JOURNAL OF GEOPHYSICAL
RESEARCH, VOL. 101, NO. D18, PAGES 23,475-23,489, OCTOBER 27, 1996
Variability in the mesosphere
observedby the Nimbus6
pressure modulator radiometer
BryanN. Lawrence
•
SpaceSciences
Division,Rutherford
Appleton
Laboratory,
Didcot,England,
U.K.
Department
of Physics,
University
of Oxford,Oxford,England,U.K.
William
J. Randel
National
Center
forAtmospheric
Research,
Bouide;,
Colorado
Abstract.
The pressuremodulator radiometer flew aboard Nimbus 6, collecting radiance
data from June 1975 until June 1978. These data have been processedto yield
daily temperatures,geopotentialheights, and balance wind estimatesfrom the
stratosphereand mesosphere
(30-85 km) for the entire period. We usethesedata
to examine the variability of both the zonal-mean and the nonzonal disturbances
present in the data. In terms of the zonal-mean, we show that the COSPAR
International ReferenceAtmosphere(1986), developedfrom these data, can be
misleadingin the mesospheredue to the underlying interannual variability and the
mergingwith other data sets. Daily variability of the zonal-meanflow is examined,
and these data show strong evidence of coupling between the stratosphere and
mesospherefor suchvariations. We also find evidencefor significantcoupling of
wave-like events in the stratosphere and the mesospherebut that not all such
disturbancesseen in the mesosphereare due to simple propagation from below.
Space-timespectral analysisis used to searchfor traveling planetary waves: Results
show a weak 5-day normal mode present in the equinox seasons,which appears
to correspondto the first symmetric mode predicted by theory. These data also
show clear evidenceof the 4-day wave in all three southern winters examined and
are consistentwith the hypothesisthat this mode grows due to instability of the
background zonal flow.
the stratosphere
(LIMS) andthe stratosphere
andmesospheresounder(SAMS) instrumentsaboardNimbus7;
and the ultraviolet spectrometeraboardthe Solar MesoSince the advent of satellite instrumentation
for resphericExplorer(SME). Of theseinstruments,
onlythe
mote soundingof the middle atmosphere, there have
P MR providedglobaldaily coverageof the mesosphere
1. Introduction
been many generations of instruments and retrieval
during all seasons.
techniques. In the stratosphere, data acquisition has
The data from the P MR provide an excellent conproceededfrom scientificexperimentation(for exam-
ple, the early selectivechopperradiometer (SCR) instrumentsaboard Nimbus 3 and later satellites)to the
productionof standardmeteorological
data (for example, the stratospheric
soundingunit (SSU) instruments
aboard the NOAA seriesof satellites). However, prior
to the Upper AtmosphereResearchSatellite (UARS)
the mesospherehas only been sampled extensively by
four instruments: the pressure modulator radiometer
(PMR) aboard Nimbus6, the limb infrared monitor of
text for interpretingmeasurements
from the U ARS in
terms of the typical structure and variability of the
mesosphere.Although the instrumentflew nearly 20
yearsago,the datahavenot yet beenfully analyzed.In
this study, we review the typical zonal-meantemperature structure of the mesosphereand highlight some
aspectsof mesospheric
interannualvariability by contrastingit with the COSPAR
InternationalReference
At-
mosphere
(1986)(CIRA86)[Fleminget al., 1990]climatology[Barnettand Corney,1985]. We then examine
the temporalvariabilityof both the zonal-meanand the
in thesedataø
•Department
of Physics
andAstronomy,
University
of Canterbury, eddy components
Beforeproceeding,
it is importantto recalltitat the
Christchurch,New Zealand.
Nimbus 6 PMR was a nadir instrument with very deep
Copyright
1996bythe3anerican
Geophysical
Union.
Papernumber96JD01652.
0148-0227/96/96JD-01652509.00
verticalweightingfunctions[Curti• et al., 1973;Labitzke
and Barnett,1981]. The effectof suchbroadweighting functionsis both to selectonly phenomena
with
23,475
23,476
LAWRENCEAND RANDEL: OBSERVEDMESOSPHERICVARIABILITY
largeverticalscalesandto underestimate
the amplitude tential field. The normal procedureto follow would
of any perturbationsassociated
with suchphenomena, be to take boundary fields from a suitable meteorologparticularly in the upper mesosphere.Becausethe raw ical analysisor those derivedfrom another instrument.
data collected were radiance, the further retrieval of
temperature generally introduced yet more smoothing
of the real atmosphericperturbationsvia the introduction of the retrieval climatology.
However, for the PMR data, this was not possible,
as contemporaneousglobal lower stratospheric analyseswere not available. Accordingly,where we present
time seriesof geopotentialdata, we havesimplystacked
Early analysesof thesedata focusedon temperature the data upon a monthly mean 50-hPa field from a 12retrievalsof monthly meansand a limited set of dayse year (1979-1990)climatologyof National Meteorologianalyses
[Randel,1992].
These monthly mean data provided the basisfor the cal Center(NMC) stratospheric
In addition, we were able to produce zonal-mean
mesospheric
sectionof the CIRA86 climatologyand of
the extendeddiagnostics
of Marks [1989]. (Actually, zonal winds using the gradient wind approximation.
for the heightswe are interestedin, CIRA86 is based This approximation provides winds which are in excelon an earlier climatology,that of Barnett and Corney lent agreementwith observationsthroughoutthe strato[1985],whocombined
PMR datawith SCRdata.) Sub- sphere. In the mesosphere,the gradient wind approxsequently,the entire PMR data set (from mid-June imation is probably still a good measureof the zonal1975 until the middle of June 1978) has been rere- mean zonal wind in the extratropics but not in the troptrieved by Clive Rodgers(Oxford University) to pro- ics. The tropical winds shown are simple interpolations
vide daily along-trackprofilesof temperaturesthrough- between 15 øN and 15øS.
out the stratosphereand mesosphere. These profiles
State
have been griddedby John Gille and Dan Packmanat 2. The Mean
the National Center for Atmospheric Researchusing a
Kalman filter algorithm to provide daily estimates of
zonal-mean and zonal waves 1-6 on a 4ø latitude grid
over 80øS to 80øN. The true vertical
resolution
of the
In this section, we review three aspectsof the zonalmean view of the mesosphere. We first examine the
interannual variability of the January monthly mean
temperature fields in order to highlight the differences
obscuredby the averaging processinherent in making
a climatology. Second, we show that when individual years are examined, the separated stratopause is
more prominent in the northern hemispherethan hitherto suggested.Finally we show the seasonalevolution
rection for nonlocalthermodynamicequilibrium(non- of the zonal wind, partly in order to show the backLTE), there are now somequestionsas to whether the ground upon which we seelarge wave eventsand partly
influence of non-LTE on temperature retrievals is as to demonstrate typical examples of evolution for later
great as was expected at the time of calculation. This comparisonwith middle atmospheremodels.
means that at some of the higher altitudes where the 2.1. Zonal-Mean Temperatures
assumptionmay have been in error, the mean temperMonthly mean zonal-mean temperature sectionsfrom
atures may have been underestimated especially near
the coldsummermesopause
(C. Rodgers,personalcom- the three individual PMR years for January are dismunication, 1995). It should also be noted that due played along with the resulting CIRA86 temperatures
to the configuration of the weighting functions, there in Figure 1. It is worth restating here that the very cold
is no real height discrimination in the lower and mid- summer mesopauseis probably exaggerated in these
dle stratosphere. However, the temperature thickness data (the individual years,and the climatology),since
of this region is well captured by the lower weighting all the retrievals assumednon-LTE in the upper mesofunctions.
