GEOPHYSICAL RESEARCH LETTERS, VOL. 25, NO. 9, PAGES 1297-1300, MAY 1, 1998
The Arctic Oscillation signature in the wintertime
geopotential height and temperature fields
David W. J. Thompsonand JohnM. Wallace
Departmentof AtmosphericSciences,Universityof Washington,Seattle
Abstract. The leading empirical orthogonal function of the
wintertimesea-levelpressurefield is more strongly coupledto
surfaceair temperaturefluctuationsover the Eurasiancontinent
than the North Atlantic Oscillation (NAO). It resembles the
variablein the analysisis SLP(p), expressed
in termsof the
equivalentheight of the 1000-hPasurface(Z•000)abovesealevel using the approximaterelation Z•000= 8(/9-1000) where
Z•000
is expressed
in metersandp in hPa.
NAO in many respects;
but its primary centerof action covers
As the reference variable we use the leading principal
more of the Arctic, giving it a more zonally symmetric componentof the wintertime(November-April)monthly mean
appearance.Coupledto strongfluctuationsat the 50-hPa level SLP anomaly field over the domain polewardof 20øN, which
on the intraseasonal, interannual, and interdecadal time scales, accountsfor 22% of the varianceand is well separatedfrom the
this "Arctic Oscillation"(AO)can be interpretedas the surface other eigenvaluesas per the criterion of North et al. [ 1982].
signatureof modulationsin the strength of the polar vortex The associated loading vector or empirical orthogonal
aloft. It is proposedthat the zonally asymmetricsurfaceair function (EOF)is shown in the lower right panel of Fig. 1.
temperature and mid-tropospheric circulation anomalies First identifiedby Lorenz [1951] in zonally averagedSLP data
observed in association with the AO may be secondary andsubsequently
by Kutzbach [1970], Wallaceand Gutzler
baroclinicfeaturesinducedby the land-seacontrasts.The same [1981], and Trenberthand Paolino [1981] in griddeddata, this
modalstructureis mirroredin the pronouncedtrendsin winter mode involvesa seesawbetween the Arctic basin and parts of
and springtimesurfaceair temperature,sea-levelpressure,and the surrounding zonal ring. It is a robust pattern that
50-hPa height over the past 30 years: parts of Eurasiahave dominates both the intraseasonal (month-to-month) and the
warmedby asmuchas severalK, sea-levelpressureover parts interannual variability throughout the entire 98-year SLP
of the Arctic has fallen by 4 hPa, and the core of the lower record,as documentedin Fig. 2. In seasonallyaverageddatafor
stratosphericpolar vortex has cooled by several K. These the past 30 years the Pacific center of action is partially
trendscan be interpretedas the developmentof a systematic obscuredby interdecadaltrends(discussedlater). Howeverthis
biasin one of the atmosphere'sdominant,naturallyoccurring featurecanbe recoveredby removingthe linear trendfrom the
modes of variability.
data at each gridpoint before performing the EOF analysis
(Fig. 2, right panel). Althoughthis pattern incorporatesmany
Introduction
of the featuresof the NAO, its slightly larger horizontal scale
There is a growing body of evidenceindicating that the and higherdegreeof zonal symmetryrenderit more analogous
stratospheric polar vortex is implicated in some of the to the leadingEOF of SLP in the SouthernHemisphere[Rogers
interannual and secular variability of climate at the earth's and van Loon, 1982], and more like a surfacesignatureof the
polar vortex aloft. To distinguishthe pattern in Fig. 1 from
surface.Baldwin et al. [1994], Perlwitz and Graf [1995],
the more regional NAO we will herein refer to it as the Arctic
Chengand Dunkerton[1995], andKitoh et al. [1996] have all
(AO) andthe associatedprincipal componenttime
documentedthe existenceof coupling betweenthe strengthof Oscillation
series as the AO index. This subtle distinction, first hinted at
the nearly zonally symmetric polar night jet at the 50-hPa
level and a more wavelike pattern reminiscent of the North in the analysis of Baldwin et al. [1994], is essential for
appreciating the strength of the linkages discussedin the
Atlantic Oscillation (NAO)at the 500-hPa level. Hurrell
[1995] has shown that the NAO modulates wintertime surface following paragraphs.
