The Polarization of Auroral Radio Emissions - Semantic Scholar
GEOPHYSICAL RESEARCH LETTERS, VOL. 24, NO. 24, PAGES 3161-3164, DECEMBER 15, 1997
The polarization of auroral radio emissions S. G. Shepherd,J. LaBelle, and M. L. Trimpi Department of Physicsand Astronomy,Darmouth College,Hanover, New Hampshire
To determine the polarization of various radio emisAbstract. Ground level observationsusing two vertical loop antennasoriented at 900 to each other reveal sions crossedloop antennas and associatedelectronics the senseof polarization of severaltypes of auroral ra- were installed at Churchill, Manitoba. The first obserdio emissionsin the frequencyrange 30-5000 kHz. Au- vations with this system are describedbelow. roral hiss is observedto be right elliptically polarized (RP) with respectto thelocalmagneticfield,consistent Instrumentation with theoretical expectationfor the whistler mode and with earlier measurements. Two less well-understood For several years auroral radio emissionshave been auroral emissions,are found to be left elliptically polar- continuouslymonitored at various northern and southized (LP). This polarizationi• inconsistentwith their ern hemispheresitesusinga programmablesteppedfregenerationin the X-mode as suggestedby sometheor- quencyreceiver(PSFR). This receiveris usuallyproies. grammed to sweep from 30 kHz to 5 MHz every 2 secondsin 10 kHz steps. Data are collected,digitized, Introduction and stored by a local computer and sent back to Dartmouth Collegemonthly. The auroral ionsphereis an abundant sourceof radio In March 1997, the PSFR at Churchill (58.76øN, emissions,someof which are detectableon the ground. 265.92øE,69.20invariantlatitude)wasmodifiedto meas-
At LF/MF/HF, theseincludeauroralhiss at 1 kHz-1 MHz [e.g.,Helliwell, 1905],MF-burst at •1.4-4.5 MHz [ Weatherwaxet al., 1994;LaBelleet al., 1997],and 2fc•
ure the polarization of received signals. The modified
antennasystemconsists of two vertical2.5 m2 loopantennasoriented 900 to eachother in an approximatelyN-
and 3fc• auroral roar at .-• 3 and .-• 4.5 MHz, respectively S/E-W position. Figure i showsa schematic of the two [Kelloggand Mortson,1979; Weatherwaxet al., 1993]. antennasand a simplifiedblock diagram of the system. The principal generationmechanismfor auroral hisshas (In the figure,the two antennasare separatedfor clarbeen described;however,the generationmechanismsof ity, but in fact they are physically located on the same MF-burst
and auroral
roar remain
unknown.
These au-
verticalmast.) The polarizationdetector,receiver,and
roral emissionsare signaturesof important phenomena computer are located .-• 100 meters from the antenna to suchas wave/particleinteractionsand energyexchange minimize noise pickup. processes. In the polarization detector, a 900 phase lag is inPolarization measurementsprovide an important clue troduced into the signal from the N-S loop. On alternabout the generationmechanismof these auroral ra- ate sweepsthis signalis inverted, effectivelyshiftingthe dio emissionsby placingconstraintson the propagation phase of the N-S loop signal from -900 to +900 relatmode in the ionosphere. For example, several research- ive to the E-W loop. The input to the receiver is the ers have suggestedthat auroral roar may be generated shifted and switched N-S loop signal summed with the in the X-mode by the cyclotron maser mechanismop- signalfrom the E-W loop. If the original signalsinduced eratingat F-regionaltitudes[Weatherwaxet al., 1995; in the antennaloopsare equal in amplitude but differ in Yoonet al., 1996].This mechanism predictsthat auroral phaseby exactly 900, as would be expectedfor vertically roar shouldbe right elliptically polarized. Another posincidentright- or left-circularlypolarizedwaves,the insible mechanism
of auroral roar involves the conversion
of upperhybridwavesto electromagnetic waves[Gough and Urban,1983; Weatherwax et al., 1995].This mechanism predicts either left-ellipticalpolarization(LP) or right-ellipticalpolarization(RP), dependingon the mode conversionprocess. It is therefore possible to eliminatesometheoriesbasedon the observationof signals propagatingin a mode which is forbiddenby that
anl•• p I
mechanism.
Copyright1997by theAmericanGeophysical Union. Papernumber97GL03160. 0094-8534/97/97GL-03160505.00
Figure 1. Schematicof polarizationdetectionelectronics. Phase shifting is representedas a 900 lag of the signal in the N-S antennaloop. In reality, the phasesof both signalsare shifted to producethe 900 lag. 3161
3162
SHEPHERD
ET AL.- POLARIZATION
put signalto the receiveralternatesbetweenzero and twice the inducedsignalstrength.On the otherhand,a linearly polarizedsignalinducesin-phasesignalsin the antennaloops which result in a constantinput signal to the receiver,there beingno differencebetweenshifting the N-S signalforwardor backwardin phasein this
OF AURORAL
EMISSIONS
Churchill, Manitoba
600 ......
March 16, 1997
..".:::: :.: ....................:..';•.•'..
