JOURNAL OF GEOPHYSICAL
GRAVITY
WAVES
RESEARCH, VOL. 96, NO. D3, PAGES 5153-5167, MARCH 20, 1991
IN THE MIDDLE
ATMOSPHERE
OBSERVED
BY RAYLEIGH
LIDAR
1. CASE STUDIES
R. Wilson, M. L. Chanin, and A. Hauchecorne
Service D'Atronomie du CNRS, Verrieres le Buisson,France
Abstract. Densityandtemperature
mesoscale
fluctuations
asobservedin the stratosphere
andmesosphere
by meansof
two Rayleighlidarswith highresolutionin time (15 min) and
space(300 m), have beenanalyzedin someparticularcases
correspondingto different seasonalconditions.These case
studiesarecharacteristic
of recurrentlyobservedpatternsand
thusprovidea descriptionof the mesoscalefluctuationfield
in the middleatmosphere.
The spatial,temporal,andspectral
characteristics of the fluctuations are described and discussed
in the framework of the gravity wave interpretation.
Dominantwave modeswith largeperiodand largevertical
wavelength(inertia-gravitywaves)are frequentlyobserved
in the stratosphereand lower mesosphere.These lowfrequencymodesare not generallyobservedabove50- to 55km altitude,suggestinga strongdampingof suchwavesin
the mesosphere.
The vertical growth of potentialenergy
densityindicatesthat the wave motionsare generallynot
conservativein the middle atmosphere.The gravity waves
amplitude appears too small to produce convective
instabilities in the stratosphere.On the contrary, the
amplitude of the fluctuations is close to the convective
saturation limit deduced from the linear theory for
wavelengths
up to 3-5 km in thelowermesosphere,
andup to
6-8 km above 60 km altitude.Furthermore,convectively
instablelayers,which can persistfor periodslongerthan 1
hour,havebeenfrequentlyobserved
in themesosphere.
1. Introduction
It nowappears
fmuly established
thatgravitywavesplaya
major role in the momentumand thermal budgetof the
mesosphere[Lindzen, 1981; Holton, 1982; Fritts, 1984;
Garcia and Solomon,1985]. Dissipationprocesses,which
induce a convergenceof the vertical fluxes of horizontal
momentumand energy,lead to an accelerationof the mean
flow, and to the productionof turbulence.The momentum
convergence
due to wavedissipationinducesthereversalof
the verticalmeanwind gradient,thusa meridionalflow and
thereforethe mesospheric
temperature
distributionknownto
befar fromradiativeequilibrium[MurgatroydandSingleton,
1961]. More recently, it has been suggestedthat gravity
wavesalsoplay an importantrole in the momentumbudget
of the stratosphere
[Miyaharaet al., 1986, Palmeret al.,
1986].
The mean flow
acceleration
and turbulent
diffusion
inducedby thewavesaturation
havebeenfarstparameterized
by Lindzen[ 1981] in the frameworkof the lineartheoryof
gravitywave.The wave saturation
refersto thoseprocesses
limiting the amplitudeof the waves due to instabilities
arisingfrom the large amplitudeof the fluctuations(for a
review see Fritts [1984] and Fritts and Rastogi [1985]).
Despitethe successencounteredby the linear theory,the
involvedschemeof the wave-meanflow interactionappears
to berelativelysimpleandrequiresimportanttheoreticaland
observational
efforts in order to be better understood and
parameterized.Numerousquestionsremain: what are the
typical energy levels of the gravity wave field in the
stratosphere
and mesosphere?
What is the vertical energy
growthrate of the waves?Is therea significantseasonalor
geographicalvariability of the wave activity?What are the
dominantdissipativeprocesses:convectiveor dynamical
instabilitiesor otherprocesses
This paperwill bringelements
of answers for some of these issues.
Up to the present,mostof the mesoscalefluctuationsdata
in the middleatmosphere
havebeenobtainedby usingradar
techniques,rocketssounding,or balloon-borneinstruments.
The horizontalandverticalwind fluctuations
areobservedby
radarbothin the troposphere
andlowerstratosphere
[Balsley
and Garello, 1985; Fritts et al., 1988] and in the upper
mesosphere
[Vincent, 1984; Meek et al., 1985; Vincent and
Fritts, 1987] but radars are blind in part of the middle
atmosphere.Rocket data, on the other hand, have allowed
the studyof wind and temperaturefluctuationsin the 20- to
65-km altituderangefrom a largenumberof rocketranges
[Hirota, 1984; Hirota and Niki, 1985; Hass and Meyer,
1987], but the data are obtained in a sporadicmanner.
Balloonborneinstruments
give accessto thefine structure
of
the fluctuation field [Barat, 1982; Cot and Barat, 1986], the
Papernumber90JD02231.
altituderangebeinglimited to about25 km. Rayleighlidar
offers the unique feature of high resolution routine
measurements
of the densityor temperaturefluctuations,in
the 30- to 75-km altituderange, where radarsand in situ
measurements are not possible and rocket soundings
sporadic. The use of lidar in this height range thus
complements
to the othertechniquesandhasnow provedto
be quitepowerfulto measurethe mesoscale
fluctuationsin
the middle atmosphere.Preliminaryresultswere reported
duringthe early phaseof the developmentof this technique
0148-0227/91/90JD-02231$05.00
[Chanin and Hauchecorne,1981, 1984; Shibataet al., 1986,
Copyright
1991bytheAmerican
Geophysical
Union.
5153
5154
Wilsonet al.: Gravitywavesobservedby RayleighLidar, 1
1988] and from a relatively low performance system
[Gardner et al., 1989]. The purpose of this paper is to
pressure
estimated
at height Zsup
from the CIRA 1986
describe and to discuss the main features of the mesoscale
Thereare now two operatinglidar stationsin the southof
France, located in two sites different with respect to
orography:
theObservatoire
deHauteProvence(OHP) in the
foothillsof theAlps (44øN,6øE)andthe Centred'Essaisdes
Landes at Biscarosse(BIS) on the Atlantic Coast (44øN,
1øW). Routinemeasurements
havebeenperformedon clear
nightssinceJune1981at OHP andsinceMarch 1986at BIS.
High spatialandtemporalresolutiondataallow the studyof
gravitywavemotionswith 300 m verticalintegration
and 15
min integrationtime. A very large data base has been
acquired:it consistsof about100 nightsa year, eachnight
duration lasting between 3 and 14 hours,i.e. more than
10,000 verticaldensityprofries,in an altituderangewhich
hasbeenthusfar barelyexplored.
Only a part of these nights, about 100, have been
individuallystudiedin greatdetails.From this limited but
alreadylarge numberof casestudies,it appearsthat some
wavepatternsare very frequentlyobserved.The few cases
presentedhere are characteristicof recurrentobservations
and thus provide a meaningfuldescriptionof the gravity
wavefield in the middleatmosphere.
Theseparticularcases
fluctuation field from observations obtained by two
performingRayleighlidars.The fluctuations
are interpreted
in the frameworkof the gravitywave theory.Casestudies,
characteristicof recurrentlyobservedwave patterns,are
analyzed,thusprovidinga descriptionof the wave field in
the stratosphereand mesosphere.A climatologicalstudy,
performedover a large data set, is reportedin a separate
paper.
This paperis organizedas follows.In the secondsection
the databaseand the relevantparametersare described.The
differentdataprocessingmethods,giving accessto various
wave features, are reviewed in section 3. From typical
particularcasesthe main characteristics
of the gravitywave
field in the middle atmosphereare describedin section4:
dominant modes, vertical anisotropyof the wave field,
spectralenergydensityversusverticalwavenumber,vertical
growth of energy density per unit mass. The frequent
occurrenceof unstablelapserate in the mesosphereis also
shown. The results and conclusions
are summarized
in
section 5.
model.
have
2. The Data Base
The backscatteredlight from a pulsedlaser beam sent
verticallyinto the atmosphere,
providedthat Mie scattering
couldbe neglected(i.e., above30 km) andin the absenceof
any resonantline, is due to Rayleighscatteringfrom neutral
molecules and is thus proportional to the atmospheric
density. The uncertaintyon the density measurementis
simplyrelatedto the numberof receivedphotons,n(z), and
dependsuponthe spatialandtemporalintegrationof thelidar
signal.The statisticof receivedphotonsbeing a Gaussian
function (asymptoticlimit of a Poissondistribution)the
standarddeviation of the density measurementis simply
proportional
ton(z)
-1/2.