The lower left panel of Fig. 1 shows wintertime monthly
air temperature
(SAT) overmuchof Eurasia:negative sea-level
pressure(SLP) anomaliesover Iceland and enhancedwesterlies mean SAT anomaliesover land regressedupon the AO index.
pattern is similar to Fig. 3 of Hurrell [1995], but the
acrossthe North Atlantic at 50øN (i.e., the positivepolarity of The
correlations in Table 2 indicate that the AO accounts for a
the NAO) are associated with positive SAT anomalies
extendingfrom Great Britain and Scandinaviafar into Siberia. substantially larger fraction of the variance of Northern
Hurrell [1996] wenton to demonstrate
thatthe upwardtrendin HemisphereSAT than the NAO.
Patterns derivedby regressingwintertime monthly mean
the NAO during the past 30 years accountsfor much of the
geopotential
height fields onto the normalized AO index are
warmingin SAT averagedover the domainpolewardof 20øN.
Koderaand Yamazaki[1994], Graf et al. [1995], andKodera also shown in Fig. 1, together with the corresponding
and Koide [1997] have suggestedthe possibility of a "expansioncoefficient time series"generatedby projecting
dynamicallinkage betweenthe recent wintertime warming monthly fields upon the respectiveregressionpatterns and
over Eurasiaand a strengtheningof the polar night jet and averaging them over winter seasons. The SLP and 50-hPa
Robockand Mao [1992], Graf et al. [1994], andKodera [1994] height patternsare remarkablysimilar and their expansion
have invoked similar linkages to explain the observed coefficienttime seriesare strongly correlated,as indicatedin
positive SAT anomalies over Eurasia during the winters Table 3. The 50-hPapatternis indicativeof fluctuationsin the
following major volcaniceruptions.In this short contribution strength of the polar vortex (Fig. 3). This 50-hPa height
we will offer our own analysis and interpretation of these regressionpattern is virtually identicalto the leadingEOF of
50-hPaheightshownin Fig. 4. In fact, a SLPpatternvirtually
relationshipsbasedon the datasetslisted in Table 1.
identicalto the leadingEOF in Fig. 1 can be recoveredin the
inverse
manner,by regressingSLP uponthe leadingprincipal
Results
component of 50-hPa height. The geopotential height
In this sectionwe constructa three-dimensional
pictureof anomaliesat the 50-hPa level are almost five times as strong
the primarymodeof wintertimevariability over the Northern as the 1000-hPa height anomaliesassociatedwith this mode.
Hemisphereworking from the groundup. The primary field Since the spatial patterns at the two levels are similar, it
followsthat energydensity(the productof densityandsquared
amplitude) is nearly invariant over this height range.
Copyright1998 by the AmericanGeophysical
Union.
Regressionpatterns basedon subsequentEOFs of SLP (not
shown) do not exhibit any upwardamplification from the
Papernumber98GL00950.
troposphereto the stratosphere.
0094-8534/98/98GL-00950505.00
The 500-hPaheightpatternin Fig. 1 is the sumof the SLP
1297
1298
THOMPSON AND WALLACE: THE ARCTIC OSCILLA•ON
Table 1. Datasetsused in this study.
Resolution
Source
Period of record
See Jones[1994] and Parker et al. [1995].
5øx 5 ø
1900-6/1997
Sea level pressure(SLP)
5ø x 5ø
1900-4/1997
NCAR Data SupportSection
Geopotential
heightand
2.5øx 2.5 ø
1958-6/1997
NCEP/NCARReanalysis
via NOAA ClimateDiagnostics
Center
: Surfaceair temperature(SAT)
i
tropopause pressure
Trop.
Z',d •'•
•
:A• .. • •
1960 1970
V
v -/"x•
••
X/•-z•
1980 19•0
Figure 1. Leftsixpanels:
Regression
mapsforgeopotential
height(Z), tropopause
pressure,
1000-500-hPa
thickness
and
surface
airtemperature
(SAT)anomalies,
asindicated,
based
uponthe leading
principal
component
of wintertime
(Nov-April)
monthlymeansea-level
pressure
anomalies
(theAOindex)for 1947-1997.
