GEOPHYSICAL RESEARCH LETTERS, VOL. 28, NO. 9, PAGES 1903-1906, MAY 1, 2001
Mesospheric planetary waves at northern hemisphere
fall equinox
H.-L. Liu, R. G. Roble
NCAR High Altitude Observatory,Boulder, Colorado
M. J. Taylor, W. R. Pendleton, Jr.
SpaceDynamicsLaboratory & Department of Physics,Utah State University,Logan, Utah
Abstract.
Northern hemisphereplanetary wavesare strong in the
winter and weak in the summer, and they go through
a fast transition around equinox. This transition is
studied here using NCAR Thermosphere-IonosphereMesosphere-Electrodynamicsgeneral circulation model
tions with gravity wavesduring winter (Smith, 1996).
More recently, Wang et al. (2000) inferred planetary
wave structures
in horizontal
wind in the lower thermo-
sphere from the UARS Wind Imaging Interferometer
(WINDII), and it is foundthat the stationaryplanetary
wave I is strong at solsticeand wave 2 is strong around
(TIME-GCM) simulationswith 1997 National Centers equinox. Because the extratropical strato-mesospheric
for EnvironmentalPrediction (NCEP) analysis. The zonal mean wind changesdirection around equinox, it
planetary wave variability during the transition and its is a transitional period for the planetary wave propaeffect on the temperature and winds in the mesosphere gation betweenthe summercharacteristics(weak planare examined. The simulated planetary wave struc- etary wave) and winter characteristics(strong planeture agrees with climatological studies, and the fast tary wave). The climatologicalstudiesby Barnett and
transition of the planetary waves is captured by the Labitzke (1990) indeedshowthat the planetary waves
model. The wave variability produces large tempera- (especiallywave 1) changesignificantlyaround both
ture changesin the upper atmosphereabove local sta- spring and fall equinoxes. Furthermore, the in situ
tions in middle and high latitudes. The qualitative be- forcing of planetary waves should also change as a rehavior of the model is in excellentagreementwith recent sult of changesin the mean winds as well as wind filobservationsof a major perturbation in OH mesospheric tering of gravity waves. The equinox transition may
temperaturesfrom Ft. Collins(Taylor et al., 2001), al- produce large wave transience and variability in wind,
though the smaller calculated magnitude suggeststhat temperature, and hence airglow emissions. Large varithe planetary wave amplitude might be underestimated ations of the OI 557.7nm airglow emissions at midby the model.
dle and high latitudes are observedaround the spring
equinoxby ground based instrumentation and WINDII
(Shepherdet al., 1999). More recently, an exception1.
Introduction
ally large amplitude perturbation in mesospherictemperature was observedaround the fall equinox from Ft.
The propagation and growth of planetary waves are
Collins(41øN, 105øW) by Taylor et al. (2001). They
strongly dependenton the backgroundatmosphericwind
also found that a large departure from the nominal seasystem. Theoretical studies show that the westerly
sonaltemperature(•-25-30 K) occursoveran intervalof
wind in the middle atmosphere during winter is faapproximately 3-4 weeksshortly after the fall equinox.
vorable for the propagation of the small wave numIn this study, the fall equinoxperiod for 1997 is simuber planetary waves(cf. Charney and Drazin, 1961). lated by the NCAR TIME-GCM with 1997 NCEP forcThis is supported by climatological studies from both
ing at the lowerboundaryof the model (•-10 mb). The
ground based and satellite measurements.Barnett and
model simulations predict transition of the planetary
Labitzke (1990) show that the planetary wave per- wavesconsistentwith the climatology and temperature
turbations of geopotentialheight and temperature are
variations qualitatively similar to the aforementioned
strongest during the winter seasonbetween •-20 km observations.
and •-80 km. Wind measurements from the Upper
AtmosphericResearchSatellite(UARS) High Resolu2. Numerical
Model
tion Doppler Imager (HRDI) suggeststrong planetary
wave activity as well as large variations due to interacThe TIME-GCM (Robleand Ridley, 1994) is a threedimensionalgeneral circulation model that includesthe
Copyright2001 by the AmericanGeophysical
Union.
