Sie sind auf Seite 1von 10

JOURNAL OF GEOPHYSICAL

RESEARCH, VOL. 104, NO. D24, PAGES 30,937-30,946, DECEMBER 27, 1999

Propagation theArcticOscillationfrom the stratosphere of to the troposphere


Mark P.BaldwinandTimothyJ.Dunkerton
Northwest Research Associates, Bellevue, Washington

Abstract. Geopotential anomalies ranging from the Earth's surface the middlestratosphere to in the northernhemisphere dominated a modeof variability are by knownastheArctic Oscillation (AO). TheAO isrepresented hereinby theleading mode(thefirstempirical orthogonal function)of low-frequency variability wintertimegeopotential of between1000 and 10hPa.In the middlestratosphere signature theAO is a nearlyzonallysymmetric the of patternrepresentingstrong weakpolarvortex. 1000hPatheAO is similar the North a or At to AtlanticOscillation, with morezonalsymmetry, but especially highlatitudes. zonalat In meanzonalwind theAO is seen a north-south as dipolecentered 40-45N; zonalon in meantemperature isseen a deep it as warmor coldpolaranomaly fromthe upper troposphere --10hPa.Theassociation theAO pattern thetroposphere to of in with modulation thestrength thestratospheric vortex of of polar provides perhaps best the measure coupling of between stratosphere thetroposphere. examining the and By separately time series AO signatures tropospheric stratospheric of at and levels, isshown it thatAO anomalies typically appear in thestratosphere propagate first and downward. midwinter The correlation between 90-daylow-pass-filtered the 10-hPa anomaly the 1000-hPa and anomaly exceeds whenthe surface 0.65 anomaly time series lagged aboutthreeweeks. is by The tropospheric signature theAO anomaly characterized substantial of is by changes the to stormtracks andstrength the midtropospheric especially the North Atlanticand of flow, over Europe. implications large The of stratospheric anomalies precursors changes as to in tropospheric weather patterns discussed. are

also that part of the tropospheric pattern may represent downward influence from the stratosphere. A coupled modeof variability between northernwinter the Thompson Wallace and [1998]examined leading the EOF of stratosphere troposphere discussed Nigam [1990], northernwintersea-level and was by pressure (SLP)and founda winterwho examined rotatedempirical orthogonal functions (EOFs) timepattern resembling NAO,but with theprimarycenter the of zonal-mean wind.Nigam's result showed the dominant of actioncovering that more of the Arctic,givinga more zonally mode of variabilityin zonal-mean wind appears a deep symmetric as appearance. called leading They the EOFof SLPthe north-south dipole, with a nodenear40-45N. poleward ArcticOscillation The (AO). The AO is remarkably similarto the part of thedipolerepresents fluctuations the strength the spatial in of pattern produced using original by the definition the of polar vortex. Couplingbetweenthe stratosphere and troNAO [Walkerand Bliss,1932]. The NAO index was first defiposphere further explored Baldwinet al. [1994],who nedasa linearcombination normalized was by of pressures temand examined geopotential patterns themiddletroposphere in that peratures from nine stations covering of longitude 90 from werelinked to the stratosphere. usingsingular By valuede- Washington, DC to Vienna. The resultingspatialpattern is composition (SVD,Bretherton al., [1992])between et 500-and substantially morestrongly correlated with the AO indexthan 50-hPageopotential, showed theleading they that modehad a with "Atlanti-centric" indices such as the normalized SLP strongdipolesignature zonal-mean in wind, extending from difference betweenPortugaland Iceland(J.M. Wallace,Unithe surface above10 hPa. They found that the dominant versityof Washington, to personal communication). Neverthe500-hPa patternwasa modesimilarto the North AtlanticOs- less,a comparison the NAO (as illustrated Hurrell of in cillation (NAO), with centersof action near Greenland and [1995])with theAO shows nearly a identical patternin middle justwestof Spain, with a slightly but largerspatial extent than to high latitudes from North Americato Europe.The main theNAO anda largedegree circular of symmetry abouta cen- differencesare 1) the NAO has no center of action in the ter point displaced toward Greenland. They suggested that Pacific,and 2) the AO has a more broad centerof action over elements this patternmay actto forcethe stratosphere of but the polar cap.The closematchbetween AO and NAO the
Copyright bytheAmerican 1999 Geophysical Union.
PaperNumber 1999JD900445. 0148-0227/99/ 1999JD900445$09.00

1. Introduction

across Atlanticregionimply that the weather the and climate implications theAO will be similar thatof theNAO. of to Eventhoughthe AO was definedon the basisof surface data, it showedremarkablystrong couplingto the stratosphere. Thompson Wallace and alsoshowed 500-hPa the and
30,937

