Sporadic Narrowband Radio Emissions from Uranus

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 91, NO. All, PAGES 11,958-11,964, NOVEMBER 1, 1986

SporadicNarrowband Radio EmissionsFrom Uranus

W. S. KURTH AND D. A. GURNETT

Departmentof Physicsand Astronomy,The Universityof Iowa, Iowa City

F. L. SCARF

TRW Spaceand Technolo•TyGroup, Redondo Beach, California Instituteof Geophysicsand Planetary Physics, University of California,Los Anodes

Amongseveral different types of radioemissions discovered at Uranusduring the Voyager2 encounter in January1986 is a verysporadic, bursty signa I which consists of verynarrow bands lying in the frequencyrange from about 3 to 10 kHz. The burstyemission was virtuallyundetectable from the daysideportion of the Voyager 2 trajectory but was observed out to beyond300 Rv during the outbound trajectorythrough the predawnsector. While the narrowbandtones making up this emissionare remi- niscentof escapingcontinuum radiation observed near earth, Jupiter, and Saturn,the Uraniansignals showlarge amplitude variations on time scalesof 1 s, suggestinga very differenttype of generation mechanism.

1. INTRODUCTION at 3.11 kHz, and some emissions spread upward into the During the Voyager 2 flyby of Uranus in January 1986, 10-kHz channelearly in the plotted interval.It is important to severaltypes of radio emissionswere discovered by the plasma point out, however, that during the interval from late on Janu- wave investigation['Gurnett et al., 1986,] and the planetary ary 27 through January 29 the spacecrafttraversed the bow radio astronomyinvestigation ['Warwick et al., 1986].The shock severaltimes [Bridge et al., 1986; Ness et al., 1986], and bulk of the radio spectrum of Uranus lies above about 30 the response at 3 kHz is primarily due to electron plasma kHz; however, there is evidence of nonthermal continuum oscillation (Langmuir wave) activity in the upstream regions radiation at frequenciesdown to about 1 kHz. As Voyager 2 associatedwith theseshock crossings.The activity at 3.11 kHz left Uranus in the predawn sector,still another type of emis- prior to late January 27 is the lower frequencyextent of the sion was detected in a band centered near 5 kHz [Gurnett et waves of interest here. Very little activity is observed after al., 1986]. January 31; however, there is evidencefor the bursts on Feb- The new band of radio emissionfound on the nightside of ruary 5 at about 570 Re. Uranus is remarkable in that while it displays a spectrum The bursty nature of the emissionis apparent in Figure 1. similar to escapingnonthermal continuum radiation detected In addition to the highly sporadic fluctuations in the peak at earth [Kurth et al., 1981], Jupiter [Gurnett et al., 1983], and amplitudes,the very large peak-to-averageratios, which are in Saturn [Gurnett et al., 1981], it is much more bursty and many casesmuch greater than 10,point to the highlyvariable sporadicin nature than the continuumradiation. In fact, the amplitudesof the emission.In fact, that one cannot discerna time scale.for amplitude variations of the order of 30 dB is differencebetween the temporal character of the radio emis- typically 1 s. sion and the electron plasma oscillationsseen on days 28 and In this report we investigatethe detailsof the sporadicnar- 29 of Figure 1 underscoresthe temporal variability, since rowband emissionfrom the magnetosphereof Uranus. We will plasma oscillationsare typically one of the most sporadic emphasizethe peculiar bursty nature of the emissionand types of plasma waves. argue that a rather exotic sourcemechanism may be respon- Figure 2 shows several examples of high-resolution sible for the emission.The observationspresented herein are frequency-timespectrograms, which detail both the temporal all taken from the Voyager2 plasmawave receiver,which has and spectralcharacteristics of the radio emission.Each spec- beendescribed in detail by Scarfand Gurnett[1977]. trogram showsthe intensity of wavesas a function of both frequency(ordinate) and time (abscissa).The most intense 2. OBSERVATIONS waves are darkest. Each panel representsa 10-s interval, and An overview of the observationsof the bursty radio emis- the frequencyrange coveredis 0-12 kHz. The Fourier trans- sion in the frequencyrange of a few kilohertzis presentedin forms used to calculatethe spectraprovide about 28-Hz reso- Figure 1. Illustratedare the peak and averagepower fluxes lution, and a new spectrumis calculatedevery 60 ms. Panels detectedby the spectrumanalyzer channels centered at 3.11, A-E in Figure 2 show examplesof the spectrumof the radio 5.62, and 10.0 kHz as a function of time. The height of the emissionresponsible for the activity shownin Figure 1 at 5.62 solid black areas representsapproximately l 1-rain average kHz and sometimesthe surroundingchannels. valuesin each channel,while the peak power fluxesobserved The radio emissionappears in the spectrogramsin Figure 2 during the same 11-min averagingintervals are plotted as a as numerousnarrowband signals,many of which have dura- line above the averages. tions of only a few seconds.The narrowbandtone at 2.4 kHz It is clear in Figure 1 that the bulk of the emissionis in the is interferencefrom the Voyager power supply, and the band 5.62-kHz channel,although considerable activity can be seen at 1.8 kHz is also thought to be interference,since it also appearedin the Saturndata set;however, the sourceof this lower-frequencyinterference is not known. The radio emis- Copyright1986 by theAmerican Geophysical Union. sions of interest seemto be composedof both the brief bursts Paper number6A8566. and an underlying,more continuoussystem of narrow bands. 0148-0227/86/006A-8566505.00 In some spectrogramsthe weaker, more stable bands tend to 11,958 KURTH ET AL.' SPORADICRADIO BURSTS FROM URANUS 11,959