....
:.....
.:...
.
..
• ........
::
:..
,..•..
.
..
400
case. To assist in data analysis, a marker is recorded
300
during part of the noninvertedsweepto allow sweep identification,and a calibrationsignalthat simulatesa receivedright-cirularlypolarizedsignalis periodically
200
inserted into the antennas. In order to determine polar-
lOO
ization usingthis technique,the measuredsignalsmust be relativelyconstantin amplitudeand polarizationduring twoconsecutive meaurements (~ 2 s); signalswhose amplitudevariesfaster than that may registera falseor indeterminant polarization. The senseof polarization (right or left) is determinedby notingthe relativesignal strengthsof the two sweepsand comparingthat to
RADIO
...................................... ""'""'-• '""•- '-"•:'••••' /:'_L.::.:7.L::.:.•,:
••.••:.::•:•._•: 0544
0546
' ...%::•....;....;;;...................`.................•...•••.•••.•...•`.."........•
•,..:.. .................. ..................................................................• .:..••.... 0548
0550
0552
0554
0556
0558
Time (U•
Figure 2. A raw spectrogram recordedby the swept frequencyreceiverduringa time whenauroralhissoccursat frequencies from 30 kHz to above500 kHz. Alternatesweepsare dark or light, showingthat the waves areellipticallypolarized.Carefulinspection revealsthat the marker. theyareright-hand polarized,asexpected ontheoretical The interpretation aboveassumesthat the electronics groundsand on the basisof previousmeasurements.
are perfect. It is difficult to shift a single signal by 900 over a wide bandwidth, but it is easy to shift both the N-S and E-W loop signalssuch that the phase of the N-S loop signallags the E-W loop signal by ~ 900 over a range of approximatelytwo decades.The error in phaseshift overthe 0.05-5.0 MHz frequencyrangeis less than ten degrees,implyingthat the maximumamplitude differencebetweenconsecutivesweepsfor either rightor left-circularlypolarizedverticallyincidentsignalsis about 20 dB which is less than the observed differences
sphere, which implies RP, consistentwith the observation. This observationof auroralhissprovidesa natural calibrationof the polarizationdetector.
Figure3a (toppanel) is a spectrogram showing three types of auroral radio emissionsrecorded 0445-0459 UT on April 4, 1997. Auroral hiss occurs below ~ 500 kHz
beginningnear 0452 UT lasting for ~ 5 minutes and again during the last ~ 30 secondsof the record. MF-
of the real signalsdescribedbelow. (~ 2 MHz) emission above~ 1.5 In this paper all wavepolarizationsare measuredwith burstis the broadband MHz and correlated with the auroral hiss. Auroral roar respectto the local magneticfield. The electricfield vec-
at tor of a right-handedcircularlypolarized(RCP) wave is therelativelynarrowband(dr ~ 200kHz) emission ~ 3 MHz beginning near 0448 UT and ending near 0453 rotates clockwise in time as viewed in the direction of UT. Horizontal dark bands are fixed-frequencytransthe magneticfield. In this case, for a receiver in the missions;the band from 550-1600 kHz is the AM broadNorthern Hemisphere the electric field vector rotates cast band. The sweepidentificationmarker is at 1-1.25 clockwiseas seenby an observerlookingdown on the MHz. Clearly, alternatesweepsare light or dark, implyantennas.This definitionis standardin plasmaphysics ing that they are elliptically polarized. Careful inspec-
[e.g.,Chen,1984]and is usedin previousauoralradio tion shows that the MF-burst and auroral roar are LP emission polarizationstudies[Tanakaet al., 1976]. and auroral hissis RP. Figure 3b (bottompanel)showsthe polarizationof
the signalsas a grayscale. In this display, white and blackpixelscorrespondto LCP and RCP wavesrespectFigure 2 showsan auroral hissevent recordedstart- ively. Elliptically polarized waves are representedas ing at 0548 UT on March 16, 1997, lasting approxim- gray pixels betweenthe two extremes,with linear poately 3 minutes,and extendingfrom the lowerboundof larizationbeingat themiddleof thegrayband(halfway Observations
the instrument(30 kHz) into the AM-broadcastband betweenwhiteandblack). As expected,the auroralhiss above 500 kHz. Dark horizontallines on this spec- showsup as dark pixels implying right-hand polarizatrogram representfixed-frequencyman-madetransmis-
tion. In contrast, both MF-burst and auroral roar are
sions. In contrastto the fixed-frequencysignals,the left-hand polarized. There are two sweepsat the onauroral hiss is clearly elliptically polarized becauseal- set of an auroralsubstormnear 0453 UT during which ternate receiver sweepsdiffer in amplitude. Careful the polarizationmeasurementmomentarilyimpliesthat inspectionrevealsthat the observeddark-lightpattern the auroral roar is RP, but at this time the auroral roar correspondsto RP. This result is consistentwith pre- amplitudeis probablyhighly time variable,as is known viousauroralhisspolarizationmeasurements usingsev- from fine structuremeasurements [e.g.,LaBelle et al., eraltechniques [Tanakaet al., 1976].Also,auroralhiss 1997;Shepherdet al., 1996],and undersuchconditions is believedto propagatein the whistlermodein the iono- the polarization measurementcannot be trusted.