Contour
intervals
(expressed
in unitsperstandard
deviation
of theAO index)are:10m (-5, 5, 15...)for SLP(expressed
asZ•0oo),
Zsoo,
andZsoo-Z•0oo;
30 m (-45, -15, 15...)for Z,,; 5-
hPa(-2.5,2.5, 7.5_.)for tropopause
pressure;
0.5 K (-0.75, -0.25, 0,25...) forSAT.Negative
contours
aredashed.
Extrema
are
labeled
in theappropriate
units.
Topfight,timeseries:
(Toptobottom)
Normalized
expansion
coefficient
timeSeries
fortheZ,,,,
Z5oo,
Z•oo-Z•ooo,
andSIP regression
mapsdepicted
at left;normalized
Eurasian
mean(40 -70øN,0 -140øE)
SATanomalies.
Time
seriesarewintertime
(Nov-April)seasonal
means
based
on monthly
datafrom1958-1997.Correlation
statisticsarepresented
in
Table 3. Horizontalaxesrepresent
the meansfor years1958-1967.
(or 1000-hPa) pattern below it plus the 1000-500-hPa
thicknesspatternto the left of it. The featuresin the thickness
field are more clearly reflectedin the 500-hPaheight field than
in the more zonally symmetricSLP and 50-hPa height fields. It
follows that the temperatureanomalies in the 500-50-hPa
layer mustlargelycancelthosein the lower troposphere.This
compensation
is reflectedin the patternof tropopausepressure
regressed
upon the AO index (Fig. 1, upper left panel). The
1000-500-hPathicknessanomaliesbear a strong resemblance
Intraseasonal
Interannual
Interannual
1947-97
1919-68
1968-97
to the SAT anomalies in the panel below. The interrelationshipsbetweenthe variouspatternsshownin Fig. 1 are
entirely consistentwith the resultsof the previous studies
cited in the Introduction, but our inclusion of the additional
fields in the analysisreveals more clearly a three-dimensional
modal structureconsisting of two components: a deep
equivalent barotropic signature extending high into the
stratosphere and a tropospherically confined baroclinic
signature.The equivalentbarotropicsignatureis dominatedby
Table 2. Correlationstatistics(r) for the NAO a and AO from
1900-1995 (Nov-April monthly values).r (NAO, AO) = 0.69.
T•HL.nd
T•HLand+Ocean T•ur•i.
b
NAO
0.17
0.16
0.2•
AO
0.39
0.36
0.55
a Thenormalized
monthly
meansealevelpressure
(SLP)difference
Figure2. Left'to right:Normalized
leading
EOFs
of
betweenPontaDelgada,Azoresand Stykkisholmur,Iceland.
detrended as described in text.
the AO.
b If Lisbon
is substituted
for theNAOAzores
station
(i.e.,see
wintertime (Nov-April) SLP anomalies for: intraseasonal
Hurrell and Van Loon, 1997), and Dec-March means are used, then
(month-to-month variability within winters) 1947-1997;
correlations with Eurasian mean SAT are 0.57 for the NAO and 0.65 for
seasonalaverages1919-1968; seasonalaverages1968-1997
THOMPSON AND WALLACE: THE ARCTIC OSC2LI•TION
1299
Table 3. Correlations(r) with AO index (Nov-Apr 1958-97).
,.
AO
Monthly
Interannual
Intraseasonal
Zsoo
0.95
0.95
0.94
Zso
0.63
0.80
0.53
Tœ=•i
a
0.62
0.77
0.53
0.88
0.94
0.85
Zsoo-Z•ooo
SAT
AOindex
v v
1900
19•0
19•0
1960
1980
2•
the zonally symmetriccomponent,
•whereas
the baroclinic
signatureis more wavelike.Thesetwo signaturesare not Figure •. Top to bottom:No•alized winteaimeexpansion
separatemodes,but components
of a singlemodalstructure,as
evidencedby the high correlationsbetweenthe AO index and
the expansion coefficient'time series of the 1000-500-hPa
thicknessand 500-hPaheightpatterns(Table 3).
