
,'
•
i'
•
January1994
.... '•........ , ':"•- .......... "'"%/4>'; .... .'"'xS•/,;': .....
'" '
i
i
Figure 1. (continued).
Dethloffetal. [1996]performed winterandsummer simulations of thislargeensemblereflectsa significantfractionof theobserved of theyear1990for thewholeArcticandshowed thatthesimula- interannualwintertimevariabilityandthuspermitsanexamination ted spatialpatternsare consistent with observations, although of the model'sperformanceunder a wide range of conditions, deviations occurred which are due to deficiencies in the cloud includingits representation of the interannualdifferences. Figure 1 showsthe mean synopticconditionsduringthe 11 radiationscheme,surfaceprocesses, andboundarylayerscheme. The resultsfor the Arctic havebeenclearlyimprovedusingthe individualJanuary casesin theformof chartsforthemonthlymean sea level pressure(SLP) and 500 hPa geopotential height.In newparameterization package ECHAM4 [Rinkeet al., 1997]. January1985, 1986, 1988, and 1994 a well-developed high
3. January Ensemble
pressurecenter was generatedover the Arctic Ocean which is high 500 hPaheights.The IcelandicLow In this study,11 monthlongsimulationsof the January1985- coupledwith anomalous Cyclones(or troughs)emerging 1995datahavebeencarriedout.The multiplicityin the circulation waspushedawayto far southwest.
19,030
RINKE ET AL.: ARCTIC WINTER CLIMATE
(c) ...........
; ...... ;•. ...... .• --..•.•...........
o
,; ...........
ß
/
(e)
Figure 2. (a-d) Geographical distribution of monthlymeandynamicalfieldsfor theJanuaryensemble1985-1995. Simulatedandobservedfieldsare shownassolidanddashedlines,respectively. Figure2a showssealevel pressure in hectopascals (contourintervals:3 hPa),Figure2b shows850hPatemperature in degrees celsius(contourintervals: 3øC),Figure2c shows500 hPageopotential heightin meters(contourintervals:50 m), andFigure2d showsstandard deviationof sealevel pressurein hectopascals (contourintervals:1 hPa). (e-f) Differences"modelminusECMWF analyses" areshownin Figure2e for thesealevelpressure (in hectopascals, contourintervals:1 hPa)andin Figure2f for the500 hPageopotential height(in meters,contourintervals:10 m).
RINKE ET AL.' ARCTIC WINTER CLIMATE
19,031
slightlyunderestimate the500hPaheightof about5 m, from Baffin Bay oftencrossedGreenland.In January1989, 1993, simulations is up to 20 m (seeFigure and 1995 the SLP was dominatedby strongcyclonicconditions butovertheArcticOceanthediscrepancy over the North Atlantic/Greenland Sea far to the north. In these 2f forthe500hPaheightdifference "modelminusanalyses"). The (seeFigure3) offset monthsa zonallysymmetriccirculationwas foundin 500 hPa toocoldmeansurfaceto 500hPatemperatures thatanoverestimated SLPnormallyalsoleadsto an heightwhichwascoupledwith a deeppolarvortexcenteredover thetendency of 500 hPaheight.Figure2d showsthe temporal the pole.More or lessclimatologically normalconditions with a overestimation developed IcelandicLow andhighpressure centeroverSiberiaand standarddeviationof SLP whichgivesan indicationof the synopthe Arctic Oceanhavebeen foundin January1987, 1990, 1991, tic variabilityof the simulatedfield. This field was derivedby and 1992. calculating thestandard deviationfor eachseparate1-monthsimulationandthenaveragingoverthe Januaryensemble. The spatial featuresof theanalyzedvariabilityarewell reproduced; thediffe-