SHEPHERD ET AL.' POLARIZATION
OF AURORAL RADIO EMISSIONS
Churchill• Manitoba
(a)
3163
April4, 1997
intensity (nV/m
40003500-
•'3000-
•-.•-2500-
•**?•:.•js:*-'•:*****::*s;4j:•*•'½,25' *;;5•: ':.' :::•:; '::'***'2.:L?.'•-:'•, ::{*"4:":•: 4'"'½: .:;:-4:.
:::•:*:½'*•-:*•:-•:::•$•*•**' ':½*"•* ":'
2000-
1500
....... .--..... .
500 ..................................... -?, ........................... •:;.:...;:;•::• .......................... ?•?.•;:•?;:,.,;• ...................................... -::Q':X ;•:::L.:.:'.:.. .:.'.. ,:.'.'.:. ..L,.. ,,'. ;.................... ;.............................................................. :• ....................... ::,.;;,,;;.,•; ............................. :.................... ;•E•..•:.:*•?•;•:•:;•*•;•*•.•:M:•;;•.*•;&M;•:•?;:;:X;•?::•:..•.:.;?•;•;.S::"`"•% ........................ •;•:•;;;•.•.•;•:?.•;;•.•:;a•?•`•;•:•:.:•;.•::: 12.5
polarization
4000 •, ............................................................. :.-; .................. *•;?'"' .............................................. '"': ........................... ;:•':"• ............................. '............. "':"" ........ "":" ......................................... '"? ............. ............ •...................... '"'""' ........ ":'"':' ........................................................................ ':"":'?:' right ...........
;:.x .....
•
2500 2000-
1500
.•:.,
.•..
;:;.::
.
'"•:••
linear
5oo 0446
0448
0450
0452
0454
0456
0458
Time (UT)
•o•1
•o•,
•ol•i•e•. (b) •ol•i•io• •-b•,
•o•1
of e•i•io•
•o•i•
•o•,
from strong atmospherics. An atmosphericproduces an output to the receiver which resemblesa broadband purposesof accumulating statistics,an eventis defined signal that is strongly polarized in either direction, deas an auroral emissionthat is detectedfor longer than pendingon the signof the inducedphaseshift duringthe 30 secondsand is separatedby at least 10 minutesfrom event. The impulsivenature of thesesignalsobscurethe other events. Table i showsthe senseof polarizationof polarization of coincidentwavesand illustrate the effect all eventswhichoccurredduringthe observationperiod, of a signal which varies in amplitude faster than the asdeterminedfrom spectrograms similarto thoseshown sweepperiod of the receiver. Auroral emissionswere seen on 38 of the 78 days of observation between March 15 and June 1, 1997. For
in Figure3b. Table i showsthat auroralhissis right ellipticallypolarizedand MF-burstsand 2f• auroralroar Interpretation are both left elliptically polarized. While the statistics
The experiment described above establishesthat the MF-burst and 2f• auroral roar emissionsare left-hand hand polarized, more statisticsare neededto exclude establish that these emissionsare almost entirely left-
the possibilitythat a few percentof theseemissions are polarized,in contrastto auroralhisswhichis right-hand right-handor linearlypolarized.The few eventsin the polarized. For all three emissions,the mean power differencebetweenalternatesweeps,during whichthe sigunknown column of the table are due to interference nals constructivelyand destructivelysum, averages-10 dB.
The
instrumental
error
in this measurement
due
Table 1. Polarization of auroral hiss, auroral roar, and to phase inaccuraciesin the polarization detectionelecMF-burst events from Churchill, Manitoba, between tronics is the order of-20 dB. The departurefrom circuMarch 15 and June 1, 1997. lar polarization inferred from our measurementscan in principle be interpreted either in terms of finite elliptiType of Event Right Left Unknown city of the polarization or in terms of off-zenith incidence of a perfectly circularly polarized wave. In practice, Hiss 17 0 4 MF-Burst 0 23 4 otherfactorslimit the observeddegreeof polarizationso Roar 0 46 3 that it is not possibleto quantitativelyinfer more from
thedatathanthesenseof polarization.Foremost among
3164
SHEPHERD ET AL.: POLARIZATION
thesefactorsis the variabilityof the signals.High time resolutionmeasurements showthat both MF-burst [La-
OF AURORAL RADIO EMISSIONS
Gough,M.P., and A. Urban, Auroralbeam/plasmainteraction observed directly, Planet. SpaceSci., 31,875, 1983.
