Observations of Convective and Dynamical Instabilities in Tropopause Folds and Their Contribution to Stratosphere-Troposphere Exchange

Home , Col (meteorology)

Observations of convective and dynamical instabilities in tropopause folds and their contribution to stratosphere-troposphere exchange

The MIT Faculty has made this article openly available. Please share how this access benefits you. Your story matters.

Citation Cho, John Y. N. et al. “Observations of Convective and Dynamical Instabilities in Tropopause Folds and Their Contribution to Stratosphere-Troposphere Exchange.” Journal of Geophysical Research: Atmospheres 104, D17 (September 1999): 21549–21568 © 1999 American Geophysical Union (AGU)

As Published http://dx.doi.org/10.1029/1999JD900430

Publisher American Geophysical Union (AGU)

Version Final published version

Citable link http://hdl.handle.net/1721.1/111136

Terms of Use Article is made available in accordance with the publisher's policy and may be subject to US copyright law. Please refer to the publisher's site for terms of use. JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 104, NO. D17, PAGES 21,549-21,568, SEPTEMBER 20, 1999

Observations of convective and dynamical instabilities in tropopause folds and their contribution to stratosphere-troposphere exchange JohnY. N. Cho,1 ReginaldE. Newell,1 T. Paul Bui,2 EdwardV. Browell,3 Marta A. Fenn,4 MichaelJ. Mahoney,s Gerald L. Gregory,3 Glen W. Sachse,3 StephanieA. Vay,3 Tom L. Kucsera,6 and Anne M. Thompson6

Abstract. With aircraft-mounted in situ and remote sensinginstruments for dynamical, thermal, and chemicalmeasurements we studiedtwo casesof tropopause folding. In both foldswe foundKelvin-Helmholtz billows with horizontalwavelength of -0900 m and thicknessof -•120 m. In one case the instability was effectively mixing the bottomsideof the fold, leadingto the transferof stratosphericair into the troposphere. Also, we discoveredin both casessmall-scale secondary ozone maxima shortly after the aircraft ascendedpast the topsideof the fold that correspondedto regionsof convectiveinstability. We interpreted this phenomenonas convectively breaking gravity waves. Therefore we posit that convectivelybreaking gravity wavesacting on tropopausefolds must be added to the list of important irreversible mixing mechanismsleading to stratosphere-troposphereexchange.

1. Introduction sonicsAssessment (SASS) Ozone and NitrogenExper- iment (SONEX) [Singhet al., 1999]. Although the At midlatitudes, tropopause folds are believed to be primary purpose of this mission was not the study of the major source of stratosphere-to-tropospheretrans- stratosphere-troposphereexchange, the extensive col- fer of air. These V-shaped layers with high ozone and lection of chemical, dynamical, and thermal measure- potential vorticity content angle down from the strato- ments in the upper troposphereenabled us to examine sphere for several kilometers and, at times, reach the examples of tracer transport around tropopause folds. planetary boundary layer. However, the fold itself is a reversible process, so for permanent stratosphere- troposphere exchangeto take place the air in the fold 2. Instrument Descriptions must mix irreversibly with the outside environment. This necessarymixing has been attributed to Kelvin- SONEX was conducted between October 7, 1997, and Helmholtz-inducedturbulence [Shapiro, 1980], bound- November 12, 1997, and was aimed at studying the ary layerturbulence [Johnson and Viezee,1981], strato- chemical effects of aircraft emissions in the North At- sphericstreamer fragmentation and roll up [Appenzeller lantic flight corridor. The measurementplatform was et al., 1996],and localizedwet convection[Langford and a DC-8 equipped with 16 scientific instruments. Below Reid, 1998]. However,few directobservations of the ex- we briefly describe the instruments relevant for this pa- change processexist, and it is far from clear which, if per: the meteorologicalmeasurement system (MMS), any, of these proposedmixing mechanismsis dominant. the differentialabsorption lidar (DIAL), the microwave In this paper we present observationsmade through temperatureprofiler (MTP), the in situ ozoneprobe, two tropopausefolds with in situ and remote sensing the differentialabsorption CO measurement(DACOM), instruments aboard a DC-8 during the NASA Sub- and the laser hygrometer. The MMS measured the air velocity with respect to the aircraft and the aircraft velocity with respect to the •Departmentof Earth, Atmospheric,and PlanetarySciences, Earth and combined the two vectors to get the zonal MassachusettsInstitute of Technology,Cambridge. u, meridional v, and vertical w wind components. The 2NASAAmes Research Center, Moffett Field, California. air motion system consistedof two 5-hole differential 3NASALangley Research Center, Hampton, Virginia. 4ScienceApplications International Corporation, Hampton, Virginia. pressureprobes (Rosemount 858Y), threetotal temper- 5NASAJet Propulsion Laboratory, Pasadena, California. ature probeswith different responsetimes, a pitot static 6NASAGoddard Space Flight Center, Greenbelt, Maryland. pressure probe, and a dedicated static pressure system. The aircraft motion sensingsystem consisted of a high-resolutioninertial navigationdevice (Litton LTN- Copyright 1999 by the American GeophysicalUnion. 72RH), an embeddedGPS ring laserinertial navigation Paper number 1999JD900430. instrument(Litton LN-100G), and a multiple-antenna 0148-0227/99/1999JD900430509.00 GPS attitude referencesystem (Trimble TANS Vector). 21,549 21,550 CHO ET AL.: TROPOPAUSE FOLD INSTABILITIES