As described in earlier publications[Hauchecorneand
Chanin, 1980; Chanin and Hauchecorne,1984], the absolute
temperature
profilecanbe deducedfrom therelativedensity
measurements
by assumingthe hydrostaticequilibrium,
applyingtheidealgaslaw andfittingthetemperature
profile
with an atmospheric
model(CosparInternational
Reference
Atmosphere(CIRA), 1986) at the upper altitudeof the
measurement.
Thisupperlimitaltitude,
Zsup,
isdefined
asthe
heightwhere the relative statisticalerror on the density
reaches15%, typically about80 to 85 km for a temporal
integrationof 15 min. The temperature
T(zi)in a layerof
thickness
Az centeredon thealtitudezi is givenby
T(zi)
=
giAz
(1)
))
RLn(1
+pigiaz/(P(zsup)
+zw
Lpzgzaz
z=zi
wherePi andgi are respectivelythe densityand the Earth
acceleration
at altitudezi, R theair constant
andP(zsup)
the
been selected
either
because measurements
were
obtained simultaneouslyat both sites (May 29, 1986;
December 17, 1987) or becausethey correspondto long
measurement
periods(August7, 1985 at OHP, January28,
1989 at BIS). In a single case,on May 29 at BIS, wind
measurements
from
tracked rocketsondes
were obtained
simultaneously
with the temperaturemeasurements
from
Rayleighlidar.
Let us brieflydescribetheparameters
of interestfor wave
studieswhicharemadeavailableby Rayleighlidar.It should
be firstrecalledthat,with regardsto gravitywavemotions,it
is equivalentto considerthe densityor the temperature
relative fluctuations (the pressureperturbationsbeing
negligible) provided that the vertical scalesof the
fluctuations
arenotmuchlargerthantheatmospheric
density
scaleheight,Hp. The relativedensityfluctuations,
P'/Po,are
estimatedas the first orderperturbation
termof thevertical
profileof log(p).Therelativeperturbations
of density
(P'/Po)
ortemperature
(T'/To)areextracted
bysubtracting
fromthe
raw vertical profile (or from its logarithm) a smoothed
profileor a third-order
polynomial
fit. The smoothed
profile
is obtainedby applyinga finite impulseresponsefilter,
symmetricand nonrecursive,
the so-calleddiscreteprolate
spheroidal
filter(DPSF)[Mathews
et al., 1983].Examples
of
temperaturevertical profiles, resulting from 15-min
integrationtime and 1-kmverticalresolution,are givenin
Figure 1. The shadedareacorrespond
to plus or minus1
standarddeviation.The relativeuncertaintyfor suchspatial
andtemporalresolution,whichcouldvary from onenightto
another depending upon the laser power and upon the
atmospheric
transmission,
is typicallyaround0.2% at 30 kin,
1% at 55 km, and 5% at 70 km. Vertical profilesof relative
temperature
perturbations
corresponding
to the temperature
profilesshownin Figure1 areplottedin Figure2.
Wilsonet al.:Gravitywavesobserved
by RayleighLidar,1
Mean
75'
5155
Tem')erature
70
65
[
•o
'ID
50
40
22•:2i0
I
)/'it'>r[
•
t"I
/ [
OHP 7 AUG1985
)::!9
0:,4
ß
L
ii
,
Temperoture
(K)
Fig. 1. Temperature
profilesobtainedby Rayleighlidar on August7, 1985, at OHP. The temporal
resolution is 15 min, the vertical resolution is 1 km.
Relative
Temperature
OHP 7 AUGl•g5
Pertubations
(T'/To) (g)
Fig.2. Relativefluctuations
of temperature
corresponding
to thetemperature
profilesshownin Figure1.
The perturbations
areextractedby removinga smoothed
profile(DPS filter, cutoffwavenumber1/12000
m-1)
From the mean densityor temperatureprofilesdeduced
from a low-passfiltering of the data, the static stability
characterized
by the local Brunt-Viiisiiliifrequency,N(z), is
givenby
$2(z)=_
g(po•Z
1•p- -•2
•z+-$Cs)=T•o
C•) (2)
where
g istheEarthacceleration,
Csthesound
velocity,
Cp
the air specificheat at constantpressure,and z the vertical
coordinate.
Density or temperaturefluctuationsare associatedwith
verticaldisplacement
of the atmospheric
fluid andtherefore
are related to the available potential energy of the
fluctuations.
Indeed,for a verticaldisplacement
relativeto an
equilibrium
position,•, thepotential
energyperunitmass,
Ep, is givenby
g2
'1"2
>=•1(N•)2<(p')2>___
<(•oo)
> (3)
Ep
=•1N2<•2
Po •1(N)
wherebracketsindicatea spatialor temporalaverage.
The temporal characteristicsof the temperature (or
density) fluctuations are obtained from the successive
verticalprofilesof the perturbations
whichgive the apparent
frequencyof thewave,o. Indeed,it is notpossibleto directly
estimate from ground based measurementsthe intrinsic
frequency,co(z),in the referenceframeof the meanflow, as
it dependsuponthe meanwind velocity,u(z), the horizontal
wavenumberkh, andtherelativedirectionof propagation
of
the wave with respectto the meanflow:
co(z)= 0- kh. u(z)
(4)
3. Data Processing
The distribution
of gravitywaveenergyas a functionof
altitude, time, vertical wave number,and (apparent)
frequency
is described
by means
of various
dataprocessing
5156
Wilsonet al.: Gravitywavesobservedby RayleighLidar, 1
methodswhich are being presentedhere. Each of these
methodsgivesaccessto differentwaveparameters
in several
altitudeand spectralranges.The grossfeaturesof the wave
field, i.e., the vertical and temporalscalesof the dominant
wavemodes,are viewedby time-heightperturbation
plots.
The spectralenergydistributionis givenby the verticaland
frequencyspectra,whereasthe spatial (vertical) energy
distributionis estimatedby a complexdemodulation
of the
perturbation
profilesin severalwavelengthbands.For each
of thesemethods,
thedefinitionof theperturbation
termswill
be slightlydifferentasdescribedbelow.
Fromdimensional
arguments,
it hasbeenshownby Dewan
averageduponthe individualspectra(resultingfrom 15-min
integratedprofiles) obtainedfor a given day. It has been
verified that the noise of the lidar signal due to the
uncertaintyon the photon counting has a white spectral
response.The white noiselevel is evaluatedin the high wave
numberportionof the meanPSD (wherethe signal-to-noise
x:atio
is low),thatis to say,forwavelengths
smaller
than!
andGood[1986]thatthepowerspectral
density(PSD)of the
relativedensity(or temperature)
fluctuations
of a saturated
km andis subtracted
from theraw spectrum.The uncertainty
on theraw powerspectrum,estimatedin the standardway, is
proportionalto the estimatedPSD. However,becauseof the
variablesignal-to-noiseratio versuswave number(the noise
being subtracted),the uncertaintyon the resultingpower
spectrumshouldbe proportionalto the signal-to-noise
ratio.
The relative uncertainty on the resulting PSD is thus
assumed to be of the form
gravity
wavefieldis proportional
to N4/(g2m3).
The
perturbed
densities
arethus
normalized
as(g•2)(p7Po)
in
orderto providea comparison
betweenthe verticalspectra
obtainedin variouslocations,times,andheightranges,with
respectto the saturation
limit. In orderto reduceedgeeffects
a Welchwindowis appliedto the spatialseries.The power
spectrum,estimatedby a simpleFourier transform,is then
normalizedto the raw variance.The daily meanspectrumis
A(S(k)-B)
1
S(k)
(5)
S(k)-B nsl/2S(k)-B
whereS(k) is the powerspectraldensityof the raw data,B
the whitenoiselevel, andns the numberof spectraover
whichtheaverageis performed.
65
60-
6O
• 50
.•=•
ß•
45
.<
40
40-
35
:3O
18
19
2.0
œ1
a2.
a3
Tin-ie
0
I
2.
3
4
5
6
{ H)
T'/To
I
.•ovE
I
0.0:35•
0_025"-•--.'• 0.015•
o 005 [•3
-0.005 [•3
-0.015 [---]
-0.0•5 ':;i•
-0.035 •
-0.045 ':'•
BELOW
o,o45
0.045
0.035
0.035
0,015
0.005
-0.005
-0•015
-0 0..25
-0.035
--0.045
21
22_,
23
o
Tinge
I
2,
{ H)
Fig. 3. Time-heightcontoursof temperature
relativeperturbations,
on (a) January28, 1989, at BIS; (b) May
29, 1986, at OHP; (c) August 7, 1985, at OHP; and(d) December17, 1987, at BIS.