VOYAGER 2 JANUARY25- FEBRUARY I, 1986

I0 kHz

5.6 kHz

ic•14

ic?17

DAY 25 26, 27 28 29 30 31 32 I I I I I I I R(Ru) 50 I00 150 200 250 300 350

Fig. l. Power flux as a function of time for three of the Voyager 2 spectrum analyzer channels showing the low- frequencyradio emissionas the spacecraftleft Uranus in the predawn direction.Notice that the bulk of the activity occurs in the 5.6-kHz channel but occasionallyspreads into the adjoining channels.Much of the activity at 3.1 kHz after the middle of day 27 is electronplasma oscillations associated with numerousbow shockcrossings.

coalesce into weak, diffuse bands. The more intense bursts are while showing random variations in detail, remains constant occasionallyseen superimposed upon the diffuse bands. in general, even though the spacecrafthas moved through the The observationspresented in Figures 1 and 2 provide solid magnetotail, magnetosheath, and solar wind from January 25 evidencethat the bursty emissionsare freely propagating elec- through 31. While uncertainties in the receiver calibration tromagnetic waves.The radio emissionsare at frequenciessuf- caused by a failure in the spacecraftdata system and the very ficiently high that the waves can freely propagate into the bursty nature of the emissionsmake it difficult to detect a solarwind. The solarwind plasmafrequency fp is typicallyin 1/R•- intensity variation, it is clearthat both the amplitudeand the range of 2-3 kHz at the orbit of Uranus, based on the the occurrence rate of the emissionsdrops dramatically over frequency of upstream plasma oscillations [Gurnett et al., the seven days illustrated. 19861,and fp in the magnetotailis even less.The spectrum, The frequency range of the narrowband radio emission lies