The data stream was sampled at 65 Hz, then subse- usinga differentialabsorption technique [Sachse et al., quently reducedto 5 and 1 Hz for postmissionanalysis. 1987;Collins et al., 1996]. Again, the responsetime was Pressure p and temperature T were also obtained from estimated to be ~1 s basedon the sampleexchange rate. these probes, and then potential temperature 0 was cal- There were quasiperiodicgaps in the data of order 30 s culated from thosevalues. The turbulent energydissi- every ~10 min becauseof the real-time calibration pro- pation rate e was estimated from a frequency band of cedure, which was necessaryto account for slow drifts 0.8 to 1.5 Hz. For data accuracy, resolution, and the in instrument sensitivity. methodfor estimatinge, see (fhap.,et al. [1998]. The laser hygrometer consistedof a compact laser The DIAL system was used to remotely measure transceiver mounted to an aircraft window plus a ozone and aerosols in the zenith and nadir directions. sheet of retroreflecting"road sign" material affixedto Two frequency-doubled Nd:YAG lasers were used to an outboard engineenclosure to completethe optical pump two frequency-doubled,tunable dye lasers. One loop. Usingdifferential absorption techniques, H20 was of the dye lasers was operated at 292 nm for the DIAL sensedalong this external path to an estimatedpreci- on-line wavelengthof ozone, and the other one was op- sionof 2% in mixingratio with a responsetime of 50 ms. erated at 300 nm for the off-line wavelength.The out- See S. Vay et al. (Troposphericwater vapor measure- put beams were transmitted out of the aircraft coax- ments over the North Atlantic during SONEX, submit- ially with the receiver telescopes. The backscattered ted to the Journalof GeophysicalResearch, 1999) for laser energy was collected by two back-to-back 36-cm- further details. diameter telescopes. The analog signalsfrom the de- tectors were digitized at 5 MHz to a 12-bit accuracy, 3. Observations and the averageddigitized signalswere stored every 2 s. The data sampling resolutions used in our study are After two test flights out of its home baseat Moffett 90 m in altitude and 60 s in time. An earlier version Field, California, the DC-8 flew cross-countryto Ban- of this instrument made the first remote measurements gor, Maine, on October 13, 1997. A prominent strato- of ozoneand aerosolsacross a tropopausefold [Brow- spheric intrusion was present over the northcentral ell et al., 1987]. More details on the systemare avail- plains,and the DC-8 traversedits southend (Plate1). able elsewhere[Browell et al., 1998; W. B. Grant et The sectionof the tropopausefold penetratedby the al., Observationsof tropical boundary layer air masses aircraft was downstreamof a trough and on the west in the northern midlatitude upper troposphereduring sideof a jet at the aircraftlevel. Satelliteimages showed SONEX, submitted to the Journal of GeophysicalRe- a clear columnof air along the easternflank of the in- search,1999]. trusion bordered farther east by a line of storm clouds The MTP, a passivemicrowave radiometer, measured associatedwith an accompanyinglower-level cold front. the natural thermal emissionfrom oxygenmolecules at Details will be shown and discussed in section 3.1. three frequencies(55.51, 56.66, and 58.79 GHz). It is During the next flight on October 15, 1997, the DC- viewed at 10 elevation angles between -800 and 800 8 encounteredanother stratosphericintrusion over the using a scanning mirror, yielding a vertical tempera- mid-Atlantic as it flew from Bangorto Shannon,Ire- ture profile from near the surface to 24 km for each land (Plate 2). This time the fold existedon both the 16-s cycle. The altitude resolution was thus variable, upstreamand downstreamsides of the trough. The air- beingbest (-,150 m) near the aircraft altitude, and de- craft, flying at --•3 km, passedthrough the lowertip of grading with distance from the aircraft. The postflight the westernbranch of the fold, then againtraversed the results were calibrated against 42 rawinsondeslaunched samefold as it ascendedsteeply to over ? km. It then near the DC-8 flight track during the two transit flights entered the main body of the intrusion and ascendedin between California and Maine. These provided an ab- stepsas it crossedthrough to the easternside. The ver- solute accuracy of 0.5 K rms near the aircraft altitude; tical cross-sectional view will be shown and discussed this degradeswith distance, becoming<<1.5 K rms 3 km in section 3.2. away. SeeDenning et al. [1989]for further information. Someof the subsequentflights during SONEX also Ozone was measured in situ with the NO + 03 chemi- sampled stratosphericintrusions, but we will focusour luminescenceprinciple [Clough and Thrush,1967]. On analysis on these two events becauseonly in these the basisof the sampleexchange rate at sea level the re- casesdid the aircraft ascend(or descend)through a sponsetime of the probe was estimated to be ,-•1 s with tropopausefold giving us vertical profilesof data nec- faster responseat higher altitudes. Sampleswere taken essaryfor estimatinghigh-resolution gradient and flux at 6 Hz then averaged to I Hz. The data were further quantities. corrected for water vapor quenching effects using the 3.1. Case 1: October 13, 1997 hygrometer measurements. The accuracy is estimated to be 5%, or 2 ppbv, and the precisionis estimatedto Plate 3 is a "curtain plot': of ozone concentration be 2%, or 0.8 ppbv. measured by the DIAL as the DC-8 flew under then The DACOM measured CO, N20, and CH4 in situ up through the tropopausefold. The DC-8 was fly- with a folded path, tunable diode laser spectrometer ing at a fairly constantair speedof •220 m s-1 (the CHO ET AL.: TROPOPAUSE FOLD INSTABILITIES 21,551

EPV (K m^2/kgs)

1.0e-05 8.5e-06 7.0e-06 5.5e-06 4.0e-06 2.5e-06 1.0e-06 -5.0e-07 -2.0e-06

Plate 1. Map of geopotentialheight in decameters(red contour lines), Ertel's potential vorticity (colorscale), and horizontal winds (yellow vectors) on the 330-Ksurface for 1200UT, October 13,1997. The scalingon the windvectors is standard:one flag is 50 m s-•, onebar is 10rn s -•, and one half bar is 5 m s-• The plotted field valueswere output from a model run by the DataAssimilation Office of t•e NASAGoddard Space Flight Center (A.M. Thompson et al., The Goddard/Amesmeteorological support and modelingsystem during SONEX: Evaluationof analysesand trajectory-basedmodel products, submitted to the Journalof GeophysicalResearch, 1999). The longitudinaland latitudinal grid spacing used by the modelwas 2.5 ø by 2o . The white line showsthe flight path of the DC-8, whichdeparted Moffett Field, California,at 1519UT and landed in Bangor, Maine, at 2241 UT. groundspeed was slightly higher), so roughly 1 hour (Wheneverthe time is givenin decimalhour notation correspondsto •800 km. Becauseof the incomplete we will also give it in standard notation in parenthe- overlap of the transmitted beam and the receiverfield ses. All figures and plates with time axes are labeled of view at closerange, there were no data from the DIAL in decimalhours UT.) We deducethis from the obser- in the height rangedelineated by the dashedlines. In vation that ozone remained steady as the aircraft was this regionthe data werelinearly interpolated by also at level flight in the first half, while it increasedas the includingthe in situ measurementof ozoneat the air- aircraft ascended,and also from the positive correlation craft level, hence the vertical blurring of the image in between small-scale fluctuations of ozone and potential this zone. The apparentwidening of the fold wherethe temperaturein Figure 2. Just after 18.06(1803:36) UT aircraft cut acrossit was probably not real. when the aircraft began to ascend, there was a rapid In situ measurementsof trace gasesreveal the fine oscillation in the trace constituent concentrations. We structurepresent within the fold (Figure 1). The de- will argue in section3.1.1 that this was a manifestation tailed anticorrelation of Oa versusCO, N20, and H20 of a Kelvin-Helmholtz instability. shows the intermingling of stratosphericand tropo- As the aircraft continued to climb, there was a sec- sphericair. (The CH4 trace,not shownhere, was virtu- ondarymaximum in the ozonemeasurement (and min- ally identicalto that of CO.) Within what appearedto ima in the other shownconstituents) just after 18.11 be the main body of the fold from ~18.04 (1802:24)to (1806:36)UT. A possiblethird localmaximum in ozone ~18.09 (1805:24)UT the gradientin the ozoneconcen- began •18.14 (1808:24) UT. The sharp edge on the tration was mainly in the vertical (upward)direction. topside of the main fold and this secondary maxima 21,552 CHO ET AL.: TROPOPAUSE FOLD INSTABILITIES