R. A., Whistlers and Related IonosphericPhenomBelleetal.,1997] andauroral roar[LaBelie etal.,1995; Helliwell, ena, Stanford Univ. Press, Stanford, Calif., 1965. Shepherd et al., 1996]amplitudes varyontimescales Kaufmann, R. L., Electrostatic wave growth: Secondary
far shorter than the 2 s time scaleneededto make a polarization measurement. The fast time-scalevariations
peaks in a measured auroral electron distribution function, J. Geophys.Res., 85, 1713, 1980. effectively impartnoiseintotheSystem whichbiases the Kellogg, P. J., and S. J. Monson, Radio emissionsfrom the aurora, Geophys.Res. Left., 6,297, 1979. resultawayfromperfectcircularity.Anotherpossible LaBelle, J., M. L. Trimpi, R. Brittain, and A. T. Weatherf•ctor istheeffectof thecomponent of thewavereflected wax, Fine structure of auroal roar emissions,J. Geophys. from the conductive groundunderlyingthe antenna. Res., 100, 21953, 1995. As mentionedin the introduction,in the caseof au- LaBelle, J., S. G. Shepherd,and M. L. Trimpi, Observations roral roar the polarizationmeasurement has signific- of auroral medium frequency bursts, J. Geophys. Res.,
ant implications for theories.Severalauthors havecalculated that for realistic loss-conedistribution func-
tionsthe X-modecyclotronmaserinstabilityoperating
102, 22221, 1997.
Shepherd,S. G., J. LaBelle, M. L. Trimpi, and R. Brittain,
Furtherinvestigationof auroralroar structure,Eos Trans. AGU, p. S190, 1996.
at F-regionaltitudescan havea growthrate exceed- Tanaka, Y., M. Hayakawa, and M. Nishino, Study of auroral ing electron-neutral collision frequencies [Weatherwax VLF hiss observed at Syowa Station, Antartica, Mere. Natl. Inst. Polar Res., A, 1976. et al., 1995;Yoonet al., 1996]. However,Yoonet al. Weatherwax, A. T., J. LaBelle, M. L. Trimpi, and R. BritI1996]pointout that onlytheX-modewill reachthe tain, Ground-based observationsof radio emissionsnear groundfromthismechanism, because the O-modere2fc• and 3fc• in the auroral zone, Geophys.Res. Lett., œ0, mainstrappedin the ionosphere. Hencethis mechan- 1447, 1993. ism predictsthat the wavesshouldbe right-hand el- Weatherwax, A. T., J. LaBelle, and M. L. Trimpi, A new type of auroral radio emission observed at medium frelipticallypolarized.The measurement of LP for auquencies(•0 1350-3700kHz) usingground-based receivers, roral roar emissions excludes this mechanism. The
Geophys. Res. Left., 21, 2753, 1994.
cyclotron instabilityalsoexcitestrappedZ-modewaves withhighgrowthratesat locations wheretheupperhybrid frequency matches the cyclotron harmonics [e.g., Kaufmann,1980; Yoonet al., 1997],and it hasbeen suggested that thesewavesmayconvertby a variety
Weatherwax, A. T., J. LaBelle, M. L. Trimpi, R. A. Treumann, J. Minow, and C. Deehr, Statistical and case studies of radio emissionsobserved near 2fee and 3fee in the auroral zone, J. Geophys. Res., 100, 7745, 1995. Willes, A. J., and S. D. Bale, An alternative electroncyclotron maser model for bottomside ionospheric harof mechanismsto either L-O mode or R-X mode elecmonic radio emissions,submitted to J. Geophys.Res. tromagnetic waves[Goughand Urban,1983;Weather- Winglee, R. M., and G. A. Dulk, The electron-cyclotron waxet al., 1995]. This mechanism therefore predicts maser instability as the sourceof solar type V continuum, Astrophys. J., 310,432, 1986. eitherLP or RP radiation depending ontheconverslonYoon, P. H., A. T. Weatherwax, T. J. Rosenberg,and J. Lamechansismand henceis not excludedby the polarizBelle, Lower ionosphericcyclotron maser theory: A posation measurements presentedabove.Anothersugges- sible source of 2fce and 3fce auroral radio emissions,J. ted mechanisminvolvesthe interactionof upper hybrid Geophys.Res., 101, 27015, 1996.
wavesnearharmonics of the gyrofrequency [Willesand Yoon,P. H., A. T. Weatherwax,T. J. Rosenberg,J. LaBelle, S. G. Shepherd,P. Stauning,and R. Doe, Ray tracing Bale, 1997; Wingleeand Dulk, 1986]However,the poand spatialgrowth/dampingof 2f•, and 3f•, auroralradio larization measurements place a constrainton the conemissions,Eos Trans. AGU, p. S221, 1997. versionmechanismswhich may be consideredin these theories.
S. G. Shepherd, J. LaBelle, and M. L. Trimpi, Hinman
Acknowledgments. This researchwas supportedby Box 6127, Wilder Hall, Department of Physics and National ScienceFoundationgrant ATM-9316126 to Dart- Astronomy, Dartmouth College, Hanover, NH 03755. jlabelle@einstein. mouthCollege.The authorsacknowledge helpfuldiscussions(e-mail: [email protected]; with A.T. Weatherwax and R.A. Treumann. dartmouth.edu; [email protected]) References
Chen, F. F., Introductionto Plasma Physicsand Controlled Fusion, 2nd ed., Plenum Press, New York, New York, 1984.