sphere. The main features of the typical temperature
Although the instrument collected data for the en- structures
shown
herearedescribed
elsewhere
[e.g.,Antire period from June 1975 until June 1978, there was drewset al., 1987]. Althoughit is not obviousfrom
a period of about 9 months (from October 1975 un- these figures,differencesfrom the CIRA86 climatology
til July 1976) when the top soundingchannelwas not demonstratethat for January, the CIRA86 atmosphere
scanning correctly, leading to poor height discrimina- underestimates
the northern hemisphere(NH) midlattion (J. Barnett, personalcommunication,1995). Ac- itude temperatures in the region 60-80 km by 5-10 K
cordingly, some of the data from this period probably in all years• By contrast, in the tropics, the lower refurther underestimatethe true variability in the data, gions(50-60km) are overestimated
by CIRA86 and the
and we focus away from this time period for detailed region 60-80 km underestimated, in both casesby 5retrievals is probably in excessof 15 km, although they
have been oversampledto provide a vertical grid at 3.5km intervals. The resulting data are the subject of this
study.
Above 80 km, our results shouldbe assessed
with caution, since although the data was retrieved with a cor-
studies.
10 K. Examination
of the differences for other months
Geopotential fields have been calculated from the and years suggeststhat the differencescan be up to
PMR temperatures using the hydrostatic equation to 10ø of either sign and are not simply due to interannual
stack temperature thicknesseson a boundary geopo- variability, but probably rather that the CIRA86 cli-
LAWRENCE AND RANDEL: OBSERVED MESOSPHERIC VARIABILITY
Zonal Mean Temp
Jan 76
80
Zonal Mean Temp
.01 80
.,o• '•ø
•-•o
23,477
Jan 77
• •
.01
•_ •o
60S
30S
0
30N
60N
60S
30S
Lafifude
0
30N
60N
Lafifude
Zonal Mean Temp
Jan 78
,• 60
CIRA January climatology
.•0• • 60
.•0•
,7o
.••,• • ..Q
•70 •
[
• •o
•
605
5OS
0
50N
60N
•o
605
Lafifude
50S
0
50N
60N
Lafifude
Figure 1. Zonal-mean
monthlymeanJanuarytemperaturefrom the 3 pressuremodulator
radiometer(PMR) yearsand the Barnett and Corney[1985]climatology.
2.2. The Separated
matologyis lesssensitiveto verticalgradientsthan the
Stratopause
reretrieved PMR data. This conclusionis supported by
comparisonsbetween CIRA86 and other climatologies
The presenceof a warm stratopausein the polar night
[e.g.,ClancyandRusch,1989;Clancyet al., 1994]and is of interest sinceit must be of dynamicalorigin (in
must be due to either the poorer temperature retrievals the polar night, there can be no solar heating to proused originally or the inadequaciesin the merging and duce a stratopause). The separated stratopause was
averagingprocedureusedto createthe CIRA86 clima- first describedby Kanzawa[1989]using CIRA climatologicaldata and examinedmore closelyby Hitchman
tology.
80 N
Zona
Mean Temp
80 S
Zonal Mean Temp
BO
7O
•
60
• so
40
.Ol
70
30
1975
1976
1977
1978
Figure 2. Zonal-meantemperatureat 80øN (tipper panel) and 80øS (lower panel) for the
duration
of the PMR
data collection.
23,478
LAWRENCE AND RANDEL: OBSERVEDMESOSPHERICVARIABILITY
et al. [1989],whopointout the salientfeatures.Both hibits intense winter variations. It can be seen that in
papersstate that it is a muchstrongerfeaturein the February 1977, the region of elevatedtemperaturesin
southernhemisphere.However,examinationof the individual yearsleadsone to concludethat on occasion
it can be a very pronouncedfeature in the northern
hemisphereas well.
If we comparethe Januarymeansfor the three individual years (Figure 1), we can seethat there is a
the NH was much higher and strongerthan the other
yearsøIn the climatology,it can be seenthat the averagingprocesshashighlightedthe southernseparated
stratopausedue to the interannualrepeatability,in contrast with the north, where the averagingof the interannual variability has hidden the occasionallyintense
significantamountof variability:January1976shows feature.
a polar stratopause
at least10 K warmerthan midlati- 2.3. The Zonal-Mean
Zonal Wind
tudes,and January1978showsa weakerbut similarpoWe begin by showingthe seasonalevolutionof the
lar maximum. January1977is very differentfromthese
otheryears,exhibitingpolar temperatures
muchcolder zonal wind at 1, 0.1, and 0.01 hPa as a functionof latthan midlatitudes and a weak, low stratopause. For itude (Figure 3). At 1 hPa, the contrastbetweenthe
the corresponding
monthsin the southernhemisphere, three different northern hemispherewinters is apparent;
the separatedstratopauseis very prominenteachyearø eachyearhasdifferenttiming and positionof winterjet
The CIRA86 climatologyshowsa somewhatstronger
feature, probably due to the fact that the separated
stratopauseoccurspreciselywherethe SCR and PMR
data setsare mergedin the climatology(it shouldbe
recalledthat the SCR data are from differentyears).
Figure2 showsaltitude-timesections
of the temperatures at 80øN and 80øS illustrating evolution of the
polar stratopause. A repeatableannual cycleis seen
in the southernhemisphere(SH), whereasthe NH ex-
.01
mb
Zonal
30N
maxima, which is evidenceof quite differentwinter vortex structure. The relatively regular seasonalmarch in
the southern hemisphereis also clear. A comparison
from one year to the next of the distanceof penetration of the southernhemispheresummereasterliesinto
the northernhemisphereseemsto imply high repeata-
bility. At 0.1hPaandabove,the difference
between
the
northernwintersis evenmore apparent,with markedly
differentjet structures
fromoneyearto the next. In the
Mean
Wind
81
km
•
.1 mb
1 mb
Zonal
Mean Wind
65 km
Zonel
Meen
48
Wind
km
•i•u•e 3. •o•l-me•
•o•] wi•d o•e• the P•E period •t (top to bottom) O.O1,O.1, and 1
hP•. •stefiies •e lightly shaded,•d the d•k sh•di• de•otes•e•io•s where the •die•t of
qu•si-•eost•ophicpotential •o•QciW is st•o•l• •e•ti•e (seetext).
LAWRENCE AND RANDEL: OBSERVED MESOSPHERIC VARIABILITY
23,479
south,the mesaspheric
windregimeis not sorepeatable in January of eachyear of the Nimbus 6 period are disbetween
yearsasit is in the stratosphere.
Althoughthe playedin Figure 4. For comparison,the winds derived
balanceassumptionusedto derive the wind at 80 km from CIRA86 using the same computational domain
is probablyquite poorin the tropics,onecan seethat and schemeare also displayed. The major features in
in the extratropicsthe mean flow is characterizedby January are the summermesasphericjet, the subtropical summer stratasphericjet, the winter mesaspheric
much smaller scalesand more rapid variation. There is
clearlyconsiderably
moreinterannualvariabilitythan jet, and the stratasphericpolar night jet. The evolulowerdown,althoughthe strengthof the circulationas tion of these features is well characterized in the preindicatedby the jet maximais fairly consistentin the vious time series. These figuresallow one to conclude
that the main structures are relatively repeatable from
summer regions.