The intraseasonal
andthe detrended
interannualvariability
(not shown)exhibit virtually identicalvertical structures.The
couplingbetweenthe 50-hPa height andSLP fields is not as
strong in the intraseasonalvariability (Table 3), but it is no
less statisticallysignificantowing to the largernumber'of
degrees of freedom. Both intraseasonal and interannual
coefficienttime seriesfor the SAT and SLP regressionmaps of
Fig. 1 for 1900-1997. •e
light lines denoteunsmoothed
seasonalaverages(r = 0.86); heavy lines denotefive-year
running means.
relative
amplitudes
ofthetrends
ofthethree
fields
inFig.6 are
alsoquite comparableto those in Fig. 1: in eachcasethe 30year trend amountsto about 1.25 stand•d deviationsof the
month-to-monthvariability. Hence,the observedtrendscan
be viewed as a bias in a prefe•
mode of month-to-month
variability.
variabilityexhibit substantial
upwardamplification.
Figure5 showstheAO indexandthe expansioncoefficient
time seriesof the SAT field in Fig. 1 for the more extended
period of record 1900-1997. Strong coupling is evident Interpretation
between the AO and the time series of its attendant SAT field
Anecdotal evidence of deep vertical coupling in the
on the interannual timescale, and both time series exhibit a
trendover the past 30 yearswith falling SLPover the Arctic
andwarmingoverSiberia.Thesetrendsarealsoclearlyevident
in Fig. 1.
wintertime polar vortex includesevents such as the dramatic
stratospheric warming of January 1977, which was
accompanied
by the formationof a strongsurfaceanticyclone
over the Arctic [Quiroz, 1977]. However it has not been until
the relatively recent studies cited in the Introduction that
Thirty-year(1968-1997) linear trendmapsfor wintertime
averagedSAT, SLP, and 50-hPa height (Fig. 6) strongly convincing statistical evidenceof this coupling has been
resemblethe modalstructuredepictedin Fig. 1, but for one forthcoming.
notable disagreement:the 30-year SLP trend is negative over
both the Arctic and the North Pacific, whereas in the modal
structurein Fig. 1 the SLP anomaliesin the two regions are of
opposite sign. The pressuredecreasesover the North Pacific
were largeenoughto cancelthe weak center of action over the
North Pacific in the leading EOF of the seasonal-meanSLP
field duringthis period:hencethe needfor the removal'of the
linear trendto recoverthe AO patternin the fight panel of Fig.
2. ENSO-like interdecadal variability documented by
Trenberth and Hurtell [1994] and Zhang et al. [1997] has
contributed
to this anomalous
behavior
in the Pacific
sector.
Despite this discrepancy,the spatial correlation between the
SLP trend and the EOF in Fig. 1 is 0.77. The observed
wintertime warming over Eurasia and the cooling over
Labrador[IPCC, 1995], the persistentcyclonic surfacewind
anomalies over the Beaufort Gyre [Walsh, 1996], the
increasing prevalenceof the high index phase of the NAO
[Hurrell, 1995], and the strengtheningof the stratospheric
polar night jet [Graf et al., 1995; Kodera and Koide, 1997] in
recentyearsare all relatedto the deepeningof the polar vortex
The analysisin the previoussectionconfirmsthe existence
of deepverticalcoupling
in thewintertime
polarvortexandits
relation
to SAT anomalies
over Eurasia and the Northwest
Atlantic duringan extended(November-April)winter season.
Whereas
thestudies
of Baldwinet al. [1994],PerlwitzandGraf
[1995], Chengand Dunkerton [1995] and Kitoh et al. [1996]
have representedthe troposphericcirculation in terms of the
500- or 850-hPa height fields, this studyhas emphasizedthe
SLP field. We haveshownthat fluctuationsin the intensity of
the stratosphericcirculationare linked to the leadingEOF of
SLP, a robust, quasi-zonallysymmetricmodeof variability
that has received much less attention than the more wavelike
teleconnection
patternsthat dominatethe leadingEOFsof the
mid-troposphericgeopotentialheight field. This deepvertical
coupling evidently prevails through a wide range of
frequencies
andit is simulatedin recent modeling studiesof
Kitoh et al. [1996], Kodera et al. [1996], and Volodin and
Galin [ 1998].