30,938

BALDWIN AND DUNKERTON: ARCTIC OSCILLATION

50-hPasignatures the AO (regressions their AO index of of with geopotential fields). comparison 500 and 50 hPa) A (at withtheleading SVDmodeofBaldwin al. [1994]shows et that the two modes nearlyidentical. are SVD analysis between 500 and50 hPacaptures same the modeof variability does as EOF analysis surface of pressure. AO represents dominant, The a robust,naturallyoccurring modeof variability, whichmaybe recovered using by tropospheric or a combination trodata of pospheric and stratospheric data. The AO is also seenin stratosphere-troposphere generalcirculation models,and its trend revealsa possiblelink to greenhouse gas increases
[Shindell al., 1999]. For a brief overviewof the AO and cliet mate connections, Kerr [ 1999]. see

March with little effecton the results.)We treat the five levels

Downwardpropagation stratospheric of disturbances into the troposphere beenobserved dataand simulated has in in generalcirculationmodels.Stratospheric suddenwarmings standard deviation fluctuation in the nondimensional AO inoftenbeginin the lower mesosphere (-60 km) and develop dex. downward,and casestudiesof warmingshave documented penetration the Earth'ssurface. to Kodera al. [1990] ob- 3. Results et served that strong upper-stratospheric westerlies December in of tend to propagate polewardand downward,so that tro- 3.1. Signatures Arctic Oscillation pospheric westerlies high latitudes stronger at are duringthe The Arctic Oscillation accounts for 23% of the variance in following February. Thisbehavior wasalsosimulated a gen- the five-level in data set (as compared with 22% in Thompson eral circulationmodel. Koderaand Koide [1997] discussed the and Wallace's [1998] SLPAO). Figure 1 illustrates signathe formationof an AO-like circulation the troposphere as- turesof the leading in in modeat nine levels, including fivelevthe sociationwith the downwardpropagation zonal wind elsusedto definethe AO. The panels contoured meters of are in anomalies from the stratosphere. per standard deviation AO index.We now compare of the We take the view that the AO is the largest and most fun1000-, 500-, and 50-hPapanelsin Figure 1 with the corredamental mode of variabilityin the northernhemisphere tro- sponding panels Figure of Thompson Wallace in 1 and [1998]. posphere-stratosphere system. During winter, when the The 1000-hPa panelis nearlyidentical Thompson Walto and stratosphere large-amplitude has disturbances, AO has a lace's the SLPAO regression. mode hasbeen discussed This by strong signature the stratosphere. in Outside winterthe AO several of authors beginning, with Kutzbach [1970],andis charis confined the troposphere, to although with greatly dimin- acterized one centerof actionover the Arcticregion,disby ishedamplitude (-1/3) duringsummer. placedtowardGreenland, and an opposing ring at midlatiSection is a glossary terms.Eachterm is italicized 5 of when tudeswith prominentfeatures over both the Atlanticand first used. Pacific Oceans. Atlanticregionresembles NAO,but the The the pattern is more zonallysymmetricand has a largerspatial scale. The agreement is remarkable consideringthat 2. Data SetandAnalysis Technique Thompson Wallace and usedmonthly-mean November-April We use the National Centers for Environmental Predicdataonlyat 1000hPa,andweuse90-day low-pass-filtered Detion/National Center for AtmosphericResearch (NCEP/- cember-February spanning data 1000-10hPa. NCAR) Reanalysis [Kalnay al., 1996].Thesegeopotential, The 500-hPa panel again capturesthe same mode as et wind, and temperature data are global and are archivedas Thompson Wallace and [1998],with onecenter action of over dailyaverages a 2.5x 2.5gridandareavailable 17levels southern on at Greenland opposing broadmidlatitude a band.Figure 3a from Baldwin et al. [1994] illustratesthat the 500-hPa from 1000to 10hPafor theperiod1958-1997. Our approach differssomewhat from eitherBaldwinet al. patternresulting from SVD between and 500-hPa essen50 is [ 1994] or Thompson and Wallace[1998]; we examinethe tially similar to that shownin Figure 1 or that shownby leading EOF of a deeplayerof the wintertime stratosphere- Thompson Wallace. the stratosphere 50-hPa and In the pattern troposphere, represented thefivegeopotential by fields 1000, is dominated a centerof actiondisplaced at by slightlyoff the 300, 100,30 and 10 hPa,using90-daylow-pass-filtered data pole towardGreenland, with a nearlyzonallysymmetric ring north of 20N. The data at each of the five levels were normalin midlatitudes. with the 500-hPa As pattern,the 50-hPa patizedby dividing thespatial by standard deviation each at level. tern is nearlythe sameaseitherthe SVD modeof Baldwin et Asnoted Thompson Wallace by and [1998], energy density of al. [1994] or the AO signatureof Thompson and Wallace theAO (theproduct density squared of and amplitude) be [1998].The 10-hPa will panelshows patterncentered a nearly over roughly invariant overthe five datalevels, whichargues for the pole, but with lesszonal symmetryin midlatitudes, feaequalweighting eachlevel.In practice, details the turingan opposing of the of center actionoverthe eastern of Pacific. weighting unimportant; are Thompson andWallace obtained The above comparisons thattheAO is extraordinarily show a similarleading mode with monthly-mean surface data.We robust, thatit maybe recovered in usinga varietyof datalevels confine EOF analysis wintertime the to (December-February) and techniques, and it is not sensitive the number of to sothat stratospheric windstendto be westerly, allowing up- months included in the winter season. Other than the seasonal wardpropagation large-scale of tropospheric waves [Charhey cycle, AO is the largest the modeof variability the exin and Drazin,1961].(We alsotried including November and tratropical northern troposphere-stratosphere.