(Received September 5, 1997;accepted October 28,1997.)
The polarization of auroral radio emissions S. G. Shepherd,J. LaBelle, and M. L. Trimpi Department of Physicsand Astronomy,Darmouth College,Hanover, New Hampshire
To determine the polarization of various radio emisAbstract. Ground level observationsusing two vertical loop antennasoriented at 900 to each other reveal sions crossedloop antennas and associatedelectronics the senseof polarization of severaltypes of auroral ra- were installed at Churchill, Manitoba. The first obserdio emissionsin the frequencyrange 30-5000 kHz. Au- vations with this system are describedbelow. roral hiss is observedto be right elliptically polarized (RP) with respectto thelocalmagneticfield,consistent Instrumentation with theoretical expectationfor the whistler mode and with earlier measurements. Two less well-understood For several years auroral radio emissionshave been auroral emissions,are found to be left elliptically polar- continuouslymonitored at various northern and southized (LP). This polarizationi• inconsistentwith their ern hemispheresitesusinga programmablesteppedfregenerationin the X-mode as suggestedby sometheor- quencyreceiver(PSFR). This receiveris usuallyproies. grammed to sweep from 30 kHz to 5 MHz every 2 secondsin 10 kHz steps. Data are collected,digitized, Introduction and stored by a local computer and sent back to Dartmouth Collegemonthly. The auroral ionsphereis an abundant sourceof radio In March 1997, the PSFR at Churchill (58.76øN, emissions,someof which are detectableon the ground. 265.92øE,69.20invariantlatitude)wasmodifiedto meas-
At LF/MF/HF, theseincludeauroralhiss at 1 kHz-1 MHz [e.g.,Helliwell, 1905],MF-burst at •1.4-4.5 MHz [ Weatherwaxet al., 1994;LaBelleet al., 1997],and 2fc•
ure the polarization of received signals. The modified
antennasystemconsists of two vertical2.5 m2 loopantennasoriented 900 to eachother in an approximatelyN-
and 3fc• auroral roar at .-• 3 and .-• 4.5 MHz, respectively S/E-W position. Figure i showsa schematic of the two [Kelloggand Mortson,1979; Weatherwaxet al., 1993]. antennasand a simplifiedblock diagram of the system. The principal generationmechanismfor auroral hisshas (In the figure,the two antennasare separatedfor clarbeen described;however,the generationmechanismsof ity, but in fact they are physically located on the same MF-burst
and auroral
roar remain
unknown.
These au-
verticalmast.) The polarizationdetector,receiver,and
roral emissionsare signaturesof important phenomena computer are located .-• 100 meters from the antenna to suchas wave/particleinteractionsand energyexchange minimize noise pickup. processes. In the polarization detector, a 900 phase lag is inPolarization measurementsprovide an important clue troduced into the signal from the N-S loop. On alternabout the generationmechanismof these auroral ra- ate sweepsthis signalis inverted, effectivelyshiftingthe dio emissionsby placingconstraintson the propagation phase of the N-S loop signal from -900 to +900 relatmode in the ionosphere. For example, several research- ive to the E-W loop. The input to the receiver is the ers have suggestedthat auroral roar may be generated shifted and switched N-S loop signal summed with the in the X-mode by the cyclotron maser mechanismop- signalfrom the E-W loop. If the original signalsinduced eratingat F-regionaltitudes[Weatherwaxet al., 1995; in the antennaloopsare equal in amplitude but differ in Yoonet al., 1996].This mechanism predictsthat auroral phaseby exactly 900, as would be expectedfor vertically roar shouldbe right elliptically polarized. Another posincidentright- or left-circularlypolarizedwaves,the insible mechanism
of auroral roar involves the conversion
of upperhybridwavesto electromagnetic waves[Gough and Urban,1983; Weatherwax et al., 1995].This mechanism predicts either left-ellipticalpolarization(LP) or right-ellipticalpolarization(RP), dependingon the mode conversionprocess. It is therefore possible to eliminatesometheoriesbasedon the observationof signals propagatingin a mode which is forbiddenby that
anl•• p I
mechanism.
Copyright1997by theAmericanGeophysical Union. Papernumber97GL03160. 0094-8534/97/97GL-03160505.00
Figure 1. Schematicof polarizationdetectionelectronics. Phase shifting is representedas a 900 lag of the signal in the N-S antennaloop. In reality, the phasesof both signalsare shifted to producethe 900 lag. 3161
3162
SHEPHERD
ET AL.- POLARIZATION
put signalto the receiveralternatesbetweenzero and twice the inducedsignalstrength.On the otherhand,a linearly polarizedsignalinducesin-phasesignalsin the antennaloops which result in a constantinput signal to the receiver,there beingno differencebetweenshifting the N-S signalforwardor backwardin phasein this
OF AURORAL
EMISSIONS
Churchill, Manitoba
600 ......
March 16, 1997
..".:::: :.: ....................:..';•.•'..
....
:.....
.:...
.
..
• ........
::
:..
,..•..
.