EPV (K m^2/kgs) 1.0e-05 8.5e-06 7.0e-06 5.5e-06 4.0e-06 2.5e-06 1.0e-06 -5.0e-07 -2.0e-06

Plate 2. Sameas Plate 1, except,for October15, 1997. The DC-8 departedBangor at 1211UT and landed in Shannon, Ireland, at 1821 UT.

are reminiscentof past in situ observations[Danielsen observedin both chemical and meteorologicalquanti- et al., 1970; Shapiro, 1974; Danielsen and Mohnen, ties shortly after 18.06 (1803:36) UT. Figure 2 shows 1977; Johnsonand Viezee, 1981]. Unfortunately,the the periodic undulations in tracer and dynamical mea- resolution and the near-field blind zone of the DIAL did surementssuggestive of Kelvin-Helmholtz(KH) billows. not allow us to observe the two-dimensional character The perturbations in v were much smaller than in u, of these substructures. which is consistentwith the plane of the vertical shear The dynamicalmeasurements are displayedin Fig- (estimatedfrom the data to be 59ø , measuredclockwise ure 3. The increasein horizontalwind speed,especially from the north) being in closeralignment in the zonal in the meridional,as the aircraftascended through then direction than in the meridional direction. away from the fold is consistentwith the classicpicture Assuming that the oscillations were, indeed, KH bil- of the jet stationedabove the topsideof the tropopause lows, we can estimate their horizontal wavelengthand fold, inducingthe secondarycirculation necessary for vertical thickness. Along the aircraft headingthe pe- folding. Clearly, the backgroundwind shearwas quite riod was •5.5 s, which translatesto a lengthof 1.23 km strong in this regionbeneath the jet. The rapid oscil- at the aircraft speed of 224 m s-1. Since the air- lation seenjust after 18.06 (1803:36)UT in the tracers craft heading was 102ø, this yields a true horizontal are also visiblehere, especiallyin 0 and u. Turbulence, wavelengthof hh = 1230cos(102 ø--59 ø) = 900 m. as indicatedby loge, peakedat the edgeof the main The vertical thicknessis then given by the relation- fold'stopside (18.09 (1805:24)UT), aroundthe •e•o•d ship Az = Aa/7.5 = 120 m [Milesand Howard,1964]. ozonemaximum (•18.12 (1807:12)UT), and at the be- These numbersare quite similar to thoseof rapid os- ginningof the third ozonemaximum (just before18.14 cillations observednear the tropopauseby a uniquely (1808:24)UT). There appearsto be somecorrelation high-resolutionradar experiment[Cho et al., 1996]. between •, relative upward motion, and a decreasein However,the calculatedgradient Richardsonnumber dO/dz,which is suggestiveof convectiveinstability. for the height interval acrossthe presumedKH layer is 3.1.1. Dynamical (Kelvin-Helmholtz) instabil- Ri = 0.52, which is larger than the classicinstability ity. Let us now look closerat the rapid oscillations criterion of 0.25. It could be that the aircraft was al- CHO ET AL' TROPOPAUSE FOLD INSTABILITIES 21,553

SONEX DC-8 1997/10/13

200 i i i ! i i

150 . .

•100

'•' 80

0 60

314 ,-, 312 .•.• 310 ø•308 z 306

120 100 80 60

E •-7 N•

ß , 6 i , i i , , 18.02 18.04 18.06 18.08 18.1 18.12 18.14 18.16 UT (h)

Figure 1. Plots,from top to bottom,of O3, CO, N20, andH20 mixingratios through the tropopausefold regionon October 13, 1997, and the pressurealtitude of the aircraft.

readywithin the instabilitylayer as it beganits ascent, 3.1.2. Convective (Rayleigh-Taylor) instabil- sincethe oscillationin the data beganjust as the climb ity. We now turn our attention to the topsideof the began. If this were true, then the extra wind shearnec- tropopausefold and to the secondarymaxima in ozone essaryto bring Ri down to 0.25 may have existedjust that followed shortly. The bottomside of the fold was belowthe aircraft heightat 18.06(1803:36) UT. Also, not associatedwith any appreciableturbulence, so we the instability may have been in a decay phasewhere will not discuss it here. Ri could be larger than 0.25. Finally, becausethe air- Figure 4 shows the changesin the squares of the craft flew more horizontallythan vertically, horizontal Brunt-V&tishl//frequency N and vertical shear in hor- gradientscould have mapped into the estimatedvertical izontal winds dU/dz as well as the resultingRi. For gradients and causedthem to be underestimated. easeof comparison,we displaylog e, w, and ozonecon- Becausethe apparent KH instability occurredin the centrationbelow them. To calculatethe vertical gra- middle of the tropopausefold, it would not havebeen ef- dients, we fit a straight line in a least squaressense fectivein mixing the air insidethe fold with the air out- to the quantity versus altitude for an interval 4-60 rn side. Mainly, it wouldhave smoothed the verticalgradi- from the data point. This 120 m representsthe outer ents of quantities within the instability layer. In terms scaleof turbulenceLo, whichvaries according to L0 - of stratosphere-troposphereexchange we are moreinter- (2rr/O.62)eZ/2-3/2[Weinstock, 1978]. We get Lo - estedin turbulenceoccurring at the edgeof the fold and 120 rn if we take e - 4 x 10-4 W kg-z (the high also around the secondary maxima observed in ozone valuemeasured by the MMS in thisvicinity) and N 2: just outsidethe main fold region. 2 x 10-4 rad2 s-2 (a typicalvalue calculated from MTP 21,554 CHO ET AL.- TROPOPAUSEFOLD INSTABILITIES

SONEX DC-8 1997/10/13 •' 170 •- 160 • 150 0•140 •, 65 •. 60 0 55 o 50

,-, 311 •310 0•309 ...... z

> , E 70 0 60

•' 29 ,•E28

•, 20 E,•18 • 16

,_, 0.5

• -0.5

317 ...... 315

313

•-7 0)-8 o 18.059 18.06 18.061 18.062 18.063 18.064 18.065 18.066 18.067 18.068 18.069 UT (h)

Figure 2. Expandedtime plots, from top to bottom, of O3, CO, N20, H20, the zonal wind componentu, meridional wind componentv, vertical wind componentw, potential temperature •, and turbulence log e from October 13, 1997.