4. Mean Climatology
renceis below 2 hPa.It is evidentthatmaximumvariabilityoccurs in theNorthAtlantic.This is consistent with theinterpretation that
In this sectionthe simulatedmean Januaryclimatologyis
discussed andcompared with observed climatologies. Figure2 showsvarioussimulated andanalyzed meandynamicfieldsfor the January ensemble 1985-1995. All fieldshavebeencomputed as arithmetic averages of the11corresponding monthlymeans. The modelcaptures themonthlymeandistribution of theSLP(Figure 2a) well withtheoverallpatternof theSiberiananticyclone, the Icelandic Low,andthetroughof theAleutianLow.OvertheArctic OceantheSLPisrelativelyhigh,appearing asa saddle of relatively
thewintertime year-to-year changes resultprimarilyfromvariation in the Atlantic storm tracks. Serreze et al. [1993] showed,on the
basisef -!climatological analysisof Arcticcyclonecharacteristics, that under wintertime conditions the Atlantic side of the Arctic is
the mostsynoptically activeregionandthatthe ArcticOceanis characterized by thefrequentoccurrence of anticyclones. Plate1 compares thesimulated January 2 m temperature field with observationaldata, that is, with the Arctic gridded2 m air
fromthePolarExchange attheSeaSurface (POLES) highpressure between theSiberian andtheAlaskahighs.The temperature modelreproduces themonthly meandistribution of theSLPvery program[Martinand Munoz,1997]andwith the surfaceair climatology of Legatesand Willmort[1990a].The wellwithdifferences fromtheECMWFanalyses of lessthan3 hPa temperature temperature structures agreeamong Platesl a, lb, and overmostof the areasandan area-averaged differenceof 1.5 hPa. large-scale The modelhasa slighttendencyto overestimate theSLP overthe l c, the modelis colderover the Barentsand Kara Seasthanthe wholearea.The largestdeviationsin the SLP are foundoverthe POLES data. Anothermeasure of themodel'scapabilityto represent a good Kara Seawith up to 5 hPa(seeFigure2e for the SLP difference
isthedomain meanmodelbias,defined asthemonth"modelminusanalyses"). Thisproblemseemsto be relatedto the climatology mean(averaged overthewholemodelarea crudeseaicethickness representation; seaice thickness is crudely ly meanandhorizontal represented in themodelas2 m for eachgridboxwith seaice.A excludingthe boundaryzone)valueof the deviationbetween simulation andanalyses. The verticalprofilesof thistemperature realistic sea ice thickness distribution in the Arctic exhibits much and specific humidity bias are shownin Figure3 for the entire largervariety.The850hPatemperature (Figure2b) is reproduced reasonablywell. The deviationsaccordingto the analysesare domainof model,andfor land,ocean,andseaice areasseparately. biasis generallysmallerthan1 K, exceptnearthe maximally1 K. At 500 hPathegeopotential patterns (Figure2c) Thetemperature arein goodagreement withtheanalyses; overmostof theareasthe surfacewith the modelbeingtoo coldup to 3 K. The largestdevi-3
-2
-1
0
1
2
3
-0.3
100'
i ,'
200'
200
300'
300 t
400e
500
•' 400 J•
500-•
600
•. 600-
700-
700-
800
800900-
1000'
................
-
1000-
19,032
RINKE ET AL.: ARCTIC WINTER CLIMATE
(a)
(b)
Figure4. Monthly meanprecipitable waterinmillimeters forJanuary: (a)simulation oftheJanuary ensemble 19851995and(b) Serrezeclimatology1974-1991.Contourintervalsare 1 mm.