of geopotential a single as fieldfor thepurpose theEOFcalof culation.EachNCEP latitude-longitude grid wasresampled ontoan equal-area grid.The AO patternwasidentified the as leadingEOF of the temporalcovariance matrix of the fivelevelgeopotential data.The AO indexis defined the daily as time series the first mode (often calledthe principalcomof ponenttime series) based the 90-daylow-pass-filtered on geopotential data.We thenshowthe spatial patterns theAO by of regressing AO indexwith various90-daylow-pass the filtered datafields; call the resulting we patternsthe signatures the of AO. Asnotedby Thompson Wallace and [1999],theregression mapsare not precisely sameasthe EOFsbecause the the of weightings usedin the calculation. The regression mapscorrespond anomalyvaluesin a field associated to with a one

BALDWIN AND DUNKERTON: ARCTIC OSCILLATION

30,939

1000 laPa

FigureSignatures Arctic 1. ofthe Oscillation at10, 50, 200, 500, and hPa. (AO) 30, 100, 300, 700, 1000 Each panelis produced by regressing index geopotential. valuesmeters, the AO with Contour are corresponding toa
one standard deviation anomaly AOindex. contour inthe The intervalthe of 1000-hPa is+5,+15 for panel ....
ease comparison Thompson Wallace of with and [1998].

is highly correlated (0.93)withtheAO index. The Figure illustrates 2a correlations between AOindex the and which also regression AOindex zonal-mean (Figure ofthe with wind 2b) that exceeds m s in themiddle 10 ' theregion northof 20N. dipole The structure similar is to shows thewindoscillation If the calculation isperformed angular for thatillustrated Baldwin al. [1994] SVDbetween by et for 500- stratosphere. same the and parts thedipole of are and 50-hPa heightfieldsandhascorrelations to 0.96 at momentum, northern southern up
zonal-mean wind. The AO accountsfor 34% of the variance in

60N,100hPa.Kodera al. [1999]defined "polarnightjet et the

verysimilar magnitude. in
Correlations between the AO index and zonal-mean tem-

(PNJ)index" the zonal-mean as wind at 65Nand 50 hPa,

30,940
C i
hPa

BALDWIN AND DUNKERTON:

ARCTIC OSCILLATION

100

t I
1958 1959

, 1960

, ., , ,l 1961

1962

hPa !

100i

1000[
1963

1964

1965

1966

1967
C

10 -hPa
lOO

-rll
1972

I
hPa

1968

1969

'

1970

1971

10 i ..... t'' I
I

100

lOOO

193 ' ' '


C

1974

1975

1976
C

1977

10 hPa lOO

lOOO 1978

1979

1980

1981

1982

lO I
hPa

C _1

lOO

1000

1983
10 hPa lOO

1984

1985
_ I ! i i

1986
, . .

1987

lOOO

1988
10 hPa
100

199

I
:

1990

1991

1992

!,

1000

1993

1994

1995

1996

1997

Plate The10-day 1. low-pass filtered signature series 1958-1997. AO time for Contours +0.5,+1.0,+1.5, are withvalues between and0.5notshaded. correspondsa weak, -0.5 Red to warmpolar vortex, while blueindi-

catesstrong, vortex. a cold Vertical denote November-March season. ticks lines the winter Red indicate major midwinter early warmings or final during December-March. red"C" Each indicates aCanadian warming.

BALDWIN

AND DUNKERTON:

ARCTIC

OSCILLATION

30,941

hPa

100

1000 .......
1958

1959

1960

1961

1962

10
hPa
100

"

,i. nil

. C] '

C/I II

el

1000 ..... &


........

1963
hPa

1964

' I '

1965

1'966

1967

10'i a..___.
i
1969 1970

1000, , ,,, ..... / 1968


10 hPa 100

100q
1973

1971

1972

10001k', i......,I i
hPa

1974

1975

'

'1976'
,C i

1977

1 t
,

1ooo /
hPa lOO

100 t
1978

..
I ' ' 1979 ...... I
1980

1981

1982

1000 ......

