imaging radar observations and theory of type i and type ii...

To appearin JGR,2002

Imaging radar observations and theory of type I and type IIquasiperiodic echoes

D. L. Hysell�, M. Yamamoto,andS.Fukao

RadioScienceCenterfor SpaceandAtmosphere,Kyoto University, Uji, Japan

Abstract.MidlatitudeE regionplasmairregularitieshavebeeninvestigatedusingtheMiddleandUpperAtmosphere(MU) radarin Shigaraki,Japan,and the Clemson30MHz radarin SouthCarolina,USA. A new in-beamimagingtechniquehasbeenincorporatedin thedataanalysis.Radarimagesrevealthatthecoherentbackscatterassociatedwith quasiperiodic(QP)echoesmainly arrivesfrom spatiallylocalized,elongatedscatteringregionsandthatstriationsin radarrange-time-intensity(RTI)mapsarethesignaturesof themigrationsof theseregionsthroughthesparsely-filled illuminatedvolume. Thescatteringregionsin questionappearto maintainaltitudeasthey drift. Refractionandfinite aspectanglesensitivity permit fieldalignedirregularitiesin theregionsto bedetectedby radarsover a broadrangeofzenithangles.Circulationobservedwithin thescatteringregionsis consistentwithsimulationsof polarizedE region plasmacloudsdescribedin a companionpaper,asis theoccasionalappearanceof typeI echoes.Severallong,continuousbandsofscatterersarealsoevidentin theradarimages.

1. Introduction

In thecompanionpaper(Hysellet al. [2002], henceforthreferredto as paper1), we investigatedthe electrodynam-ics of irregular, patchysporadicE layersandplasmacloudsin the nighttime midlatitude E region. Such layers havebeenobserveddirectly at Arecibo [Miller andSmith, 1978;Smithand Miller , 1980] and inferred from radio scintilla-tion measurements[Bowman, 1989;Maruyama, 1991],andthe morphologyof the relatedquasi-periodicscintillations(QPS)hasbeenshown to matchthat of QP echoesclosely[Maruyamaet al., 2000]. The sourceof the patchinessisnot known, althoughLarsen[2000] hasarguedthat it maybe driven by neutralshearinstabilities. Larsen[2000] fur-therarguedthatthesepatchylayersarethedirectsourcesofquasiperiodicechoes.Our theoreticalinvestigationshowedthat large polarizationelectricfields andHall currentscanarisespontaneouslyin elongatedlayersand that the layersthemselvesareunstableto kilometer-scaleprimary plasmawaves.Theingredientsnecessaryfor formingsmall-scaleir-�

On leave from the Departmentof Physicsand Astronomy, ClemsonUniversity, Clemson,SouthCarolina,USA

regularitieswithin theclouds,eitherdirectlybyFarley Bune-maninstabilitiesor indirectlyby modecouplingandplasmaturbulencedrivenby theprimarywaves,seemto bepresent.However, links betweenthesetheoreticalfindings andob-servationsof coherentscatterfrom midlatitudeirregularitiesandQPechoesin particularneedto beestablished.

Numeroustheoriesintendedto accountfor the charac-teristicsof midlatitudeE region coherentscatterhave beenadvancedin recentyears[Woodmanet al., 1991;Tsunodaet al., 1994;Haldoupiset al., 1996;Shalimov et al., 1998;Tsunoda, 1998; Kagan and Kelley, 1998; Rosado-Romanet al., 1999; Kagan and Kelley, 2000; Maruyamaet al.,2000]. Validating thesetheorieswith conventional radartechniqueslike range-time-intensity(RTI) analysis,spectralanalysis,single baselineinterferometry, and beamswing-ing alonemay not be possible,however. Thesetechniquesdo not alwaysdistinguishclearly betweenspatialandtem-poral variationswithin the illuminated volume and cannotresolve fine structurein the directionsperpendicularto theradarbeam.Theexisting empiricaldatabasefor QPechoesis somewhatambiguousin thatregard,andmultiple theoriesappearto bebroadlyconsistentwith it.

1

Hysell etal.: IMAGING RADAR OBSERVATIONS OFQPECHOES 2

This paperpresentsobservationsof plasmairregularitiesin themidlatitudeE regionmadewith theMU andClemsonradarsusinga new in-beamradarimagingtechnique.Thetechnique,introducedat Jicamarcaby Kudeki and Surucu[1991] anddevelopedby Woodman[1997] andHysell andWoodman[1997], utilizesmultiple interferometrybaselinesto resolve spatial-temporalambiguity and to discernfinestructurewithin the radarilluminatedvolume. Imagespro-ducedwith the techniquewill favor an interpretationof QPechoesaslocalized,drifting scatteringregionssparselydis-tributedthroughouttheradarilluminatedvolume.Striationsin RTI plots reflectthe motion of the scatteringregionsto-ward or away from the radar. Telltale signsof circulationwithin the scatteringregionssupportthe ideathat they areassociatedwith elongated,polarizedE regionplasmacloudsdrifting with the backgroundwind. The detectionof typeI echoesimmediatelyabove a very elongatedscatteringre-gion alsosupportthis picture.We furtherarguethatplasmairregularitieswithin localized,horizontallydrifting plasmacloudscanremainvisible to coherentscatterradarsfor longperiodsof timeduein largepartto refraction.

Webeginwith anoverview of aseriesof experimentsper-formedwith the Clemson30 MHz radarandthe MU VHFradarwhich wereusedto studyQP echoesandoperatedinimagingmodein thesummerof 2001.After briefly describ-ing the principlesof radarimaging,we presenta seriesofimagesconstructedfrom the radardataand evaluatetheirconnectionto the theory outlined in paper1. Finally, wesummarizeour findingsan suggestdirectionsfor future re-search.

2. Experimental overview

Midlatitude E region plasmairregularitieswere studiedwith theClemson30MHz coherentscatterradaronapprox-imately 50 eveningsbetweenMay and August, 2001 andwith theMU VHF radaronthreesuccessiveeveningsin Au-gust,2001. StrongE region echoesweredetectedat Clem-son on abouthalf of all eveningsand at the MU radaronall threeeveningsof operation.Theseradarsaresituatedatsimilar geographiclatitudesbut producedqualitatively dif-ferentdatasetsdue to differencesin their geomagneticlat-itude, frequency, sensitivity, andhoursof operation.Infor-mationabouttheprocessesresponsiblefor quasiperiodicandotherformsof echoescanbededucedfrom a comparisonofthedatasets.Table1. lists theexperimentparameters.Themain beamsof the MU andClemsonradarsweredirectedtowardgeographicandgeomagneticnorth,respectively. Ex-perimentsat both sitesweresupportedlocally by ionoson-des.

Figure1 showsarepresentativeexampleof theMU radar

Table 1. Tableof radarparametersfor theQPechoexperiments.