profiles).Also, the observationof a nearbydynamically (1806:58)and -18.124 (1807:26)UT whenN 2 and Ri unstable layer of the same thicknessindicates that this became negative. Note that the vertical shear was is a reasonable value. very low during this time, which is consistentwith the For regions with Ri < 0.25, there was a clear corre- convectiveinstability of high-frequencygravity waves spondencebetween high thermalstability N 2 and high in both theory and laboratory observationsJOvianski, vertical shear. This preference for dynamic instability 1972]. The same regionexperienced strong turbulence to occur acrosslayers of enhancedthermal stability is in and large perturbations in upward velocity, which also keepingwith both theory [Phillips,1966] and past ob- points to convective instability. Furthermore, the grav- servations[e.g., Fritts and Rastogi,1985]. There was ity wave breakdown interpretation is supported by a also a convectively unstable period between •18.116 wavelike perturbation in u, v, and 0 with maxima in CHO ET AL' TROPOPAUSE FOLD INSTABILITIES 21,555

SONEX DC-8 1997/10/13 4O

o• 35

• 30

m 40 E v • 20

-1

-3

33O 325 320 315

18.02 18.04 18.06 18.08 18.1 18.12 18.14 18.16 UT (h)

Figure 3. Plots, from top to bottom, of u, v, w, log e, and 0 through the tropopausefold region on October 13, 1997.

the wind componentsand minima in dO/dz at •018.09 dimensions, we can draw upon past numerical mod- (1805:24)and ,-•18.12(1807:12) UT. Eventhe enhance- els of convectively breaking gravity waves. Figure 5 ments in fine-scaleu and v perturbations at the "crests" is a sequential schematic of three potential tempera- are suggestiveof wavebreakdown. Although N 2 did ture isopleths perturbed by a gravity wave until con- not becomenegative •18.09 (1805:24)UT, it did reach vective overturning occurs. The heavy lines indicate a local minimum. If the gravity wave breakdownexpla- the hypothetical paths of an ascendingaircraft and po- nation is correct, then the secondary ozone maximum tential temperature increaseswith height as indicated. ,-•18.12(1807:12) UT wasprobably the resultof either Note that dO/dl, the gradient along the aircraft path, vertical wave advection of the tropopause fold or up- only becomesnegative as the isoplethstilt more steeply ward, irreversibleturbulent mixing of ozoneor both. than the aircraft ascent angle. Although only a quali- Of course,vertical gradientquantities calculated from tative sketch, an important feature illustrated here that a slantwiseflight are only approximations. Remotely is seen in models is the stretching of the isentrope senseddata from the MTP did not indicate a supera- spacing in the horizontal direction as the wave grows diabatic lapse rate--•18.12 (1807:12) UT, but the in- [e.g., Walterscheidand Schubert,1990]. Typically,as strument only gave a reading every 16 s with a vertical the overturning begins, the potential temperature hor- resolution of about •150 m, so it could have missed a izontal gradient is an order of magnitude lessthan the localized instability. To examine the situation in two background vertical gradient. If the aircraft could as- 21,556 CHO ET AL.' TROPOPAUSEFOLD INSTABILITIES

x 10-4 SONEX DC-8 1997/10/13

O.75 0.5 0.25 0 -0.25 •-3 •-5 •-7 o_ 9

.• 2oo > '" 150 "'•100 O 50 18.07 18.08 18.09 18.1 18.11 18.12 18.13 18.14 18.15 18.16 UT (h)

Figure 4. Plots,from top to bottom,of the Brunt-V•iis•il•ifrequency squared N 2, verticalshear squared(dU/dz) 2, Richardsonnumber Ri, loge, w, andOa mixingratio from October 13, 1997.

cend purely vertically, then convectiveinstability would be detected as dO/dl = dO/dz < 0; however,since it Kz- O:N---• , (1) wasmoving mostly horizontally in this case(at a slope where c• is a proportionality factor that varies with the of-•1:50), if the magnitudeof dO/dl was more than, flow conditions[e.g., Itsweire et al., 1986]. We invoke say,one tenth of the backgroundvertical gradient, then the expressionderived by We•nstock[1992] there would be strong evidencefor convectiveinstability. Around 18.12 (1807:12)UT we estimatethe background potential temperature vertical gradient to have been -•0.007 K m -• from the MTP, while MMS data showedthat IdO/dll (negativegradient) reached as high as 0.003 K m-•, thus supportingthe conclusionthat whereL• = -2(2)X/2p'(Op/Oz)-•is the vertical energy- there was an area of convective instability there. containing scale of turbulence with p• the rms air 3.1.3. Tropopause fold •urbulen• ozone flux. density fluctuation and p the backgroundair density, We can now estimate the vertical tracer flux from the L•, - 2•reX/2N-a/2 is the buoyancyscale, and Lv = tropopausefold to the ambient troposphereinduced by 31•,a/4e-•/4is the viscousdissipation scale with kine- turbulence generated by convectiveand dynamical in- maticviscosity •, = 1.458x 10-6Ta/2/[(T+ 110.4)p] stabilities. We will use ozone as the example tracer. (Sutherland 's formula). Here p can be calculatedfrom For a stablystratified (N 2 > 0) environmentwe ap- p and T through the ideal gas law p = pRT, where R ply the vertical eddy diffusivity is the gas constantfor air. CHO ET AL.- TROPOPAUSE FOLD INSTABILITIES 21,557