ationsoccurnearthesurfaceoverseaicewhichmaybe referredto corresponding mapof Serrezeet al. [1994](Figure4b).Duringthe the crudesea ice descriptionin the model and the insufficient coldseasons the temperatures at all levelsarehighestoverthe
description of thewintertime verylow andstableplanetary boun- Atlantic side of the Arctic between Iceland and Scandinavia.This darylayer[Abe$$et al., 1997].Additionally, wehaveto keepin areashowsthelargestvaluesof precipitable water(upto 7 mm). mindthatthe ECMWF analyseshavea coarserhorizontalresolu- Thecomparatively abund. antprecipitable waterlargelyreflects the tion(1.5ø)thanthemodelandthereforetheyaremoreinaccurate, abilityof a warmer atmosphere toholdmorewatervapor.Overthe especially nearthesurface withits inhomogeneities. Thespecific ArcticOceanbothfiguresshowa meanprecipitable watervalueof humiditydeviations aregenerallyabout+0.05g/kg,whichcorres- about2 mm.The simulated meanspatialdistribution of theaccupondsto differences of up to 8%. mulatedtotalJanuary precipitation, presented in Figure5, shows To showthemodel'scapability/deficiencies in thesimulation of theArcticOceanwaterbudgetcomponents, Figure4 and5 show the Januaryclimatologyof precipitablewaterand precipitation from simulations andobservations. Figure4 showsthatthe simulated field of meanprecipitablewater agreesvery well with the
precipitationmaximaon the southeastcoastof Greenlandandwest
coastof Scandinavia. TheLegates/Willmott climatology [Lesate• andWillmott,1990b](Figure5b) showstheprecipitation maxima over the sameregionsand of aboutthe sameamounts.Over the
Arctic Oceanboth figuresshowa meanprecipitation valueof
Figure5. Monthly meanprecipitation inmillimeters forJanuary: (a)simulation oftheJanuary ensemble 1985-1995 and(b) Legates/Willmott climatology1920-1980.Contourintervalsare20 mm.
(a)
-
-f
(c)
Plate 1. Monthly mean2 m air temperaturefor January:(a) Polar Exchangeat the Sea Surface(POLES) for 19851994, (b) simulationfor 1985-1994,and(c) Legates/Willmottclimatology1920-1980.Contourintervalsare 3øC.
about 10 mm. The validation of the precipitationis difficult because of thelack of adequately denseobservational dataandthe absenceof essentiallyany data for the central Arctic Ocean. Kattsovet al. [ 1999]comparedHIRHAM simulations of monthly (January)andyearly(1990) precipitation withtheGCM ECHAM, reanalysis, andLegates/Willmott andBryazginclimatologies and showedthatthe HIRHAM modelproducescrediblesimulationof Arctic precipitation. The Surface Heat Budget of the Arctic Ocean (SHEBA) campaign phase1 includedtheanalysisof existingdatasets(land andairborneobservations, globalmodels)resultingin an atlasof monthlyspatialfieldsfor the ArcticOcean[Serrezeet al., 1997]. Thefieldsin thepresentstudyagreequitewell withthelarge-scale patterns presented in thisatlas,butthehighhorizontalRCM simulationsgenerallyproduceregionaldetailsof surfaceclimateas forcedby topography andnarrowlandmasses.
1995. Th: data are extracted from the Arctic rawinsonde archive
[NationalSnowand Ice Data Center,1996]. The figure shows largeobservedinterannual amplitudes, in temperature up to 6 K andin SLP up to 20 hPa.Lookingat the simulatedtemperatures andSLP,whichwereinterpolated fromthesurrounding gridpoints ontothe stationpoint,it seemsthatthe modelis ableto reproduce the largeinterannualfluctuationsin the monthlymeanvariables very well.
To conr, ider thedynamicalvariabilityin theJanuaryensemble, we comparethe signalin the form of the interannualvariability with the noise in the form of the error in the model simulations. The ensemble mean model bias is defined as the deviation between
simulation and analysescalculated for each month and than averagedoverthe ensembleof the 11 months.Figures7a and7b comparethe error of the model, that is, the ensemblemean model bias for the 500 hPa height with the simulatedand observed
standarddeviationof the 500 hPa heightbasedon the January ECMWF climatologyfor theyears1985-1995.The clearinference 5. Interannual Variability from this figure is that the dynamicalaspectsof the interannual Figure6 showsasan exampleof theyear-to-yearvariabilitythe variabilitycan be adequatelycapturedby the modelsimulations. variationin the 850 hPa temperatureand SLP at the Canadian The maximumof the modelbiasof the 500 hPaheightfield is ~ 30 stationEureka(80øN.86øW)duringJanuaryof the 11 years1985- m, whereasthe interannualvariability is significantlylarger ß
.