Parameter MU radar ClemsonradarLatitude(deg) 34.9N 34.7NLongitude(deg) 136.1E 82.8WDip angle(deg) 51 66.5Frequency (MHz) 46.5 29.8Peakpower (kW) 1000 8IPP(ms) 1.5 4Coherentint. 4 1Baudlength(km) 0.3 1.5Codelength(n) 32 28Receivers 4 4Baselines 11 9��

( � ) 3.04 1.25��( � ) 5.27 5

HPFB(deg) 3.6 10

datafrom August 9, 2001, plotted in range-time-intensity(RTI) format.Thefigureis dominatedby two distinctgroupsof QPechostriationsseparatedby a periodof relative inac-tivity lastingaboutanhour. The rangerates,rangeextents,slopesigns,and pseudoperiodsof the striationsare con-sistentwith whathasbeendocumentedby Yamamotoet al.[1991, 1992]. Echoesobserved on August 7 weresimilarto thoseshown hereexcept that the striationsoccurredinthreedistinct groups,separatedagainby hour-long periodsof inactivity. In contrast,Figure 2 shows echoesreceivedby the MU radaron August8, 2001. Theseechoesdiffermorphologicallyfrom thosein Figure 1, failing to resem-blequasiperiodicstriationsandinsteadappearingfrom 2115to 2320 LT as a long, unbroken successionof amorphous“blobs”.

Figure3, meanwhile,shows representativedatafrom theClemson30 MHz radartaken on June11, 2001(UT date).This figure shows four distinct groupingsof echoessepa-ratedin time by about50 min. Thefirst two of theseexhibitthecharacteristicsof QPechostriationswith nearlyuniform,negative rangeratesandwith large rangeextentsin a fewcases.In thethird andfourthgroupings,however, theechoesappearmorelikeblobsor elongatedblobswith positive,neg-ative,or indeterminaterangerates.Theterm“quasiperiodic”appliesequallypoorly to theseechoesandtheechoesregis-teredby theMU radaron August8. Althoughthemajorityof echoesseenby theClemsonradarin thesummersof 1998and1999resembledquasiperiodicstriationswith predomi-natelynegative andpositive rangeratesin the eveningandearlymorning,respectively, mostof theechoesobservedin


Figure 1. Range-time-intensityplot showingquasiperiodicechoesobservedwith theMU radaronAugust9,2001.Grayscalesdepictsignal-to-noiseratiosin dB. A coarseapproximationof theechoaltitudein kilometersappearon theright axisof theplot (seetext).


Figure 2. Range-time-intensityplot showing echoesobservedwith theMU radaronAugust8,2001.

Figure 3. Range-time-intensityplot showing quasiperiodicechoesobservedwith theClemsonradaronJune11,2001.Here,UT=LT+5 hours.


Figure 4. Range-time-intensityplot showing quasiperiodicechoesobservedwith theClemsonradaronJune15,2001.Here,UT=LT+5 hours.Notethattheechoespresentat 0210UT in many rangegatescorrespondto groundclutter.

2001wereblob-like [Hysell and Burcham, 2000]. Further-more,thosestriatedechoesseenin 2001wereequallylikelyto have positive andnegative rangerates. Figure 4 showsan exampleof particularly intenseechoesreceived by theClemsonradar. QP echoeswith bothpositive andnegativerangeratescanbe seento overlapin this figure, ruling outthe idea that they are somehow relatedto multiple, tilted,planarsporadicE layers.Blob-like echoesarealsopresent.It is noteworthy thattheobservationsshown here,which in-cludetheonly caseof typeI echoesreceivedin 2001,weremadeduringstrongspreadF conditions.

The right axesof Figures1–4 indicatethe approximatealtitudesfrom which theechoescouldhave arisen.Theap-proximationin questionis thatradarsignalspropagatealongstraightlinesthroughtheazimuthalcenterof theradarbeamat the elevation anglewherefield alignedirregularitiesex-actly satisfythe conditionfor Braggscatter. However, Hy-sell and Burcham [2000] arguedthat theseapproximationsareunreliableandthat refraction,finite aspectanglesensi-tivity, andoff-azimuthpropagationmakeit impossibleto as-sign a scatteringaltitude to an echosolely on the basisofrange. The echoesthat appearto have originatedfrom al-titudesaslow as90 km in Figure3 and80 km in Figure4actuallyoriginatedat higheraltitudebut underwentsignif-

icant refraction. (SporadicE layersover Clemsonhad topfrequenciesof 6 MHz at0400UT onJune11and10MHz atvarioustimeson June15, andtop frequenciesashigh as12MHz werenot uncommonduring the eveningsof the sum-merof 2001thereandat theMU radar.) Likewise,thelongrangeechoesthat might seemto have comefrom altitudesabove about120 km in Figure4 actuallycamefrom loweraltitudesbut from largeazimuthangles.In orderto quantifythevariouseffectsthatdisassociateechorangeandaltitude,to studytheconfigurationof theirregularitiesresponsibleforQPandotherechoes,andto learnwhattheQPstriationsandblobs in RTI mapssignify, we must turn to more incisivediagnostictoolslike in-beamradarimaging.

True imagesof the backscatter“brightness”distribution(thedistributionof backscatterintensityversusbearing)withinthe radarilluminatedvolumecanbe formed from interfer-ometrydatatakenwith multiple baselines.It is well knownthat interferometrywith a single baselineyields two mo-mentsof the brightnessdistribution [Farley et al., 1981].Morebaselinesyieldmoremoments,andasufficientnumberof momentsdefineanimagein oneor two dimensions.(For-mally, theinterferometrycrossspectrumor “visibility” is re-latedto the brightnessdistribution by an integral transformresemblinga Fourier transform[Thompson, 1986].) Radar


Clemson Radar

MU radar

dx

dy

dx

dy

O

O

Figure 5. Configurationof interferometrybaselinesusedwith the MU andClemsonradars.Baselinelengthsarein-teger multiples of the

��and

��factorslisted in Table 1.

Arrowheaddirectionsindicatehow thereceiversweremul-tiplexed.

rangegatingaddsanotherspatialdimension.An importantfeatureof in-beamradarimagingthat distinguishesit frombeam-swingingapproachesis that the angularresolutionofthe techniqueis limited by the lengthof the longestinter-ferometrybaselinesratherthanthesizeof themainantennaarray. The formercangenerallybe increasedeconomicallywhereasthelattergenerallycannot,andit is possiblein prac-tice to form veryhighresolutionimageswith interferometryevenusingsmall radarsystemslike theClemsonradar. Im-agesformedfromdifferentDopplerspectralcomponentscanbecombinedinto compositeimages,conveying informationaboutthe spectralcharacteristicsof differentregionsof theilluminatedvolume. Animatedsequencesof imagesrevealhow thescatterersevolveovertimeasthey travel throughtheilluminatedvolume.

Both theMU radarandtheClemsonradarcanuseup tofour receiversfor interferometry. By receiving signalsfromfour spatiallyseparatedantennagroups,cross-spectralmea-surementscanbe madeon up to six nonredundantinterfer-ometrybaselines.By multiplexing thefour receiversto mul-tiple set of antennagroups,this numbercan be increased.Crossspectrafrom 11 and9 nonredundant,nonzerobase-lineswerecomputedfrom theMU andClemsonradardata,respectively, in this way. The baselineconfigurationsforbothradarsareillustratedin Figure5. Here,thecomponentsof thebaselinesin thedirectionstransverseto andparallelto

theradarmainbeamsareintegermultiplesof the��

and��

factorsgivenin Table1, respectively.