(^qdd)o!l•l:J õu!x!lAI ouozo o o ßq' Od 0 0 0 0 0

(W>l)epnl!llV 21,558 CHO ET AL.: TROPOPAUSEFOLD INSTABILITIES

I --•18.115(1806:54)and 0-•18.125 (1807:30) UT whenN 2

dO/dl > 0 _<0,we took IzXOl- 1.5 K ZXz- 225m for (4). dO/dz>0• Kz is relatively constantin this regionbecause the T and p fluctuationsin (4) were muchsmaller than the e fluctuationsin (1). Althoughit may be counterintu- itive that convectiveinstability would not yield a much dO/dl = 0 largerKz, the reductionof the local verticalgradient by the breaking wave limits the turbulent vertical con- stituentflux [Frittsand Dunkerton, 1985]. The largest ozone flux is upward from the topside of the "main" body of the tropopausefold, whichis interestingsince d0/dl < 0 z one usuallythinks of stratosphere-to-tropospheretrans- fer in terms of downward flux. The values are in excel- lent agreementwith thoseestimated by Shapiro[1980] (q)o•= 1.3-1.4m s-• ppbvon the topsideof folds), which is remarkableconsidering that we useda com- Figure 5. Sequentialschematic of threepotential tem- pletelydifferent estimation method. The averagevalue perature isoplethsperturbed by a gravity wave until ofK• = 4.4m 2 s -• calculatedby Danielsen et al.[1987] convectiveoverturning occurs. The heavylines repre- sent the hypotheticalpaths of an ascendingaircraft. from a simplethree-dimensional model of a tropopause The lengthvariable along these paths is l, andz is up- fold is somewhatlarger than our valuesbut not very ward and x is horizontal.The dO/dlvalues are for the different consideringthe variability of the fold struc- center dot area. ture. We did not calculateq)o• for the bottomsideof the fold becausethe aircraftwas flying levelthere and we couldnot estimatethe verticalgradient quantities. However,as seenin Figures I and 3, the turbulencewas Forunstable and neutral conditions (N 2 _<0), weuse very much weakeron the bottomside,so we believethe verticalmixing there wasnegligible. We must,however, insert a cautionarynote regard- ing our estimatesof Kz and q)o•. First, as mentioned 0•, az, (3) before,horizontal gradients may be mappedinto verti- calgradients estimated from slant path data. Although Kz is dominatedby e variations,q)o• is linearlypro- where/7is an empiricalconstant, g is the gravitational portionalto the verticalozone gradient. Thus the ver- acceleration,A0 and Az are the verticaltemperature tical ozonegradient is a crucialquantity, especially its differenceand thickness across the unstableregion, and sign. The "vertical"structure of the tropopausefold K,• is the moleculardiffusivity [Kraichnan, 1962; Or- up to 18.1 (1806:00)UT matchesnicely with the DIAL lanskiand Ross, 1973]. We choose/•= 0.08,following the laboratoryresults of Ingersoll[1966]. Usingthe ozonedata aboveand below.In fact, a comparisonwith relationK,•/u = 1.34[Chapman and Cowling, 1970], q)o• calculatedwith the interpolatedDIAL ozonepro- Sutherland'sformula, and the idealgas law, (3) be- filessupports the negativeozone gradient (and upward comes flux) just after 18.09(1805:24) UT. However,the up- wardgradients (and downward fluxes) •018.11 (1806:36) and -,-18.14(1808:24) UT are not observedin the DIAL- derived results. The associated ozone features were too smallscale and localto showup in the DIAL data, and Kz_1.57 x10_3 [O(T+ g[AOI110.4) p T•Az, (4) thereforewe cannotconfirm that the verticalgradient featuresthere were not mainlyproduced by horizontal structures. where p is in hPa. The vertical flux of trace constituent Second,(3) is not the only eddy diffusionformula- mixing ratio •( is then givenby tion proposedfor convectiveturbulence. For example, dx INI has alsobeen suggested [Priestley, 1954; ß• ---•c• d-7 (•) Lilly, 1962], whichyields much higher Kz valuesfor our two cases,so there is considerableuncertainty in eddy diffusivity under convective turbulence. How- Figure6 showsthe resultingKz and (I)o3calculated ever,there are theoreticalreasons and experimentalev- for the sameperiod as coveredby Figure4. The ozone idencefavoring (3) [Globeand Dropkin,1959; Kraich- mixing ratio and its vertical gradient are also plotted nan, 1962; Ingersoll, 1966]. Third, the methodof es- for reference.The backgroundgradient quantities were timating e from 5-Hz dynamicalmeasurements as out- calculatedin the sameway as for Figure 4. Between lined by Chan et al. [1998]has not beenfully verified CHO ET AL.- TROPOPAUSE FOLD INSTABILITIES 21,559

(^qdd)o!•eEI õu!x!• eu0zo c) o o

(W)l) epn•,P, IV 21,560 CHO ET AL' TROPOPAUSE ];'OLD INSTABILITIES

(•s/•pm)aN

(trot)opnl!11¾ oJrl$$oJcl CHO ET AL.: TROPOPAUSE FOLD INSTABILITIES 21,561

SONEX DC-8 1997/10/13 200

,-, 150

0•100

• 0.5 E

,', 0

o '• -1

2.5 ß i i i ß i

2 ß . ß

1.5 1

0.5

1.5

-0.5 18.t }7 18.08 18.09 18.1 18.11 18.12 18.13 18.14 18.15 18.16 UT (h)

Figure6. Plots,from top to bottom,of 03 mixingratio, vertical gradient of 03 mixingratio, verticaleddy diffusivity Kz, andvertical flux of 03 mixingratio from October 13, 1997.

by comparisonwith other instruments.We decidedto fold,one can see that the aircraftcrossed it horizontally usethe Kz approachbecause the 1-sresolution of the from topsideto bottomside.-•14.9 (1454:00) UT, then tracer measurementswas not fine enoughfor a direct ascendedthrough it from bottomsideto topside•015.1 eddycorrelation (w/x •, wherethe primed quantities are (1506:00)UT. This scenariois supportedby the plot perturbationsfrom the means)approach. In i s the air- of stabilityas characterized by N 2 calculatedfrom the craft traversesover 200 m, so the Nyquist length scale MTP temperatureprofiles (Plate 5). Althoughinstru- wouldbe greaterthan the turbulenteddy scale, so w•x • mentalproblems cut offthe data at 15.16(1509:36) UT, would not representturbulent flux, which is what we we can still seethe layer of very stablestratospheric air wantedin our investigationof irreversibletransport. extendingdown from the right-handedge of the plot, leftwardacross the flightline •15.1 (1506:00)UT, then 3.2. Case 2: October 15, 1997 downagain through the flight path .-•14.9 (1454:00) UT. Plate 4 shows in color-scale form the vertical pro- Figure7 againshows the anticorrelationof O3 versus files of ozone concentration measured by the DIAL CO and CH4, clearly indicatingthe encounterwith a as the aircraft encountered the western branch of the lowerportion (height .-•3 km) of the tropopausefold tropopausefold (see also Plate 2). Theapparent discon- at .-•14.9(1454:00) UT and a highersection (height tinuityin the foldwas caused by the DIAL'snear-field •4 km) at 15.1(1506:00) UT. Theaircraft began enter- blind zone,marked by dashedlines in Plate 4. Thus ingthe mainbody of the stratosphericintrusion .-•15.4 the fold was not visible inside the blind zone except (1524:00)UT. We notethe similarityof the upperpart where the aircraft intersectedit with the in situ probe. of the fold to case i in that the sharply defined top- Followingthis "connect-the-dots"mental picture of the side was followedby a coupleof weakerperturbations. 21,562 CHO ET AL.' TROPOPAUSE FOLD INSTABILITIES

SONEX DC-8 1997/10/15 120

100

o o 80

..• 800 •11780 (,•1760 1740

2 14.8 14.9 15 15.1 15.2 15.3 15.4 15.5 UT (h)

Figure 7. Plots, from top to bottom, of 03, CO, and CH4 mixing ratios through the two intersectionswith the westernbranch of the tropopausefold regionon October 15, 1997, and the pressurealtitude of the aircraft.