Figure 3 also includesheavy shadingindicatingre- one year to the next, but the strengthand positionof
gionswhere the quasi-geostrophic
potential vorticity both the winter mesasphericjet and the strataspheric
gradient
(•y)issignificantly
negative
(•y_•-1.5x 10-• polar night jet vary accordingto the dynamicsof the
m-• s-1), potentiallyindicativeof dynamicinstability winter in question(in particular, due to the timing and
of the zonal-mean
flow[Andrews
et al., 1987].Regions strength of any stratasphericwarmings). The situation
of persistent
negative
• arefoundin the SH winter in the southern winter August is similar to that of the
northern January, as can be seen in Figure 5, which
mesasphere
on the polewardsideof the subtropicaljet
(near50øS)and alsoat highlatitudesthroughout
the displayssectionsfor the three Augusts(and that of the
stratosphereand mesasphere.
Theseregionsare associ- CIRA86 climatology). It is apparent from these secated with the appearanceof the 4-day waveoverpolar tions that the summer mesasphericjet is much weaker
latitudes in SH winter, as discussed
in more detail be- in the NH than in the SH and also that the SH subtrop-
low.A second
regionof persistent
negative
• is found ical mesasphericwesterlyjet is more narrow and intense
in the summermesasphere
at high altitudes (near 65 than the NH counterpart. There is lessinterannual vari-
km) onthepoleward
sideoftheeasterly
jet, particularly ability in the SH strataspheric polar night jet than in
the NH, so that the climatology is closerto individual
i• •h• N•. •t•
[•OSS]
•d m•o• [•01 h•ve•lyzed
instabilitiesof jet structuressimilar to thosefound in years. However, the patch of easterliesin August 1975
this region,findingstronggrowthrates. Nonetheless, at high southern latitudes and high altitudes is a saluwe do not find evidence of transient waves in the sum- tary reminder that there is still considerablevariability
mer mesaspherein the PMR data, possiblybecauseof in the SH (seealsoFigure 6). One can alsoseethat the
the low vertical resolution.
CIRA86 climatology(which in the mesasphereis solely
Latitude-heightsections
of the zonal-meanzonalwind due to an averageof the 3 yearsof the PMR data), has
Zonal
Mean
Wind
Jan
Zonal
76
8O
.01
0
30N
.•..
Mean
60
•-
4o
3o
60S
60N
30S
0
30N
60N
Latitude
Wind
Jan
78
CIRA January climatology
.• .Ol80
80
77
.Ol
Latitude
Zonal
Jan
70
.•o"-'E••..
lO
30S
Wind
80
_o
60S
Mean
.Ol
.10 •... •.. 60
•_
lO
60S
30S
0
30N
30
lO
60N
Latitude
Figure4. Zonal-mean
monthly
meanJanuary
zonalwindfromthe3 PMRyearsandtheBarnett
and Carney[1985]climatology.
23,480
LAWRENCE
AND RANDEL'
Zonal Mean Wind
60S
3OS
0
OBSERVED
Aug 75
3ON
60N
MESOSPHERIC
VARIABILITY
Zonal Mean Wind
Aug 76
60S
SOS
Latitude
Zonal
Mean
Wind
Aug 77
10
SOS
0
3ON
60N
CIRA August climatology
.Ol
60S
0
Latitude
,.30N
80
.Ol
30
60N
Latitude
Figure 5. Zonal-meanmonthly-meanAugustzonalwind from the three PMR yearsand the
Barnett and Corney[1985]climatology.
introduced spuriousstructuresin the mesosphere(such [e.g., Labitzkeand Barnett, 1981;Hirota and Barnett,
as the westerly jet peak found at about 20øN and 60 1977;HassandEbel,1986].Herewe concentrate
on the
km). Such artifacts can also be found in other months. variability occurringon faster than seasonaltimescales
in temperature, wind, and derived quantities associated
with planetary waves.
3. Daily Variability
The broad scopeof variability over the period is well
Previous workers have looked at the planetary wave characterized by the root-mean-squaredeviation from
structure in the mesosphereobservedby the PMR in- the zonal-mean, and for an overview, we display this
strument by concentrating on the raw radiance data for the geopotentialheight of two pressurelevels,3 hPa
.05 mb
RMS Geopofential waves
69
km
30N
•_
0
30S
3 mb
RMS Geopofenfial waves
41
km
""i;':i'?"!?,
30N
. ,,,/,..,,
,-,,
1975
,.,?,,,.,,
,.,,,
'.....
1976
..",.-,
.-''-2
1977
1978
Figure 6. Latitude-time sectionsof the rms geopotentialheight around the zonal-meanat 3 hPa
(-• 41 km) and 0.05 hPa (-• 69 km). Contourinterval(solidlines)is 300 m, with values:•bove
600 m hatched.
The dashed line is the 150-m contourø
LAWRENCE
AND RANDEL:
Geopotential variance
OBSERVED
MESOSPHERIC
VARIABILITY
23,481
each season. Such wave growth events are well corre-
50-70N
lated with the zonal wind decelerations
8,3
km
76
_
69
62
_
55
48
_
41
35
i
S
0
N
D
J
F
M
km
i
A
over 30-65 km
seenin Figure 3 (this correlationis lessevidentfor the
0.01 hPa windsin Figure 3). At first inspection,many of
the individual pulsesin Figure 6 are approximatelycorrelated betweenthe 41 and 69 km levels,although the
detailed evolution is clearly different at each altitude.
To examine vertical correlationin more detail, Figure 7
showstime variation of geopotentialwave variancenear
60øN during NH winter 1976-1977, at eight altitude levels spanning35-83 km (eachlevel is approximatelyone
scaleheight, 7 km, apart)ø For most of the eventsin Figure 7, wave variance maxima are observed at all levels
throughout the stratosphere and mesosphere,demonstrating that these transient events span the entire mid-
1976--77
Figure 7. Time variationof rms geopotentialheight dle atmosphere(•- 30-90 km). One notableexceptionis
amplitudeover50ø-70øNat eightaltitudelevels(separated by approximatelyonescaleheight)duringnorth- the wave event near the middle of January in Figure 7,
(above
ern hemispherewinter 1976-1977. Amplitudes at each whichreacheslargeamplitudein the mesosphere
50 km) but not in the stratosphere.
The spatial patterns of the zonal-mean temperature
and wind changesduring these eventsshowstrong cou(•-41 km) and 0.05 hPa (•-69 km), representative
of the pling between the stratosphere and mesosphere,as is
upper stratosphereand middle mesospherein Figure 6. expected during periods of massivepolar vortex disrupMuch of the variability in the mesosphereis of strato- tion. Figure 8 showsthe temperature and wind changes
spheric origin, and we begin below by assessingthe associated with the beginning of the major warming
importance of direct dynamic coupling between the during late December 1976. The temperature change
stratosphereand the mesosphere.In addition, a signifi- betweenthe December 24, 1976, and January 1, 1977,
cant contribution to mesosphericvariability comesfrom exhibits a clean quadrapolepattern, with high-latitude
within' Two well-knownexamples(the 4-day wave and warminF and tropical cooling in the stratosphere and
normal mode oscillations)are examined in detail be- reversed patterns in the mesosphere. Such temperaiow. However, much o• the var]abd]ty seen in r•gure ture patten•s have been shownpreviously[e.g., Fritz
6 is probably ultimately due to the affect of breaking and Soules•1970;Labitzke,1972;Gille et al., 1983]and
level have been normalized to unit variance.
gravity waves[e.g., Smith, 1996]whichwe cannotexamine
here.