Becausethe visibility is incompletelysampledandsuf-fersfrom statisticalfluctuationsin practice,andbecausethedirecttransformationfrom visibility to brightnessspacemaybe ill-conditioned, regularizationtechniquesandstatisticalinversetheoryareappliedto thereconstructionof thebright-nessdistribution from the interferometrydata.Theparticu-lar algorithmusedhereemploys entropy asthe regulariza-tion metric [Ables, 1974;Jaynes, 1982;Skilling andBryan,1984].A detaileddescriptionof thealgorithmwasgivenbyHysellandWoodman[1997].

3. Radar images

ImageshavebeenconstructedfromtheE regionbackscat-ter data in three dimensions;azimuth, zenith, and range.Therebeingno expedientmeansof presentingthreedimen-sional imagesin print, the zenithangledependenceof theimageshasbeenintegratedout, althoughthe momentsoftheseintegralshavebeenretained(seebelow). Theresultingtwo-dimensionalimagesarecomputedasfunctionsof timeandareessentiallyfree of the spatial-temporalambiguitiesinherentin RTI diagrams.

Figure6 andFigure7 show sequencesof in-beamradarimagesof E region field alignedirregularities.Thehorizon-tal andverticalaxesof the imageframesrepresentazimuthanglein degreesandrangein kilometers,respectively. Ata rangeof 170 km (260 km), a 1� anglecorrespondsto atransversedistanceof about3 km (4.5 km). Note that thetransmittingantennasmainly illuminatethecentralportionsof the imaging fields of view, explaining why the imageperipheriesare frequentlyempty. Occasionally, however,strongechoescanbe seenarriving from large azimuthan-glesthroughantennasidelobes.

Interferometriccross-spectrawere computedfrom theMU andClemsonradardatausing4-and8-pointfastFouriertransforms,respectively. Eachimagepanelin Figure6 andFigure7 is reallyacompositeof threeseparateimages,com-putedfrom threeof theavailableDopplercomponentsof thecross-spectra.Thethreecomponentswereselectedandcolorcodedaccordingto thekey in Figure8 andsuperimposedtoform thecompositeimages.Red,blue,andgreencolorsde-notered-shiftedechocomponents,blue-shiftedcomponents,andcomponentswith smallDopplershifts.Whencombinedthis way, the lightnessof eachpixel becomesa represen-tation of the signal-to-noiseratio (on a log scale,typicallyfrom 5–35dB), thehueof thefirst momentDopplerveloc-ity, and the saturationof the Doppler spectralwidth; pure(pastel)colorsindicatenarrow (broad)spectra.

Plottersymbolsin the small graphsto the right of each


Figure 6. Sequencesof four consecutive in-beamradarimagescomputedfrom MU radardata. (top row) August8,2001.(remainingrows)August9, 2001.Timesshown areLT.


Figure 7. Sequencesof two or four consecutive in-beamradarimagescomputedfrom Clemsonradardata.(top threerows)June11,2001.(fourth row) June16,2001.(bottomrow) July 15,2001.In thebottomrow, imageswith redhuescorrespondto typeI echoes.Timesshown areUT. (UT=LT+5 hours).


625 m/s

270 m/s

R G B

R G B

R G B

Clemson radar

MU radar

(type I only)

Figure 8. CorrespondencebetweenDoppler spectralbinsandimagecolorcomponentsfor theMU andClemsonradarexperiments.The respective Nyquist velocitiesfor the ex-perimentsat bothradarsaregiven.

imagepanel in Figure 6 and Figure 7 show the elevationangle of the echo in degrees. The dashedlines drawnthroughthesegraphsmark the elevation anglesanticipatedfor straightline propagationandperfectlyfield-alignedbackscat-ter. Sometimes,theplottersymbolsfall preciselyalongthedashedline. At othertimes,the measuredelevation anglesfall a few degreesoff thedashedlines. Largediscrepancies,thoughtto bemainly indicativeof refraction,aremorecom-mon in theClemsonthantheMU radardata.TheClemsonradaroperatesata lowerfrequency andalowerelevationan-gle thantheMU radar, andsignalsfrom it aremorepronetoexperiencetheeffectsof refraction(seebelow).

3.1. Type II echoes

The imageson the top row of Figure6 typify the radarimagingdataset.In them,echoesareseento arisefrom alocalizedregion of space.The region is slightly elongated,beingabout10 km deepin rangebut narrower in azimuth.Moreover, the Doppler shifts of the echoesare inhomoge-neousand vary acrossthe scatteringregion. The Dopplershiftsaresmall at theedgesof thescatteringregion but aredistinctly red-shiftednearthe azimuthalcenter. Theseim-agessuggestcirculation within the scatteringregion, withelectronsrecedingrapidly from the radarin the centerbutnot at theperiphery. Localizedscatteringregionswith tell-talesignsof internalcirculationarecharacteristicfeaturesofbothradardatasets.

Over time, the scatteringregion depictedin the top row

of Figure6 drifted westward (toward the left of the image)while nearlymaintainingits shape.In fact, all of the scat-terersseenby the MU radaron August8 drifted westwardandhadonly verysmallmeridionaldrift rates.Thattheechosignaturesin theRTI plot in Figure2 areamorphousratherthan striatedcan be attributed to the predominatelyzonalmotionsof theunderlying,localizedscattererswhichbehavelikepoint targetsratherthanvolumescatter.

Note also that the backscatterin this casearrived fromelevationanglescloseto whatwould beanticipatedfor per-fectly field-alignedbackscatter, asdesignatedby thedashedlinesin thesmallgraphsto theright of theimagepanels.Infact,closeinspectionshows that themeasuredelevationan-gles fall just above the dashedline on the nearsideof thescatteringregionandjust below it on thefar side.Themea-suredelevation anglesimply that the backscatterin rangesbetween170 and180 km actuallyoriginatedat a commonaltitudeasif from a thin, horizontal,patchylayer. Exami-nationof theentireimagingdatasetrevealsthat it is typicalfor echoesfrom range-extendedscatteringregionsto arrivefrom asinglealtitudeandfor thataltitudeto remainapprox-imatelyconstantasthescatteringregionsdrift.

The secondand third rows of imagesin Figure 6 showhow multiple localizedscatteringregionscancoexist withinthe radarilluminatedvolume. Thoughsmaller, the scatter-ing regionsin thesepanelsmainly sharethe characteristicsof theonediscussedabove.Themaindifferenceis thatthesescatterers,observedon August9, drifted rapidly towardthesouthwest.Thenumerousstriationsevident in Figure1 be-tween 2030 and 2200 LT are manifestationsof multiple,sparselypackedscatteringregionsclosingon theMU radarasthey movedsouthwestward throughthe illuminatedvol-ume.Thecloselyspacedscatteringregionsat 180km rangeandbeyondin thesecondrow of imagesin Figure6, for ex-ample,were responsiblefor the “short period” QP echoesapparentjustprior to 2100in theRTI map.

Whereasall of the scatteringregions discussedthusfarcanbe regardedas localizedif elongatedblobs, the fourthrow of imagesin Figure 6 depictsa qualitatively differentsituation.Here,anextremelyelongatedribbonof backscat-ter is evident. The ribbon extendsfrom 0� azimuthat 165km rangeto thewestwardedgeof theimagingfield of viewat 175 km. Becauseof azimuth angle aliasing, the rib-bonre-emergeson theeastwardedgeof the imagesandex-tendsa few degreesfurtherwestwardat greaterandgreaterrange.Over time,theribboncanbeseento propagatesouth-westward in the direction perpendicularto the front. Thescatteringintensityalongtheribbonis modulated,andeachbright spotalongit produceda striatedsignaturein theRTImapin Figure1. TheDopplercharacteristicsof theribbonsarecomplicatedbut suggestredshifts in thebodyof there-


gionsandblue-shiftsor smallDopplershiftson theperiph-ery.