Papernumber2000GL012689.
0094-8276/01/2000GL012689505.00
relevant physics and chemistry to self-consistentlypredict the responseof the neutral atmosphereto tides and
other atmospheric wavesfrom 30 to 500 km. The model
1903
1904
LIU ET AL.: MESOSPHERIC PLANETARY
180
180
90
90E 90
90E
WAVES AT EQUINOX
wave forcing in driving the wind reversalin the mesosphereand lower thermospherearound the fall equinox.
The planetary wave I growth aroundfall equinoxcan
also be seenfrom the temperature perturbations. Figure 3(a) and (b) showlongitude-latitudecrosssection
of the planetary wave I componentof the temperature
perturbation at about 87 km on days 260 and 285. On
day 260, the dominant source of the planetary wave
I is in the southern hemisphere, and its amplitude at
0
0
(a)
(b)
middle/highlatitudesin the northernhemisphere
is less
than 1 K. The planetary wave 1 amplitude in the northern
hemisphereis much larger on day 285 with a peak
(a)260and (b) 280at 10 mb (NCEP 1997)from30øN
value
of 12 K between 60øN and 70øN with phase lines
to north pole. Solid contoursare positive valuesand
Figure I.
Geopotentialheight perturbationson day
contour interval
is 20 m
setup is similar to that describedin Waltercheid et al.
tilting westwardwith decreasinglatitudes(and increasing altitudes,not shown). These agreewith the climatological features of the planetary wave under winter
conditions(Barnett and Labitzke, 1990). Therefore,the
(2000), exceptthat the daily variation of geopotential model reproducesthe rapid transition of planetary wave
and temperatureaccordingto NCEP analysis(Randel, I from summer characteristics to winter characteristics
1992) is specifiedat the lowerboundaryfor the period around the fall equinox within lessthan a month. This
July 1-October 31 of 1997 for the run usedin this paper. fast transition is consistent with the September and
October climatologicalresults (Barnett and Labitzke,
1990). Figure3(c) and (d) are the temperaturechanges
3. Results and Analysis
at the same altitude
as a function
of time
above two
Analysisof the N CEP geopotentialheight at 10 mb locations:105øWand 70øEand both e•t42.5øN.They
indicatessignificantvariationsin the planetary waves show a superposition of variations due to tidal waves
from day 260 to day 290, with the generaltrend of a and planetary waves. The crossesindicate temperature
growingplanetary wave I component.Figure I shows valuesat local night (2000LT-0400LT)as measuredby
the geopotentialheight perturbation on day 260 and ground based imagers. It should be pointed out that
280. The wave I componenton day 280 has increased phasesof the tidal wave perturbations are different in
significantlyand becomesthe dominantfeature. The Figure3(c) and (d), with temperatureminimumat midphase of the wave I component also changesduring night in (c) and at 1800LT in (d). This is due to the
the sameperiod of time. Zonal wavenumber-frequencypresenceof relatively large non-migratingtides (espeanalysis(not shown)showsthat the planetarywave 1 cially westwarddiurnal wave2 componentin this case).
has an averageamplitudeof •100 m betweenday 260
The amplitude and phase changesof the planetary
and 273 and then increasesto a peak value of -•270 m wave I between day 260 and day 290 at 62.5øN and
on day 280. It decreasesslightly and stays above 200 42.5øN are plotted in Figure 4. There are large variam betweenday 280 and day 290. The phaseshifts by tions in the planetary wave amplitude, but the general
about 120ø eastward between day 260 and 280.