3.1. Stratosphere-Mesosphere
The zonal wind changesin Figure 8 show deceleration
of the winter hemispherejet from the middle strato-
Coupling
As seen in Figure 6 large-amplitude, episodicwavelike eventsare observedin eachwinter hemisphere,with
typically three to six individual events,or "pulses,"seen
Dec 24-Jan
80
,-,
70
are in good agreement with theoretical expectations for
winterhemisphere
planetarywaveevents[Garcia,1987].
1
sphereto the middle mesosphere(over 30-70 kin, in
accordwith the subjectiveimpressiondescribedabove)
andeasterlyaccelerations
in the tropics(in balancewith
TBAR difference
' ,'/"'•,
''
"-
Dec 24-Jan
-
' '• ......," '-•i
,
•,'"E 60
g
•_
.01
•, ,,,,,,•
,,,,,,. .-•.10 •
:.:,, ,,,'
"
/,'
-;J
,
, '-•.-•,;',
,,'.:-:.;:q •
'• "........... -'•.,..-..'..-..•.•.'..'" '-:';'.•
,,. ,......., /
,,,1•
',,
,,
:
, , ,
605
305
0
Lafifude
30N
•'
60N
•
UBAR difference
• • ••
IIIi'1
80 -
•'•I"'•'"/;••
60 _
,". i., ', Ill _•
,,', ,, ,,',ill
,', ,'.'.: ,''/ "',
] 10•'E
, .....:
,,
, -.'., \'..
1
,. "!
,', ,'
-
•_ ....
-
60S
,
,, - •,',,
,,'
305
0
,
ql •1.01
,' ;;,',,, ',,' ',111,-I
,
' '
,,
• ',
• '.
',, IIIII -I
,-,.'
,I•L./ 1
, ''","
....
',,,,,•
, .-,
,
, •-'
,.: , .,,,
30N
' I•.
J
60N
Lafifude
Figure 8. Changesin the zonal-meanstate betweenDecember24, 1976, and January l, 1977:
zonal-meantemperaturedifferences
(left panel) and zonal-meanwind differences
(right panel).
23,482
LAWRENCE
Augusf 6-14
,•
8O
,
!
,
AND RANDEL:
OBSERVED
TBAR difference
,
,
i
,
,
i
UBAR
,
difference
,
tl
.01
•
70 "
•
,'",'•',
60 ',,'..
._-...
),, ',,
.---:•_•
VARIABILITY
August 6-14
,
.... .
-3.
MESOSPHERIC
',
8O
--',
,, ',
.01
7O
.10
-
50
40
,-'-
30
f,
,••
60S
,','
'"" "
30S
0
30N
60N
lO 30
60S
10
30S
Lotitude
0
30N
60N
Lotitude
Figure 9. Changesin the zonal-meanstate betweenAugust 6 and 14, 1976' zonal-meantemperature differences(left panel) and zonal-meanwind differences(right panel).
the tropical temperaturechanges).Figure 9 showssim- are governedby the resonantpropertiesof the atmoilar diagramsfor a waveeventduringSH winter (Au- sphere,rather than by the details of any forcingmechagust 6-15, 1976). The patternsof temperatureand wind nisms. The fundamental signaturesof these modes are
changesare similarto the NH example(Figure8), but planetary-scalehorizontal structures,regular westward
with typiare somewhatweakerin amplitude(consistentwith the propagationat (nearly) discretefrequencies
weaker-amplitudeplanetary wavesobservedin the SH; cal periods of 5-20 days, and small vertical phase tilts
seebelow). (Examinationof time seriesand otheryears with height.(seethe reviewby Madden[1979]).Numerof data showsthat the NH high-latitude temperature ous studies have identified normal-mode structures in
changesin Figure 9 are related to the seasonalcycle stratospheric data derived from satellite measurements
[e.g.,Rodgers,1976;Hirota and Hirooka,1984;Hirooka
rather than to planetary-wave-induced
transience.)
In the NH upper stratosphere,large-amplitudewave and Hirota, 1985, 1989; Prata, 1989; Venne, 1989; Rantransience is usually associated with massive distur- del,1993a].Typically,identification
asnormalmodesin
bancesof the polar vortex as anticyclonicvorticesform these studies is based both on enhanced spectral power
in the upper stratosphere which are associated with
wave 1 Temp cohsq 32 S-52 N
stratosphericwarmings of various strengths. Analysis
of the pulsesseenin Figure 7 showsthat each of them
is associatedwith strong vortex disturbancesand consequentlymost of the variance is concentratedin zonal
wave numbers 1 and 2. The large pulse at the end of
December is associatedwith the beginning of the major midwinter warming and the destructionof the lower
stratosphericvortex describedby Labitzke and Barnett
7/77ß' --
_
-
[1981]and Labitzkeet al. [1993]. During the period
while there was no lower stratosphericpolar vortex, the
mesospherestill maintained a strong vortex, probably
due to the strongradiative coolingthere. Followingthis
period, it appearsthat air brought polewardbecameas-
sociatedwith a mesospheric
high (similar to Figure 17
of Dunkertonand Delisi [1985]). This latter episode,
which didn't involve simplewave propagationfrom below, resulted in the pulse-like event seen only in the
mesospherein Figure 7.
3.2.
Normal
Modes
7/75
2
3
4 5
wesTword
l0
l0
5 4
3
2
eosTword
Period (days)
Figure 10. Coherencesquared spectra between 32øN
and 32øS for zonal wave 1 temperature fluctuations at
0.1 hPa (-• 65 km) throughoutthe PMR record. Spec-
Normal-mode Rossby waves are westward propagat- tra are calculatedfrom overlapping60-day time series.
ing planetary waveswhich are the free or resonant os- Note the coherencemaxima for 5- to 10-day westward
cillations of the atmosphere. Their basic characteristics traveling wavesnear March and October of eachyear.
LAWRENCE AND RANDEL' OBSERVED MESOSPHERIC VARIABILITY
Sepf-Oct
60N-
6ON-
3ON
SON-
1977
23,483
k-1
Temp .01 mb
Ee-
EQ-
30S-
30S-
60S-
60S-
I I I I I I I I I I II I I I I II • I II I I I I I [I I I I I I Itlll
I
2
3
4 5
westword
10
IIIIIII111111111111
I
I
I
I
10
5 4
3
II111 i
2
eosfword
Period (days)
Figure 11. Latitude-frequency
spectrafor zonalwave1 temperature
fluctuations
at 0.01hPa (•
81 km) duringOctober1977.Left panelshowspowerspectra(contours
arearbitrary;dashedline
is one half lowestlevel), and right panel showscoherencesquaredat eachlatitude with respect
to 32øS,
and global coherencefor westwardpropagating plane- coherence. The combination of enhanced power and
for discrete(westward)frequency
tary wavesnear the appropriatenormal-modefrequen- near-globalcoherence
cies and on horizontal and vertical structures consistent
bands is strong evidenceof normal-modeoscillations.
with theoretical predictions(calculatedstructuresare The spectral power density for these wavesis generally
givenby Kasahara[1980],Schoeberl
and Clark [1980], a small fractionof the total (i.e., the wavesare usually
and Salby[1981]). We followthis methodby search- of small amplitude, althoughperiodsof large amplitude
(seenin Figures11 and
ing for normal-modesignaturesin power and coherence do occur). The globalcoherence
spectraand then examiningthe spatial structure of the 12) is often a more sensitivediagnosticfor the presence
modes during periods of large wave amplitude. Our of normal modes than the occurrenceof peaks in power
focus is on mesosphericwaves and zonal wave number spectra alone.