Finally, moreexamplesof localizedscatteringregionsareshown in the bottomrow of Figure6. (Note: the echobe-tween160and165km is a meteorechoandlies outsidetheboundsof this discussion.)Theseregionsareelongatedanddo not sharea commonorientation.Over time, they driftedtowardthesouthwestand,consequently, left striationsin theRTI mapin Figure1.

The Clemsonradarimagesshown in Figure 7 are sub-stantiallysimilar to the MU radarimages,allowing for theinferior rangeresolutionof thedatafrom theformer. Thetoptwo rowsof Figure7 show imagescorrespondingto someofthe striationsapparentin Figure3. Eachof thosestriationsarosefrom a localized,elongated,southward drifting scat-teringregion. Theirregularitiesindicatedin theimageshereoccupy aregionabout10km deepin rangeandabout1� widein azimuth.As with theMU radarimages,thescatteringre-gionsshown herehave inhomogeneousDopplershifts andtend to be red shifted in the centerandblue shiftedat theedges.

The third row of Figure7 presentsimagesfrom later inthe eveningon June11, 2001. By this time, the scatteringregionshadbegunto drift predominantlywestward,exhibit-ing only smallmeridionaldrifts of eithersign. Thenumer-ousscatteringregionsevidentin theimageswereresponsiblefor theblob-like featuresin thecorrespondingRTI diagram.Thefourthrow of Figure7,meanwhile,showsanexampleofavery longribbonof backscatterseenby theClemsonradar.This ribbonextendedabout30 and70 km in themeridionalandzonaldirections,respectively, andwaslessthan10 kmwide at any point. The Doppler shifts of the echoeswereuniformly smallin thiscase.Liketheother, thisribbonprop-agatedsouthwestwardovertimein adirectionnormalto thefront.

Animatedsequencesof imagesfrom all of the radarob-servationsdemonstratethatboththestriationsandtheamor-phousblobs in the radarRTI mapscanbe associatedwithlocalized,long-livedscatteringregionsdrifting throughtheradarilluminatedvolume.Striationsresultwhentheregionsdrift in the direction of the radar. Negative and positiveQPechostriationslopescorrespondto southwardandnorth-warddrifting scatterers,respectively. Mostoften,thescatter-ing regionsareobserveddrifting southwestbeforemidnight.However, thefield alignednatureof thescatteringirregular-ities must imposeboundarieson the scatteringvolumeandlimit therangesfrom whichechoescanbereceived.We caninvestigatetheseboundariesby examiningthezenithanglesof theQPechoes.

Figure 9 shows examplesof Clemsonradar data fromJune11, 2001, and MU radardatafrom August 9, 2001,

duringperiodswhenmultiple quasiperiodicechoeswerere-ceived. The echoesin questionarrivedfrom small azimuthangles.Theechoeshave beensortedby rangeandelevationangleandplottedin a grayscaleformatrepresentingtherel-ative signal-to-noiselevel on a dB scale.Eachgroupingofplottedpixelscorrespondsto echoesfrom anindividual QPstriationaccumulatedover time. Thedashedlinesin thefig-urespredicttherelationshipbetweenrangeandelevationan-gle for echoesoriginatingat a fixedaltitude.Thesolid linesrepresentthelocusof perpendicularity. Theevidenceof thefigure is that thescatterersin questionarebetterorderedbyaltitudethanby magneticaspectangle.

In thecaseof theClemsonradarobservations,theechoesonly spannedoneor two rangegatesat any giventime, andtheir rangefollowedcurvesof constantaltitudeasthey ap-proachedtheradar. In thecaseof thehigherresolutionMUradardata,the echoesalsooccupiedseveral adjacentrangegatesin every integrationtime. Even so, all of the echoesassociatedwith eachgivenstriationarrivedfrom a commonaltitude.Evidently, thescatteringregionsresponsiblefor theechoesarenot only localizedin rangeandazimuthbut alsoin altitude,andthey maintainaltitudeasthey drift. Mean-while, theelevationanglesof thebackscatterdetectedby theMU and Clemsonradarscan be seenhereto departfromwhatis expectedfor field-alignedbackscatterby asmuchasabout1.5� and2� , respectively. Similar observationswerepresentedby Yamamotoet al. [1994].

Using radarinterferometryto probethe equatorialelec-trojet at Jicamarca,KudekiandFarley [1989] measuredas-pectanglehalf widthsdecreasinglinearlywith altitudefromabout0.35� at98km to about0.1� at114km. Repeatingtheiranalysison theMU radardatashown in Figure9, we mea-sureaspectanglehalf widths betweenabout0.25� – 1.0� .While the smallestof theseanglesare roughly consistentwith theJicamarcaresultsfor thealtitudesindicatedin Fig-ure9, the largest,which weremeasuredin rangegatescon-tainingmultiple targets,reflectthefact thatthosetargetsre-sideat differentaltitudes.Moreover, asFigure9 shows, theintensitiesof the echoesfrom the targetspeakat elevationanglesotherthanthepresumptiveloci of perpendicularity.

3.2. Type I echoes

TypeI echoesarearelatively rarephenomenonatmidlat-itudesandwereobservedby theClemsonradaron a singleoccasionin 2001on July 15, 2001for an interval of about2 minutesbeginning at about0348 UT in rangesbetween280and295km. Thespectraof theechoeswerenarrow andhadDopplershiftsof about250m/swith thesenseof prop-agationaway from theradar. TypeII echoeswereobservedsimultaneouslybut at closerrange.Imagesof all theechoesare presentedin the bottom row of Figure 7. Thesewere


June 11, 2001 0150-0205 UT August 9, 2001 2112-2114 LT August 9, 2001 2116-2117 LT

Figure 9. Comparisonbetweenechorangesandelevationanglesfor groupingsof scatterers.Grayscalesindicaterelativeechointensityon a dB scale.Horizontalcurvestracetheanticipatedrelationshipbetweenrangeandelevationanglefor scatterersat the altitudesindicated. Vertical curvesrepresentthe locusof perpendicularityfor field-alignedbackscatter. (left panel)Clemsonradardata.(centerandright panels)MU radardata.

calculatedin thesamemannerastheotherimageswith oneexception;theDopplerspectralbinsutilized wereshiftedasshown in thekey in Figure8. Shifting thebins this way in-cludedthe type I echoesin the red imagecomponentandalsoexcludedsomeinterferencepresentat positive Dopplerfrequenciesat thetime in question.

Thestrongechoesin theimagesat rangesbetweenabout260and275km have typeII spectra.Thescatteringregionthat gave rise to the echoesis very elongatedand is tiltedwestward.TheDopplerspectraof theechoesfrom thehori-zontalcenterof theregionarebroadandredshiftedwhile theechoesfrom the peripheryhave narrow spectrawith smallDopplershifts.Again,theelongatedscatteringregionshowssignsof internalcirculation.Theregionsfrom wheretypeIechoeswerereceived, meanwhile,resolve aspoint targets.While they resideat higheraltitudes,they fall alongthelineof the elongated,tilted scatteringregion. All of the echoesarrivedfrom elevationanglescloseto thepresumptivelocusof perpendicularity. Notethat theweak,greenvertical linesat thecenterof theimagesareartifactsfrom therangeside-lobesof thecodedradarpulseandshouldbeignored.