19,034
RINKE ET AL.' ARCTIC WINTER CLIMATE EOF 2 explains18%, andEOF 3 explains12%.EOF 1 describes a regimewith a well-pronounced polarvortex,whereasEOFs 2 and 3 showwavestructures. A polarvortexpatternsimilarto EOF 1 is also found in the filtered (low passbelow 6 days) standard deviationof the 500 hPa height field, whereasEOFs 2 and 3 approximately reflect the patternof the band passfiltered (2-6 days)standarddeviationwhich describesthe stormtracks(not
' T8•50'
[øC] -22 -
x
-24 x
-26-
shown).The presented EOF 1 structure of the 500 hPa height (Figure8) is similarto thespatialstructure of theleadingEOF for winter (1947-1994)500 hPa height foundby Slonoskyet al. [1997], andcorresponds well to the "ArcticOscillation"described by Thompson and Wallace [1998]. The EOF 1 structureshowsa generalhigh or low pressureall over the polar and subarctic regions,with a pronounced centerof oppositesignoverwestern
-28 -
-30-
-32
i
i
i
i
I
I
I
i
i
i
[hPa] 1025
I
I
I
I
I
I
I
I
Europe,which is characteristicof the North Atlantic Oscillation [Hurrel, 1995].
The totalvarianceof the500 hPatemperature is explained by EOF 1 with 39%, by EOF 2 with 27%, andby EOF 3 with 12%.
SLP 1020-
(a) 1015-
1010-
x
1005-
x
l
1000
I
1985
1986
1987
1988
i
19_.89 1990
i
1991
1992
1993
1994
1995
Figure 6. Time seriesof monthlymeansealevelpressure and850 hPa temperatureat the stationEureka (80øN, 86øW) for January 1985-1995. Columns indicate observations from rawinsondes and crosses indicate simulated values.
throughout theentiredomain.We alsonotefromFigure7bthatthe HIRHAM simulatedstandarddeviationreproduces the observed interannualvariabilitywell. To gainmoreinsightintothespatialandtemporalstructures of the model's variability, we applied an empiricalorthogonal function (EOF) analysis to determine the most significant structures in thefluctuations of themonthlymeandynamicalfields (500hPaheightandtemperature, SLP,850hPatemperature). The timeseriescanberepresented compactly by projectingthedynamical fieldsat any time level on the setof EOFs;the EOF patterns represent anoptimalpotentialtocompress thedataintoa minimum numberof patterns.A setof EOFsfor the time seriesof the 11
monthlymeansof thefourvariables is generated by determing the eigenvalues andeigenfunctions of thecovariance matrix.Thetime seriesof amplitudes of theith EOF arecalledtheithprincipalcomponent(PC/) of the time series.The EOFsand PCshavebeen normalized by thesquarerootandthereciprocal of thesquareroot of theeigenvalues, respectively Iron Storch,1995].In thisnormalization the PCs have variance1, and the relativestrengthof the signalis in the EOF patterns.
Figures8 and9 showthe calculated EOF patternsof the first threeEOFs for monthlymean500 hPa heightand temperature. EOF 1 represents 53% of thetotalvarianceof the500 hPaheight field andexhibits"typicalanomalies"of the orderof 40-100 m,
Figure 7. The 500 hPa geopotential heightfield (in meters, contourintervals:10 m) of the Januaryensemble:(a) ensemble mean model bias and (b) observed(solid lines) and simulated (dashedlines)standarddeviation.
RINKE ET AL.: ARCTIC WINTER CLIMATE
EOF
19,035
1
EOF 3
of the firstthreeEOFsof themonthlymean Figure8. Structures of thefirstthreeempirical orthogonal func- Figure 9. Structures tions(EOFs)of themonthlymean500 hPaheightof theJanuary 500 hPa temperatureof the Januaryensemble1985-1995.The ensemble 1985-1995. The simulated fields are shown as solid
simulated fields are shown as solid lines; the observed fields
areshownasdashedlines.Unitsarekelvins, lines;theobserved fields(ECMWF analyses) areshownasdashed (ECMWF analyses) lines.Units are meters,and contourintervalsare 20 m.
and contour intervals are 0.5 K.