,,,,, , , ,

, , ', , ,',,, , ,,

30,942

BALDWIN

AND

DUNKERTON:

ARCTIC

OSCILLATION

formedfor everyday of the 40-yeardata set at eachlevel, yielding 17 x 14609arrayof the dailyAO ocoefficients a at eachlevel.The time series eachlevelrepresents daily at the 30 values theAO signature of (positive, corresponding a strong to hPa polarvortex,or negative, representing weakvortex)and is a calculated independently the otherlevels. of The resulting ar100 ray,whichwe callthe AO signature series, be usedto time may examine whether signals AO tendto appear the surface at and 3OO propagate upward thestratosphere viceversa. to or The AO signature time series usedto calculate cliwas the 1000 matological amplitude, AO illustrated Figure4. Beforethe in 70 80 NP 2b 3 4b 5 60 ' EQ originalEOF calculation wasperformed, datawere northe malized dividing the standard by by deviation the Decemof ber-February90-day low-pass geopotential fields.Thus the RegressionAO Index of withZonal MeanWind AO signaturetime seriesis nondimensional each level. at Contours showthe climatological amplitude, AO definedas the rms valueof the AO signature time series eachdayof for theyearand eachdatalevel.The climatological amplitude AO peaks nearFebruary at all levels. relatively 1 The large values in thetroposphere near10hPaindicate theAO accounts and that for a largerfractionof the spatial variance therethan at levels in the lowerstratosphere. Thereis a cleardistinction between thebehavior the AO in the troposphere the AO in the of and stratosphere, the climatological with stratospheric ampliAO EQ 10 20 30 40 56 66 760 86 NP tude droppingrapidly in springto very low levelsduring lune-October. LargeAO amplitudes found when stratoare Figure (a) Correlations 2. between zonal-mean zonal wind winds are westerly, allowingthe upwardpropagation andthe AO indexduringDecember-February. minimum spheric The correlation is -0.72, and the maximum is 0.96. (b) Same as of planetary waves from the troposphere. contrast, cliIn the Figure except 2a, regression between zonal-mean zonalwind matological tropospheric amplitude AO diminishes gradually andthe AO index.Contourvalues meters second, are per cor- in the spring,reaching minimum of-1/3 of its wintertime a
lO
i i i

Correlation of AO Index with Zonal Mean Wind

10 'i!iii':iiii'iiiii!iiii ' ' ii:'.8 'I ,,3 %&' '''

responding a onestandard to deviation anomaly theAO inin


dex.

Corq, ofAQIndex, Zonal elation, with Mean Temperature

perature 3a) 0.90150 and The a (Figure exceed hPa 80N.AO at


accounts 22% of the variancein temperature for north of

20N. the southern of the'dipole, In half correlations are


strongnear the midlatitudetropopause and extendto the equator. Figure3b illustrates regression the between zonalmeantemperature the AO time series and and shows that the AO couldbe described a deeppolar temperature as anomaly extending from the uppertroposphere the middlestratoto sphere, with a compensating anomaly southof-50N. In the tropics anomaly confined the stratosphere. pattern the is to A essentially similarto Figure3b wasshownin Dunkerton and Baldwin [1992]asthefirstEOFof zonal-mean temperature.

30
hPa 100
300

1000

EQ

10

20

30

40

50

60

70

80

NP

The aboveanalysis the AO describes spatially of a fixed patternand doesnot allow for the analysis verticaldevelof

3.2. Downward of AO PropagationSignature the

b1
3O

Regression AO Indexwith ZonalMean Temperature of ......... ! ........... !............. l::.u.u.:.'..'... . '

opment theAO or suggest of cause effect. and However, the


regressions result different in signatures (maps)for each the of 17pressure levels. These signature maps characterize AO at the

hPa
00

each level, thesignature canbe compared the and maps to


daily anomalies the data to measure in how closea daily anomaly mapisto thepattern representing AO.If ZlO^o the is the 10-hPasignature the AO, andZI 0 is a dailymap of deof
seasoned10 hPa Z, then

300
E 16 26 36 4b 56 6b 76 8b N

the AO signature the dailymap.This calculation per- onestandard to was deviation anomaly theAO index. in

Figure (a)Correlations zonal-mean 3. between temperature and AOindex the during December-February. The minimum correlation is-0.90 tlae and maximum 0.65. Same Figis (b) as ure except 3a, regression zonal-mean between temperature and defines asthemultiple Z10^o results a "best of theAO index. o of that in fit" Contour values kelvins, are corresponding to a

minZ10-czZ10^o l l2

BALDWIN AND DUNKERTON: ARCTIC OSCILLATION

30,943

Climatological Amplitudeof the A O


lO

30
hPa

100

300

1000

is no obvious preference anomalies eithersignwhenthe for of dataare90-daylow-pass filtered. Thereis alsoa tendency for morerapidvertical propagation whenthe anomalies reach the troposphere (e.g.,winters1958-1959and 1968-1969). The AO events which correspond a weakpolar vortex to showa remarkable correspondence stratospheric to sudden warmings. majormidwinterwarmings all final warmAll and ingsduringor beforeFebruary indicated Plates1 and 2 are in by red ticks at the top of each panel. The occurrences of warmings weredefined[K. Labitzke, University Berlin, Frei of personal communication, 1998] on the basis the synoptic of
situations and the standard definition of the World Meteoro-