4. Analysis

ConventionalradarRTI analysismay not be sufficientlyincisive to differentiatebetweencompetingtheoriesof mid-latitudeE regionplasmairregularities.However, someof theambiguityinherentin coherentscatterradarexperimentscanbemitigatedwith in-beamradarimaging.Theprocessesre-sponsiblefor QPechoesbecomeclearerwhenradarimagesareinterpretedin light of thetheoreticalframework outlined

in paper1 andof findingsfrom otherremotesensingandinsitu experiments.

The electrodynamicconsequencesof ionizationpatchesor cloudsin the midlatitudeE region were investigatedinpaper1. SuchpatchysporadicE layershave beenseenatArecibo by Miller and Smith[1978] andSmithand Miller[1980] andhave alsobeeninferredfrom radioscintillations[Bowman, 1989;Maruyama, 1991]. In paper1, we showedthat the clouds can develop very large polarizationelec-tric fields if elongatedandthat strongcurrentscanflow inthe cloudsanremainsolenoidalby closingin theF region.Driversfor thepolarizationincludethebackgroundhorizon-tal electricfield imposedat night by the F region dynamoalongwith horizontalE region winds. Climatologically, themidlatitudemeridionalelectricfield is expectedto besouth-wardandhaveanamplitudeof theorderof 1 mV/m atnightwhile the zonalelectricfield is expectedto be significantlysmaller[Richmondetal., 1980].Thehorizontalwindsin thelower thermosphere,meanwhile,exhibit considerablevari-ability, asdemonstratedrecentlybyLarsen[2002]. TheHor-izontalWind Model (HWM) indicatesa broadtendency forthehorizontalwindsin themidlatitudeE region to besouthwestwardin theearlyevening,shifting to northeastwardbymidnight [Hedin et al., 1996]. In the premidnightsector,therefore,we would expectE region plasmacloudsto driftsouthwestward and to experiencea predominantlysouth-ward or south eastward backgroundelectric field in theirframeof reference,giving riseto a strongeastwardor northeastwardpolarizationelectricfield within. Polarizationfieldstrengthof up to 10–20mV/m areconsistentwith modelre-sultsandhavebeendetectedin ground-basedandin situex-


periments[Pfaff etal., 1998;HaldoupisandSchlegel, 1994].Theconjunctionof plasmadensityinhomogeneities,strongelectricfields,andstrongcurrentsin plasmacloudssuggestsan associationwith plasmainstabilities,plasmaturbulence,small-scaleplasmairregularities,andcoherentscatter.

Becauseof theresemblanceof midlatitudecoherentscat-ter spectrato typeII spectraobservedin theequatorialelec-trojet, gradientdrift instabilitiesarecommonlyassumedtobe presentin the sporadicE layers. Rosado-Romanet al.[1999] showedhow intermediate-scale(tensof meters)gra-dientdrift modeswith finite parallelwavenumbercanbeun-stablein the midlatitudeE region. In paper1, it wasalsoshown that large-scale( 1 km) collisional drift modesin-volving finite parallelwavenumberscanalsobeunstableinE regionplasmaclouds.Evidencethatthelatterarefoundinnaturewas provided recentlyby Kelley et al. [1995] whoconducteda rocket experimentin a volume in which co-herentechoesweresimultaneouslydetectedby radar. Therocket payloaddetectedclearsignaturesof horizontal,kilo-metricplasmadensitystructuringin a plasmacloudsituatedaboveasporadicE layerin whichintermediate-scaleplasmawaveswerealsoembedded.

Whetherthe primary waves in questionare kilometriccollisional drift waves or intermediate-scalegradientdriftwaves,we canassumethatsmall-scaleplasmairregularitiesdetectedby radarsarisefromthemthroughtheturbulentpro-cessesdescribedby Sudan[1983]. Thephasevelocity, �� ,for thesmall-scalewavesis generallytaken to be governedby � ��! "$#&% # �'�(��

(1)

where%

is the anisotropy factor (the ratio of the electronto ion transversemobility), is the frequency, is thewavenumber, and

� ��and

� ��aretheelectronandion drift

velocities,respectively. While thisexpressionis strictly truefor primarygradientdrift andFarley Bunemanwaves,expe-riencehasshown that it describesthepropagationof small-scaleirregularitiesin the equatorialelectrojetas well (seeFejer andKelley [1980] for example).Theelectrondrift ve-locity in (1) is approximatelythe )+*-, drift speedwhichreflectsthe effects if the backgroundelectricfield, the po-larizationelectricfield that may arisein the plasmacloud,and the polarizationelectric field associatedwith the pri-mary waves. The ion drift velocity, meanwhile,is mainlycontrolledby the neutralwind. The simulationstudiesinpaper1 indicatethat,of these,thepolarizationelectricfieldsetup by theplasmacloud itself is likely to dominateat al-titudeswherethe

%factor is small. The polarizationfield

setup by primary wavesmay well becomevery strongun-der somecircumstances,but its rapid variation in azimuth

will generallybebeyondtheabilitiesof theradarto resolve.Neutralwind effectsmaydominate(1) at altitudeswhere

%is comparableto unity andbelow, but thisdomainis asourceof only asmallminority of theechoesweencountered.Mostof the scatterersobservedoccupiedaltitudesabove 100kmwhere

%is muchlessthanunity.

Thecombinationof a southwardmeridionalelectricfieldanda westward wind will setup polarizationelectricfieldsandelectroncirculationin E regionplasmacloudslike thoseshown in Figure3 of paper1. There,theelectronstreamlinesindicaterapidnorthwarddrifts in thecenterof thecloudsandslowersouthwarddrifts onthecloudedges,in therestframeof thecloud.In astationaryframe,theelectronmotionat thecloudedgescouldappearto benorthwardor southwardde-pendingon themeridionalneutralwind speed.To theextentthatsmall-scalewavesareavailableastracersthroughoutthebodyof thecloud,significantlyred-shiftedechoesshouldbeobserved from the centerof the cloudsin accordancewith(1) whereasrelatively small Doppler shifts shouldbe ob-servedat thecloudperipheries.SuchcharacteristicDopplershift patternsappearthroughoutour imagingdata,andwethereforeassociatetheelongatedscatteringregionswith di-mensionsof a few kilometersin theimageswith theplasmacloudssimulatedin paper1.

It is importantto recognize,however, thatnothingguar-anteesthat small-scaleirregularities will be presentuni-formly throughoutthe plasmacloud to “paint” the entirecirculation cell for the radar. Just as the clouds are dis-tributedsparselythroughoutthe radarilluminatedvolume,the irregularitiesaredistributedinhomogeneouslythrough-out the cloud, and only part of the circulation is likely tobevisible at any giventime. Often,however, enoughof thecirculationis visible to make it recognizable.