19,036
RINKE ET AL.' ARCTIC WINTER CLIMATE
500 hPaheight
500 hPatemperature
ß .
'
985
'
'
lg•86 1987
'
1988
' 1& 1; 1 1; 2 1; 3 1; 4 1; 5
3
1989
'
1985
'
1986
'
1987
'
1988
' 1;•o1; 1 1; 2 1993 ' 1; 4 19;5
1989
Figure10. Timeseries oftheprincipal components ofthefirstthreeEOFsof thesimulated monthly mean500hPa heightandtemperature of theJanuary ensemble 1985-1995.
Co)
Plate2. Monthly mean 2 mairtemperature (a)fromPOLES and(b)fromsimulations forJanuary 1989and1985. Contour intervals are 3øC.
RINKE ET AL.: ARCTIC WINTER CLIMATE
19,037
EOF 1 exhibits"typicalanomalies" of-0.5 to 2 K andrepresents ton. We also benefited from discussions with T. Kandlbinder and from an east-westgradient,whereasEOF 2 describesa wintertime helpful commentsby the anonymousreviewers.This work was partly
temperature patternwitha warmAtlanticregionanda coldcenter
supported by GermanMinistry of Educationand Researchundercontract
07VKV01/1.
This is AWI
contribution
1461.
over Alaska and the Arctic Ocean.We alsonote from the Figures 8 and 9 that the EOF structurescalculated from the HIRHAM
data
agreewith the observedstructures calculatedfrom ECMWF analyses.
The time seriesof the corresponding PCs are shownin Figure
10.A pronounced interannual variabilityis noticed,especially in PC 1, but alsoin the time seriesof PCs2 and3. The PCsvary most of the time between+1 but also show large negativevalues.
References
Abegg, C.,K.Dethloff, A.Rinke, andV. Romanov, Influence ofplanetary boundary layerparameterization onArcticclimate simulations, paper presented at Conference on PolarProcesses andGlobalClimate,Rosario, Wash., 1997.
Bromwich, D. H., R.-Y. Tzeng,andT. R. Parish,Simulation of themodern Arcticclimateby theNCAR CCM1, J. Clint.,7, 1050-1069,1994. of the Togetherwith the informationprovidedby the EOF patterns Chen,B., D. H. Bromwich,K. M. Hines,andX. Plan,Simulations 1979-1988 polarclimates by globalclimatemodels, Ann.Glaciol.,21, (Figure9) suchlargenegativecoefficients in temperature PCs
represent ananomalously coldJanuary, suchasin 1989and1993 with mean negativeanomaliesof the order of-2 K. For example, the observationsshow that January 1989 was characterizedby anomalously low temperatures and500 hPaheights.The simulated broadpolar vortexin the 500 hPa geopotentialin this month(see Figure 1) centeredjust off the pole was found as the mostdetermined EOF structureand resultsin a more zonally symmetric circulation, which is also found in the GEOS 1 reanalysis [Overlandet al., 1997]. In contrast,Figure 10 showsthat January 1985 and January1994 are characterizedby anomalouslyhigh temperatures.Also, the polar vortex over the centralArctic is not developedat all. A look at the 500 hPaheightsof January1985and 1994 (see Figure 1) showsthat thesestructuresare similar to the patternof EOFs2 and3. As a demonstration of themodel'sability to describea wide range of circulationstates,Plate 2 showsthe simulatedandobserved(Arctic griddedPOLES 2 m air temperature data set) [Martin and Munoz, 1997] 2 m temperaturefor the two extrememonthsof January1989/1985.The observedanomalouslylow/hightemperatures over mostof the Arctic are captured quitewell by the modelsimulation.