The structure of the dominant mode during Septem1 features (higher wave numbersdo not exhibit clear
evidence as that seen for wave 1, possibly due to the ber-October1977 (the time period of Fig 11) is shown
in more detail in Figure 13. Here the amplitude and
vertical resolution).
The occurrenceof global normal modes in the meso- phase of the westward 5- to 10-day-period wavesseenin
sphere throughout the PMR record is investigated in geopotentialheight amplitude and phaseare displayed
time seriesfrom windowedspecFigure 10, which showsthe coherencesquared between (we haveresynthesised
32øN and 32øSfor zonal wave 1 temperature at 0.1 hPa, tral coefficientsand then averagedin time over several
constructed from overlapping 60-day spectra through- weeksof largewaveamplitude). The amplitudeexhibits
out the record. Figure 10 showsoccurrenceof coher- broad maxima over midlatitudes of both hemispheres,
ent global westwardpropagating wavesnear eachMarch with nearly in-phase (symmetric) horizontal structure
and October(equinoxseasons),
with periodsin the 5- to between the NH and SH and weak westward phase tilts
10-day range. The appearanceeach March and October with height. Figure 13 also shows the temperature
of the coherencemaxima in Figure 10 is quite striking, structure for this spectral band, showing in-phase latshowing that such modes are a regular feature of the itudinal structure and vertically out-of-phase tempermesosphere.
ature maxima near the stratopause(• 50 km) and in
Two individual periods are shownin more detail: Oc- the middle-upper mesosphere.The in-phaselatitudinal
tober 1977 at 0.01 hPa (Figure 11) and March-April structure seenin Figure JL3is a characteristicobserved
1976 at 0.1 hPa (Figure 12). Both power spectraand of each of the equinox coherencemaxima identified in
coherencesquared are shown for temperature fluctua- Figure 10. The amplitude peaks are roughly colocated
tions and geopotential fluctuations, respectively. Clear with the background zonal jet maxima, as can be seen
spectral peaks are visible in both hemispheresfor both in Fig 14, where we have superimposedthe geopotencases;both the 5- to 10-day periodicity in October and tial maxima from Figure 13 as bold lines upon the zonal
the 5-day periodicity in March-April show near global wind fields. The slight NH-SH asymmetry is mirrored in
23,484
LAWRENCE AND RANDEL' OBSERVED MESOSPHERIC VARIABILITY
Mar-Apr
1976
k-1
Geop .05 mb
Mar-Apr
6ON-
6ON-
5ON
3ON-
2
EQ-
k-1
Geop .05 mb
EQ-
30S-
30S
60S-
60S
ililil;llllllll,il;llllil,llil
I
2
Period (days)
1976
I
I
I
I
3
4 5
westward
10
i,illlliillliilll,Jillllll,il
I
10
I
i
5 4
3
eastward
I
I
2
Period (days)
Figure 12. Latitude-frequency
sections
of powerandcoherence
squared(withrespectto 32øN)
for zonalwave 1 geopotentialfluctuationsat 0.05 hPa (,.• 69 km) duringMarch-April 1976.
both the zonal wind and wave amplitude structure. It is day wave. Power spectraover polar latitudes are conlikely that occurrenceof similar symmetricbackground structedfrom overlapping60-day time seriesand plotwinds is associatedwith the presenceof normal-mode ted versustime (Figure 15). In addition to the low-
oscillationseachequinoxseason(seenin Figure10).
frequencymaximsin the respectivewinter seasons
and
3.3. 4-Day Wave
near-zero wave v•riance in summer, the 72øS spectra
show spectral maxima correspondingto an eastward
A zonal wave number I feature, with a period near
4 days, was first observedin SCR temperature data by
Venneand Stanford[1979]. Further analysisby those
authors and others has shown that the wave is almost
an ubiquitous feature of the winter polar stratosphere
and that it occurs in the winter of both hemispheres,
although the signal is usually stronger in the south.
Prata [1984]and Lait and Stanford[1988]showedthat
travelingwavewith a periodnear 4 daysduringSH midwinter in 1976and 1977 (early SH winter 1977alsoexhibits eastwardpropagatingvarianceat slightlylonger
periodsof 5-10 days). The NH spectrain Figure 15 do
not exhibit similar strongpeaks near the 4-day period,
although weak 4- to 5-day maxima are observednear
November of each year.
The structure of the wave is demonstratedby highlighting the SH event of August-September
1977. Fig-
the wave is associatedwith higher zonal wave number disturbanceswhich are phase-lockedto the wave 1 ure 16 shows meridional cross sections of variables charcomponent. They also showedthat the resultant wave acterizingthe event averagedover September1-7, 1977.
structure is associatedwith warm pools of air rotating The 4-day wave signature is extracted from the data
around the polar vortex. The vertical extent of these by resynthesisof the space-timeharmonic coefficients,
waveswasdemonstrated
by Lawrenceet al. [1995],who retaining only componentscorrespondingto eastward
showedevidencefor their penetration up to 90 km by traveling waveswith periods near 4 days. Eddy wind
examiningradar wind data from Scott Base (79øS)o coefficientsare then calculatedusing balancewind as(EP) flux divergence
Theoretical analyses of the mean flow in the winter sumptions,andthe Eliasssen-Palm
polar middleatmosphere[Hartmann,1983;Randeland is calculatedneglectingthe vertical velocity term. The
Lait, 1991;ManneyandRandel,1993]alongwith obser- EP flux divergenceshowsa north-south dipole pattern
vations[Manney,1991]havedemonstrated
that insta- which is mostly attributable to the horizontal EP flux
bility mechanismsin the upper stratosphereand meso- component(proportionalto the momentumflux, which
sphereprovide a plausible explanation for the develop- is equatorward).
Figure 16 also shows the zonal wind for this time
ment of the 4-day wave.
The 4-day wave has been observedin P MR data be- period, with shaded regions denoting negative quasi-
fore [Prata,1984];[,erewe concentrate
on the structure geostrophic
potentialvorticitygradients
(•y < 0). The
and v•riability of the 4-day wave observedin the PMR zonal wind structureis dominatedby a strongjet in the
data. We begin by using space-timespectral analysis SH subtropicalmesosphere,apparently decoupledfrom
to demonstratethe occurrenceand strength of the 4- the upward extensionof the high-latitude stratospheric
LAWRENCE AND RANDEL: OBSERVED MESaSPHERIC VARIABILITY
wave 1 Geop 5-10
• t,xg• ....
days west
•'/
,
wave 1 Geop 5-10
v/\it
,,,,.,,,._,
....
?
• 40
•
10
30S
0
30N
40
lO
30 )
60N
i
i
60S
i
i
days west
i
i
wave 1
i [,
i
i
30N
80 "',
•'
•, 70
so
g
r> 50
•: 40
Temp
5-10
, ]i, , i , , i
' 01
80
•.
i
0
i
60N
Latitude
5-10
, [,,ii
i
30S
Latitude
wave 1 Temp
.Ol
•• •
i
60S
,,,
, .,"
•o•••o
30
days west
,, -. ,..
....
•o
23,485
•
"',
70
• 60
,,
days west
i i ' '! ]
,•J] • ]:
..... 'Z
.....