3
Wilsonet al.: Gravitywavesobservedby RayleighLidar, 1
Spectralanalysisas a functionof vertical wave numberis
performed
in threealtituderanges:
theupperstratosphere
(30
to 45 km), the lowermesosphere
(45 to 60 km), andthe 55to 70-kmheightrange.The potentialenergydensityperunit
massis estimatedby integratingthePSD oververticalwave
numbers
between
1/15and1/0.6km-1,subtracting
thenoise
variance
andscaling
bytheaveraged
value
ofN2.
5157
changesignificantly
in theheightrangeunderstudy.For the
demodulationanalysisthe perturbationterm is definedas
(g/N)(p'/po), the varianceof which is twice the available
potentialenergy.As for the spectralanalysis,thenoiselevel
is estimatedby estimatingthe meanvarianceof shortscale
fluctuations, smaller than 1 km, and is removed from the raw
variance. The uncertainty on the resulting variance is
evaluated in the same way as for the spectral analysis
(equation(5)).
functionof verticalwavenumberand(apparent)
frequency.
The spectraland demodulationanalysesare performed
A frequency
Fouriertransform
is applieduponthesuccessive from densityprofiles which are simply proportionalto the
verticalwavenumberspectragivinga qualitativeinsightinto
lidar signal (i.e., the number of backscatteredphotons),
the temporal characteristicsof the fluctuations.The 2D
whereastemperature
profilesresultfrom a differentiationof
spectrum
shows,for eachwavenumber,thephaseevolution the densityprofiles.Indeed, the white noiselevel is better
of the successiveFourier Transformsof the density defined from the density data, becauseit could be biased
perturbations
profiles.The zerofrequencyterm represents from the temperaturedataowingto the differentiationandto
on the pressureat the topof theprofile.
thepowerspectrum
of thefluctuations
with periodslonger theinitial assumption
In any case,whethertemperatureor densityfluctuationsare
than the measurement
duration,suchfluctuationsbeing
used for such analyses the results are not significantly
verticallyresolved.The integration
of the2D spectrum
over
frequenciesgives of coursethe power spectrumversus different,providedthattheyareobtainedfar enoughfrom the
topof theprofile,asis the casein thisstudy.
verticalwavenumber.According
to therelativesignof the
The time evolution of the density fluctuations are
described
by a two dimensional
(2D) spectralanalysisas a
frequency
andwavenumbertermsof thephase,expressed
as
exp(mz+ cot),the upwardand downwardverticalphase
velocitiescan be identified.Providedthat co<<N(asit is the
case in this study), the dispersion relation of a
monochromatic
gravitywave is
4. Gravity Wave Characteristics
The main features of the gravity wave field in the
stratosphere
and mesosphere,
as observedby two Rayleigh
lidars at mid-latitude, are now describedfrom a few selected
o,)2
=f2+N2kh
2
m2
(6)
where f is the inertial frequency,kh and m being the
horizontaland vertical wave numbers,respectively.The
verticalgroupvelocity,Cg,is thus
;9o•
o•
4.1. A DominantGravityWave Mode in the Stratosphere
f2
Cg-3m-m(1-•)
cases,characteristicof recurrentlyobservedwave patterns.
The presenceof dominantlow-frequencywave modesin the
stratosphere,
the potentialenergydistributionversusvertical
wave number,apparentfrequencyand altitude,and finally
thefrequentoccurrence
of convectivelyinstablelayersin the
mesosphere
will be shown.
(7)
The verticalgroupvelocity,Cg,andtheverticalphase
velocity,Cz (=co/m),arethusin opposite
direction.Therefore
the energy flux of a wave the vertical phasevelocity of
which propagatesdownward is directed upward. Even
thoughthe result couldbe biased,owing to the frequency
Dopplershift, an observedrecurrentbehaviorwouldbe a
clearindicationof a verticalasymmetryof thewavefield.
A complexdemodulationof the densityprofilesis also
performedin threewavelengthbandscenteredat 6, 5, and4
km. The purposeof the complexdemodulation
is to estimate
the variationof the varianceversusaltitudein a givenwave
numberband of the wave spectrum.The spatial seriesis
multiplied by exp(-imc) before applyinga low-passfilter
(DPSF), thereforeprovidingthe varianceof the fluctuations
in a spectralband centeredon rnc. The spectralwidth is a
function of the cutoff wave number of the filter. The used
Most often,a dominantoscillatorymodeis observedin the
stratosphere
andlower mesosphere.
This dominantmodehas
a large apparentperiodrangingtypically from 6 hoursto
infinity, a downwardphasevelocity, and a large vertical
wavelength,from 5 to 10 km. Examplesof time-height
contoursof relative temperatureperturbations
are shownin
Figure3. Thesedata havebeenobtainedduringdifferent
seasons
andthusin variousmeanflow conditions:
January
28, 1989, at BIS; August7, 1985, at OHP; May 29, 1986,
and December 17, 1987, at both sites. From these four
examples,which illustratea recurrentpattern,a dominant
oscillationof severalkilometers'vertical wavelengthand
severalhours'periodcan be clearlyidentified.Excepton
August 7 (Figure 3c), this low-frequencymode seemsto
disorganize
andto disappear
in thelowermesosphere,
above
55 or 60 km altitude.The temperatureperturbationplot,
shownin Figure 3d, for the measurements
of December17,
averageof the varianceover the threeneighboringspectral
intervals(partiallyoverlapping)will be presented,
takinginto
1987, at BIS, exhibitsa somewhatmore complexpattern
which probablyresultsfrom the superposition
of several
largeamplitudeoscillatorymodes.
In a single case,on May 29, 1986, at Biscarosse,wind
account the fact that the vertical
measurements, obtained from tracked rocketsonde, were
DPS filter has a cutoff wave numberat 1/10 Km-1. An
scales of the waves could
5158
Wilsonet al.:Gravitywavesobserved
byRayleigh
Lidar,1
collectedon the site simultaneously
with the lidar
Thereforethe amplitudesof the temperatureand wind
measurements.
The time-heightplot of the temperature perturbationsare alsorelatedto to. The directionof the wave
fluctuations
as well as thehorizontal
windperturbation propagation
is estimated,
assuming
a quasi-monochromatic
profile (Figure 4) showsthe presenceof a dominant wave,by determining
analytically
themajoraxisorientation
oscillatory
modein thestratosphere
andlowermesosphere of the wind elliptical hodograph[Kundu, 1976; Cot and
between
35-and55-kinaltitude
with6-to12-kin
wavelength Barat, 1986]. The direction towards which the wave
androughly10hours'period.Thezonalandmeridional
wind
components
havea roughlyconstant
phaselagupto 55 km
altitude.Abovethisheightthepatternis somewhat
more
complicated,
resulting
probably
fromwavesuperposition.
In
thissingle
opportunity
it hasbeenpossible
toclearly
identify
a gravitywavefromthepolarization
relationships,
i.e.,the
relationships
between
thedifferentperturbed
variables.
From
theWKBsolutions
of thelinearized
perturbations
equations,
propagatesis found to be closeto-0.8 tad (anticlockwise
from east). The propagationdirectionof the wave with
respectto themeanflow (directedtowardswest-south
West)
makeanangleof roughly•/3 (anticlockwise).
Figure5 shows
thehorizontal
windfluctuations
projected
in the directionof the wave propagationand in the
perpendicular
directionjust as thenormalized
temperature
perturbations
(g/N)(T'/To)
between
33and
50-km
altitude.
the horizontal wind fluctuations in the direction toward
The
fluctuations
are
extracted
by
a
band
pass
filtering
(cutoff
whichthewaveis travelling,
U', andin theperpendicular
wavelengths
at
3
and
12
km)
of
the
vertical
profiles.
The
direction,
V', andthetemperature
relative
fluctuations,
T'/To,
phase
relationships
between
the
two
horizontal
wind
of a monochromatic
gravitywavearerelatedasfollows:
componentsare consistentwith the polarizationrelations
(equation(8a)). The temperature
fluctuationsare about180ø
V'=-i
fU'=(1-to2/N2.•1/2
(•)
outof phasewithV' between
33 and50 km (equation
(8b)).
Below and abovetheseheightsthe phaserelationship
(8a)
betweenthe temperatureandwind fluctuationsdo not seem
U'=-i(11to2]N2.,•l/2
f2-'•2'!