The strengtheningof the polar vortex over the past 30
years,unrelatedto any knowntropospheric
forcing, has led to
from the earth's surfaceto the lower stratosphere.The nearly speculation that anthropogenically induced temperature
five-fold amplification of the geopotential height anomalies changes at stratospheric levels might somehow be
from the troposphereto the 50-hPa level implies that the responsible.
Mindful of the numerousunsuccessful
attemptsto
lower stratosphereover the Arctic has cooled by several establish a dynamical basis for "downward control" of the
degrees.Not only are the spatial patternsof the trend similar troposphericcirculation, we are not convinced that the trends
to those associated'with the year-to-year variability, the originate in the stratosphere.Nevertheless,it is of interest to
Figure 6. Leftto right:30-year(1968-1997)lineartrends
(as indicated)for: SAT anomalies;
SLP (expressed
as Z]000)
hPaheight field for 1958-1997. Contourinterval is 150 m. anomalies;50-hPageopotentialheight (Z) anomalies.All
Figure 4 (right). The leadingEOFof wintertime50-hPa mapsarebaseduponwintertime
averaged
(Nov-April)data.
Figure 3 (left). Climatological wintertime (Nov-April) 50-
geopotentialheight anomaliesfor 1958-1997, which accounts Contourintervalsare(expressed
in unitsper 30 years):0.5 K (for 50% of the total variance. Contour interval 40 m (-60, -20,
20...). Negative contoursare dashed.
0.25, 0.25, 0.75...) for SAT; 10 m (-15,-5, 5...) for Z,o00;
30m
(-45,-15, 15...)for Z50.Negativecontours
aredashed.
1300
THOMPSON AND WALLACE: THE ARCTIC OSCR,I.ATION
considerthe ways in which sucha "top down" linkage could
conceivably operate.
Kodera and Yamazaki[1994] and Perlwitz and Graf[1995]
have hypothesizedthat the SAT anomaliesover Eurasiaand
Labrador are induced by an anomalous mid-tropospheric
planetary-wavepattern which develops in responseto the
intensificationof the stratosphericpolar vortex. Here we offer
an alternative interpretation motivated by the threedimensional modal structuredepictedin Fig. 1, which is the
only suchmode that exhibitsnearly constantenergydensityat
levels rangingfrom the surfaceto the middle stratosphere.We
hypothesize that strong coupling between troposphereand
stratosphereis intrinsic to the dynamics of the zonally
symmetric polar vortex; i.e., that undercertain conditions,
dynamical processesat stratosphericlevels can affect the
strengthof the polar vortex all the way down to the earth's
surfacethroughthe combinedeffectsof an induced,thermally
indirect mean meridionalcirculationanalogousto the Ferrell
cell and inducedchanges.
in the polewardeddyfluxesof zonal
momentum
at intermediate
levels.
We furtherspeculatethat the tropospheric
signatureof these
inducedfluctuationswouldbe zonallysymmetricwereit not for
the existenceof land-seacontrasts.A strongerzonal flow
Swift, Long-termcoordinatedchangesin the convectiveactivityof
theNorth Atlantic, Progressin Oceanography,
38, 241-295, 1996.
Graf, H. -F., J. Perlwitz, and I. Kirchner, Northern Hemisphere
tropospheric
mid-latitudecirculationafterviolentvolcaniceruptions,
Contr.Atm. Phys.,67, 3-13, 1994.
Graf, H.-F,, J. Perlwitz, I. Kirchner, and I. Schult, Recent northern
winter climate trends,ozone changes,and increased greenhouse
forcing,Contr.Atm. Phys.,68, 233-248, 1995.
_Hurtell,J. W., Decadaltrendsin the North AtlanticOscillation:Regional
temperatures
andprecipitation,
Science,269, 676-679,1995.