,,' / /
1 .01
..•
..----.' '-9• .10•
,• ..50
40
t
,
50
I
,
i
i
60S
t
t
30S
i
I
0
Latitude
•o •o •, •
I
/• •• •o
I
30N
60N
riOS
30S
I
0
30N
B0N
katitude
Figure 13. Meridionalcrosssections
of geopotential
(upperpanels)and temperature(lower
panels)amplitudeand phasefor westwardtravelingzonal wave 1 with 5- to 10-dayperiods
during October 1977. The phasesare calculatetlwith respectto the mean altitude maxima as
markedo
polar night jet (compare with the monthly averaged negative
•y, whicharein turn relatedto opposing
mean
structure shown in Figure 15). The resulting "double flow changesat 32øS and 56øS. The zonal winds at 56øS
jet" structure has marked curvature and strongly neg- exhibit decelerationsjust prior to growth of the 4-day
ative potential vorticity (PV) gradientsover 415ø-60øS wave, while the jet near 32ø intensifies. These changes
throughout the mesosphereo
stronglyincreasethe horizontalwind shears(and curFigure 17 showstime seriesof 4-day wave geopoten- vature), resulting in the strongly negative PV gradient
tial amplitude and EP flux divergenceat 60øS, 0.2 hPa near 56øS occurringjust prior to wave growth.
during 1977o These time seriesshow severalepisodesof
The temporal coincidence of wave growth with apwave growth and decay throughout the SH winter, with pearanceof negativePV gradients(Figure 17) strongly
a large event in early September(the time period sam- suggestsinstability of the mesosphericjet structure as
pled in Figure 16). Also shownin Figure 17 is evolution the cause of the 4-day wave. The source of the wave
of the 0.1-hPa zonal wind at 32øS and 156øS,together from instability is also suggestedby the spatial coinciwith the 0.1-hPa potential vorticity (PV) gradient at denceof negativePV gradient and positiveEP flux di156øS.
The wave and mean flow time seriesin Figure 17 vergenceseenin Figure 16 (near 50ø-60øS,throughout
show clear temporal coupling: the wave growth events the mesosphere).The spatial and temporal coincidence
(as seenin the EP-fiux divergence)
are coincidentwith of these quantitiescan be a signatureof an internal
23,486
LAWRENCE AND RANDEL: OBSERVED MESOSPHERIC VARIABILITY
Sept-Oct
well describedby Clancyet al. [1994]. In their work,
1977 Zonal wind
they describedifferencesof 5-20 K between the SME
and the CIRA data in the lower mesosphere,with the
SME being warmer. These results are consistentwith
.Ol
ours (i.e., the newerPMR temperatureretrievalsbeing
warmer) for somelatitudes and years;however,differencesof both signswere found, highlighting the difficulty of comparingdata with different vertical resolutions. The additional problem for comparingclimatolom giesfrom differentperiodsis that they cantime average
TM
data with high internal interannual variability; it is exemplifiedby the separatedstratopausein the northern
•0
hemisphereand is highlighted here for the benefit of
middle atmospheric modelers.
Much of the rapid transienceseenin the zonal-mean
8o
• .50
30
60S
30S
0
30N
60N
fields(e.g., Figure 3) mustbe dueto fluctuationsin the
gravity-wavedriving, as conjecturedby Fritts and Vin-
Lafifude
Figure 14. Zonal-mean
zonalwindsfor October1977. cent[1987]amongothers.Here we showthat thereis
Boldlinesdenote
geopotential
maxima
forthewestwardalsoa significant
component
offastzonal-mean
variabil-
traveling
waveisolated
in Figure13.
ity resulting
fromlarge-scale
transience
in thestrato-
sphere. The changesin zonal-meantemperature shown
in Figures 8 and 9 are also consistentwith changesin
instability[seeRandelandLair, 1991].The wavestruc- the residualmean circulation (e.g., Figure 8 of Ranture observedin Figure 16 (includingthe presenceof del [1993b])and are evidenceof globalcirculationcells
downward pointing vectorsnear 60-65 km, which indi- which accompany winter hemisphere planetary wave
cate a nonnegligiblecontribution from baroclinic wave events[Garcia,1987].
processes)
is alsoconsistentwith the instability calculaIn our analysisof the couplingof large-scalepertur-
tionsof ManneyandRandel[1993],performedfor back- bations in the stratosphereto those in the mesosphere,
ground zonal flows similar to that in Figure 16.
interpretation in terms of wave propagation is complicated by the large vortex interactionshigh in the up4. Discussion
per stratospherewhere linear theory must break down.
The well-known propensity of the polar vortex to sufThe difficulties inherent in producing mesospheric fer pulse-like distortions associatedwith sequencesof
climatologies, and in particular, comparing such cli- warmingsthroughoutwinter [e.g.,Rosieret al., 1994]
matologies from different periods and instruments are is seenhere. As has beenfoundbefore[Pierceet al.,
wove 1
Geop
power
72 S .4 mb
wave 1
1/78-
1/78-
7/77-
7/77-
1/77-
7/76-
1/76-
1/76-
,,,,,i,,,,,,,,,,,,
i
3
i
i
4 5
westward
72 N .4 mb
7/75 -
,,,,,,,,,,,,,,,,,,,,,,
i
2
power
1/77-
i
7/76-
7/75-
Geop
i
i
10
10
I
i
5 4
eastward
Period (days)
i
i
i
i
3
2
2
3
i
i
4 5
westward
i
i
10
10
i
i
5 4
i
3
eastward
Period (doys)
Figure 15. East-westpowerspectrafor zonalwave 1 geopotentialheight fluctuationsat 0.4 hPa
(•- 55 km) for data at 72øS(left) and 72øN (right). Spectraare calculatedfrom overlapping60day time seriesthroughout the PMR record. Note the spectralpeaks correspondingto eastward
traveling waveswith periodsnear 4 days during southernhemispherewinter.
2
LAWRENCE
AND RANDEL:
OBSERVED
MESOSPHERIC
Zonal wind
•/,I/
•
80
-'-
.01
80
• 60
.10• • 60
•5o
• ••
40
•
30
,
0
30N
,
,
,
.Ol
5o
40
10
30S
30
,
,
60N
•
,
,
60S
30S
Latitude
4-day
23,487
4-day wave amp
......
60S
VARIABILITY
0
30N
60N
Latitude
wave mom flux
8o
4-day
.01
wave EP flux div
80
_o
.Ol
•70
.10E ••
.•o E
60
L
o• 50
10
60S 5OS '0 '5ON 60N
1
•_
•40
50
60s
Latitude
•n
10
5os
0
50N
60N
Latitude
Figure 16. Meridional
sections
of zonal-mean
wind(upperleft) and4-daywavegeopotential
height(upperright)andnorthward
momentum
flux(lowerleft)andEliassen-Palm
(EP) flux
signature
(lower
right)duringa largeamplitude
waveeventin September
1977.Shading
in the
zonalwinddenotes
regionsof negativepotentialvorticitygradient.Contours
of the EP flux
divergence
areq- 1, 2, .... m s-1 d-1.
ubar
at
.1
mb
1993],suchpulsesare not alwaysobviouslyassociated
120
with pulsesfrom the troposphereand cannot easily be
interpreted as Rossby-waveevents.
The actual mesosphereis even more disturbed than
.-• 100
60
40
•
._
20
-20
the PMR observations
suggest.Both rocketdata [e.g.,
Leovy and Ackerman,1973] and satellite data (Miles
and Grose[1989]and the more recenthigh-resolution
temperature retrievals provided by ISAMS) showphe-
................
qsuby af 56 S
.1 mb
nomena with periods of a day to a few days and small
vertical
scales which would not be visible in the PMR
data. Indeed,Leovyand Ackerman[1973]statein their
4-day
wave DF at 60 S
analysis of rocket data that the main feature of interest is the changefrom remarkable steady flow below 45
km to strongly oscillating flow above 50 km, and this is
.2 mb
5
consistentwith the more recentsatellitedata (including direct wind observationsfrom the high resolution
Dopplerinterferometeraboard UARS). The PMR data
don't showsuchan obviouschange,which again is probably due to the poor vertical resolutionof the retrievals.