NK
(•----o) (8b)
consistent
withthepolarization
relationships
(equation
(8a),
(8b)), suggesting
a superposition
of otherwavemodes.An
inertia-gravity
waveis thusclearlyidentifiedoverabouttwo
where
i=x/-1,
theother
notations
being
defined
aspreviously.verticalwavelengths,eventhoughotherwave modesare
Theinertial
frequency,
f, isclose
to2n/17hours
'1 at44øN.
probably superimposed.The main characteristicsof this
inertio-gravity wave, as deduced from the wind and
temperaturemeasurements
over two verticalwavelengths,
The windperturbations
perpendicular
to thedirectionof the
wave propagation,V', is 180ø out of phasewith the
temperature
fluctuations
andin quadrature
withU' (equation
aresummarized
in Table1. Thehorizontal
windhodograph
(8a)),U'andT' being
ofcourse
inquadrature
(equation
(8b)).
(Figure 6) shows a clockwise rotation of the wind vector
Theratio,U7V',isproportional
totheintrinsic
frequency,
to.
RELATIVE
•O
-"':"'
'" •:
FLUCTUATIONS
":-".•
' '
"
•: '"
OF TEMPERATURE
'•-"•z--•.•] ..................
•
.....
6 0 '-•:
' •; -%> •. ' -
WIND
FLUCTUATIONS
....................................
•' •. •.• .......... (a)
•
' "--' ..•c:•
',:•-•_•--.:,
29MAY
1986•sz'""'•½/••-
20:29- 1:15 •5o•
with altitude(thusan upwardenergyflux). The ratio f/to,
,' [-•-•'
•
60
......
......
•
• 55
•-•'
%46
I
AaOVE
0.045
....**.••
0.035- 0.045 38*' •--
•-0.005-
0.005 32
• -0.015
--0.005
•,:• -0.035
-0.025
--o.o15
--0.025
::;':•-0.045
--0.035
•:.• BELOW -0.045
•';"
'
•t
'
....
••'....
I
4o
o k:' ,•,,-
2z '
23" '
Time(H)
0
'
Wind In•ensi[y
BIS22:55
(M/S)
ZonalWind
........
Meridional
Fig.4. Time-height
conto•s
of(a)•e tem•rature
relative
fluctuations
obse•ed
byRayleigh
lidar•d (b)
ß e hofizon•windfluctuations
ob•• from•ack• r•ketsonde
onMay29,1986,atBIS.
Wind
Wilson et al.: Gravity wavesobservedby RayleighLidar, 1
W
&
T
5159
closeto 0.75, indicatesan intrinsicperiodof theorderof 13
hours. The horizontal wavelength, deducedfrom the
FLUCTIIATIONS
5O
dispersion
relationship
(equation
(6)), shouldbe therefore
48
close to 1500-2000 km. The vertical phase velocity, as
observedfrom the temperature
fluctuations(Figure4), is
directed
downwards
(roughly
-0.2 to-0.4 m s'l), as
expected
fora wavewitha positive
intrinsic
horizontal
phase
velocity (equation(7)). Note also that the wave pattern
observed
the samedayat OHP (Figure3b), 550 km eastof
BIS, is quite similarabove40 km altitude: it showsa
verticalwavelength
of 6 to 8 km anda downwardphase
velocity
ofabout
-0.3ms'1.
Theverticalcomponent
of theperturbation
velocity,w', as
well asthemeridionalcomponent,
v', arerelatedto thezonal
windcomponent
[Gossard
andHooke,1975]by
36
W'=kh2 1 ')U'
- •- (1+ia.tg(x)' (9a)
34
v'=(1iatg(x)
) u'
+ ia.tg(x)
BIS
g9 MAY 86
THETA--0.8 rad
where x is the angle defining the directionof the wave
propagation(anticlockwise
from east),tg(x)=l/k,1 andk
beingthe meridionalandzonalwavenumberrespectively,
anda=f/t0.Followingthe methoddescribedby Eckermann
U
22:55
V
........
(9b)
T
Fig. 5. Verticalprofilesof thehorizontal
windfluctuations and Vincent [1989], the vertical fluxes of the zonal and
meridional horizontal momentum (<u'w'> and <v'w'>,
(projected
in thedirection
of thewavepropagation
andin the
perpendiculardirection) and normalized temperature respectively)are written
fluctuations
(g/N)(T'/To)obtained
witha bandpassfilter
kh
<U'W'> = - --
<u'v'>
msin•(1-a
2)
centered on 6 km between 33- and 50-km altitude.
Table1. TheInertia-Gravity
WaveCharacteristics
Observed
onMay 29, 1986,at BIS
From 33 to 39 km
Polarization f/co
Directionfrom east,rad
Verticalwavelength,
km
Horizontal
wavelength,
km
0.78
-0.82
6
1640
From 39 to 49 km
0.73
-0.75
10
2208
Momentum
flux,N m-2
po<U'W'>
po<V'W'>
2.47x10
-5
- 2.17x
10-5
0.67x10
-5
- 0.63x
10-5
Induced
acceleration,
m s-1d-1
-1/poA/Az(po<u'w'>)
-1/poA/Az
(po
<v'w'>
)
Vertical
group
velocity,
ms-1
4.lx10
-2
- 3.6x
10-2
0.49x10
-1
1.02x10
-1
Horizontal
group
velocity,
ms-1
Kinetic
energy
density,
j kg-1
Potential
energy
density,
j kg-1
Total
energy
density,
j kg-1
Vertical
energy
flux,
j m-2s-1
34.8
4.2
1.9
6.1
2.18x10
-3
50.0
12.1
5.6
17.7
4.26x10
-3
Horizontal
energy
flux,j m-2s-1
1.55
2.1
(10a)
5160
Wilsonet al.- Gravitywavesobserved
by RayleighLidar, 1
HODOGRAPE
<V'W'>
-- kh <U'V'>
mcosx(1-a
2)
--i
(10b)
the bracketsdenotingherea verticalaverage.On insertingin
theserelations the inferred values for f/co,kdm, <u'v'>, and
% and the atmosphericdensityPo(CIRA, 1986), the mean
values for po<U'W'> and po<V'W'> can be estimated
[Eckermann
acceleration
evaluated
•
and Vincent,
between
33-
1989].
The wave-induced
and 50-kin
from these momentum
altitude
have been
fluxes estimations.
These
valuesare reportedin Table 1. Even thoughvery small, the
orderof magnitudeof the inducedaccelerationwas foundto
be nearlythe samein the zonalandmeridionaldirection.The
wave-inducedaccelerationis in a directionoppositeto the
zonal componentof the mean flow (eastward)and in the
samedirectionthanthe (weak) meridionalcomponentof the
-3
mean flow (southward).
4.2. The VerticalAnisotropyof theWave Field
Zonal Wind (rn/s)
BIS
29 MAY
86
22:55
Fig. 6. Hodograph of the horizontal wind fluctuations
between33- (indicatedby x) and50-km altitude.The wind
vectoris turningclockwise.
BIS
28
JAN
30--45
PERIOD (H)
89
--2
--4-•
•
4
OHP
2
!a)
km
I
StEp/PrEp = 0.45
O.4
I
[•]
r--]
11.811.4-
3
PERIOD (H)
86
-,• •
30-45
km
10.4
10.1
9.7
B[LOW
,.•UpEp/DnEp
=2.33
(rn3s)
4 Z•
';:--• li.t- n.4
02
5
•
DEE]
0.1
8
10.7
10.4
9.7
•
07
AUG
30-45
85
StEp/PrEp = 0.26
UpEp/DnEp = 2.72
-z
-
km
'
PERIOD (H)
-•
• •' -
,
4
-r.!....
:-'
-
(b)
/A,:%:."':-":-4
Z
9.7- 10.3
10.0
• 1o.o9.4
•
--
0.7
0J.
-- 0.4
BIgLOW
0.1
•?:":::•i
.....
:!::!?!':ij?
!:.:i?,,/(;>.,::
.':.,,.... ,,, , ....•
8
"::::,. 'AM:'"':::.L',
FREQUENCY(i/H)
z
*--
(t-•
BIS 17 DEC 87
.,\•'/
•
-•
•
]
,-,';•m:.•
:--• .......
. ............................
PERIOD (H)
30-45 km
•le •
•
o3 •
•R•u•cY
•
•
•
0.:•
'• . 7•:::.;{,•
-•/:•x_•.,•-__--::
•:,_:..