whichrequires, 10 hPa,easterly at winds Figure Climatological 4. amplitude theArctic of Oscilla- logicalOrganization, tion. The contouredvaluesare the rms valuesof the AO sig- at 60N and warmer temperatures the pole than at 60N. at nature time series. Althoughwarmings appearto happenrapidly,thereis typicallya periodbeforethe warmingduringwhichthe vortexis reduced size(preconditioned). in Recovery thevortexaftera of warmingis largelya radiativeprocess and is relatively slow. peakamplitude, before beginning increase September. to in are magnitude theirlife cyand Thisbehavior seen is clearly a comparison thevariability Thewarmings of largeenough in of for events be seen to clearly 90-day in of 1000-hPaand 50-hPa geopotential over the polar cap cleislongenough these low-pass-filtered data. In the stratosphere, suddenwarmings [Thompson Wallace, and 1999]. the negative, a returnto but The height-time development AO signature of time series, cause AO indexto rapidlybecome positive values alsorelatively is slow.(The warmings tend to for40years data, illustrated Plate Thedata of is in 1. have been 10-day low-pass filtered eliminate to synoptic-scale Red occurrather abruptly at the beginningof the positiveAO events. and the eventsare sometimes followedby lingering shading corresponds a weakpolar vortex,as during a events, to anomalies.) skewness the unfiltered10-hPa The of AO stratospheric warming, whileblueshading indicates strong, positive a signature time series -0.70,whilethat of the 90-daylow-pass is coldvortex. valueof 1.0corresponds thedefinition the A to of is filter tendsto make sudden AO signature each at level(Figure Thecontour 1). interval is time series -0.07. The low-pass warmingsappearmore symmetric time. in 0.5 with values between -0.5 and 0.5 not shaded, and values Every major or early final warming during Decemberexceeding-1.5 1.5 in solidred or blueshading, and respec2 tively. Vertical gray linesdefine November-March the winter February(markedin Plates and 3 by a wide red tick) corresponds a largenegative to deviation the AO signature in time season. largemagnitudes the AO are,in the stratoThe of In sphere, confined almost entirely theperiod to October-April, series. addition,many,but not all, of the earlywinterevents to warmings," markedby a red "C," whileoccasional large-amplitude events be foundin the correspond "Canadian may to troposphereanytimeof theyear, at consistent theclima- duringwhich the Aleutianhigh develops disruptand diswith placethe vortex.If the vortexis displaced from the pole,but tology illustrated figure Manyof thestratospheric in 4. events the in are connected tropospheric to events, the connection but is withoutdiminutionof the circulation, signature the AO time series minimal. is typically intermittent, involving fairly rapid vertical bursts magnitude The low-frequency vertical development the AO wasalso of whichtendto descend fartherinto the troposphere overa peinvestigated usinga lag correlation technique. Figure5 illusriodof several weeks (e.g.1960-1961, 1968-1969, 1973-1974). between 90-daylow-pass signathe AO In somecases, downward penetration positive of (red) AO tratesthe correlation at 1 anomalies occurred continuously a periodof N2 weeks ture time series 10 hPa on January (the key datefor the over calculation) with the AO signature time series all otherlevat (e.g., 1958-1959, 1984-1985). elsduringNovember-March. The dominantfeatureis strong In the stratosphere, winterhastypically each onlyone or through the lower stratosphere into two sustained large-amplitude anomalies, whilethe timescale downwardpropagation with exceeding at 1000 0.65 in thetroposphere appears muchshorter. isolate low- the troposphere, lag correlations To the

frequency vertical propagation visible Plate thedatawere in 1, 90-day low-pass filtered. Above hPa, 150 66%of theunfiltered variance retained the 90-day is by low-pass filter,whilebelow 10 200hPa,only39% is retained. Plate shows manyevents 2 that (typically or twoperwinter) connected thevertical one are in fromthe middlestratosphere the surface. troposphere 30 to The hasevents withlarge amplitudes AO whicharenot connected hPa to thestratosphere, especially outside winterseason. the Some lOO stratospheric donotreach troposphere, in most events the but cases thereis some penetration the troposphere. of Typically, theAO anomalies beginat 10hPaor above movedown- 300 and wardovera period a fewweeks. timedelay quite of The is vari-

LagCorrelation AOSignature Series with Time '

able,rangingfrom near zero (e.g.,December1996) to more 1000 November December March January February than a month (e.g.,December 1968).Thisreflects sporadic the natureof the burstsof largeAO magnitude. Only when lowFigure5. Correlations between 90-daylow-pass the AO pass filtereddoesthe downward propagation appearsmooth. signature series 10hPaon January withtheAO sigtime at 1 Anomalies both signs of appear propagate to downward; there nature timeseries alllevels at duringNovember-March.