Such small-scale phenomena are probably due to a
number of causes:Obviously,fluctuations in the source
strength and propagation conditions will lead to variability in gravity wave driving. The further influenceof
,
4-day
wave amp af 72 S
.4 mb
600
400
200
0
May
June
July
August
September
Figure 17. Time series of background mean flow
and 4-day wave quantities during May-September 1977.
Shown from top to bottom are the 0.1-hPa zonal wind
at 32øS and 56øS; quasi-geostrophicpotential gradient
at 56øS, 0.1 hPa; EP flux divergenceat 60øS, 0.2 hPa;
and geopotential height amplitude at 72øS, 0.4 hPa.
nonlinearity[e.g,DunkertonandDelisi,1985]uponupward propagatingdisturbances(suchasthosedescribed
above) and both local barotropic and baroclinic instability must also contribute. This latter prospect is consistent with our calculations that the meridional gradient of zonal-mean potential vorticity often changes
sign on the poleward flank of the mesosphericjet, and
towardthe pole, duringwinter (thussatisfyingthe nec-
23,488
LAWRENCE
AND RANDEL: OBSERVED MESOSPHERIC
VARIABILITY
waves
essaryconditions
for meanflowsto be unstable).Since that of Salby[1981].Hisresultsshowsymmetric
near5 daysandin therangeof 11-20days,
the zonal-meanflow itself exists in the presenceof per- with periods
mode(latitudinallyoutof
turbations,it is highlylikely that a localcriterionfor whilethefirstantisymmetric
hasperiodsof 8-11 days.
instabilityis satisfiedthroughoutthe polarwintermeso- phasebetweenhemispheres)
sphere(Pedlosky
[1979]presents
an argument
that sim- The 5- to 10-daymodesseenin the PMR data appear
symmetric
5-day
pleinstabilityanalyses
areprobablybiasedtowardsta- verysimilaroverallto the calculated
little spreadof
bility)ø Further,giventhat the staticstabilityin the mode,althoughSalby'sresultssuggest
mesosphere
is assmallasandoftenlocallysmallerthan enhancedresponsebeyondperiodsof 4.4-5.7 days,in
seenin the PMR data.
that in the troposphere,disturbances
with horizontal contrastto the lowerfrequencies
scalesof the order of 1-2000 km might be expected.
However,unlike the troposphere,disturbances
in the
mesosphere
are subjectto rapid radiativeand dynamical damping,somanysuchdisturbances
may not developintoplanetary-scale
waves,evenwhentheymight
be locallylargefor shortperiodsof time. The manifestationof these sort of phenomenain the PMR data
may well be the smallanticyclones
that can be seen
persisting
for shortperiodsof time in synopticmapsof
geopotentialheightin the mesosphere.
Acknowledgments. Thisworkisbasedonthe considerableeffortof Clive Rodgers,JohnBarnett, and the original PMR team with the PMR instrument. We are indebted
to Clive for providing the new PMR temperature retrievals
and to John for providing the CIRA86 data in numerical
form•
Both Clive and John have been invaluable
sources of
information about the data. We are also grateful to John
Gille and Dan Packman for taking the along-track data and
convertingit into a global grid. B.N.L. acknowledges
the
supportof the Natural EnvironmentResearchCouncil,the
inspiration of Sean Fitzpatrick and his team, and a num-
Although the PMR data don't often directly show ber of useful conversations with David Andrews and Alan
such small-scale phenomena, the effect of these scales O'Neill. W.J.R. acknowledgessupport from NASA grants
is manifestin the calculations
of Hassand Ebel[1986],
W-16215
and W-18181.
who in an attempt to assessthe efficiencyof transport
due to waves carried out an analysis of the integral References
scalespresentin the data. They found considerablespatial and temporal variability and, in particular, found Andrews, D. G., J. R. Holton, and C. B. Leovy, Middle
timescaleslessthan 5 daysnearly everywherein the upAtmosphereDynamics. Academic Press, Inc., San Diego,
Calif., 1987.
per mesosphere. This, coupled with their observation
that if radiative processeswere dominating then spatial Barnett, J., and M. Corney, Middle atmosphere reference
model derived from satellite data, in Atmospheric Strucfields should have been smoother than observed, leads
ture and Its Variation in the Region 20 to 120 kin, edited
one to conclude
that
the effect of the small scales can
by K. Labitzke, J. Barnett, and B. Edwards, vol. 16 of
be found in the data even if it is not directly observed.
Handbookfor MAP, pp. 47-85. Univ. of Ill., Urbana, 1985.
The major wave phenomenawhich are directly ob- Clancy, R., and D. W. Rusch, Climatology and trends of
mesospheric(58-90 km) temperaturesbasedupon 1982servablein the PMR data are the 4-day wave and the
1986 SME limb scattering profiles, J. Geophys.Res., 9•,
normal modes discussedabove. The 4-day wave in the
3377-3393, 1989.
PMR data is prevalent throughout the southern win- Clancy, R. T., D. Wo Rusch, and M. T. Ca]lan, Temperature
ters observed, consistent with the repeated occurrence
minima in the average thermal structure of the middle
mesosphere
(70-80 km) from analysisof 40- to 92-km SME
of negative gradients of potential vorticity that are obglobal temperature profiles, J. Geophys.Res., 99, 19,001servedin the winter mesosphere.Although the change
in signof •y is not a sufficient
conditionto indicate
instability mechanisms,the coincidenceof the negative PV gradients and positive EP flux divergencesreported above are further indicators of internal instabilities. The lack of an obvious 4-day wave in the NH
is consistent
with the lessnegative
•y andis probably
19,020, 1994.
Curtis, P., J. Houghton, G. Peskett, and C. Rodgers, Remote sounding of atmospheric temperature from satellites Vo The pressure modulator radiometer for Nimbus
F, Proc. R. SocoLondon A, 337, 135-150, 1973.
Dunkerton, T. J., and D. P. Delisi, The subtropical mesospheric jet observed by the Nimbus 7 limb infrared monitor of the stratosphere, J. Geophys. Res., 90, 10,681-
due to the fact that the typical NH winter is character10,692, 1985.
ized by massivetransiencein the backgroundstructure, Elson, L. S., Satellite observationsof instability in the mid-
associated
with the seriesof minor and/or major warmings.
The power and coherence spectra calculated from
P MR data show evidence of normal modes with pe-
dle atmosphere, J. Atmos. Sci., •7, 1065-1074, 1990.
Fleming, E. L., S. Chandra, J. Barnett, and M. Corney,
Zonal mean temperature, pressure,zonal wind and geopotential height as functionsof latitude, Adv. SpaceRes., I O,
1211-1259, 1990.
riods in the (broad) 5- to 10-day band, all with sym- Fritts, D.C., and R. A. Vincent, Mesosphericmomentum
metric
latitudinal
structure.
The modes observed here
flux studies at Adelaide, Australia: Observationsand a
gravity wave-tidal interaction model, J. A tmos. Sci., •,
with periods near 5 days appear to be manifestations
605-619, 1987.
of the first symmetricmode (often termed the "5-day Fritz, S., and S. Soules,Large-scaletemperature changesin
wave," first analyzed in P MR radiance data by Prata
the stratosphere, J. Atmos. Sci., 27, 1091-1097, 1970.