,]
StEp/PrEp
= 0.44
0•
•8
.,•
0.3
O/n)
•• --.,:.,•._ ?- ::---:•
[--• _?•.:.:.- ....
0.4
lO.6- 10.G
•.o.s-
t•
'e •-0.4-o'.•
o.o
oa0.4o.•
•u•c•
.•
StEp/PrEp= 0.37
m
12.1
11.8
z
OHP
MAY
A•OVS
.•...• 1•.4-12.8
•
29
0.4
'"
UpEp/DnEp = 1.50
From most of the case studiesperformed,the observed
inertia-gravitywaves exhibit a downwardvertical phase
velocity of the order of a few tens of centimentersper
second.Four examplesof two-dimensional
spectra,function
of vertical wave number,m, and apparentfrequency,o,
resulting from density measurements in the upper
stratosphere(30-45 km), are shown in Figure 7. The
•
•-
.•
•0.?
- •.•
•.•
'-'• to.o-to.3
o.• o.• o.•
01•)
Fig. 7. Two dimensional
spectra,functionof the verticalwavenumberandapparentfrequency,
of the
densityfluctuations
in thestratosphere
(30-45km). Theratiosof upwardto downward
propagating
energy
(UpEp/DnEp)andof stationary
to propagating
energy(StEp/PrEp)
areindicated.
(a) BIS, January
28, 1989;
(b) OHP, May 29, 1986;(c) OHP, August7, 1985;(d) BIS, December17, 1987.
Wilsonet al.:Gravitywavesobserved
by RayleighLidar,1
dominant
low-frequency
andlow wavenumberwavemodes
are easilyidentifiedfrom eachof theseexamples.On
December17, a low-frequency
modewitha relativelysmall
verticalwavelength
(3 km)is observed
whereas
fortheother
casestheverticalwavelengthof the dominantmoderanges
between5 and 8 km. The frequencyterm of the phase
(expressed
asexp(mz+ (•t) with(•>0)of thisdominant
wave
modesis generallypositive.It impliesa downward
vertical
phasevelocity,indicativeof an upwardenergyflux. The
energyratioof thefluctuations
withupwards
anddownwards
verticalphasevelocityrespectively
isreported
in Figure(7).
This ratio rangesbetween1.5 and 2.7. The PSDsof the
fluctuationswith vertical phase velocities propagating
upwardanddownward
(resulting
froman integration
over
the positiveand negativefrequencies,
respectively)
are
shown
in Figure8. A verticalasymmetry
of thewavefieldis
observed,
65 to 80 % of theenergycontentof theprogressive
waves(i.e.,resolvedfrequencies)
havingdownward
vertical
phasevelocity.
5161
about0.3-0.4 in the stratosphere
to 0.1 in the mesosphere.
Thisfrequently
observed
featuresuggests
thatthewaves
should
haverelatively
higherfrequencies
in themesosphere
thanin the stratosphere.
It appears
frommostcasestudies
thattheverticalphase
velocityof thedominant
inertia-gravity
wavesobserved
in
the upper stratosphere
and lower mesosphere
is very
generally
directed
downward.
Above50- or 55-kmaltitude
the lower-frequencywaves frequentlyseempartially
dissipated.
A verticalasymmetry
of the wavefield (for
verticalscalesbetween1 and 15 km) is generallyobserved,
mostpartof the waveshavingdownward
verticalphase
velocityindicating
anupwards
energyflux.Sucha vertical
asymmetry
of thewavefieldhasbeenobserved
in thisheight
rangefrom rocket soundings[Hirota and Niki, 1985,
EckermannandVincent, 1989].Thesefindingsarebasically
in agreement
withtheresultsshownabove,suggesting
that
mostpartof thewavesaregenerated
below30-kmaltitude,
in thetroposphere
andlowerstratosphere.
The two-dimensionalspectraresultingfrom density
measurements
in themesosphere,
from 55 to 70 km (Figure
9), showthat the dominantwavemodeshave somewhat
4.3. The PowerSpectralDensity
higherfrequencies
thanat lowerlevel.Theenergyratioof
the stationarywaves (unresolvedfrequencies)to the
progressive
waves(resolved
frequencies)
decreases
from
Before discussingthe resultsconcerningthe density
fluctuationPSDs, we first briefly recall someof the main
resultson the gravity wave saturationtheory.Numerous
WAVELENGTH(KM)
WAVELENGTH(KM)
1,2
9 9
9•
f;
5
4
3
?
2
5
4
S
2
t
• toTM
ld ø
•
6
•o•
m
lo
c•
lo
-4•0
-$•
--S.0
•
10•
a•
10
a.
10
-4•
-3,4
•.6
Up Ep= 4.8 J/kg
Dn Ep= $.• J/kg
•w.•
.........
DOWNWARD
•ER•Y
Up Ep= 1.8 J/kg
Dn Ep= 0.8 J/kg
9 8
Y
8
5
4
-•4
.........
DOWNWARD
•N•RGY
WAVELENGTH (KM)
WAVELENGTH
•
-3.8
LOGWAVEN•MBER
hOG W•NLrM•E•
S
•
9 8
?
O
5
4
S
Fig.8. ThePSDversus
vertical
wave
number
ofthefluctuations
forwhich
thevertical
phase
velocity
propagates
downward
(i.e.,upward
energy
flux)andupward
(i.e.downward
energy
flux)obtained
by
integrating
the2Dspectra
overthepositive
andnegative
frequencies,
respectively.
(a)BIS,January
28,
1989;(b)OHP,May29,1986;(c)OHP,August
7, 1985;(d)BIS,December
17,1987.
5162
Wilsonet al.:Gravitywavesobserved
by Rayleigh
Lidar,1
BIS
28
JAN
PERIOD(H)
89
55-70
kin
ß -.
StEp/PrEp = 0.O7
UpEp/DnEp = 0.72
I
ABOVE
•
'•
{2_9
-4-6
6 4
....
• UpEp/DnEp
=1.02
13.2
Lz .6
{2.9
{2-3 {2.0
{2.6
{2.3
11.5
,
•
•
ii-8
• ii8-{2.0
11.2
-2
13.4
:.• 13.2
- 13.4
•
•
PERIOD(H)
-4•8
86 4
2
-'• ............
!............
,(a) SIS17DEC
87
...- • •..... %,• ,"i:ff•.•,•;.
,55-70km
' , ' •'•
StEp/PrEp
=0.13
-
it.5
iO 9 -
ii.2
BELOW
la6- ta.8
.'
• {2.•{2.,
10.9
-0.8•.6-0.4-0.a
0.0 0.a 0.4 0.6 0.8
vr•u•nc•
-0.8-0.6 •.4-0.a
(l/•)
0 0.g 0.4 0.• 0.8
v•u•ncv
(l/•)
Fig. 9. Two dimensional
spectraresultingfrom densitymeasurements
in the mesosphere
(55-70 km)
performedon (a) January28, 1989, and(b) December17, 1987, at BIS.
experimental
evidencesindicatea largesimilarityin both
shapeand amplitudeof the mesoscalefluctuationPSDs as a
functionof wave numberand frequencyin the middle
atmosphere
[Dewanet al., 1984;BalsleyandGarello, 1985;
Nastrom and Gage, 1985; Smith et al., 1987; Hass and
Meyer, 1987; Sidi et al., 1988; Fritts et al., 1988]. It hasbeen
WAVELENGTH
12
9
8
7
6
5
suggested
by Van Zandt [1982] that the observedmesoscale
fluctuationsspectraare due, as in the ocean [Garrett and
Munk, 1975], to a randomsuperposition
of gravitywaves,
theso-called"universal
spectrum."
The shapeandmagnitude
of the wind and temperature fluctuations PSDs versus
verticalwave numberare thoughtto be the consequence
of
(KM)
4
WA':ELET,IGTH
3
2
I
12
td ø
9
•-
10
•
10
m
" '-
-30,4-5km
•-
I
10
<
I
I
EP= 11
6
1
+ 0_8l/kg
-4.0
-3.8
-3.6
6
5
(KM)
4
I
3
I
{
tO8
•
lo ?
•
1o
El:'= 3.
+ 03
/kg
-3.8
-3.6
-3.4
(KM)
WAVELENGTH
2
12
9
8
7
6
5
4
I.KM)
3
I(d)II't,-..,
1..•1
II
""-
HP 07 UG 85
•
I
{
•
2
I
I I3o-.45km
.
"'-. -,1
-3.2
(CY, M)
3HP
-...