•--,.• ' •.'-L..--:::-• .. •::•.... -

L.Z • ...... '•' ..• -----

•:•:•:• .-.•A•...... •:•: • --

.•::=•:- •½•..... •.... •:•, :•::•:. voYAGER• .•;•;• -,-?;• 2. . •...... •:•:•

VOYAGER 2 at about 5.5 kHz. The diffuse component is not present; how- 0.6-SECOND AVERAGE SPECTRA ever, Voyager 2 was greater than 300 Rt• from Uranus at this 10-13 I • I I I - time, and the amplitude of the diffuse emission may have JANUARY 25, 1986 simply dropped below the detection threshold becauseof the i(•14 0635:50.3SCET_ I/R e effect.We shouldpoint out, however,that the diffuse componentdoes not necessarilyvary only as I/Re; there are iC•15 _ spectrograms much closer to Uranus which show evidence for •d•6 only the bursty component; hence it is likely that there are temporal variations in the diffuse component with time scales iC•17 perhaps much greater than a few seconds. The narrowband nature of these Uranian radio emissions is iC•14 similar to escapingcontinuum radiation at the earth and simi- dANUARY 26, 1986 lar emissionsobserved at Jupiter and Saturn. However, the 0217:28.6SCET -F 15 very bursty nature of the Uranian emissionsshown in Figures _ 1 and 2 set the Uranian emissionsapart from those with which we are familiar. Figure 4 illustrates the very short time x let16 scalesof intensity variations characteristicof the bursty emission. In Figure 4 we have plotted the intensity (power flux) of radio waves as a function of frequency and time in a three- ,,, IC•I dimensional perspectiveplot to show the rapid variations in o time for this narrowband burst. The frequency range is 4-6 kHz, and the entire time representedis 12 s. The original data iC•14 _ dANUARY 31, 1986 have a temporal resolution of 60 ms and a spectral resolution 0004:02.7 SCET of about 28 Hz; however, we have averaged each point over iC•15 _ the five surrounding spectral and five surrounding temporal components(a two-dimensionalboxcar average)to obtain the

•¾•6 _ surfacerepresented in Figure 4. Figure 4 demonstratesthat the power flux for this burst can increase or decreaseby 30 dB within a second or so. Similar i(•17 analyses of the temporal variations of narrowband electro- magneticemissions from Jupiter show variations in power flux 0 2 4 6 8 I0 12 of a few decibelson the same time scale.Figure 5 is an exam- FREQUENCY (kHz) ple of narrowband electromagneticradiation at 11 kHz escaping from the Jovian magnetosphereas observedby Voyager 1. Fig. 3. High-resolution spectra from three selectedtime intervals which show the rich spectral structure of the Uranian radio emission Note that the only apparent variations in amplitude during from 3 to 10 kHz. Some evidence for underlying diffuse bands is the 48-s interval occur about 6 and 30 s into the frame. These present in the upper two panels, while only the intense, narrow band two decreasesin signal level are actually due to decreasesin at 5.5 kHz is presentin the bottom panel. the gain of the automatic gain controlledreceiver in response to interferencefrom a steppermotor on board the spacecraft. The band shows significant amplitude variations only on a time scale of several minutes. The Uranian radio burst in between about 3 and 10 kHz. The more diffuse portion of the Figure 4 will not be fully understood unlessthe generation spectrum is limited to the range from about 3 to 6 kHz. If there is a preferred frequency for the bursty emissions,it is 5 kHz, but the bursts can be found over the entire range from 3 to 10 kHz. More details of the spectrum of the Uranian radio emission VOYAGER 2 near 5 kHz can be seenin Figure 3, which showspower flux as JANUARY 31, 1986 a function of frequencyfor three selectedtime periods. Each 0004:11 SCET spectrum is a 0.6-s average, and the strong band of noise below 1 kHz is interference thought to be associatedwith the I operation of the tape recorder on board the spacecraft.The interference tones at 1.8 and 2.4 kHz have been removed. In the top panel of Figure 3 there is a general rise in wave amplitudes from about 3 to 7 kHz correspondingto the more diffuse component of the emission. Superimposed upon this o generally elevated spectrum are a few, very narrow lines with power fluxes at least a factor of 10 greater than the back- ground levels. The middle panel of Figure 3 again shows some evidence • •,s•c• for a diffuse component between 3 and 5 kHz but also strong lines which extend from just above 3 kHz to almost 10 kHz. The bandwidths of the lines often do not exceed 100 Hz. For a line near 10 kHz the bandwidth is only 1% of the center Fig. 4. A perspectiveplot showing intensity as a function of fre- frequency. quency and time illustrating the extremely narrowband and bursty The bottom panel of Figure 3 shows a single intense spike nature of this radio emission at about 5 kHz. KURTH ET AL..' SPORADICRADIO BURSTSFROM URANUS l 1,961