83-92, 1995.
Christensen, J.H., O. B.Christensen, P.Lopez, E. vanMeijgaard, andM. Botzet,TheHIRHAM4 regionalatmospheric climatemodel,DMI Sci. Rep.96-4,Dan.Meteorol. Inst.,Copenhagen, 51pp.,1996. Davies,H. C., A lateralboundary formulation for multilevel prediction models,Q. J. R. Metorol. Soc.,102, 405-418, 1976.
Dethloff,K., A. Rinke,R. Lehmann, J. H. Christensen, M. Botzet,andB.
Machenhauer, Regional climatemodelof the Arcticatmosphere, J. Geophys.Res.,101, 23401-23422, 1996.
Hostetler, S.W.,F.Giorgi,G.T. Bates, andP.J.Bartlein, Lake-atmosphere feedbacksassociated with paleolakesBonnevilleand Lahontan, Science,263, 665-668, 1994.
Hurrel,J. W., Decadaltrendsin theNorthAtlanticOscillation regional temperatures andprecipitation,Science,269, 676-679, 1995.
Kattsov, V. M., J. E. Walsh,A. Rinke,andK. Dethloff,Atmospheric climatemodels: Simulation of theArcticOceanfreshwaterbudget components, in TheFreshWaterBudget oftheArcticOcean, editedby L. Lewis,in press,1999.
Legates, D. R.,andC.J.Willmott, Meanseasonal andspatial variability in globalsurfaceair temperature, Theor.Appl. Clintatol.,41, 11-21, 1990a.
Legates, D. R.,andC.J.Willmott,Meanseasonal andspatial variability in gauged-corrected globalprecipitation, J. Clintatol.,10, 111-127,1990b. Luethi,D., A. Cress,H. C. Davies,C. Frei, and C. Schaer,Interannual variabilityandregionalsimulations, Theor.Appl.Clintatol.,53, 185209, 1996.
6. Summary Applying a RCM of the Arctic, it has been shownthat it is
Lynch, A. H.,W. L. Chapman, J.E.Walsh,andG.Weller,Development of a regionalclimatemodelof thewestern Arctic,J. Clint.,8, 1555-1570, 1995.
Lynch,A. H., M. F. Glueck, W. L. Chapman, D. A. Baily,andJ.E. Walsh, possible to realistically simulate themeanArcticclimatology and Satelliteobservation andclimatesystemmodelsimulation of the St.
interannual variabilitywith thismodelingmethod.The presented LawrenceIslandPolynya,Tellus,2, 277-297,1997. of theArctic2 m airtemperature Januarydynamical andthermodynamical fieldsagreequitewell Martin,S.E., andE. Munoz,Properties for 1979-1993 derived froma newgridded dataset,J. Clint.,10, 1420with thelarge-scale patterns knownfromglobalmodelsanddata 1440, 1997.
analysesand additionallyshow regionaldetailsof the surface McGregor, J.L., andK. J.Walsh, Nested simulations ofperpetual January climateasforcedby topography andcoastlines. climateovertheAustralian region,J. Geophys. Res.,98, 23283-23290, 1993. Thestudyshowsthatthemodelcanreproduce a widerangeof observed climaticstates.It hasbeenshownthatthe dynamical NationalSnowand Ice Data Center(NSIDC), HistoricalArctic RawinsondeArchive,Vol. 3-5, CD-ROM, Boulder,Colo., 1996. aspects of theinterannual variabilitycanbeadequately captured by Nomura, A., Global seaice concentrationdata set for use with the ECMWF
the model simulations;that is, the maximum of the model bias of
re-analysis system,ECMWFTech.Rep. 76, 40 pp., Eur. Cent.for thedynamicalvariablesis significantly smallerthantheinterannuMedium-Range Weather Forecasts, Reading, England, 1995. al variabilitythroughoutthe entiredomain.This resultunderlines Overland, J.E., J.M. Adams,andN. A. Bond,Regional variation of winter temperatures in the Arctic,J. Clint., 10, 821-837, 1997. the justificationand usefulnessof the RCM as a tool for climate variability ofarcticseachangestudiesin theArctic.As Luethiet al. [ 1996]noted,theissue Power,S.B.,andL. A. Mysak,Ontheinterannual levelpressure andseaice,Atrnos.Ocean,30, 551-557,1992. of regionalclimatepredictability iscentralto anyassessment of the Rinke,A., K. Dethloff,J.H. Christensen, M. Botzet,andB. Machenhauer, potentialand the limitationsof the regionalclimatemodeling Simulation andvalidation of Arcticradiation andclouds in a regional climatemodel,J. Geophys. Res.,102, 29833-29847,1997. approach, andsuchstudiesaspresented hereprovidesomebasic information on this focal issue.