4.1. Refraction

The finite aspectanglesensitivity of field alignedirreg-ularitiesmakesit possiblefor themto be observed for sig-nificant periodsof time asthey drift. Hysell and Burcham[2000]calculatedthatirregularitieswith anaspectanglehalfwidth of 0.5� couldbedetectedby theClemsonradarovera50km spanof ranges,thelengthof thelongestQPechostri-ation ever seenat Clemson.However, Figure9 shows thatthe coherentechoesarenot only spreadin magneticaspectanglebeyondwhatis expectedonthebasisof experimentsatJicamarca,but alsothatthey canbemostintenseatelevationangleswell off the presumptive locus of perpendicularity.We attributethesephenomenamainly to refractionandseekto quantifytheeffectsbelow.

We work in the domainof geometricopticsandneglectbirefringence,taking the radarfrequency to be well abovethecritical frequency of theplasma.Considerthecaseof a


Figure 10. Pathsof rayspenetratinga plasmacloudwith aGaussianspheroiddensitydistributionwith . =0.5.Theaxesrepresentnondimensionaldistances.Thecritical frequencyis closeto theradarfrequency in thiscase,which is exagger-atedfor demonstrationpurposes.

sphericallysymmetricplasmacloud with densitygiven by/=/ �103254 �6�17(8 ��9�. 8� where

7is theradialdistancefrom

the center. It is well known that, in the caseof a mediumwith sphericalsymmetry, the ray characteristicin polar co-ordinatesis governedby theordinarydifferentialequation��7��: � 7 ; < =?> 7 > � ; > (2)

where=

is theindex of refractionand ; is anintegrationcon-stant(e.g. Born andWolf [1964]). We canintegrate(2) nu-mericallyusingthe specifiedplasmacloudmodelandkeeptrackof thecharacteristicsof raysaimedat thecloudfrom agreatdistance.Integrationis performedhereusinga predic-tor correctormethod.

An examplecalculationis picturedin Figure10. In thisfigure,we mayregardtheraysashaving enteredtheregionfrom theright in parallelasif from a distantsource.Whenthe raysexit the cloudon the left, they do sowith a spreadof trajectories.This spreadmaypermittheBraggscatteringconditionto be met at variousplacesinsidethe cloud evenif theincidentraysarenot orthogonalto themagneticfield.We haveperformednumerouscalculationsof this kind for arangeof radarfrequencies@ , peakplasmadensities

/ � , andordersA . It wasfoundthat themaximumangulardeflectionof theraysfrom thehorizontal( B ) in eachcaseis predictedby thesimpleempiricallaw:B � 9DC�E�F@HG > / �" F�I �KJML > A #N"

(3)

wheretheunitsof B , @ , and/ � aredegrees,MHz, andccO?P ,

respectively. This formulaassumesthat theradarfrequencyis muchgreaterthanthepeakcritical frequency andis accu-ratetowithin afew percentoverabroadrangeof parameters.

Notethattheproblemis scaleinvariantsuchthat . doesnotenterinto (3). Themaximumdeviation angle B providesanindicationof theinfluenceirregular, patchysporadicE layersarelikely to have on themeasuredelevationanglesof field-alignedbackscatterfrom irregularitiesinsidethelayers.

Ionogramstaken throughoutthe evening of August 9,2001,at the MU radarshowed sporadicE layerswith topfrequenciesbetween10and13MHz. Theblankingfrequen-cies, meanwhile,were only about2 MHz, supportingthenotion that the layerswerepatchyor cloud like. AlthoughMU radarandionosondedo not probea commonvolume,we may take the vertical soundingsasan indicationof theenvironmentfrom which thequasiperiodicechoesemerged.Accordingto (3), sphericallysymmetricplasmacloudswithGaussiandistribution functions(A = 2) andwith peakden-sitiesof just

" * " F I1QSR OUT couldgive rise to deflectionsat46.5 MHz of up to 1.66� . This angleis closeto the max-imum departurefrom the locusof perpendicularityfor theMU radarbackscatteranalyzedin Figure 9. Cloudswithsteeperboundariescouldcauseevengreaterdeflection,andthedeflectionsassociatedwith Clemsonradarsignalsshouldbecorrespondinglylarger. With

/ � = VXWZY[* " F�\ Q3R O]T andA= 2, (3) predictsa maximumdeflectionof 1.8� for Clemsonradarsignals,in closeagreementwith themaximumdepar-turesfrom perpendicularityapparentin Figure9. Althougha systematicstudyof the plasmadistribution within patchysporadicE layersremainsto be performed,plasmacloudswith ratherdiffuseboundariesandwith peakdensitiescon-sistentwith ionosphericsoundingscanaccountfor therangeof magneticaspectanglesobservedin thecoherentbackscat-ter. Thisphenomenonallowsthelocalizedscatteringregionsunderlyingquasiperiodicechoesto betrackedby theradarasthey drift overconsiderabledistances.

Allowing thatrefractioncanaccountfor ^ 1.5� deviationsin the radarray pathsand taking 0.25� as a representativevalueof theaspectanglehalf widthsof thefield alignedir-regularities,we canexpectcoherentechoesto be receivedfrom anglesas far as ^ 2.5� from the locus of perpendicu-larity while suffering lessthan35 dB of attenuationat theextremes.Striationsassociatedwith suchechoescouldhavedetectablerangeextentsasgreatasabout20 km in typicalMU radarRTI plots, the approximatelengthof the longeststriationsvisible in Figure1 andFigure2. This figure be-comesabout60 km for theClemsonradar. Echoesreceivedfrom large azimuthanglescould have considerablylongerrangeextentsthanthese,dependingon thescatteringgeom-etry. Echoesreceivedthroughsidelobeswouldhaveparticu-larly largerangesandrangeextents.


4.2. Continuous bands (ribbons)

In theMU andClemsonradarimages,wefoundexamplesof extremely long bandsor ribbonsof backscatter. Theserepresenttheminority of thescatterersobservedatbothsitesbut werenotedto occurseveral timesin our dataset.Sincethe ribbonsare modulatedin intensity along their length,they are not easily distinguishedin RTI mapsfrom other,morecommonkindsof scatterersandleavestriationsassig-natures.Theribbonshaveall beenobservedpropagatingto-wardthesouthwest.

A possibleinterpretationfor theribbonsis that they rep-resentdiscontinuitiesin ordinary, planarsporadicE layers.In paper1, it was found that ripple discontinuitiesin ex-tendedlayerscanbecomepolarizedin abackgroundelectricfield, althoughthe resultingpolarizationfield is not neces-sarily very intense. How sucha ripple would be inducedin a sporadicE layer is unknown. The morphologyof theribbons is reminiscentof the ripples in the total electroncontentobservedby Saitoet al. [1998,2001]with the GSInetwork of global positioning satellite receivers in Japan(GEONET)duringthepassageof traveling ionosphericdis-turbances(TIDs). We mayhypothesizethattheribbonscat-terersreflecttheE regionresponseto electrifiedTIDs propa-gatingin theF regionacrosscommonmagneticfield lines.Asimilar hypothesiswasadvancedrecentlyin detailby Tsun-odaandCosgrove[2001]. In thefuture,radarimagescanbecomparedwith imagesfrom theGEONETnetwork to inves-tigatecommonalitiesin thetwo phenomena.