"-'
k to
•,HP
2.9
MAY86
30-.45
km
LOG WAVENUMBER
-"'•-
I
{
-......
•.•II1
•
-4.0
3
2
I
I
"['-
(CY
4
5
I
"'""'
-3.4
WAVELENGTH
6
1o
I
LOG WAVENUMBER
7
(b) i
'"' •111B
i
8
II I 1.[ I I I
-
1otø
(a) !
I
9
17
DEC 87
i_o
--.•
1o8-
•
•'--._
..
1o8
i2::
r....
cc
•
to
7
'
!
t
'
c.
I
i
½
..t_
1o
..
.o •o
I ' •I
-4
0
-3 8
-3 (
LGG WAvENUMEElq
-3 4
(CY :M)
-3 2
IIII
I
o
i
.
,
.
/
10
-4 0
--3.8
-3.6
LOG WAVENuMBER
-3 4
-3 2
ICY: M)
Fig.10.PSD
versus
vertical
wave
number
ofthenormalized
density
perturbations
g/N2(p'/po)
intheupper
stratosphere
(30-45km). Thespectra
areplottedupto thelargerwavenumbers
for whichthesignal-to-noise
ratiois largerthanunity.(a) BIS, January
28, 1989;(b) OHP,May 29, 1986;(c) OHP, August7, 1985;(d)
OHP, December 17, 1987.
Wilsonet al.'Gravitywavesobserved
byRayleigh
Lidar,1
the saturationprocesses[Dewan and Good, 1986]: the
convectiveand/ordynamicalinstabilitiesinducedby large
amplitudewaves acts to limit this amplitudeto an upper
value
related
to the vertical
wave
number.
With
these
assumptions, the relative density (or temperature)
fluctuations PSD versus vertical wave number of a saturated
5163
the upper stratosphere
are 1 or 2 ordersof magnitudeless
than
the1/2m
3limitforvertical
wavelengths
between
1and
15 km (1/200<•x<1/20, i.e., much below the expectedvalue
for saturated waves). The standard deviation of the
temperaturefluctuationsappearsto be too small (between
1.5 and 2.5 øK) to produceconvectiveinstabilitiesin the
upperstratosphere.
Excepton January28, 1989, the power
wave field is scaled as
spectra
arenotscaled
asm'3in thelowwavenumber
partof
N4
FST'/To,p,/po(m)•
c• 2 3
(11)
gm
where• is a proportionalityfactorrangingbetween1/2 and
1/20, dependingupon the spectralwidth of the saturated
the spectrum(i.e., for wavelengthlargerthan2-3 km). The
spectralindex(thepowerdependency
of the spectrumversus
vertical wave number) ranges between -2 and +1. The
potential energy density per unit mass, estimated by
integratingthe PSD over wave numbers,varies within a
wave field [Dewan and Good, 1986; Smith et al., 1987, Sidi
factor
of 6,ranging
between
2 and12Jkg'1.It appears
from
et al. 1988]. This estimateof the powerspectraof saturated
densityfluctuationsdoes not dependupon the frequency
distribution of the waves owing to the assumptionof
convectiveinstabilitiesastheonly saturation
mechanism.
thesefew examplesthatthepotentialenergydensityis larger
duringwinter than duringother seasons.
This resultwill be
confumedfrom the statisticalstudy.
The PSD of the normalizeddensity fluctuationsin the
lower mesosphere
(45 to 60 km) increaseswith altitudefor
all wave numbers(Figure 1la). The power spectraof the
densityfluctuationsappearto be compatiblewith the spectral
limit (11) for vertical wavelengthssmallerthan 3 or 4 km
(1/10<a< 1/4). For largerverticalscalefluctuationsthe PSDs
PSDs versus vertical
wave number of the normalized
densityfluctuationsin the upperstratosphere
(30-45 km) are
shown in Figure 10. With the chosennormalization,the
straight
line1/2m
3 indicates
thespectral
limit(11)with
a=l/2. An importantvariabilityin the spectralshapeand
magnitude is observed from a day to another in the
accessible
spectraldomain.The densityfluctuations
PSDsin
WAVELENGTH
12
9
B
7
6
5
are much smaller than the saturationlimit (11) (a< 1/20), the
spectral index being also larger than -3. The standard
WAVELENGTH
(I_-M)
4
3
l()tø I I I i I I I I {I I
12
I
I
I
I
O
I'I
I
8
7
I
6
I
.5
I
4
I
(I-IM)
3
2
I
I
I
.
101ø
45- 60
km
""'
3HP 07 AUG8.5
E
--....
4.5--60km
_L_
[-
-'
lo
•_ii13
•
10
rf.,
2;
•
108--
.
• 10
8
ffi
a• 107
EP= 19 I
ñ 2,8 J kg
-4 0
-3 8
-3 6
-:3 4
LOG WAVENUMBER
WAVELEIIGTH
12
9
8
7
6
5
4
EP= 9.,I :t: 1.7 J/kg
I0
-3 2
-4 0
iotO
•
II I I [ I I I I I
-3 6
WAVELENGTH
(KM }
:1
2
I
-
-3 8
•IS
-3.4
-3.2
LOG WAVENUMBER (CY/M)
(CY/M)
12
I
I
•
.-.
'
•
c•
•
7
I55-70
28
lat.[
I
9
I
'1,.
8
I
7
I
6
I
õ
I
4
I
(KM)
3
2
I
I
,
1olO
89
"
3HP
07
AUG
55-.70
•
lo
a.
1o
..
(b)
,
•
,•:
¸
7
EP= 8,3
ñ /2.2 J
¸
a•
-4
0
-3
,q
-3
6
LOG WAVEHUMBER
-3
4
(CY
-3
6
_
EP= 60. • + 12.5 J/'kg
1o
2
l•l)
-4
0
-3 8
-3 6
LOG WAVEHUMBER
-3
4
-3.2
(CY/M)
Fig. 11. PSD versusverticalwavenumberof thenormalized
densityperturbation
(a) in thelower
mesosphere
(45-60kin) and(b) in themiddlemesosphere
(55-70km).
85
km
5164
Wilsonet al.: Gravitywavesobservedby RayleighLidar, 1
deviationof the temperature
fluctuations
is of theorderof 2ø
to 3 øK,thepotentialenergydensityperunitmassincreasing
by a factor 2 to 5 from the upperstratosphere
to the lower
mesosphere.
The fact that the densityPSDs are not scaledas
between 10 and 20 km. Using the WKB solutionsof the
linearized perturbationsequations,the potential energy
density,Ep(z), of a conservativemonochromaticgravity
wave should be scaled as
N4 fromthestratosphere
tothemesosphere
(asN4 decreases
by abouta factor of 2 to 3) seemsto indicatethat the wave
field is not convectively saturated(i.e., saturatedthrough
inducedconvectiveinstabilities)in the stratosphere.
Normalized density fluctuations PSDs in the middle
mesosphere
(55 to 70 km), on January28 andAugust7, are
shown in Figure 1lb. At these heights, waves amplitude
reachthe saturationlimit as definedin equation(11) (et• 1/2)
strongly suggestingthat the wave field is convectively
saturatedabove60 km altitudefor wavelengthsup to 6 km.
The spectralindex seemsto be consistent,within the error
bars,with a -3 slopefor wavelengths
between6 and2-3 km.
Furthermore, the standard deviation of the temperature
fluctuationsrangingbetween3ø and 5 øK, appearsto be
large enoughto induceconvectiveinstabilitiesin an height
rangewherethe temperature
gradientis negative.
The normalizeddensityfluctuationsPSDs, interpretedin
the frameworkof the gravity wave saturationtheory, thus
indicatethat the wave field is not convectivelysaturatedin
the upper stratosphere(30-45 km) within the accessible
spectraldomain(from 1- to 15-km verticalwavelength).As
the altitudeincreases,the power spectraldensityincreases
for all wave numbers,reachinga spectrallimit close to
N4/292m
3inthemiddle
mesosphere.
It strongly
suggests
that
gravity waves reach saturation inducing convective
instabilitiesfor vertical wavelengthsup to 6 km. A growth
from the stratosphere to the mesosphereof a mean
temperaturePSD has also beenobservedby Shibataet al.