trend of an increase in wave I activity is evident. At
The increasedplanetary wave activity evident in the 62.5øN near where the wave amplitude peaks, the phase
lower stratosphericforcing penetrates into the meso- of the wave is relatively stable before day 274 and af-
sphere.This is demonstratedin Figure 2, which shows
the Eliassen-Palm(EP) flux on days260 and 285. The
EP flux on day 285 increasesat almost all altitudes in
the middle/highlatitude stratospheric
and mesospheric
region, with preferredupward/equatorwardpropagation. It is evident from Figure 2 that the planetary
wavepropagationand growth are sensitivelydependent
on the zonal wind, becausethe maximum westerlywind
correspondsroughly to the minimum refractive index
for planetarywaves.Therefore,planetarywavesrefract
away from the region of maximum westerly winds as
60
40
•O Latitug
7o
eo
40
eo LatitudeeO 7o
eo
(b)
(a)
Figure 2. EP flux on day (a)260 and (b)285. The
shownin Figure2. In addition,divergence
of the EP arrows represent direction and magnitude of EP flux
flux showsthat the planetary wave produceswestward and the length scaleis the samefor both plots. The thin
forcingabove80 km at middle/high latitudes with a contourlines are the zonal mean wind values(contour
peakvalueof •30 ms-Xday
-x at 70øNonday285,com- interval:5 m s-x), andthe thickcontourlinesare EP
pared with almost zero forcingon day 260. Therefore, forcingvalues(contourinterval:15 m s-x day-X). Solid
planetary wave forcingappearsto complementgravity
contour
lines are for eastward
direction
LIU ET AL- MESOSPHERIC PLANETARY WAVES AT EQUINOX
1905
fact that the observedtemperature changesare larger
than the model results may indicate that the planetary
wave temperature perturbations are underestimated in
................
•••Longltude
Longitude
Longitude= 70 •
Longitude= 105W
ß
Local Time
...::: :.+•::: .: •: •::
Local Time
Figure 3. Stationary planetary wave 1 temperature
perturbationson day (a)260 and (b)285 at •87 kin,
and temperature changesabove 2 longitudinal locations at 42.5øNand 87 km from the model: (c)105øW
and (d)70øE. The dotted lines are the time seriesbetween day 260 and 290, and crossesare at local nights
(2100LT-0300LT
in (c) and2000LT-0400LT
in (d)) with
2 hours intervals.
the model. This is confirmedby comparing the temperature amplitude from the model with that from the
NCEP at I mb and 60øN. They are in reasonableagreement beforeday 274, but the differenceincreasesrapidly
afterward with the NCEP amplitude more than twice
as large after day 276.
The wave I growth also showsup in the zonal wind
perturbations. Figure 5 showsthe horizontal and vertical structures of the wave I zonal wind perturbations
on days 260 and 285. Both wave amplitude and phase
structures undergo considerablechange during the fall
equinox. The peak value of the wave amplitudes at •96
km and 30øNincreases
from •5 ms-• to •15 ms-•, and
the tilting of the phase lines changes from eastward-,
equatorward to westward-equatorward. On day 260,
wave I zonal wind perturbation peaks in the southern
hemisphere and penetrates to the northern hemisphere
with a secondarypeak at the equator. This agreeswith
the global stationary wave I structures calculated by
Medvedevet al. (1991). On day 285, the wave I peak
shifts to the northern hemisphere and the structure
ter day 280, indicating the wave is stationary. Between there is similar to the stationary planetary wave unwind data
these two days, however,the phaseshifts westwardby der winter conditions as seen in the WINDII
almost 180ø. At 42.5øN, there is a similar westwardshift (Wang et al., 2000). Thereforethe rapid transition of
of about 220ø. The phaseshift is related to the phase the planetary wave around equinox is also reflected in
shift of the N CEP forcing at 10 mb and the associated its zonal wind component.
transience.
As shownin Figure 5(c) and (d), the vertical phase
The wave I growth and phasetransiencemay provide lines of the wave have a characteristicwestward tilting
a plausibleexplanation
to the observed
largetempera- on day 285 but not on day 260, and the former indicates
ture variationaroundfall equinox(Taylor et al., 2001), a forcedplanetary wave I with large vertical wavelength
where it is shownthat the nightly averagedtempera- ( > 100kin). The phaselineson day 285 alsoshowa disture at 87 km altitude increasesby about 25 K between tortion above 85 kin. This is due to the superposition
day 270 and day 280 aboveFt. Collins(41øN,105øW).