There was also a hint of a wavelike structure in the ondary ozone maxima. This oscillation is marked by ozone trace on the the more diffuse bottomside. Fortu- a horizontal bar with vertical tick marks separated by nately, the aircraft was ascendingthrough this region, 0.01 hours above the u trace in Figure 8. This combi- sowe will be able to estimatethe vertical gradientquan- nation of dynamical and tracer observations was also tities needed for instability analysesas we did for case 1. seen in case 1, which we interpreted to be convec- The correspondingdynamical measurementsare dis- tively breaking gravity waves. In this case the small- played in Figure 8. In the ascent region of interest scale wave was superimposed on a larger-scale wave (15.05-15.3 (1503:00-1518:00)UT), there wasa zoneof with period •,0.1 hours that also grew in amplitude strong vertical shear and regular oscillationsin w •,15.1 with height and/or horizontal distance and appeared (1506:00)UT (moreeasily seen in Figure9) correspond- to break-•15.3-15.4 (1518:00-1524:00)UT. (Thisoscil- ing to the secondfold encounter. Together with the lation is marked by a horizontal bar with vertical tick wavy bottomside seen in ozone, this suggeststhe pres- marks separatedby 0.1 hours below the u trace in Fig- ence of KH billows. Immediately following the shear ure 8.) Note the strong enhancementin e -•15.3-15.4 zone was a fast oscillation with period •,0.01 hours (1518:00-1524:00)UT. One can alsosee that dO/dzwas in u that grew quickly with height and/or horizontal close to zero in the areas of suspectedconvective wave distanceand appearedto break •,15.14 (1508:24) UT breaking. With one-dimensionalmeasurements such corresponding to a maximum in e and to the sec- as we have it is not possibleto unambiguouslydeter- CHO ET AL.' TROPOPAUSE FOLD INSTABILITIES 21,563

SONEX DC-8 1997/10/15 2O

E 10

-2O

-40

-60

-80

-1

330 320 310

300

14.8 14.9 15 15.1 15.2 15.3 15.4 15.5 UT (h)

Figure 8. Plots, from top to bottom, of u, v, w, loge, and 0 throughthe two intersections with the western branch of the tropopausefold region on October 15, 1997. Suspectedwave growth regions are marked on the u plot by horizontal bars and vertical tick marks. See text for explanation.

mine the wave parameters. Instead, we will focusagain ing was 82ø, this yields a true horizontalwavelength of on the instabilities themselves and their effects on the An = 5200cos(163 ø -82 ø) = 810 m. The verticalthick- ozone flux. nesswas then Az = •a/7.5 = 110 m. These numbers Figure 9 is a blowup plot of the second encounter are close to those from case 1. with the tropopause fold. The oscillation was most One can readily seethe mixing effectsof the KH insta- clear in w, while smaller perturbations with the same bilityon the fold in theozone measurement (Figure 9). period were also visible in •, v, and, to a lesser ex- The bottomside of the ozone maximum was much more tent, in ozone. The perturbations in u were not closely diffuse than the topside. The ozone flux will be esti- correlated to those in w, which is consistent with the mated shortly. planeof the vertical shear(estimated from the data to As in case1, we cancalculate N 2, the verticalshear, be 1630, measuredclockwise from the north) being in and Ri. These are shown in Figure 10. Again, there closeralignment in the meridional direction than in the was a correspondencebetween regions of high shear zonal direction. and high thermal stability, conditions typically asso- Again, assuming that the oscillations were KH bil- ciated with KH instability. Ri did fall close to or be- lows, we can estimate their horizontal wavelength and low 0.25 during the period of the suspected KH bil- vertical thickness.Along the aircraft headingthe period lows (15.085-15.115(1505:06-1506:54)UT), and, as was --•25s, which translates to a length of 5.2 km at the with case1, therewas a regionof negativeN 2 (•15.14 aircraft speedof 208 m s-1 Sincethe aircraft head- (1508:24)UT) associatedwith a secondaryozone maxi- 21,564 CHO ET AL- TROPOPAUSE FOLD INSTABILITIES

SONEX DC-8 1997/10/15

11o

-3O

E,•-40

-5O

310

300 15.08 15.09 15.1 15.11 15.12 15.13 15.14 15. t5 15.16 UT (h)

Figure 9. Expandedtime plots, from top to bottom, of Os, u, v, w, and 0 from October15, 1997.

mumand apparent wave breaking (as discussed above). 4. Summary Discussion Finally,we estimate the vertical ozone flux for thispe- riod(Figure 11). WhereN 2 < 0 •15.15 (1509:00)UT, Eventhough the two casesof tropopausefolding stud- we took IA01= 0.8 K and Az = 80 rn for (4). The ied here occurredat differenttimes and placesand were flux valuesare generally lower than for case 1. This sampledat differentaltitudes, there weresome striking is not surprisingsince the turbulencewas weakerand similarities. Both folds had sharply definedtopsides the ozonegradients were smaller. Sincethe altitude of with two foldlike substructuresshortly followingthat foldinterception was lower in case2 (-•4 km) than in contained lower ozone amounts than the fold itself. It caseI (•6.5 km), the stratosphericair insidethe fold is difficult to say whether these substructureswere hor- would have had more time to mix with the surround- izontal or vertical variationsbecause the flight paths ingtropospheric air, thusweakening the ozonegradient wereone-dimensional and slantwise.Unfortunately, the at the edges.This can alsobe seenin the peak ozone near-field blind zone and coarse time resolution of the concentrationinside the folds,-• 110ppbv for case2 ver- DIAL did not allow two-dimensionalmapping of these sus•190 ppbv for case1. However,as in case1, the small-scaleozone perturbations. If the perturbations estimatedvertical ozonegradients of the substructures were purely horizontal, then their separationdistances between15.13 (1507:48) and 15.15 (1509:00) UT (and wouldhave been ~ 10 and 8 km for case1 and case2, re- the correspondingfluxes) must be viewedwith caution spectively.If they werepurely vertical,then their sepa- becauseof possiblealiasing of horizontalgradients. ration distanceswere •400 and 200 m, respectively.Of CHO ET AL.' TROPOPAUSEFOLD INSTABILITIES 21,565

x 10-'• SONEXDC-8 1997/10/15

x10 •-'5

0.75 0.5 0.25 0 -0.25 i i i i • '•-3

oo- 7 o_ 9

0 120

80 60 40 15.08 15.09 15.1 15.11 15.12 15.13 15.14 15.15 15.16 UT (h)

Figure10. Plots,from top to bottom, of N 2, (dU/dz)•, Ri, loge, w, and O3 mixing ratio from October 15, 1997.