[ 1988] from 13 Rayleighlidar profiles.Thesefindingsseem
to be in contradictionwith those of Smith et al. [1987] and
Tsuda et al. [1989], who have shown from in situ and MST
radar measurements, that the horizontal wind PSDs are
scaled
byN2in thesaturated
range
fromthestratosphere
to
the mesosphere. This may indicate that dynamical
instabilities (Kelvin-Helmoltz instabilities) are the main
saturationprocesses
in the stratosphere
wherelow frequency
waves appearsdominantand where the amplitudeof the
density fluctuations seemsto be small comparedto any
convective
saturation
limit.
Nevertheless,
no firm
conclusionscan be reachedfrom the densitymeasurements
alone.
4.4. The Potential Energy Density Versus Altitude
The available potential energy density per unit mass
(Figures 10 and 11) increasessignificantlyfrom the upper
stratosphere,
whereit is a few joulesper kilogramme,to the
N
Ep(z)
o•I
exp(z•p) (12)
wherez is theheight,Hp thedensityscaleheight,and•
the mean horizontalwind. Thereforea backgroundwind
shearcanbethecauseof largedeparture
from anexponential
growthparticularlyif the intrinsicphasespeedof the wave,
c-u--(•-, is small.
The potentialenergydensitiesversusaltitude,asobserved
on January28 andDecember17 at BIS andon May 29 and
August7 at OHP, deducedby a complexdemodulation
of
the signalaveragedover threewavelengthbands(centered
on 6, 5 and 4 km), are plottedin Figure 12. The energy
densityis plottedup to the altitudewherethe signal-to-noise
ratio is lessthan unity. The noisevariance(dottedline in
Figure 12), which is subtractedfrom the raw variance,
increaseswith altitude as the density scale height. An
increaseof energydensityis generallyobservedfrom the
stratosphere
to the mesosphere.However,therateof energy
growthvariesconsiderably
fromonecaseto theother.In two
cases(January28 and December17) the energydensityis
almostconstantin the stratosphere
whereasin anotherone
(August7) the growthis exponential
with a scaleheightof
the order of the density scaleheight. A minimum in the
variance is observedin some casesabove the stratopause
level (January28). This variancedecreasecouldbe due to
the dampingof the inertia-gravitywavesoften observedat
that level (Figure 3a). An energy growth is quasisystematically
observedin the mesosphere
at leastup to 65
km. In a few cases,ason August7, theenergyscaleheightis
close to the density scale height in the stratosphereand
mesosphere.
From these selected case studies, on cannot clearly
concludeaboutthe energydensityscaleheightas the energy
growthis highly variablewith altitude.This variabilitymay
be due to wind shears and/or dissipative processes.
Nevertheless,
the smallenergygrowthfrequentlyobserved
in
the stratosphereseems to indicate that the waves are
dissipatedat theseheights.BecausethePSDsof thedensity
fluctuationsdo not appearcharacteristicof a convectively
saturatedwave field (section4.3), this suggestthat the waves
are dissipatedthroughdynamical instabilitiesor any other
processes.
Conclusionson the verticalgrowthof the waves
will be furtherachievedfrom the statisticalstudyon a large
number of cases.
middlemesosphere,
whereit reaches
some10j kg'1. The
4.5. Convective instabilities
energydensityper unit massgrowsroughlyby a factor2 of
to 4 over about two atmosphericscale heights(the exact
valuesfor the casespresentedhere are given in Figure 10
and 11). The vertical scaleheightof the energydensityper
unit mass, Hi• (=Az/ln{Ep(z+Az)/Ep(z)}), thus ranges
The powerspectraof the normalizeddensityfluctuations
and the inferred standarddeviation of the temperature
fluctuationsboth suggestthat convectiveinstabilitiesmay
occur in the mesospheredue to large waves amplitudes.
Wilsonet al.: Gravitywavesobserved
by RayleighLidar, 1
•
7o
•
5165
()
70
•1•65
6o
"0 55
"13 55
•iil-ii õo
•
•
50
ß•
45
/?t/H
40
40
35
log Potential •nergy
log Potential Energy (J/Kg)
BIS
OHP
E8 JAN
89
17:52-18:13
70
•
,•
E9 MAY
86
g0: 51:31
(d)
7o
,
t
.
,'0 55
ß'•
-P-I
50
45
35
log Potential Energy (J/Kg)
log Potential Energy (J/Kg)
OHP
07 AUG 85
gg:343:89
BIS
17 DEC 87
;gO:45-;g;g:13
Fig. 12. Availablepotentialenergydensityper unit massversusaltitudededucedby a complex
demodulation
of thesignalaveraged
in threewavelength
bands(centered
on 6, 5, and4 km).Thenoise
level,deduced
by demodulation
on shortscalefluctuations
(smaller
than1 km)is plottedfor comparison
(dashed
line).Thevariance
of thefluctuations
is plottedup to thealtitudewherethesignal-to-noise
ratiois
largerthanunity.(a) BIS, January
28, 1989;(b) OHP,May 29, 1986;(c) OHP,August7, 1985;(d) BIS,
December 17, 1987.
Vertical temperatureprofiles deduced from 1 hour
integrationtime (in order to reducethe uncertainty),
observedduringthe night of January28, 1989; May 29,
1986; December 17, 1987, at BIS; and August7, 1985, at
gravity wave breaking [Hauchecorneet al., 1987;
Hauchecorneand Maillard, 1990].
5. Conclusions
OHP are plotted in Figure 13. Vertical gradientsof
temperature
closeto theadiabatic
lapserate(roughly
-10ø
km'l) areobserved
in themesosphere
independently
of the
seasons,during winter (January28, December17), in
summer(August7), and in equinoctialconditions(i.e., in
weak wind conditionsas on May 29). Theseconvectively
unstable
layersareobserved
eitherjustabovethestratopause
level(May29) or athigherlevels.Fromcomparing
Figures3
and13,theunstable
layersarefrequently
produced
abovethe
altitude where the dominant wave amplitude reachesa
maximum(asonJanuary28, May 29 andAugust7). Above
this level, the dominantmodeappearsto be disorganized.
Suchinstabilitiesseemthus clearly inducedby the large
ampiit,,de.
c}fthe waves. On January28, just abovethe
adiabaticlapserate, a mesospheric
temperature
inversion
occurs.Thoseare frequentlyobservedin the mesosphere
abovelargetemperature
gradientandarelikely inducedby
Densityand temperature
fluctuations,
observedin the
stratosphere
andmesosphere
by meansof Rayleighlidars
with highspatialandtemporalresolution
(15 min and300
m), havebeenanalyzedin someparticularcasesin various
seasonal conditions. These selected cases are characteristic
of tipicallyobservedwavepatterns.
1. The presence
of a dominantfluctuating
modeof large
period, typically 10 hours period, and large vertical
wavelength,
from5 to 10 km,is verygenerally
observed
in
the stratosphere
and lower mesosphere.
In a singlecase,
when simultaneouswind and temperature measurements
have beenperformed,an inertia-gravitywave has been
clearlyidentified.The frequencies
of thewavefield seemto
be relatively higher in the mesospherethan in the
stratosphere,
suggesting
a strongdampingof the lowerfrequencywavesin thelowermesosphere.
Wilsonetal.:Gravitywavesobserved
byRayleigh
Lidar,1
5166
8O
-• õo
()•IF
)
17 DD' 87 /
<•
/
II
,½ ,
Temperoture(K)
Fig. 13. Verticaltemperature
profiles,resultingfrom 1-hourintegrationtime of the signal,observedon
January
28, 1989;May 29, 1986;andDecember17, 1987at BIS andonAugust7, 1985,at OHP.
2. The verticalscaleheightof thepotentialenergyis most
often larger than the densityscaleheight, generallylarger
than 10 km. Nevertheless,in a few casesthe energy scale
heightis closeto thedensityscaleheight.The energydensity
growthis generally,but not systematically,
smallerin the
upperstratosphere
andlowermesosphere,
thanabove55 km.
Sucha changein the energyscaleheightaccordingto the
altitude domain could be due partially to the mean wind
reversalin the mesosphere.
3. If the gravity wave field doesnot appearto produce
convectiveinstabilitiesin the stratosphere,
the amplitudeof
the density(or temperature)fluctuationsis consistentwith
the convective saturation limit of the linear theory for
wavelengths
up to 3 or 4 km in the lowermesosphere,
andup
to 6 km above 60 km altitude. Furthermore, convective
unstablelayers,whichcouldpersistoverperiodslargerthan
one hour,are currenfiyobselwedin the mesosphere.