..
400
case. To assist in data analysis, a marker is recorded
300
during part of the noninvertedsweepto allow sweep identification,and a calibrationsignalthat simulatesa receivedright-cirularlypolarizedsignalis periodically
200
inserted into the antennas. In order to determine polar-
lOO
ization usingthis technique,the measuredsignalsmust be relativelyconstantin amplitudeand polarizationduring twoconsecutive meaurements (~ 2 s); signalswhose amplitudevariesfaster than that may registera falseor indeterminant polarization. The senseof polarization (right or left) is determinedby notingthe relativesignal strengthsof the two sweepsand comparingthat to
RADIO
...................................... ""'""'-• '""•- '-"•:'••••' /:'_L.::.:7.L::.:.•,:
••.••:.::•:•._•: 0544
0546
' ...%::•....;....;;;...................`.................•...•••.•••.•...•`.."........•
•,..:.. .................. ..................................................................• .:..••.... 0548
0550
0552
0554
0556
0558
Time (U•
Figure 2. A raw spectrogram recordedby the swept frequencyreceiverduringa time whenauroralhissoccursat frequencies from 30 kHz to above500 kHz. Alternatesweepsare dark or light, showingthat the waves areellipticallypolarized.Carefulinspection revealsthat the marker. theyareright-hand polarized,asexpected ontheoretical The interpretation aboveassumesthat the electronics groundsand on the basisof previousmeasurements.
are perfect. It is difficult to shift a single signal by 900 over a wide bandwidth, but it is easy to shift both the N-S and E-W loop signalssuch that the phase of the N-S loop signallags the E-W loop signal by ~ 900 over a range of approximatelytwo decades.The error in phaseshift overthe 0.05-5.0 MHz frequencyrangeis less than ten degrees,implyingthat the maximumamplitude differencebetweenconsecutivesweepsfor either rightor left-circularlypolarizedverticallyincidentsignalsis about 20 dB which is less than the observed differences
sphere, which implies RP, consistentwith the observation. This observationof auroralhissprovidesa natural calibrationof the polarizationdetector.
Figure3a (toppanel) is a spectrogram showing three types of auroral radio emissionsrecorded 0445-0459 UT on April 4, 1997. Auroral hiss occurs below ~ 500 kHz
beginningnear 0452 UT lasting for ~ 5 minutes and again during the last ~ 30 secondsof the record. MF-
of the real signalsdescribedbelow. (~ 2 MHz) emission above~ 1.5 In this paper all wavepolarizationsare measuredwith burstis the broadband MHz and correlated with the auroral hiss. Auroral roar respectto the local magneticfield. The electricfield vec-
at tor of a right-handedcircularlypolarized(RCP) wave is therelativelynarrowband(dr ~ 200kHz) emission ~ 3 MHz beginning near 0448 UT and ending near 0453 rotates clockwise in time as viewed in the direction of UT. Horizontal dark bands are fixed-frequencytransthe magneticfield. In this case, for a receiver in the missions;the band from 550-1600 kHz is the AM broadNorthern Hemisphere the electric field vector rotates cast band. The sweepidentificationmarker is at 1-1.25 clockwiseas seenby an observerlookingdown on the MHz. Clearly, alternatesweepsare light or dark, implyantennas.This definitionis standardin plasmaphysics ing that they are elliptically polarized. Careful inspec-
[e.g.,Chen,1984]and is usedin previousauoralradio tion shows that the MF-burst and auroral roar are LP emission polarizationstudies[Tanakaet al., 1976]. and auroral hissis RP. Figure 3b (bottompanel)showsthe polarizationof
the signalsas a grayscale. In this display, white and blackpixelscorrespondto LCP and RCP wavesrespectFigure 2 showsan auroral hissevent recordedstart- ively. Elliptically polarized waves are representedas ing at 0548 UT on March 16, 1997, lasting approxim- gray pixels betweenthe two extremes,with linear poately 3 minutes,and extendingfrom the lowerboundof larizationbeingat themiddleof thegrayband(halfway Observations
the instrument(30 kHz) into the AM-broadcastband betweenwhiteandblack). As expected,the auroralhiss above 500 kHz. Dark horizontallines on this spec- showsup as dark pixels implying right-hand polarizatrogram representfixed-frequencyman-madetransmis-
tion. In contrast, both MF-burst and auroral roar are
sions. In contrastto the fixed-frequencysignals,the left-hand polarized. There are two sweepsat the onauroral hiss is clearly elliptically polarized becauseal- set of an auroralsubstormnear 0453 UT during which ternate receiver sweepsdiffer in amplitude. Careful the polarizationmeasurementmomentarilyimpliesthat inspectionrevealsthat the observeddark-lightpattern the auroral roar is RP, but at this time the auroral roar correspondsto RP. This result is consistentwith pre- amplitudeis probablyhighly time variable,as is known viousauroralhisspolarizationmeasurements usingsev- from fine structuremeasurements [e.g.,LaBelle et al., eraltechniques [Tanakaet al., 1976].Also,auroralhiss 1997;Shepherdet al., 1996],and undersuchconditions is believedto propagatein the whistlermodein the iono- the polarization measurementcannot be trusted.