Hurtell, J. W., Influence of variationsin extratropical wintertime
teleconnectionson Northern Hemisphere temperature,Geophys.
Res. Len., 23, 665-668, 1996.
Hurrell, J. W., and H. van Loon, Decadal variations in climate
associated
with the North AtlanticOscillation,Clim. Change,36, 301326, 1997.
!PCC, Climate Change 1995: The Science of Climate Change, J. T.
Houghton,F. G. Meira Filho, B. A. Callander, K. Maskell (Eds.),
CambridgeUniv. Press,Cambridge,U. K., 1995.
Jones, P. D., Hemispheric surface air temperature variations: A
reanalysisandupdateto 1993,J. Climate,7, 1794-1802,1994.
Kitoh,A., H. Koide,K. Kodera,S. YukimotoandA. Noda,Inierannual
variabilityin the stratospheric-tropospheric
circulationin a coupled
ocean-atmosphere
GCM, Geophys,Res.Lett., 23, 543-546, 1996.
Kodera,K., Influenceof volcaniceruptionson the tropospherethrough
stratosphericdynamical processesin the Northern Hemisphere
should advect more mild marine air into the interior of the
winter,J. Geophys.Res.,99, 1273-1282, 1994.
continents, cold continental air over the western oceans, and
Kodera, K., and K. Yamazaki, .A possibleinfluence of recent polar
increasethe frequencyof Atlantic cyclonestracking through
stratosphericcoolings on the troposphere in the Northern
the Kara Sea[Rogersand Thompson,
1995],therebyinducinga
Hemispherewinter, Geophys.Res.Lett., 21, 809-812, 1994.
patternof SAT anomaliesmuchlike that in Figs. 1 and6. The Kodera,K., and H. Koide, Spatialand seasonalcharacteristics
of recent
mid-troposphericplanetary-wavepattern in Fig. 1 may be
decadal trends in the northern hemispheretroposphereand
stratosphere,
J. Geophys.Res.,102, 19433-19447,1997.
regardedas the hydrostaticsignatureof these temperature
anomalies;we questionwhetherit plays an essentialrole in
the couplingwith the stratosphere.
Kodera, K., M. Chiba, H. Koide, A. Kitoh, and Y. Nikaidou, Interannual
Concluding remarks
Lorenz, E. N., Seasonaland irregular variationsof the Northern
Hemispheresea-levelpressureprofile,J. Meteor.,8, 52-59, 1951.
variabilityof the winterstratosphere
and.troposphere
in the Northern
Hemisphere,J. Met. Soc.Japan, 74, 365-382, 1996.
Kutzbach, J. E., Large-scale features of monthly mean Northern
Hemisphereanomalymapsof sea-levelpressure,Mon. Wea. Rev.,
98, 708-716, 1970.
If the deepeningof the wintertimepolar vortex continues
into the 21st century,it couldhaveramificationsbeyondthe McPhee,M. G., T. P. Stanton,J. H. Morison,and D. G. Martinson,
Fresheningof the upper ocean in the Arctic: Is perennial sea ice
wideranging(butthusfar subtle)changesin surfaceclimate
disappearing?,
Geophys.Res.Lett., in press,1998.
discussedin this article and in Hurrell [1995]. For example:
ß the weakening
(andin time,perhapsevena reversal)of the
wind-drivenBeaufortGyrecouldfurtherreducethe extent and
thickness
of theArcticpackice [McPheeet al., 1998],
ß enhancedcooling and deep convection in the seasto the
westof Greenlandcouldcontinueto produceanomalouslylarge
volumesof Labrador Sea water [Dicksonet al., 1996],
North,G. R., T. L. Be!l, R. F. Cahalan,andF. J. Moeng, Samplingerrors
in theestimationof empiricalorthogonalfunctions,Mon. Wea. Rev.,
110, 699-706, 1982.
Parker, D. E., C. K. Foiland, and M. Jackson, Marine surface
temperature:observedvariationsand data requirements,Climatic
Change,31,559-600, 1995.