Becauseof the day-to-dayvariabilityof the wavesenergy
density(whichreaches
a factorof 5 in thestratosphere,
and1
orderof magnitudein the mesosphere)
a largedata setmust
be analyzedin order to give a generalview of the wave
activity in this height range. This is the subjectof the
companion
paperwhichwill describe
anddiscuss
thegravity
wave climatology from the large data set obtained by
Rayleighlidarsat the two sites.
References
Balsley,B. B. andR. Garello,The kineticenergydensityin
the troposphere, stratosphere and mesosphere:A
preliminary studyusing the Poker Flat MST radar in
Alaska, Radio. Sci., 20, 1355-1361, 1985.
Barat, J., Initial results from the use of ionic anemometers
under stratosphericballoons:Application to the highresolution analysis of stratosphericmotions, J. Appl.
M•teorol., 21, 1489-1496, 1982.
Chanin, M. L., and A. Hauchecorne, Lidar observation of
gravity and tidal waves in the stratosphere and
mesosphere,
J. Geophys.Res.,86, 9715-9721,1981.
Chanin, M. L., and A. Hauchecorne, Lidar studies of
temperature
and densityusingRayleighscattering,
MAP
Handbook, 13, edited by R. A. Vincent, pp. 87-99,
SCOSTEP, Urbana, Illinois, 1984.
Cot, C., and J. Barat, Wave turbulence interaction in the
stratosphere'A case study, J. Geophys. Res,.,9_!_1,
2749-2756, 1986.
Dewan, E. M., and R. E. Good, Saturation and the
"universal"spectrumfor vertical profile of horizontal
scalarwinds in the stratosphere,
J. Geophys.Res.,9j_l,
2742-2748, 1986.
Dewan,E. M., N. Grossbard,A. F. Quesada,andR. E. Good,
Spectral analysis of 10 m resolution scalar velocity
profilesin thestratosphere,
Geophys.
Res.Lea.,11, 80-83,
1984(Correction,Geophys.Res.Lett.,11, 624, 1984).
Eckermann, S. D., and R. A. Vincent, Falling sphere
observationsof anisotropicwave motions in the upper
stratosphere
overAustralia,PureandAppl. Geophys,.,
130,
509-532, 1989.
Fritts, D.C., Gravity wave saturation in the middle
atmosphere:A review of theory and observations,Rev.
Geophys.SpacePhys.,22, 275-308,1984.
Fritts,D.C., and P. K. Rastogi,Convectiveanddynamical
instabilities
dueto gravitywavemotionsin thelowerand
middleatmosphere:
Theoryandobservation,
RadioSci.,
20, 1247-1278, 1985.
Fritts, D.C., T. Tsuda, T. Sato, S. Fukao, S. Kato;
Observational evidence of a saturated gravity wave
spectrumin the troposphereand lower stratosphere,
J.
Atmos. Sci., 45, 1741-1759, 1988.
Garcia, R. R., and S. Solomon, The effects of breaking
gravitywavesonthedynamical
andchemical
composition
of the mesosphere
andlower thermosphere,
J. Geophys.
Res.,90, 3850-3858, 1985.
Gardner,C. S., M. S. Miller, and C. H. Liu, Rayleighlidar
observationsof gravity wave activity in the upper
stratosphereat Urbana, Illinois, J. Atmos. Sci., 46,
1838-1854, 1989.
Garrett, C. J. R., and W. H. Munk, Space-timescalesof
Wilsonet al.' Gravitywavesobserved
by RayleighLidar,1
5167
internal waves, A progressreport, J. Geophys.Res., 80,
circulationsin the stratosphereand mesosphere,
Q. J. R.
291-297, 1975.
Meteorol. Soc., 87, 125-135, 1961.
Gossard,
E. E., andW. H. Hooke,Wavesin theAtmosphere,
Elsevier, New York, 1975.
Hass,H., andW. Meyer, Gravitywavefield aboveAndOya,
J. Atmos.Terr. Phys.,49, 705-712, 1987.
Hauchecorne,
A., andA. Maillard,A 2D dynamicalmodelof
mesospheric
temperature
inversionsin winter,Geophys.
Res. Lett., 1_27,
2197-2200, 1990.
Hauchecorne,A., and M. L. Chanin, Density and
temperature
profilesobtainedby lidar between35 and70
km, Geophys.Res.Lett., 7, 565-568, 1980.
Nastrom, G. D., and K. S. Gage, A climatology of
atmospheric wave number spectra observed by
commercial aircraft, J. Atmos. Sci., 42, 950-960, 1985.
Palmer, T. N., G. J. Shutts, and R. Swinbanck, Alleviation of
a systematicwesterly bias in general circulation and
numerical weather prediction through an orographic
gravity wave drag parameterization,Q. J. R. Meteorol.
Soc., 112, 1001-1040, 1986.
Shibata,T., T. Fukuda,and M. Maeda,Densityfluctuations
in the middle atmosphereover Fukuokaobservedby an
XeF Rayleighlidar, Geophys.Res. Lett., 13, 1121-1124,
Hauchecorne,
A., M. L. Chanin,andR. Wilson,Mesospheric
1986.
temperature inversion and gravity wave breaking,
Geophys.Res.Lett.,14, 933-936, 1987.
Shibata,T., S. Ichimori,T. Narikiyo,andM. Maeda,Spectral
Hirota, I., Climatology of gravity waves in the middle
analysisof vertical temperatureprofiles observedby a
atmosphere,
J. Atmos.Terr. Phys.,46, 767-773, 1984.
lidar in the upperstratosphere
andthe lower mesosphere,
J. Meteorol.Soc.Jpn.,66, 1001-1005,1988.
Hirota,I., andT. Niki, A statistical
studyof inertiagravity
wavesin the middleatmosphere,
J. Meteorol.Soc.Jpn.,
Sidi, C., J. Lefr•re, F. Dalaudier,andJ. Barat,An improved
63, 1055-1066, 1985.
atmospheric
buoyancywave spectrummodel,J. Geophys.
Res., 93, 774-790, 1988.
Holton,J. R., The role of gravitywave-induced
drag and
Smith, S. A., D. C Fritts, and T. E. Van Zandt, Evidence for
diffusionin the momentumbudgetof the mesosphere,
J.
Atmos. Sci., 39, 791-799, 1982.
a saturatedspectrumof atmosphericgravity waves, J.
Atmos. Sci., 44, 1404-1010, 1987.
Kundu,P. K., An analysisof inertial oscillationsobserved
Tsuda, T., T. Inoue, D.C. Fritts, T. E. Van Zandt, S. Kato, T.
nearOregoncoast,J. Phys.Oceanogr.,6,879-893,1976.
Sato, and S. Fukao, MST radar observation of a saturated
Lindzen,R. S., Turbulenceand stressdue to gravitywave
and tidal breakdown,J.Geophys.Res.,8.•6,9707-9714,
gravity wave spectrum,J. Atmos. Sci., 46, 2440-2447,
1981.
Mathews, J. D., J. K. Berakall, and G. K. Karawas, The
discrete prolate spheroidal filter as a digital signal
processing tool, MAP Handbook,9, edited by S. A.
BowhillandB. Edwards,pp. 563-571,SCOSTEP,Urbana,
Illinois, 1983.
Meek, C. E., I. M. Reid, and A. H. Manson, Observation of
mesosphericwind velocities, Cross sectionsof power
spectraldensityfor 48-8 hours,8-1 hours,and 1 hourto 10
mn over 60-110 km for 1981, Radio Sci., 20, 1363-1382,
1985.
Miyahara, S., Y. Hayashi, and J. D. Mahlman, Interactions
between gravity waves and the planetary scale flow
simulated by the GFDL "SKYHI" general circulation
model, J. Atmos. Sci., 43, 1844-1861, 1986.
Murgatroyd,R. J., and F. Singleton,Possiblemeridional
1989.
VanZandt, T. E., A universalspectrumof buoyancywaves
in the atmosphere,
Geophys.Res.Lett., 9, 575-578, 1982.
Vincent, R. A., Gravity wave motionin the mesosphere,
J.
Atmos.Terr, Phys.,46, 119-128, 1984.
Vincent, R. A., and D.C. Fritts, A climatologyof gravity
wave motion in the mesopauseregion at Adelaide,
Australia, J. Atmos. Sci., 44, 748-760, 1987.
M. L. Chanin, A. Hauchecorne and R. Wilson, Service
d'A6ronomie du CNRS, 91373 Verri•res le Buisson Cedex,
France.
(ReceivedNovember2, 1989;
revisedAugust28, 1990;
accepted
October11, 1990.)