4.3. Type I echoes

Thedispersionof Farley Bunemanwaves,of which typeI echoesarea signature,is governedby (1) alongwith thefollowing growth rateexpression:_ � % ��` �"$#&%ba � ��(� �� > � >(cd>e�f � 9�g = � (4)

where ` � is theion-neutralcollision frequency,c e is theion

acousticspeed,and g is the recombinationrate(e.g. Fejeretal. [1975].) Whencombinedwith (1), (4) impliesathresh-old velocity for theelectronsbelow which wavescannotbeexcited. Neglectingrecombination,the thresholdconditionis h ��ji h �� # c e � "k#&% (5)

wherethescalardrifts referto thecomponentof theelectronand ions in the direction of wave propagation. Favorableconditionsfor the generationof Farley Bunemanwavesin-cludelow temperaturesandmetallicion composition,whichreducethe ion acousticspeed,along with winds and iondrifts in the directionoppositethe electrondrifts. Finally,

the anisotropy factor is much lessthan unity above about100 km but increasesrapidly below, drasticallyincreasingthe electrondrift thresholdthereandmakingFarley Bune-manwavesmoredifficult to excite.

The images in the bottom row of Figure 7 depict ahighly elongatedscatteringregion betweenabout260–270km rangeshowing all thesignsof beingstronglypolarized.Theelevationanglesof this type II echoregion areclosetothelocusof perpendicularity, suggestingminimal refraction,but arelargerat thenearedgethanat thefar edge,implyingthat theelongatedscatteringregion is a horizontallayer. Atrangesbeyond270km, thereis a regiondevoid of backscat-terandthentwo compactregionswith typeI echoesthatfallalongthe line of the elongatedscatteringregion. The alti-tudesof the type II and type I echoregionsareabout100and110km, respectively.

That all the scatteringregions in the radar imagesarecollinearsuggeststhatthey arerelatedeventhoughthetypeI andII echoregionsareseparatedby 10 km in altitude.Wemaysurmisethattheelongatedplasmalayerresponsibleforthe type II echoesextendedto rangesgreaterthan270 kmbut that the Braggscatteringcondition was unmetbeyond270 km, and so no echoeswere received. Assumingthatstrongpolarizationelectricfieldswerepresentin thelayeratmoredistantranges,thesefieldswould have mappedalongmagneticfield lines to higheraltitudesat which the Braggscatteringconditionwasonceagainmet.At 110km altitude,themappedpolarizationelectricfieldcouldhaveexcitedFar-ley Bunemaninstabilitiesandgivenriseto thetypeI echoes,providedthethresholdconditionwasmetat thehigheralti-tude. Why typeI echoesappearedat thehigheraltitudebutnot thelower in thatcaseis unclear. Verticalgradientsin thewinds, composition,and the

%factorpresumablyreduced

theinstability thresholdin theupperlayerto thepointwhereinstabilitiescouldbeexcited.

5. Summary and Conclusions

We have investigatedthe scatteringregions underlyingQPechoesusingin-beamradarimaging.Theseregionsturnout to be spatially localized,to be horizontally elongated,and to maintainaltitude as they drift throughthe radaril-luminatedvolume. The trademarkquasiperiodicstriationsin radarRTI plots are the trails of the drifting scatterers,like thetrails of starsleft in long-durationphotographicex-posures.Refractioncanpermit coherentscatterfrom fieldalignedirregularitieswithin the regionsto be detectedoverawiderangeof elevationangles.Theclimatologyof thedy-namoelectricfield andtheneutralwinds in themidlatitudelower thermosphereis consistentwith the mostcommonlyobservedsignsof theechorangeratesandDopplershifts.


Plasmasarechargeneutralby definition,andtheassem-blageof free charge in a plasmais energeticallyexpensive.However, a dielectric plasmaresponseleading to the cre-ation of inducedelectric dipolesis energetically favorablein a plasmain an appliedfield anddoesnot violate chargeneutrality. Effective charge appearswherever the polariza-tion (induceddipole momentper unit volume) is divergentandtendsto concentratenearthe boundariesor surfacesofinhomogeneousplasmastructures.Thepolarizationelectricfields thatarisearedipolaror higherorder. Electrons,withstreamlinesthat follow the equipotentials,will circulateina conspicuousway when drifting throughinhomogeneousplasmas.

Circulationhasbeenuncoveredusingradarimaging.Weinterpret the circulation as evidencethat the scatteringre-gions coincide with elongated,polarized plasmaclouds.Such clouds have been observed in the ionosphereoverAreciboandhave beendetectedin otherin situ andremotesensingexperiments. Simulationsshow that polarizationelectric fields and Hall drifts large enoughto drive FarleyBunemaninstabilitiesandproducetype I echoescanarisein particularly elongatedcloudscoupledelectrically to theF region. Thetype I echoesdetectedby theClemsonradaroccurredin the vicinity of a highly elongatedscatteringre-gion. Simulationsalso show that plasmaclouds are un-stableto kilometer-scalecollisional drift instability modes,andwe presumeat this point that intermediate-andsmall-scaleirregularitiescan be generatedsubsequentlyby non-linear modecouplingandplasmaturbulence. Evidenceofhorizontal,kilometricwavesin patchysporadicE layersthatwereaccompaniedby intermediate-andsmall-scalewaveswaspresentedby Kelley et al. [1995].

In paper1, it was arguedthat E region plasmacloudsproduceall the phenomenanecessaryto accountfor small-scaleirregularitiesandcoherentradarbackscatterincludinglarge-scaleprimary waves, large electric fields, and largeHall currents. Nevertheless,simultaneousobservationsofE region plasmaclouds,of structurewithin thoseclouds,andof QPechoesfrom a commonvolumeremainunmade.Likewise,althoughthereis compellingevidencethatneutralshearinstabilitiesmaybe responsiblefor producingpatchysporadicE layersandclouds,directobservationsof thepro-cessat work are lacking. Beforea completepictureof theQP echophenomenoncandevelop,simultaneousmeasure-mentsof the backgroundplasmadensityandelectricfields(via incoherentscatteror rocket probes),intermediate-scaleplasmastructure(via rocket probesor radio scintillations),neutralwind profile (via incoherentscatteror chemicalre-leases),andcoherentscatter(via interferometryor imaging)will have to bemade.

Acknowledgments. Thiswork wassupportedby theNationalScienceFoundationthroughNSFgrantATM-0080338to ClemsonUniversity. SpecialthanksaredueM. F. Larsenfor helpingwiththe fielding and operationof the Clemsonradar. The MU radarbelongsto andis operatedby theRadioScienceCenterfor SpaceandAtmosphere(RASC)of KyotoUniversity. DLH wassupportedby RASCasavisiting associateprofessorduringthepreparationofthis manuscript.

References

Ables, J. G., Maximum entropy spectralanalaysis,Astron. Astro-phys.Suppl.Ser., 15, 383,1974.

Born,M., andE. Wolf, Principlesof Optics, Pergamon,New York,1964.

Bowman,G. G., Quasi-periodicscintillationsat mid-latitudesandtheir possibleassociationwith ionosphericsporadic-E struc-tures,Ann.Geophys., 7, 259,1989.

Farley, D. T., H. M. Ierkic,andB. G.Fejer, Radarinterferometry:Anew techniquefor studyingplasmaturbulencein theionosphere,J. Geophys.Res., 86, 1467,1981.