[1989]). The most comprehensive
numericalstudy of Garcia, R. R., On the mean meridional circulation of the
normal mode structuresin realistic backgroundflowsis
middle atmosphere,J. Atmos. Sci., •,
3599-3609, 1987.
LAWRENCE
AND RANDEL: OBSERVED MESOSPHERIC
Gille, J. G., P. L. Bailey, L. V. Lyjak, and J. M. RusselIII,
Results from the LIMS experiment for the PMP-1 winter
VARIABILITY
23,489
atmosphererevealedby Nimbus 5 and 6, J. Atmos. $ci.,
J6, 2485-2508, 1989.
Miles, T., and W. Grose, Upper stratospherepolar jet in1978/79, Adv. SpaceRes., 2, 163-167, 1983.
stability in the southernhemisphere,Geophys.Res. Left.,
Hartmann, D. L., Barotrop.•cinstability of the polar night
16, 239-242, 1989.
jet stream, J. Atmos. Sci., •0, 817-835, 1983.
Hass,H., and A. Ebel, Spaceand time scalesof large scale Pedlosky,J., GeophysicalFluid Dynamics. Springer-Verlag,
New York, 1979.
variations in the upper stratosphereand mesosphereas
deduced from the PMR of Nimbus 6, J. A tmos. Terr. Pfister, L., Baroclinicinstability of easterlyjets with appliPhys., •8, 1073-1083, 1986.
cations to the summer mesosphere,J. A tmos. Sci.,
Hirooka, T., and I. Hirota, Normal mode Rossbywavesobservedin the upper stratosphere.Part II: Secondantisymmetric and symmetric modes of zonal wavenumbers I and
2, J. Atmos. $ci., •2, 536-548, 1985.
Hirooka, T., and I. Hirota, Further evidence of normal mode
Rossby waves, Pure Applo Ceophys., 130, 277-289, 1989.
Hirota, I., and J. Barnett, Planetary waves in the winter
mesosphere- preliminary analysis of Nimbus 6 PMR results, Qo Jo Ro Meteorolo$oc.• 103, 487-498, 19770
Hirota, I., and To Hirooka, Normal mode Rossby waves observedin the upper stratosphere. Part I: First symmetric
modesof zonal wavenumbers1 and 2, J. Atmos. $ci., •1•
1253-12679 1984.
Hitchman, M. H.• J. C. Gille• C. D. Rodgers, and
G. Brasseur• The separated polar winter stratopause: A
gravity wave driven feature, J. Atmos. $ci., •6, 410-422,
1989.
Kanzawa, H., Warm stratopause in the Antarctic winter, J.
Atmos. Sci., •6, 435-438, 1989.
Kasahara• A., Effect of zonal flows on the free oscillations
of a barotropic atmosphere, J. Atmos. Sci., 37, 917-929,
1980.
Labitzke, K.• The interaction between stratosphere and
mesospherein winter, J. Atmos. Sci., 29, 1395-1399, 1972.
Labitzke, K., and J. J. Barnett, Review of the radiance distribution in the upper mesosphere as observed from the
313-330, 1985.
Pierce, R. B., W. T. Blackshear, T. D. A. Fairlie, W. Lo
Grose• and R. E. Turner, The interaction of radiative and
dynamicalprocessesduring a simulatedsuddenwarming,
J. A tmoso$ci., 50, 3829-3851, 1993.
Prata, A., The 4-day wave, J. Atmos. $ci., •1• 150-155,
1984.
Prata, A., Observations of the 5-day wave in the stratosphere and mesosphere, J. Atmos. Sci., •6, 2473-2477•
1989.
Randel, W. J., Global atmospheric circulation statistics•
1000-1 rob, Tech. Rep. TN-366+STR• Nat. Cent. for Atmos. Res., Boulder, Colo., 1992.
Randel, W. J., Global normal-mode Rossby wavesobserved
in stratosphericozone data, J. A tmos. $ci., 50• 406-420,
1993a.
Randel, W. J., Global variations of zonal mean ozone during
stratospheric warming events, J. A tmos. SCio•50, 33083321• 1993bo
Randel, W. J.• and L. R• Lait, Dynamics of the 4-day wave
in the southern hemispherepolar stratosphere•J. Atmoso
Sci., •8, 2496-2508, 1991.
Rodgers, C• D., Evidence for the five-day wave in the upper
atmosphere, Jo Atmos. $ci., 33, 710-711, 1976.
Rosier• S., B. Lawrence•D. Andrews, and F. Taylor, Dynamical evolution of the northern stratosphere in early winter
Nimbus 6 pressuremodulator radiometer (PMR), Planet.
1991/92, as observedby the improvedstratosphericand
Space2•c•., 29, 673-685, 1981.
Labitzke, K., K. Petzoldt• B. Nuaujokat, E. Klinker, and
mesosphericsounder, J. Atmos. Sci., 51, 2783-2799, 1994.
Salby• Mo L., Rossby normal modes in nonuniform background configurations. Part II: Equinox and solstice conditions, Jo Atmos. $ci., 38, 1827-1840, 1981.
Schoeberl• M. R., and J. H. E. Clark• Resonant planetary
waves in a spherical atmosphere, J. A tmos. Sci., 37, 20-
R. Lenschow,25 years of the "Berlin Phenomenon" (a
major warming in December 1976-January 1977),in Collection of Reports on the Stratospheric Circulation During
the Winters 197•/75-1991/92, editedby B. Naujokat,and
K. Labitzke, pp. 22-35. Solar Terrestrial Energy Program,
Univo of Ill., Urbana, 1993.
Lait, Lo R.• and J. L. Stanford, Fast, long-lived features
in the polar-stratosphere, Jo Atmos. Sci., •5, 3800-3809,
28, 1980.
Smith, A. K., Longitudinal variations in mesosphericwinds:
Evidence for gravity-wave filtering by planetary-waves• J.
Atmos. $ci., 53• 1156-1173, 1996.
1988.
Venne, D. E.• Normal-mode Rossby waves observed in the
wavenumber 1-5 geopotential fields of the stratosphere
Lawrence, B. N., G. J. Fraser, Ro A. Vincent, and
and troposphere, J. Atmos. $ci., d6, 1042-1056, 1989.
A. Phillips, The 4-day wave in the Antarctic mesosphere•
Venne, D. E., and J. L. Stanford, Oservation of a 4-day temJ. Ceophys. Res., 100, 18,899-18,908, 1995.
perature wave in the polar winter stratosphere, J. Atmos.
Leovy, C., and T. Ackerman, Evidence for high-frequency
$ci., 36, 2016-2019, 1979.
synoptic disturbances near the stratopause• J. Atmos.
$ci., 30, 930-942, 1973.
Madden, R. A., Observationsof large-scaletraveling Rossby
B. N. Lawrence,Department of Physicsand Astronomy,
waves, Gev. Ceophys.,17, 1935-1949, 1979.
University of Canterbury, Private Bag 4800, Christchurch,
Manney• G. L., The stratospheric4-day wave in NMC data, New Zealand(email: b.lawrence•phys.canterbury.
ac.nz)
J. Atmos. Sci., •8, 1798-1811, 1991.
W. J. Randel, AtmosphericChemistry Division, NCAR,
Manney, G. L., and W. J. Randel, Instability at the winter Boulder, CO 80307-3000.
stratopause, a mechanismfor the 4-day wave, J. Atmos.
(ReceivedOctober16, 1995;revisedMay 14, 1996;
$ci., 50, 3928-3938, 1993.
Marks• C. J., Some features of the climatology of the middle acceptedMay 14, 1996.)