30,944

BALDWIN AND DUNKERTON: ARCTIC OSCILLATION


Correlationbetweenthe AO andthe QBO
I I

influence theQBOappears beconfined early from to to win-

:::::::::::::::::::::::::::::::::::::::::::::::::: ::::' i!: .... ' 10 ..........................':::::::4::::::::: '.................... :o


30 hPa

x'Os o

ter.BaldwinandDunkerton[1998] showed that the influence

of theQBOon thezonal-mean is also early wind an winter phenomenon, thelargest with influence January greatly in and
diminished February. by The signatures the QBO and the AO arenot the same-of

300

theQBOsignaturemore is confined thestratosphere its to and influence the high-latitude on circulation negligible is after
January. The correlations betweenthe QBO and the AO are consistent correlations with between QBO andzonal-mean the
,
November December January

1000,

,
February March

Figure6. Correlation between 40-hPaquasi-biennial the oscillation (QBO) and the AO signature time seriesduring November-March. AO time series The varies dayandlevel, by whiletheQBO is defined monthly-mean by Singapore winds.

500-hPa Climatology 2 * AO Signature Z -

hPaN3 weeks later.The thickline illustrates peakcorrelathe tion at each level. Thelagcorrelations figure5 should be in not interpreted indicate precise to a downward propagation speed; rather, theyindicate overall an tendency downward for propagationon a timescale a few weeks. of The downward propagation the troposphere coninto is fined to low-frequency variability.Lag correlations using higher frequencies (e.g., 90-day high-pass data) showlittle connection the troposphere to with the highestcorrelations foundnearzerolag,sothatthetilt of the contours Figure in 5
is not present.

The same calculation shown Figure5 wasperformed as in usingkey dates rangingfrom November1 throughMarch 1 (not shown). Both earlierand later dates resulted slightly in lower correlations the troposphere similardownward in but propagation. During early winter (Novemberthroughearly December) contours the stratosphere morevertical, the in are indicating more rapid verticalpropagation. winter proAs gresses, speed downward the of propagation tends beless, to so that by March,the slopeof the contours the stratosphere in tends beslightly vertical to less thanseen Figure in 5.
3.3. Relation to the Quasi-Biennial Oscillation

500-hPa Climatology 2 * AO Signature Z -

The phase of the equatorialquasi-biennial oscillation (QBO), throughmodulationof the zonal-mean wind, affects the upwardand equatorward propagation planetary-scale of waves. Warmings morelikelywhenthe QBO is in its eastare erly phase, and Plates1 and 2 showan excellent correspondence between AO andsudden the warmings. Holtonand Tan [1980] examined northernhemisphere geopotential the on basisof the phaseof the 50-hPaQBO and showed stratoa spheric signature the QBO that is similarto that of the AO. of Theyalsoshowed, usinga 16-year dataset,a 1000-hPa signatureof the QBO duringJanuary wassimilar thatof the that to AO. An updateof that figure,usingdatafor 1964-1996(not shown), confirms the patternpersists. that Figure shows 2 that correlations between the AO amplitude and lower stratospheric equatorial zonal-mean wind are not zero,suggesting thattheQBO mayactto modulate AO. the 5800 A time-height cross section correlations of between sigAO naturetime series (for eachdayof yearandlevel)and SingaFigure Extremes theArctic 7. of Oscillation 500hPa: at (a) pore40-hPawind (whichvaries month)is shown Figure by as December-February climatology 2 x AOsignaplus 6. Correlations peak above0.45 at 10 hPa during December 500-hPa ture, (b) 500-hPa and December-February climatology minus and fall off to less than 0.20 by February, indicating that any 2 x AO signature.

BALDWIN

AND DUNKERTON:

ARCTIC OSCILLATION

30,945

anomaly shows strong a connection the stratosphere, to suggesting tropospheric that variability wouldbe less withoutinvalues below tropopause. QBOappears be a factor the The to zonalwind [Baldwin Dunkerton, and 1998],whichshow lower
that tendsto excite AO, but the influence strongest the the is in

stratosphere duringearly winter.


4. Discussion

fluence from the stratosphere. Thepattern correlation SLPandlowerstratospheric of geo-

potential greater is than that between SLP and midtropospheric geopotential (Figure Thompson Wallace 1). and suggested that the midtropospheric patternis a hydrostatic re-

the two extremes the AO is considerable. FigUre twice of In 7a

to Thedifference tropospheric in circulation patterns between sponse low-levelthermal advectionattributableto winds associated the SLPpattern.The vertically with averaged temadvection calculated Thompson Wallace by and [1999] the 500-hPa anomaly added the climatological AO is to mean perature with this interpretation. is therefore It 500-hPa flow,whilein Figure7b twicethe AO anomaly is is entirelyconsistent

subtracted from the climatology. The AO magnitudeis observed be thislargeat 500hPaon -3% of winterdays. to Since surface cyclones tend to be steered the 500-hPaflow and by typicallymove at -% the speedof the 500-hPawind, these maps suggest dramaticdifference the location of storm in tracks andsurface cyclone activity between two extremes the of
the AO. In the Atlantic sector the effects on weather will be

important understand cause the SLPanomaly its to the of and relationto the precursor signal the uppertroposphere in and lower stratosphere. Elevation polar-capSLP impliesa net of transport massinto this regionand viceversa. possible of A
mechanism for the influx of mass is an anomalous westward