Fejer, B. G.,andM. C.Kelley, Ionosphericirregularities,Rev. Geo-phys., 18, 401,1980.

Fejer, B. G.,D. T. Farley, B. B. Balsley, andR. F. Woodman,Verti-cal structureof theVHF backscatteringregion in theequatorialelectrojetandthegradientdrift instability, J. Geophys.Res., 80,1313,1975.

Haldoupis,C., andK. Schlegel, Observation of themodifiedtwo-streamplasmainstabilityin themidlatitudeE regionionosphere,J. Geophys.Res., 99, 6219,1994.

Haldoupis,C., K. Schlegel, andD. T. Farley, An explanationfortype 1 radarechoesfrom the midlatitudeE-region ionosphere,Geophys.Res.Lett., 23, 97,1996.

Hedin, A. E., et al., Empirical wind modelfor the upper, middle,andlower atmosphere,J. Atmos.Terr. Phys., 58, 1421,1996.

Hysell, D. L., andJ. Burcham,The 30 MHz radarinterferometerstudiesof midlatitudeE region irregularities,J. Geophys.Res.,105, 12,797,2000.

Hysell, D. L., andR. F. Woodman,Imagingcoherentbackscatterradarobservationsof topsideequatorialspreadF, Radio Sci.,32, 2309,1997.

Hysell,D. L., J.L. Chau,andC.G. Fesen,Effectsof largehorizon-tal windson theequatorialelectrojet,J. Geophys.Res., 2002,inpress.

Jaynes,E.T., Ontherationaleof maximum-entropy methods,Proc.IEEE, 70, 939,1982.

Kagan,L. M., andM. C.Kelley, A wind-drivengradientdrift mech-anismfor midlatitudeE-region ionosphericirregularities,Geo-phys.Res.Lett., 25, 4141,1998.

Kagan,L. M., andM. C. Kelley, A thermalmechanismfor gener-ationof small-scaleirregularitiesin theionosphericE-region,J.Geophys.Res., 105, 5291,2000.

Kelley, M. C.,D. Riggin,R.F. Pfaff, W. E.Swartz,J.F. Providakes,and C. S. Huang,Large amplitudequasi-periodicfluctuationsassociatedwith a midlatitudesporadicE layer, J. Atmos.Terr.Phys., 57, 1165,1995.


Kudeki,E., andD. T. Farley, Aspectsensitivity of equatorialelec-trojet irregularities and theoretical implications, J. Geophys.Res., 94, 426,1989.

Kudeki,E., andF. Surucu, Radarinterferometricimagingof field-alignedplasmairregularitiesin the equatorialelectrojet,Geo-phys.Res.Lett., 18, 41,1991.

Larsen,M. F., A shearinstability seedingmechanismfor quasi-periodicradarechoes,J. Geophys.Res., 105, 24,931,2000.

Larsen,M. F., Windsandshearsin themesosphereandlower ther-mosphere:resultsfrom four decadesof chemicalreleasewindmeasurements,J. Geophys.Res., 2002,in press.

Maruyama,T., Observations of quasi-periodicscintillations andtheirpossiblerelationto thedynamicsof Esplasmablobs,RadioSci., 26, 691,1991.

Maruyama,T., S. Fukao, and M. Yamamoto,A possiblemech-anismfor echo-striationgenerationof radarbackscatterfrommidlatitudesporadicE, RadioSci., 35, 1155,2000.

Miller, K. L., andL. G. Smith, Incoherentscatterradarobserva-tionsof irregularstructurein mid-latitudesporadicE layers,J.Geophys.Res., 83, 3761,1978.

Pfaff, R. F., M. Yamamoto,P. Marionnia,H. Mori, andS. Fukao,Electric field measurementsabove and within a sporadic-Elayer, Geophys.Res.Lett., 25, 1769,1998.

Richmond,A. D., et al., An empirical model of quiet-dayiono-sphericelectricfields at middle andlow latitudes,J. Geophys.Res., 85, 4658,1980.

Rosado-Roman,J.M., W. E. Swartz,andD. T. Farley, Radarstud-iesin PuertoRico of mid-latitudeE-region plasmainstabilities,paperpresentedat theXXVI URSIGeneralAssembly, Toronto,Canada,August13-21,1999.

Saito,A., S. Fukao,andS. Miyazaki, High resolutionmappingofTEC perturbationswith theGSI GPSnetwark over Japan,Geo-phys.Res.Lett., 25, 3079,1998.

Saito,A., et al., Traveling ionosphericdisturbancesdetectedin theFRONT campaign,Geophys.Res.Lett., 28, 689,2001.

Shalimov, S., C. Haldoupis,and K. Schlegel, Large polarizationelectric fields associatedwith midlatitudesporadicE, J. Geo-phys.Res., 103, 11,617,1998.

Skilling, J.,andR. K. Bryan,Maximumentropy imagereconstruc-tion: Generalalgorithm,Mon. Not. R. Astron. Soc., 211, 111,1984.

Smith,L. G., andK. L. Miller, Sporadic-layersandunstablewindshears,J. Atmos.Terr. Phys., 42, 45,1980.

Sudan,R. N., Unified theoryof type I andtype II irregularitiesintheequatorialelectrojet,J. Geophys.Res., 88, 4853,1983.

Thompson,A. R., Interferometryand Synthesisin RadioAstron-omy, JohnWiley, New York, 1986.

Tsunoda, R. T., On polarized frontal structures, type-1 andquasiperiodicechoesin midlatitudesporadicE, Geophys.Res.Lett., 25, 2641,1998.

Tsunoda,R. T., andR. B. Cosgrove, Coupledelectrodynamicsinthenighttimemidlatitudeionosphere,Geophys.Res.Lett., 2001,in press.

Tsunoda,R. T., S. Fukao, and M. Yamamoto,On the origin ofquasi-periodicbackscatterfrom sporadicE, RadioSci., 29, 349,1994.

Woodman,R. F., Coherentradarimaging: Signalprocessingandstatisticalproperties,RadioSci., 32, 2373,1997.

Woodman,R. F., M. Yamamoto,andS.Fukao,Gravity wave mod-ulation of gradientdrift instabilitiesin mid-latitudesporadicEirregularities,Geophys.Res.Lett., 18, 1197,1991.

Yamamoto,M., S. Fukao,R. F. Woodman,T. Ogawa, T. Tsuda,andK. Kato, Mid-latitude E region field-alignedirregularitiesobservedwith theMU radar, J. Geophys.Res., 96, 15,943,1991.

Yamamoto,M., S.Fukao,T. Ogawa,T. Tsuda,andS.Kato,A mor-phologicalstudyof mid-latitudeE-region field-alignedirregu-larities observed with the MU radar, J. Atmos.Terr. Phys., 54,769,1992.

Yamamoto, M., N. Komoda, S. Fukao, R. T. Tsunoda, andT. Ogawa,Spatialstructureof theE region field-alignedirregu-laritiesrevealedby theMU radar, RadioSci., 29, 337,1994.

ReceivedX; revisedX; acceptedX.

This preprintwaspreparedwith AGU’s LATEX macrosv5.01,with theextensionpackage‘AGU lXl ’ byP. W. Daly, version1.6bfrom1999/08/19.

imaging radar observations and theory of type i and type ii...

Documents