As shownby Tayloret al. (2001), the trend of this temperature variation is similar to the variations shown in
14
Figure 3(c), thoughthe observedchangesare about 3
(5)
q
times larger than the model results. In the model, the
temperature variations between day 270 and 280 corre-
spondto the wavetransience.As shownin Figure 4(c)
and (d), the amplitudeof the wave I temperatureperturbation increasedand its phase shifts westward between day 274 and day 277. On day 274 the phase of
the wave I temperature perturbation was about 0ø, so
Ft. Collins was observingthe cold sector of the wave.
On day 277 the phase of this wave shifted to about
105øW and Ft. Collins was now observingthe crest of
the warm sectorof the wave. This amplitude and phase
changeof the wave I causeda net temperature increase
of about 6 K. Detailed analysis of the model results
showsthat there was also an associatedphasechange
in the wave 2 componentwhich causeda temperature
increaseof about 2 K. The phase changealso led to
the temperature decreasingduring the same period of
time at longitude 70øE. The zonal mean temperature,
on the other hand, only increasedby about I K. The
..............
,
days
days
Figure 4. The amplitude and phase changesof stationary planetary wave I temperature perturbations
at 62.5øN and 42.5øN between day 260 and 290.
(ø)amplitude,62.5øN;(b)phase,62.5øN;(c)amplitude,
42.5øN;(b)phase,42.5øN.Somephasesare shifted360ø
for better viewing.
1906
LIU ET AL.: MESOSPHERIC PLANETARY WAVES AT EQUINOX
•___••_..
.......................
._/.bations
ortheir
transience
atmid-latitudes
might
be
,o
iiii.::11:i:-•:::i:•.•/••:.ii
underestimated
by
the
model.
'i•••
•
C.K.
Meyer
for
helpful
discussions
and
comments
Effort
]ø
-•
'•
_,•
(:•
.....
••
•
'ø
i........ I//
-'•
byHLL
and
RGR
were
supported
inpart
byNASA
grant
S-
•
,•
measurements
were
supported
byNSFgrants
ATM-9403474
and ATM-9612810. The National Center for Atmospheric
•'•••••
:-"-"•":':;•"•'•"
Rese
arch
is
sponsored
by
NSF.
•
ß
....
:..71:.!!
'"'--•••
.... .-?..•...,•..-•_.....:::.:::::::::/:::.::.::.,•....•:
97239-E
toNCAR.
TheCEDAR
mesospheric
temperature
...t ........
Lo.•,ua,
'.....
(c)
• :'•;:'
•' •":'""••' References
•,. Barnett,
ofplanetary
J.J.,
and
waves
K.
in
Labitzke,
the
middle
Climatolog
atmosphere,
distrib
Adv.
Space
Res., 10, (12)63-(12)91, 1990.
Lo•,u•, '•
Charhey,
J. G.• andP. G' Drazin• Propagation
ofplanetary
scale disturbances from the lower atmosphere into the up-
Figure 5. StationaryplanetarywaveI zonalwind perperatmosphere,
J. Geophys.
Res.,66,83-109,1961.
turbationsat -,•96km on day (a)260 and (b)285; and Medvedev,
A. S., A. I. Pogorel'tsev,
andS. A. Sukhanova,
at 32.5øNon day (c)260 and (d)285. The thick conModelingthe globalstructureof stationaryplanetary
tour linesin (d) arethe gravitywaveforcingvalues(m
waves
andtheirpenetration
across
theequator,
Izv.Russ.
s-•day-1) withsolidlinesforeastward
forcing.
Acad.
$ci.USSR
Atmos.
Oceanic
Phys.
Engl.Transl.,
31,
574-581, 1991.
Meyer, C. K., Gravity wave interactionswith mesospheric
planetary waves' A mechanismfor penetration into the
of an in situ forcedplanetarywaveI by breakinggrav-
thermosphere-ionosphere
system,J. Geophys.