similar to that of the NAO, asdiscussed Hurrell [ 1995] and in

references therein. WhentheAO indexis negative (Figure7b), the flow overcoastal Europeis weakand is from the westor west northwest. contrast, In when the AO index is positive, thereis strong westsouthwesterly flow. Large-amplitude stratospheric anomalies AO tend to precedelarge-amplitude tropospheric anomalies, AO whichhave an effect stormtracks, strength tropospheric and on the of jets, weather, especially Europe. over Therearecase studies anand ecdotal evidencethat sudden warmings precede certain

weather patterns [e.g.,Quiroz, 1986].Stratospheric events are

precursors tropospheric to events when the stratospheric AO amplitude largeand persistent. is Long-range forecast models the vertical, creates conditions favorable for maintenance of by the of musthavea realistic representation stratospheric of processes the anomaly modifying direction eddymomentum in and and planetary wavepropagation the downward if propagation transportand wavebreaking the upper troposphere lowerstratosphere [Thorncrop al., 1993].The role of eddy et is to be forecast successfully.climate If models to represent are to and accurately relationship the between stratosphere the forcingremains be demonstrated, (if important)the the and and planetary-scale waves troposphere, suchmodelsshouldshownot just the AO, but relativecontributionof synoptic Synoptic-scale waves primarilyimare also downward propagationof the AO. Both downward shouldbe evaluated. throughoutthe year,while propagation the AO signature of and realistic seasonal evolu- portantin the upper troposphere wavesare primarily importantin the stratosphere tion of lag correlations in Figure5) havebeenobserved (as in planetary [Charney Drazin, and the UKMO Unifiedmodel(P.McCloghrie, OxfordUniversity, duringwinterwhenwindsarewesterly 1961]. It appears that planetarywavesplay an important, personal communication). the stratoThe mechanism downward for propagation AO anoma- thoughindirect,role in causing large-amplitude of spheric AO anomalies whichpropagate slowlydownward to lieslikelyinvolves modulation the waveguide planetaryof for affectthe troposphere. is possible It that a positive feedback scale waves high-altitude by anomalies. zonal-mean In wind, is Theseconsiderations invite questions anomalies modifythe climatological waveguide distribu- mechanism involved. and Is forcingdirectly responsition of acceleration from convergence the Eliassen-Palm for furtherstudy. the stratospheric of blefor evolution thetropospheric anomaly winter,or of AO in flux [Kodera al., 1990].The net resultfrom an anomaly et in indirect, triggering modification transport a of the middle stratosphere appears be that the zonal-mean is its influence to waves? theinfluence stratospheric Is of forcing felt windbelowthe anomaly, overa periodof days, tends follow by synoptic to longerthan a month? Our analysis of the anomalyabove. This process may be difficultto observe primarilyon timescales lag correlationssuggests that the downward propagation because the downwardpropagation involveslow-frequency of AO is in anomalies not day-to-day and variability. understanding throughthe troposphere a "clean" signal observed An of data the mechanism involvedwill likely requireexperiments with low-pass-filtered but not in unfiltereddata. This result to fluctuations associated mechanistic climatemodels.Suchexperiments and may illu- may be attributable high-frequency with tropospheric weathersystems to modificationof the and minatewhatis required medium-range of weather forecasting wind profileaffecting vertical the propagation of models exploitstratospheric to variability improve to weather zonal-mean planetarywavesinto the lower stratosphere timescales on forecasts. The variabletime for stratospheric anomalies to

(negative) body force in the upper troposphere lower or stratosphere middle to high latitudes. in The Eulerianmean meridionalcirculationinducedby sucha force is generally polewardand downward, tendingto counteract applied the force,with an equatorward return flow in the lowermost troposphere. Haynes and Shepherd [1989]notedthat streamlines of inducedmean meridionalcirculation not completely are closed, because the proximity lowerboundary. lowof of The levelreturn flow therefore doesnot entirelycancel transthe port of massin the upper branch,and, as a result,surface pressure the polar cap increases. mechanism linear, in This is and the argument may be applied,in reverse, a positive for bodyforceanomaly. It is possible an AO anomaly, that equivalent barotropic in

penetrate troposphere be an indication the may that the troposphericanomaliesmay be largely independentof the 5. Glossary stratosphere until they are of the samesign.In this case, the appearance downward of propagation wouldbe the resultof AO index:Theprincipal component series thefirst time of amplification theAO anomalies of whentheyalignin the ver- EOFof 90-day low-pass filtered December-February geopotical.Nearlyeveryinstance a large-amplitude of tropospheric tential fivelevels at (1000,300,100,30, 10hPa).

shorter than a month.

30,946

BALDWIN AND DUNKERTON: ARCTIC OSCILLATION


Hurrell, J.W., Decadal trends in the North Atlantic Oscillation: Re-

AO signature: Regression theAO indexwith a datafield, of such zonal-mean as wind or geopotential oneof 17pressure at
levels.

gionaltemperatures precipitation, and Science, 676-679,1995. 269, Kalnay, M.E., et al., The NCEP/NCAR Reanalysis Project,Bull.Am.
Meteorol. $oc.,77, 437-471, 1996.