Roeckner, E., et al., The atmospheric generalcirculation modelECHAM-
4: Modeldescription andsimulation of present-day climate. MPI Rep. 218,90pp.,Max-Planck-Inst. forMeteorol., Hamburg, 1996. Acknowledgments. The HIRLAM systemwasdeveloped by the Serreze, M. C., J. E. Box,R. G. Barry,andJ. E. Walsh,Characteristics of HIRLAM projectgroup,a cooperative projectof the nationalweather Arcticsynoptic activity,1952-1989, Meteorol. Atrnos. Phys.,51, 147164, 1993. services in Denmark, Finland, Iceland, Ireland, theNetherlands, Norway, andSweden. Wearegrateful toI. Hebestadt forprograming support and Serreze, M. C.,M. C. Rehder, andR. G. Barry,A climatological database
preparingthe graphics.We thank A. Weisheimerand D. Handoff for
of Arcticwatervaporcharacteristics, PolarGeogr.Geol.,18, 63-75,
1994. preparing theEOFcalculations anddiscussions andS. Hagemann for providing ustheLegates/Willmott data.DatafromthePOLESprogram Serreze, M. C., J. A. Maslanik, andJ. R. Key,Atmospheric andseaice wereobtained fromthePolarScience Center attheUniversity ofWashing- characteristics of theArcticOceanandtheSHEBAfieldregionin the
19,038
RINKE ET AL.: ARCTIC WINTER CLIMATE
Beaufort Sea,NSIDCSpec. Rep.,4, 219pp.,Natl.SnowandIceData Walsh,J. E., and W. L. Chapman,Short-termclimaticvariabilityof the Cent., Boulder, Colo., 1997.
Slonosky,V. C., L. A. Mysak,andJ. Derome,LinkingArcticsea-iceand atmosphericcirculationanomalieson interannualand decadaltime-
Arctic, J. Clim., 3, 237-250, 1990.
Walsh,J.E., andR. G. Crane,A comparison of GCMsimulations ofArctic climate,Geophys. Res.Lett.,19, 29-32, 1992.
scales,Atrnos.Ocean, 35, 333-366, 1997.
Tao, X., J. E. Walsh,andW. L. Chapman,An assessment of globalclimate modelsimulationsof Arctic air temperatures, J. Clim., 9, 1060-1076,
J.H. Christensen, DanishMeteorologicalInstitute,Lyngbyvej100,DK2100 Copenhagen 0, Denmark. K. Dethloff andA. Rinke,Alfred WegenerInstitutefor PolarandMaThompson,D. W. J.,andJ. M. Wallace,The ArcticOscillationsignature in A43, D-14473 Potsdam,Germany.(e-mail: the wintertimegeopotential heightand temperature fields, Geophys. rine Research,Telegrafenberg Res. Lett., 25, 1297-1300, 1998.
[email protected]) von Storch,H., Spatialpatterns:EOFs and CCA, in Analysisof Climate Variability, editedby H. von Storchand A. Navarra,pp. 227-257, (ReceivedAugust8, 1998;revisedApril 21, 1999; acceptedApril 28, 1999.) Springer-Verlag, New York, 1995. 1996.