course,the realitycould have been somewhere in be- rolein gravitywave saturation [e.g., Dunkerton, 1989; tween the two extremes. Becausein both casesthe first Fairall et al., 1991;Andreassen et al., 1994]. Origi- of these substructureswas associatedwith convective nallyproposed by Hodges [1967], saturation theory calls instabilityand turbulenceand because there were wave- for the natural growth of upwardpropagating grav- like perturbstie"• ;" *h.... ;.A •,•cl t.•m.poraf,]]r•cc•rre- ity wavesto be limitedby variousinstabilities and/or spondingto the ozonesubstructures, weproposed con- diffusivedamping/filtering mechanisms [e.g., Gardner, vectivelybreaking gravity waves to explainthe data. 1996].Thus it is importantto obtainmore direct evi- It then follows that the substructureswere mostly hor- dencethat convectivegravity wave breakdown does oc- izontalperturbations, since convective breakdown occur in the atmosphereand to pinpointthe conditions cursin high-frequencygravity waves (KH instability underwhich it proceeds.In thisstudy we showed obser- setsin first for low-frequencygravity waves) [Fritts and vationalexamples of convectivegravity wave breaking, Rastogi,1985] and high-frequency gravity waves cannot andfor thosecases the breakingappeared to occurpref- have such short vertical wavelengths. erentiallyjust abovea tropopausefold. This couldbe Althoughlaboratory measurements and numericalbecause the 0 surfacesare movedfarther apart vertically studies confirm the existence of convective gravity (thusweakening the staticstability) on the cyclonic wavebreakdown, direct atmosphericobservations have flank of a jet correspondingto wherethe tropopause largelybeen lacking because of the difficultyin mak- foldis created [Hoskins et al., 1985;Price and Vaughan, ing the requiredhigh-resolution measurements. How- 1993]. Also,the rapidincrease in horizontalvelocity ever,convective instability is believedto playa crucial with heightunderneath the jet lowersthe amplitude 21,566 CHO ET AL.' TROPOPAUSEFOLD INSTABILITIES

SONEX DC-8 1997/10/15 120

100

.-. 0.2 E "' 0

.• -0.2

1.5

'"' 0.5

0.2

0.1 ß . .

0 o -0.1 15.1 )8 15.09 15.1 15.11 15.12 15.13 15.14 15.15 15.16 UT (h)

Figure 11. Plots, from top to bottom, of 03 mixing ratio, vertical gradient of 03 mixing ratio, Kz, and vertical flux of 03 mixing ratio from October 15, 1997.

thresholdfor wavebreaking [Thorpe, 1978] and causes The site of maximum ozone flux out of a tropopause more gravity waves with phase velocities in the same fold was on the topside of case 1, with a peak value of direction as the backgroundflow to reach critical levels •1.5 ppbvm s-x in goodagreement with previousair- and convectiveinstability [ Thorpe,1981; Fritts, 1982]. craft observationsby Shapiro[1980]. In this particular In any case,we need to add convectivelybreaking grav- case the instability was generated by a combinationof ity wavesto the list of mechanismsleading to the irre- backgroundshear and wave-inducedweakening of static versible mixing of tropopausefolds with their environ- stability. ment. KH instabilities are much more commonly observed Acknowledgments. The work at MIT was funded by in the atmosphere(see, e.g., Fritts and Rastogi[1985] NASA grants NAG2-1105, NAGl-1758, and NAGl-1901. for references). We observedKH billows inside both The work performed by M. J. Mahoney was carried out by the Jet Propulsion Laboratory, California Institute of Tech- tropopause folds encountered and noted a correlation nology, under a contract with NASA. He also acknowledges betweenstrong vertical shear and high static stability as the able assistance of Bruce L. Gary with the MTP data anticipatedtheoretically [Phillips, 1966] and frequently analysis. Finally, we would like to thank the DC-8 flight observed[e.g., Browninget al., 1973]. In case2 the crew for their hard work. KH instability had apparently acted on the bottomside ozonegradient and made it much more diffusethan the topside of the fold, whereas in case I the KH instability References located in the middle of the fold was not effective in Andreassen, (D., C. E. Wasberg, D.C. Fritts, and J. R. mixing air inside and outside the fold. Isler, Gravity wave breaking in two and three dimensions, CHO ET AL.: TROPOPAUSE FOLD INSTABILITIES 21,567