Pefiwitz, J., and H. -F. Graf, The statistical connection between
tropospheric and stratosphericcirculation of
the Northern
ß changesin ocean temperaturesand sea-iceboundaries,in
Hemispherein winter,J. Climate,8, 2281-2295, 1995.
combination with the strengtheningof the westerliesacross Quiroz, R. S., Tropospheric-stratospheric
polar vortex breakdownof
the North Atlantic, could affect fisheries recruitment,
ß changesin winter and spring precipitation patternsover
Eurasiacould affect soil moistureand vegetationduring the
subsequentgrowing season.
If the observed trends are anthropogenically induced,it
remainsto be to determinedwhetherthey are of stratospheric
origin (a responseto radiativelyinducedtemperature
changes
due to greenhousegases, as arguedby Perlwitz and Graf
[1995], or to CFC-inducedozone depletion) or whetherthey
representan unanticipateddeep barotropic responseof the
polar vortex to greenhousewarmingin the troposphere.
January1977,Geophys.Res.Lett.,4, 151-154,1977.
Robock,A., andJ. Mao, Winter warmingfrom large volcaniceruptions,
Geophys.Res.Lett., 12, 2405-2408, 1992.
Rogers,J. C., andH. van Loon, Spatialvariabilityof sea level pressure
and 500-mbheightanomaliesover the SouthernHemisphere,Mon.
Wea., Rev., 110, 1375-1392, 1982.
Rogers,J. C., and E.-M. Thompson,ArianticArctic cyclonesand the
mild Siberianwintersof the 1980s,Geophys.Res. Len., 22, 799-802,
1995.
Trenberth,K. E., andD. A. Paolino,Characteristic
patternsof variability
of sealevel pressurein the NorthernHemisphere,Mon. Wea. Rev.,
109, 1169-1189, 1981.
Trenberth, K. E., and J. W. Hurtell, Decadal atmospheric-ocean
variationsin the Pacific,ClimateDyn., 9, 303-309, 1994.
Acknowledgments.We would like to thank N.-C. Lau for pointingout
Volodin, E.M., and V. Ya. Galin, The nature of the Northern
the subfie,but importantinfluenceof the trend on the structureof the
Hemispherewinter tropospherecirculation responseto observed
leadingEOF of seasonal-mean
SLP basedon datafor the past30 years.
ozonedeplefioh
in lowstratosphere,
Quart.J. Roy.Met.Soc.,124, 1Thanksalsoto J. Hurtell, P. Rhines,C. Deser, C. Leovy, R. Reed, A.
30, 1998.
Gettelman,and two anonymous
reviewersfor their helpful comments
in the geopotenfial
and suggestions,
and to C. Gudmundson
for preparingthe manuscript. Wallace,J. M., and D. S. Gutzler,Teleconnecfions
height field during the Northern Hemispherewinter, Mon. Wea.,
This work was supportedby the National ScienceFoundationunder
Rev., 109, 784-812,. 1981.
grant9215512,andby NOAA undera grantto the HayesCenter.This is
Walsh,J. E., W. L. Chapmanand T. L. Shy,Recentdecreaseof sea
JISAO contribution number 494.
level pressurein the centralArctic,J. Climate,9 480-486, 1996.
Zhang, Y., J. M. Wallace and D. S. Battisti,ENSO-like interdecadal
References
variability:1900-93,J. Climate,10, 1004-1020,1997.
Baldwin, M.P., X. Chengand T. J. Dunkerton,Observedcorrelations
D. Thompson
andJ. Wallace,Atmospheric
Sciences,
JISA0
•,
between winter-meantroposphericand stratosphericcirculation
anomalies,Geophys.
Res.Lett.,21, 1141-1144,1994.
Universityof Washington,Box 354235, Seatfie,WA 98195-4235. ([email protected])
Cheng,X., and T. J. Dunkerton,Orthogonalrotationof spatialpatterns mail: [email protected];
derivedfrom singularvalue decomposition
analysis,J. Climate,8,
2631-2643, 1995.
(Received December 15, 1997; revisedMarch 10, 1998;
acceptedMarch 12, 1998.)
Dickson, R. R., J. R. N. Lazier, J. Meincke, and P. B. Rhines, and J.