Observations of the Quasi 2Day Wave from the High Resolution Doppler
Total Page:16
File Type:pdf, Size:1020Kb
GEOPHYSICAL RESEARCH LETTERS, VOL. 20, NO. 24, PAGES 2853-2856, DECEMBER 23, 1993 OBSERVATIONS OF THE QUASI 2-DAY WAVE FROM THE HIGH RESOLUTION DOPPLER IMAGER ON UARS D. L. Wu, P. B. Hays,W. R. Skinner,A. R. Marshall,M.D. Burrage,R. S. Lieberman,and D. A. Ortland The Universityof Michigan Departmentof Atmospheric,Oceanic and Space Sciences Abstract. A strongwestward traveling oscillation,with a January1993 are carried out to show the global distribution period of 2 days and zonal wave number3, is observedin the and temporalvariation of the wave amplitudeand phase. mesosphericand lower thermosphericwinds from the High Resolution Doppler Imager on the Upper Atmosphere Data analysis ResearchSatellite. The importantevents happen in January, July, and September/October,of which the occurrencein HRDI is a triple etalon Fabry-Perot interferometer that Januaryis the strongestwith an amplitudeover 60ms-•. determineswinds with an accuracyof 5ms4 by resoolvingthe Detailed analyses for the periods of January 1992 and Doppler shift of 02 emission lines (around 7620 A) in the January 1993 reveal a cause-and-effectrelationship in the MLT region. Details of the instrument and the calibration Wavedeveloping process at 95km. The globalstructures of procedures can be found elsewhere [Hays, et al., 1993; the wave amplitudeand phase are alsopresented. Burrage,et al., 1993]. HRDI hasbeen measuring winds since November 1, 1991 and divides the observingtime equally Introduction between the stratosphericmode (10-40km) and the MLT mode (50-115km). Until June 1993, the regular observing Over the last two decades a number of observations of the schemefor the MLT region was to scan all altitudesduring quasi-two-daywave (1.8-2.2 days;hereafter the 2-day wave) the daytime between65km-105km with a 2.5 km increment. have beenreported from ground-basedstations, rockets and Subsequently,the altitude range was extendedto 50-115km satellites. The ground-basedobservations consist mainly of with the same height resolution. In the nighttime, HRDI radar wind measurementsover a wide range of latitudes measurementsare limited to one altitude (-95km) because [Mtiller and Kingsley, 1974; Salby and Roper, 1980; Manson the observed nightglow emission is restricted to a narrow and Meek, 1990; Vincent and Lesicar, 1991; Fritts and Isler, layer. In addition to performing the altitude scanning,the 1992; Tsuda et al., 1988]. The 2-day wave amplitudeof the gimbaled HRDI telescopeis capableof moving in azimuth meridionalcomponent was foundto be quite large (50 ms-1) and allowing views of both sides of the satellite track. The in the meteor region but intermittentwith a lifetime of about latitudinal coverage is determined by the HRDI viewing 20-30 days. Sometimes,the oscillationcould reach even up direction and the UARS orbit which is circular at an altitude to the ionosphere[Chen, 1992]. At mid-latitudesa large of 585km and an inclination of 57ø. The orbital precession amplitudewas usually observedone month after the summer allows a complete samplingof local times about every 36 solstice[Craig et al. 1980; Clark, 1989] while at the tropical days.The HRDI observingplans are made on a daily basis. and subtropicalsites [Harris and Vincent, 1993; Fritts and Except duringspecial campaigns, generally, the MLT region Isler, 1992] the wave appearedin both solstice periods. In is viewedevery otherday. general, the January/Februaryevent was strongerthan one in Due to the HRDI's specialday-and-night viewing plans, July/August. The satellite data of Nimbus 5 [Rodgers and the samplessometimes are not uniformly spacedand can Prata, 1981] and Nimbus 7 [Burks and Leovy, 1986], on the have a large gap. Therefore, a least-squarespectral analysis otherhand, provided a globaland long-termstudy of the 2- methodis employedto resolvethe 2-dimensionalspace-time day wave and helped confirm its zonal wavenumber 3 spectra.The basicidea of the techniqueis to fit parametersA character. and B so as to minimize the departurebetween the desired The origin of such a transient2-day oscillationremains functions and observations [for an example, Bevington unclear. However, the eigenperiodstudies of the Rossby- 1969], which is given by gravity normal modes [Salby, 1981; Hagan et al., 1993] suggestedthat the mode (3,0) could be the major carrier of the 2-day wave. The symmetriceigenstructure of the mode, Z2=• (Acos 2n(vti +s•i)+ Bsin2n(o'ti2 +S•i)--Yi )2 in the presenceof a backgroundmean wind and temperature, i œi was slightly distorted in solstice and the amplitude was enhanced'remarkably in the summer meteor regions. An where Yi is the measurement,with an error œi, obtained at alternative mechanism proposed by Plumb [1983] emphasizedthe role of the fast growingbaroclinic instability universal time ti (days)and longitude3•i (normalized to above the summerstratospheric easterlies. In this theory, the unity). c• and s are, respectively,the frequency(days -•) and 2-day wave is believed to be a product of the instability. the zonalwavenumber of the