Res.,10•,
itywaves
ontheprimary
planetary
wave(Smith,
1996; Randel,
28,181-28,196,
1999.
W. J., Global
Meyer, 1999). The thick contourlines are the corre-
sponding
stationary
waveI component
of the gravity
atmospheric circulation statis-
tics, 1000-1 mb, NCAR/TN-366-/-STR, NCAR, Boulder,
Colo.,1992.
waveforcing
witha peakvalueof-,•25ms-1day-1 at Roble,
R.G.,andE.C.Ridley,
Athermosphere-ionosp
90 km. The peak forcingis alongthe zero wind pertur-
mesosphere-electrodynamics
general circulationmodel
bationlineoftheplanetary
waveandin theareawhere (TIME-GCM).
equinox
solar
cycle
minimum
simulations
theslopes
ofthephase
lines
arechanging,
which
clearlyShepherd,
(30-500km),
Geophys.
Res.
Left.,21,417-420,
1994.
G. G., J. Stegman, P. Espy, C. McLandress,G.
demonstrates
theinteractions
ofgravity
waves
andthe Thuillier,
andR.H.Wiens,
Springtime
transition
inlower
planetarywave.
thermospheric
atomicoxygen,
J. Geophys.
Res.,10,{,213223, 1999.
4' Conclusion
Smith,
A.K.,Longitudinal
variations
inmesospheric
winds'
Evidence for gravity wave filtering by planetary waves, J.
Themodel
results
show
thatplanetary
wave
I growsTaylor,
Atmos.
Sci.,53,1156-1173,
1996.
M. J., W. R. Pendleton, Jr., L.
C. Gardner, H.-
rapidlyin themesosphere
around
thenorthern
hemi- L. Liu,R. G. Roble,
C. Y. She,andV. Vasoli,
Large
spherefall equinoxandthe peakof the planetarywave
activity shifts from the southernhemisphereto the
amplitude
perturbations
in mesospheric
OH Meineltemperatures
aroundthe autumnalequinox
transitionperiod,
northern
hemisphere.
Thesurge
oftheplanetary
wave
Geophys.
Res.Left.,thisissue,
2001.
causes
aneasterly
mean
flow
acceleration
inthemid/high
Walterscheid,
R. L., G. G. Sivjee,
andR. G. Roble,
Mesosphericand lower thermosphericmanifestationsof a
latitudeuppermesosphere.
Thesimulated
planetary stratospheric
warming
event
over
Eureka,
Canada
(80øN),
wavestructure,in-situ planetarywaveforcingdue to
Geophys.
Res.Lett.,27, 2897-2900,
2000.
gravitywavebreaking,andthe rapidtransitionaround Wang,D. Y., W. E. Ward,G. G. Shepherd,
andD.-L. Wu,
equinox
compare
favorably
withprevious
climatologi- Stationary
planetary
waves
inferred
fromWINDIIwind
calandtheoretical
studies.
Wavetransience
associateddata
taken
within
altitudes
90-120
km
during
1991-1996,
J. Atmos. Sci., 57, 1906-1918, 2000.
with the transition can lead to fast changesin both
phase
andamplitude
oftheplanetary
waves.
These H.-L.LiuandR. G. Roble,
NCARHighAltitude
changes
maycause
largevariability
inlocaltemperatureObservatory,
POBox3000,Boulder,
CO80307-3000.
(e-
measurementsat mid/high latitudes. This providesa mail' [email protected];
[email protected])
qualitativeexplanation
for the largetemperaturevari-
M. J. Taylor and W. R. Pendleton,Jr., Space
ationsobserved
aroundthe fall equinox
by Tayloret Dynamics
Laboratory
& Department
of Physics,
Utah
State University, Logan, UT
84322-4145.
(e-mail:
al. (2001). The modeledtemperaturevariations,how- mt [email protected])
ever, are smaller than the measuredquantities,which
suggests
that the planetarywavetemperaturepertur- (Received
November
24,2000;accepted
February
1, 2001.)