AO signature time series:The dailyvalueof theAO signa- Kerr,R., A new forcein high-latitude climate,Science, 241-242, 284, 1999. turedefined themultipleof theAO signature, thatresults by cz, characteristics reof in a "best to the data.For each 17levels fit" of thereis a 40-year Kodera,K., and H. Koide,Spatialand seasonal
time series.

Climatological amplitude: The rms valueof the AO AO signature series, time whichvaries dayof yearandlevel. by
Acknowledgments. research supported the National This was by Science Foundation, grants ATM-9500613 ATM-9708026, by and and the NationalAeronautics Space and Administration, contract NAS196071. The NCEP Reanalysis data were providedby the NOAACIRES Climate Diagnostics Center,Boulder,Colorado,from their Web siteat http://www. cdc.noaa.gov. thankthe following We people for helpfuldiscussions: Allen, R. Black,N. Gillett,L.J.Gray,P.H. M. Haynes, W.A. Lahoz,P.McCloghrie, M.E. Mcintyre,W.A. Norton,A. O'Neill,D.W.J. Thompson J.M.Wallace. and

centdecadal trendsin the northernhemispheric troposphere and stratosphere,Geophys. 102,19,433-19,447, J. Res., 1997.

Kodera, K., K. Yamazaki, M. Chiba, and K. Shibata, Downward

propagation upperstratospheric of meanzonalwind perturbation to the troposphere, Geophys. Lett.,17, 1263-1266, Res. 1990. Kodera, H. Koide,andH. Yoshimura, K., Northernhemisphere winter
circulation associated with the North Atlantic Oscillation and

stratospheric polarnightjet, Geophys. Lett.,26, 443-446,1999. Res. Kutzbach,J.E., Large-scale featuresof monthly mean northern hemisphere anomalymaps of sea-level pressure, Mon. Weather
Rev., 98, 708-716, 1970.

Nigam, S., On the structure of variability of the observedtropospheric and stratospheric zonal-mean wind, J. Atmos. $ci., 47,
1799-1813, 1990.

References
Baldwin, M.P., and T.J.Dunkerton, Quasi-biennial modulation of the

Quiroz, R.S., The association stratospheric of warmings with tropospheric blocking, Geophys. 91, 5277-5285,1986. J. Res.,
Shindell,D.T., R.L. Mille, G. Schmidt, and L. Pandolfo,Simulation of

recentnorthernwinterclimate trends greenhouse-gas by forcing,


Nature, 399, 452-455, 1999.

southernhemisphere stratospheric polar vortex, Geophys. Res.

Thompson, D.W.J., J.M.Wallace, ArcticOscillation and The signature Lett., 25, 3343-3346, 1998. in thewintertime geopotential heightandtemperature fields, GeoBaldwin, M.P.,X. Cheng,and T.J.Dunkerton,Observed correlations phys. Res. Lett.,25, 1297-1300,1998. between winter-mean tropospheric stratospheric and circulation Thompson, D.W.J.,and J.M.Wallace, The structure the Arcticand of anomalies, Geophys. Lett.,21, 1141-1144,1994. Res. Antarctic oscillations, Climate, press, J. in 1999. Bretherton, C.S.,C. Smith,and J.M.Wallace, intercomparison An of Thorncroft, C.D., B.J. Hoskins andM.E. Mcintyre,Twoparadigms of methods findingcoupled for patterns climatedata,J. Clim.,5, in baroclinic-wave life-cyclebehaviour.Q. J. R. Meteorol. $oc.,119, 541-560, 1992. 17-56, 1993. Charney, andP.G.Drazin,Propagation planetary-scale J.G., of disturbances from thelowerinto the upperatmosphere, Geophys. J. Res., Walker,G.T., and E.W. Bliss,World Weather V, Mere. R. Meteorol.$oc.,
66, 83-109, 1961.
IV, 36, 1932.

Dunkerton, T.J., and M.P. Baldwin, Modes of variability in the stratosphere, Geophys. Lett.,19,49-52, 1992. Res. M.P. Baldwin and T.J. Dunkerton, Northwest Research Associates, Haynes, P.H.,andT.G. Shepherd, importance surface The of pressure NE Bellevue, 98007-3713. WA (mark@nwra.com, changes the response the atmosphere zonallysymmetric 14508 204Street, in of to thermalandmechanical forcing, J. R. Meteorol. Q. $oc.,115,1181- tim@nwra.com)
1208, 1989.

Holton, J.R.,and H.-C. Tan, The influenceof the equatorialquasibiennialoscillation the globalcirculation 50 mb, J. Atmos. on at
Sci., 37, 2200-2208, 1980.

(Received January 1999; 28, revised April 28, 1999; accepted June17, 1999.)

Das könnte Ihnen auch gefallen