1, Model description and comparison of two-dimensional Globe, S., and D. Dropkin, Natural convection heat transfer evolutions, J. Geophys. Res., 99, 8095-8108, 1994. in liquids confined by two horizontal plates and heated Appenzeller, C., H. C. Davies, and W. A. Norton, Fragmen- from below, J. Heat Transfer, 81, 156-185, 1959. tation of stratospheric intrusions, J. Geophys. Res., 101, Hodges, R., Generation of turbulence in the upper atmo- 1435-1456, 1996. sphere by internal gravity waves, J. Geophys. Res., 72, Browell, E. V., E. F. Danielsen, S. Ismail, G. L. Gregory, 3455-3458, 1967. and S. M. Beck, Tropopause fold structure determined Hoskins, B. J., M. E. Mcintyre, and A. W. Robertson, On from airborne lidar and in situ measurements, J. Geophys. the use and significance of isentropic potential vorticity Res., 92, 2112-2120, 1987. maps, Q. J. It. Meteorol. Soc., 111,877-946, 1985. Browell, E. V., S. Ismail, and W. B. Grant, Differential ab- Ingersoll, A. P., Thermal convection with shear at high sorptionlidar (DIAL) measurementsfrom air and space, Rayleigh number, J. Fluid Mech., 25,209-228, 1966. Appl. Phys. B, 67, 399-410, 1998. Itsweire, E. C., K. N. Helland, and C. W. Van Atta, The Browning, K. A., G. W. Bryant, J. R. Starr, and D. N. Ax- evolution of grid-generated turbulence in stratified fluid, ford, Air motion within Kelvin-Helmholtz billows deter- J. Fluid Mech., 126,299-338, 1986. mined from simulaneous Doppler radar and aircraft mea- Johnson, W. B., and W. Viezee, Stratospheric ozone in the surements, Q. J. R. Meteorol. Soc., 99,608-618, 1973. lower troposphere, I, Presentation and interpretation of Chan, K. R., J. Dean-Day, S. W. Bowen, and T. P. Bui, Tur- aircraft measurements, Atmos. Environ., 15, 1309-1323, bulence measurementsby the DC-8 meteorologicalmea- 1981. surement system, Geophys. Res. Lett., 25, 1355-1358, I(raichnan, R., Turbulent thermal convection at arbitrary 1998. Prandl number, Phys. Fluids, 5, 1374-1389, 1962. Chapman, S., and T. G. Cowling, The Mathematical Theory Langford, A. O., and S. J. Reid, Dissipation and mixing of of Non-Uniform Gases, 423 pp., Cambridge Univ. Press, a small-scale stratospheric intrusion in the upper tropo- New York, 1970. sphere, J. Geophys. Res., 103, 31,265-31,276, 1998. Cho, J. Y. N., R. F. Jurgens, and M. A. Slade, High- Lilly, D. K., On the numerical simulation of buoyant con- resolution stratospheric dynamics measurementswith the vection, Tellus, ld,, 148-172, 1962. NASA/JPL Goldstonesolar systemradar, Geophys.Res. Miles, J. W., and L. N. Howard, Note on a heterogeneous Lett., 23, 1909-1912, 1996. shear flow, J. Fluid Mech., 20,331-336, 1964. Clough, P. N., and B. A. Thrush, Mechanism of chemilumi- Orlanski, I., On the breaking of standing internal gravity nescent reaction between nitric oxide and ozone, Trans. waves, J. Fluid Mech., 5d,, 577-598, 1972. Faraday Soc., 63,915-925, 1967. Orlanski, I., and B. B. Ross, Numerical simulation of the Collins, J. E., G. W. Sachse, B. E. Anderson, R. C. Harriss, generation and breaking of internal gravity waves, J. Geo- K. B. Bartlett, S. T. Sandholm, L. O. Wade, L. G. Burney, phys. Res., 78, 8808-8826, 1973. and G. F. Hill, Airborne nitrous oxide observations over Phillips, O. M., The Dynamics of the Upper Ocean, 261 pp., the western Pacific Ocean: September-October 1991, J. Cambridge Univ. Press, New York, 1966. Geophys. Res., 101, 1975-1984, 1996. Price, J. D., and G. Vaughan, On the potential for Danielsen, E. F., and V. A. Mohnen, Project Duststorm stratosphere-troposphereexchange in cut-off low systems, report: Ozone transport, in situ measurements, and me- Q. J. It. Meteorol. Soc., 119,343-365, 1993. teorological analyses of tropopause folding, J. Geophys. Priestley, C. H. B., Convection from a horizontal surface, Res., 82, 5867-5877, 1977. Aust. J. Phys., 7, 176-201, 1954. Danielsen, E. F., R. Bleck, J. Shedlovsky, A. Wartburg, Sachse, G. W., G. F. Hill, L. O. Wade, and M. G. Perry, P. Haagenson, and W. Pollock, Observed distribution of Fast-response, high precision carbon monoxide sensorus- radioactivity, ozone, and potential vorticity associated ing a tunable diode laser absorption technique, J. Geo- with tropopause folding, J. Geophys.Res., 75, 2353-2361, phys. Res., 92, 2071-2081, 1987. 1970. Shapiro, M. A., A multiple-structured frontal zone-jet Danielsen, E. F., R. S. Hipskind, S. E. Gaines, G. W. Sachse, stream system as revealed by meteorologically instru- G. L. Gregory, and G. F. Hill, Three-dimensionalanaly- mented aircraft, Mon. Weather Rev., 102,244-253, 1974. sis of potential vorticity associated with tropopause folds Shapiro, M. A., Turbulent mixing within tropopause folds and observed variations of ozone and carbon monoxide, J. as a mechanism for the exchange of chemical constituents Geophys. Res., 92, 2103-2111, 1987. between the stratosphere and troposphere, J. Atmos. Sci., Denning, R. F., S. L. Guidero, G. S. Parks, and B. L. Gary, 37, 994-1004, 1980. Instrument description of the airborne microwave temper- Singh, H. B., A.M. Thompson, and H. Schlager, SONEX ature profiler, J. Geophys. Res., 9•,, 16,757-16,766, 1989. airborne mission and coordinated POLINAT-2 activity: Dunkerton, T. J., Theory of internal gravity wave satura- Overview and accomplishments, Geophys. Res. Lett., in tion, Pure Appl. Geophys., 130,373-397, 1989. press, 1999. Fairall, C. W., A. B. White, and D. W. Thomson, A stochas- Thorpe, S. A., On the shape and breaking of fmite amplitude tic model of gravity-wave-induced clear-air turbulence, J. gravity waves in a shear flow, J. Fluid Mech., 85, 7-31, Atmos. Sci., •,8, 1771-1790, 1991. 1978. Fritts, D.C., The transient critical-level interaction in a Thorpe, S. A., An experimental study of critical layers, J. Boussinesqfluid, J. Geophys. Res., 87, 7997-8013, 1982. Fluid Mech., 103,321-344, 1981. Fritts, D.C., and T. J. Dunkerton, Fluxes of heat and con- Walterscheid, R. L., and G. Schubert, Nonlinear evolution of stituents due to convectively unstable gravity waves, J. an upward propagating gravity wave: Overturning, con- Atmos. Sci., d,2, 549-556, 1985. vection, transience and turbulence, J. Atmos. Sci., 4,7, Fritts, D.C., and P. K. Rastogi, Convective and dynamical 101-125, 1990. instabilities due to gravity wave motions in the lower and Weinstock, J., On the theory of turbulence in the buoyancy middle atmosphere: Theory and observations, Radio Sci., subrange of stably stratified flows, J. Atmos. Sci., 35, 20, 1247-1277, 1985. 634-649, 1978. Gardner, C. S., Testing theoriesof atmosphericgravity wave Weinstock, J., Vertical diffusivity and overturninglength in saturation and dissipation, J. Atmos. Terr. Phys., 58, stably stratified turbulence, J. Geophys.Res., 97, 12,653- 1575-1589, 1996. 12,658, 1992. 21,568 CHO ET AL.: TROPOPAUSE FOLD INSTABILITIES

E. V. Browell,G. L. Gregory,G. W. Sachse,and S.A. T.L. Kucseraand A.M. Thompson,Code 916, NASA Vay,NASA Langley Research Center, Hampton, VA 23681- GoddardSpace Flight Center,Greenbelt Rd., Green- 2199.([email protected]; [email protected]; belt, MD 20771-0001.([email protected]; thomp- [email protected];[email protected]. gov) [email protected]) T. P.Bui, Mail Stop 245-5, NASA Ames Research Center, M. J. Mahoney,Mail Stop 246-101,NASA Jet MoffettField, CA 94035-1000.([email protected]) Propulsion Laboratory, California Institute of Technol- J. Y. N. Choand R. E. Newell,Department of Earth, ogy,4800 Oak GroveDr., Pasadena,CA 91109-8099. Atmospheric,and Planetary Sciences, Massachusetts In- ([email protected]) stitute of Technology,77 MassachusettsAve., 54-1823, Cambridge,MA 02139-4307. ([email protected]; [email protected] ) M. A. Ferm, Science Applications International Cor- poration,1 EnterpriseParkway, Hampton, VA 23666. (ReceivedJanuary 14, 1999; revised May 25, 1999; ([email protected]) acceptedJune 16, 1999.)