oscillation.Fitting is performed Reviews of the phenomenonhave been presentedby Mtiller at every latitudecircle and the parametersA and B yield the and Nelson [1978], Salby [1984] and Vincent [1984]. This paperpresents the recentresults of the 2-day wave amplitude•/A2+B 2 and the phase tan-•(B/A) for that latitude. observedby the High ResolutionDoppler Imager (HRDI) on Spectralscans have been carried out for HRDI databy tuning board the Upper Atmosphere Research Satellite (UARS). c• and s to searchsignificant wave components.The pair, The data consist of the wind measurements in the mesosphere/lowerthermosphere (MLT) regionfrom 65km to (c•,s)=(0.5,3),has a large amplituderesponse in Januaryand 105km.An intensivestudy of the eventsin January1992 and is used to stand for the 2-day wave throughoutthis paper. Becausethe 2-day wave is such a transientphenomenon, a Copyright1993 by the American Geophysical Union. running 5-day interval is used in the spectral analysis. However, this short time interval yields a coarse spectral Papernumber 93GL03008 resolution,which meansthat the 52 hrs 2-day wave can still 0094-8534/93/93GL-03008503.00 contribute80% of its amplitudeto the exact 2-day wave. 2853 2854 Wu et al.- Quasi2-day wave from HRDI on UARS At a given latitudecircle, daily observationscan be taken 90 at very different longitudesand universal times but they usually are limited in local time. The HRDI sampling schemesmay be dividedroughly into threecases in termsof 30 the samplingfrequency in local time eachday. (I) Two local times separatedapproximately by 12 hrs are sampledeach EO day, correspondingto the HRDI day-and-nightsampling in the equatorialregion at 95km. (II) Only one local time is -30 sampledeach day, correspondingto the period when only continuousdaytime or nighttimedata are available.(III) One -60 local time is sampledevery other day, which representsthe -90 situationof the regularHRDI MLT viewing.The aliasing problems associatedwith each case are different and 0 ß ' ' complicated.For a simpleand uniformsampling case the __.. aliasingbehavior can be well explainedby the so-called '•' 60 _•I r'-•5I ..... ',I ..•___•;•'-]x__ . asynopticsampling theory [Salby, 1982]. But for irregular • 30 '• o ' sampling,simulations are generallyrequired to resolvethe • , , • •,_•_•_• -- distributionand strengthof aliases.Because the satellite m 0 samplingis always accompaniedby the aliasingproblem, a spectralpeak can be interpretedas a wave componentonly =-30- • ,• - when all of its aliases are assumed not to be a realistic atmosphericwave and may be ignored[Lait and Stanford, •-60 - •,_ -.... 1988]. However, this assumptionsometimes does not hold -•• •----" (b) for every alias. In order to identify possibly destructive -90 aliases,spectral scans are made for each of the three cases Jan 1992 Feb Mar with artificialdata. As a result,all the 2-daywave aliases are ignorableunder case I and II. In caseIII, however,the 2-day Fig.2. Amplitudes of meridional (a) and zonal (b) wave can be partially affectedby non-migratingfides and componentsat 95km in •e beginningof the year of 1992. other slowly moving planetarywaves therefore any spectral Both daytime and nighttime data are used in an•ysis. Dash significance out of this sampling must be treated with lines mark the limiting latitudes of the HRDI's sampling. caution. Contours start from 20ms 4 at 10ms -• intervals. Results durationof 15-20 days, happensaround September/October The 2-day wave amplitudesare shown in Figure 1 and 1992, but it has received little attention previously. This they are derivedfrom the spectralanalysis with sliding5-day event is interestingbecause of its time of appearanceduring intervals. The equatorial (10os-10øN) amplitudesof the which the forcing mechanismcould be quite different from meridional wind are presentedbecause they have a larger that in solstice.Moreover, the October2-day wave is not only wave response[Harris and Vincent, 1993] and serve as a evidentin the HRDI winds but also in the radar data [Salby good indicator of the wave event in the meteorregion. At andRoper, 1980;Harris andVincent, 1993]. 95km the 2-day wave is better sampledwith both daytime The large data breaksin Figure 1 (Mar.31-Apr.24, 1992; and nighttimedata. During January1992 and January1993, Jun.3-jul.21, 1992; Feb.l-Feb. 11, 1993) are due to strong 2-day oscillations(over 60ms-1) occur in the MLT malfunctions in either the instrument or the satellite. At region and last about 30 days. A weaker peak (-30ms-1) is 95km, most sampling situationsthroughout Nov'91-Jul'93 observedin July 1993, which is consistentwith radar correspondto eithercase I or II. During the January1992 and observations[Harris and Vincent, 1993; Fritts and Isler, January 1993, HRDI was conducting the mesospheric 1992]. Another event, with an amplitudeof 40ms-1 and a campaigns,allowing the 2-day wave to be resolved with greaterconfidence. In someperiods, such as Nov.20-28 1991, May, late July, August,and September1992, the samplings •0 i i , i i [ i i i ! i i i , i i i i i , l correspondmainly to caseIII but the amplitudesturn out to be insignificant anyhow.