SHEPHERD ET AL.' POLARIZATION
OF AURORAL RADIO EMISSIONS
Churchill• Manitoba
(a)
3163
April4, 1997
intensity (nV/m
40003500-
•'3000-
•-.•-2500-
•**?•:.•js:*-'•:*****::*s;4j:•*•'½,25' *;;5•: ':.' :::•:; '::'***'2.:L?.'•-:'•, ::{*"4:":•: 4'"'½: .:;:-4:.
:::•:*:½'*•-:*•:-•:::•$•*•**' ':½*"•* ":'
2000-
1500
....... .--..... .
500 ..................................... -?, ........................... •:;.:...;:;•::• .......................... ?•?.•;:•?;:,.,;• ...................................... -::Q':X ;•:::L.:.:'.:.. .:.'.. ,:.'.'.:. ..L,.. ,,'. ;.................... ;.............................................................. :• ....................... ::,.;;,,;;.,•; ............................. :.................... ;•E•..•:.:*•?•;•:•:;•*•;•*•.•:M:•;;•.*•;&M;•:•?;:;:X;•?::•:..•.:.;?•;•;.S::"`"•% ........................ •;•:•;;;•.•.•;•:?.•;;•.•:;a•?•`•;•:•:.:•;.•::: 12.5
polarization
4000 •, ............................................................. :.-; .................. *•;?'"' .............................................. '"': ........................... ;:•':"• ............................. '............. "':"" ........ "":" ......................................... '"? ............. ............ •...................... '"'""' ........ ":'"':' ........................................................................ ':"":'?:' right ...........
;:.x .....
•
2500 2000-
1500
.•:.,
.•..
;:;.::
.
'"•:••
linear
5oo 0446
0448
0450
0452
0454
0456
0458
Time (UT)
•o•1
•o•,
•ol•i•e•. (b) •ol•i•io• •-b•,
•o•1
of e•i•io•
•o•i•
•o•,
from strong atmospherics. An atmosphericproduces an output to the receiver which resemblesa broadband purposesof accumulating statistics,an eventis defined signal that is strongly polarized in either direction, deas an auroral emissionthat is detectedfor longer than pendingon the signof the inducedphaseshift duringthe 30 secondsand is separatedby at least 10 minutesfrom event. The impulsivenature of thesesignalsobscurethe other events. Table i showsthe senseof polarizationof polarization of coincidentwavesand illustrate the effect all eventswhichoccurredduringthe observationperiod, of a signal which varies in amplitude faster than the asdeterminedfrom spectrograms similarto thoseshown sweepperiod of the receiver. Auroral emissionswere seen on 38 of the 78 days of observation between March 15 and June 1, 1997. For
in Figure3b. Table i showsthat auroralhissis right ellipticallypolarizedand MF-burstsand 2f• auroralroar Interpretation are both left elliptically polarized. While the statistics
The experiment described above establishesthat the MF-burst and 2f• auroral roar emissionsare left-hand hand polarized, more statisticsare neededto exclude establish that these emissionsare almost entirely left-
the possibilitythat a few percentof theseemissions are polarized,in contrastto auroralhisswhichis right-hand right-handor linearlypolarized.The few eventsin the polarized. For all three emissions,the mean power differencebetweenalternatesweeps,during whichthe sigunknown column of the table are due to interference nals constructivelyand destructivelysum, averages-10 dB.
The
instrumental
error
in this measurement
due
Table 1. Polarization of auroral hiss, auroral roar, and to phase inaccuraciesin the polarization detectionelecMF-burst events from Churchill, Manitoba, between tronics is the order of-20 dB. The departurefrom circuMarch 15 and June 1, 1997. lar polarization inferred from our measurementscan in principle be interpreted either in terms of finite elliptiType of Event Right Left Unknown city of the polarization or in terms of off-zenith incidence of a perfectly circularly polarized wave. In practice, Hiss 17 0 4 MF-Burst 0 23 4 otherfactorslimit the observeddegreeof polarizationso Roar 0 46 3 that it is not possibleto quantitativelyinfer more from
thedatathanthesenseof polarization.Foremost among
3164
SHEPHERD ET AL.: POLARIZATION
thesefactorsis the variabilityof the signals.High time resolutionmeasurements showthat both MF-burst [La-
OF AURORAL RADIO EMISSIONS
Gough,M.P., and A. Urban, Auroralbeam/plasmainteraction observed directly, Planet. SpaceSci., 31,875, 1983.
R. A., Whistlers and Related IonosphericPhenomBelleetal.,1997] andauroral roar[LaBelie etal.,1995; Helliwell, ena, Stanford Univ. Press, Stanford, Calif., 1965. Shepherd et al., 1996]amplitudes varyontimescales Kaufmann, R. L., Electrostatic wave growth: Secondary
far shorter than the 2 s time scaleneededto make a polarization measurement. The fast time-scalevariations
peaks in a measured auroral electron distribution function, J. Geophys.Res., 85, 1713, 1980. effectively impartnoiseintotheSystem whichbiases the Kellogg, P. J., and S. J. Monson, Radio emissionsfrom the aurora, Geophys.Res. Left., 6,297, 1979. resultawayfromperfectcircularity.Anotherpossible LaBelle, J., M. L. Trimpi, R. Brittain, and A. T. Weatherf•ctor istheeffectof thecomponent of thewavereflected wax, Fine structure of auroal roar emissions,J. Geophys. from the conductive groundunderlyingthe antenna. Res., 100, 21953, 1995. As mentionedin the introduction,in the caseof au- LaBelle, J., S. G. Shepherd,and M. L. Trimpi, Observations roral roar the polarizationmeasurement has signific- of auroral medium frequency bursts, J. Geophys. Res.,
ant implications for theories.Severalauthors havecalculated that for realistic loss-conedistribution func-
tionsthe X-modecyclotronmaserinstabilityoperating
102, 22221, 1997.
Shepherd,S. G., J. LaBelle, M. L. Trimpi, and R. Brittain,
Furtherinvestigationof auroralroar structure,Eos Trans. AGU, p. S190, 1996.
at F-regionaltitudescan havea growthrate exceed- Tanaka, Y., M. Hayakawa, and M. Nishino, Study of auroral ing electron-neutral collision frequencies [Weatherwax VLF hiss observed at Syowa Station, Antartica, Mere. Natl. Inst. Polar Res., A, 1976. et al., 1995;Yoonet al., 1996]. However,Yoonet al. Weatherwax, A. T., J. LaBelle, M. L. Trimpi, and R. BritI1996]pointout that onlytheX-modewill reachthe tain, Ground-based observationsof radio emissionsnear groundfromthismechanism, because the O-modere2fc• and 3fc• in the auroral zone, Geophys.Res. Lett., œ0, mainstrappedin the ionosphere. Hencethis mechan- 1447, 1993. ism predictsthat the wavesshouldbe right-hand el- Weatherwax, A. T., J. LaBelle, and M. L. Trimpi, A new type of auroral radio emission observed at medium frelipticallypolarized.The measurement of LP for auquencies(•0 1350-3700kHz) usingground-based receivers, roral roar emissions excludes this mechanism. The
Geophys. Res. Left., 21, 2753, 1994.
cyclotron instabilityalsoexcitestrappedZ-modewaves withhighgrowthratesat locations wheretheupperhybrid frequency matches the cyclotron harmonics [e.g., Kaufmann,1980; Yoonet al., 1997],and it hasbeen suggested that thesewavesmayconvertby a variety
Weatherwax, A. T., J. LaBelle, M. L. Trimpi, R. A. Treumann, J. Minow, and C. Deehr, Statistical and case studies of radio emissionsobserved near 2fee and 3fee in the auroral zone, J. Geophys. Res., 100, 7745, 1995. Willes, A. J., and S. D. Bale, An alternative electroncyclotron maser model for bottomside ionospheric harof mechanismsto either L-O mode or R-X mode elecmonic radio emissions,submitted to J. Geophys.Res. tromagnetic waves[Goughand Urban,1983;Weather- Winglee, R. M., and G. A. Dulk, The electron-cyclotron waxet al., 1995]. This mechanism therefore predicts maser instability as the sourceof solar type V continuum, Astrophys. J., 310,432, 1986. eitherLP or RP radiation depending ontheconverslonYoon, P. H., A. T. Weatherwax, T. J. Rosenberg,and J. Lamechansismand henceis not excludedby the polarizBelle, Lower ionosphericcyclotron maser theory: A posation measurements presentedabove.Anothersugges- sible source of 2fce and 3fce auroral radio emissions,J. ted mechanisminvolvesthe interactionof upper hybrid Geophys.Res., 101, 27015, 1996.
wavesnearharmonics of the gyrofrequency [Willesand Yoon,P. H., A. T. Weatherwax,T. J. Rosenberg,J. LaBelle, S. G. Shepherd,P. Stauning,and R. Doe, Ray tracing Bale, 1997; Wingleeand Dulk, 1986]However,the poand spatialgrowth/dampingof 2f•, and 3f•, auroralradio larization measurements place a constrainton the conemissions,Eos Trans. AGU, p. S221, 1997. versionmechanismswhich may be consideredin these theories.
S. G. Shepherd, J. LaBelle, and M. L. Trimpi, Hinman
Acknowledgments. This researchwas supportedby Box 6127, Wilder Hall, Department of Physics and National ScienceFoundationgrant ATM-9316126 to Dart- Astronomy, Dartmouth College, Hanover, NH 03755. jlabelle@einstein. mouthCollege.The authorsacknowledge helpfuldiscussions(e-mail: [email protected]; with A.T. Weatherwax and R.A. Treumann. dartmouth.edu; [email protected]) References
Chen, F. F., Introductionto Plasma Physicsand Controlled Fusion, 2nd ed., Plenum Press, New York, New York, 1984.
(Received September 5, 1997;accepted October 28,1997.)