Fig. 5. A spectrogramtaken by Voyager1 at Jupitershowing the very steadynature of the Joviannarrowband emission at 11 kHz. mechanism can explain both the narrowband nature and the are marked with a cross. In some casesit is possible that the rapid variations in intensity. We will attempt to addressboth signaturein the wideband frames is a different type of radio issues in the next section. emission and not the bursty emission discussedin this paper; In an attempt to understand the location of the source of hence we have placed question marks by these questionable the 5-kHz radio emissionwe have plotted the magnetic lati- events. tude of Voyager 2 from January 24 to February 1, 1986, as a After January 25, two typesof markingswere usedto identi- function of time in Figure 6, using information about the fy times (magnetic latitudes) when the sporadic, narrowband orientation of the magnetic dipole from Ness et al. [1986]. emission was present. The thin solid line represents times There is no clear evidence for the existence of the radio emis- when there is weak evidence for the emission. That is, if bursts sion in the spectrum analyzer data prior to January 25; how- were too sparseto significantlyaffect the averaged power flux ever, there is some evidence of either the diffuse emission or as plotted in Figure 1 when the spacecraftwas lessthan about the bursts in some of the wideband frames obtained on Janu- 150 Re from Uranus, a thin line was used in Figure 6. The ary 24. The locations of the spacecraftfor those times when broad lines imply the radio emission was more prominent. there is evidence in the wideband frames only on January 24 While the distinction here is mainly qualitative, it is important to provide a means of de-emphasizingthe weaker emissions observed close to Uranus, which would not be detectable at

JANUARY 24-FEBRUARY I, 1986 largerdistances, since we have not taken a 1/R•' trend into 80 ø account. The impressionleft by the pattern of detections,especially the more "prominent" detections, is that the emission was Ld observed with confidence only after encounter (when the ca 40" spacecraftwas in the predawn location) and then usually at the lowest magneticlatitudes observedby the spacecraft. ._1 In Figure 7 we have illustrated wideband data which may • 0 ø have a bearing on a possiblesource of the bursty radio emis- ß z , ß sions. This spectrogram was observed near the time when Voyager traversed the Miranda L shell. The intense nar- -40 o ß : rowband line at 9 kHz is of interest, since it is a possible ß x source for the electromagnetic emissions observed later. The

DAY 24 26 28 30 32 intensity and narrowband character make the connection to the radio emissions a reasonable one, even though the con- Fig. 6. A plot of the magneticlatitude of Voyager2 as a function nection cannot be proven. The large dot in Figure 6 late on of time based on information of the Uranian magnetic field provided January 24 marks the location of Voyager 2 at the time the by the Voyagermagnetometer. The heavydark linesshow intervals data in Figure 7 were taken. It is interestingand highly sug- when the 5-kHz emissionwas prominent.The emissionwas only very weaklydetectable in waveformsamples on the inbound(dayside) leg gestivethat the latitude of this intensenarrowband emission is of the encounter (indicated by crosses). very close to the most favorable latitude for observing the 11,962 KURTI-I ET AL.' SPORADICRADIO BURSTSFROM URANUS

radio emissions with the Miranda L shell. During the brief passage through the Uranian magnetosphere there is little chance to sample anything but the smallest region dictated by the flyby trajectory. Hence it is quite unlikely that the spacecraft passed through an active source region for the bursty radio emissions detected during the outbound leg. There are many reasons, however, to suspect that the outbound Mi- randa L shell crossingis a region which is either close to or similar to the radio emission source region. The narrowband emission at 9 kHz in Figure 7 is likely to be an electrostatic emission near the local plasma frequency or upper hybrid resonance frequencyful+ While most of the radio bursts dis- cussed herein are at lower frequencies,there are a few lines observed at 9 kHz. It is interesting to note that according to Ness et al. [1986], the range of magnetic field strengthsof the L shells traversed by Miranda correspond to electron gy- rofrequenciesranging from about 5 to 9.5 kHz. Should the line at 9 kHz in Figure 7 be a Bernstein mode, lying between the electron gyrofrequencyand its secondharmonic, then one might expect similar lines at a wide range of frequenciesex- 5 tending down to the lower end of the radio emissionfrequency range. Plasma densities from the plasma experiment (J. W. Belcher, personal communication, 1986) are consistentwith an upperhybrid resonance frequency (fUH 2 = f• 2 + fv2) in the range of about 9 kHz at 1920 spacecraftevent time (SCET) on January24;fg is theelectron gyrofrequency. Another interesting observation from the region associated with the outbound Miranda L shell crossingis the presenceof intense,low-frequency bursts of plasma wave activity. Figure 8 shows spectrum analyzer data from the 1-hour interval around the Miranda L shell crossing. The wideband data shown in Figure 7 were obtained at the time of the data gap shortly before 1920 SCET. The line crossingthrough the 5.62- kHz channelrepresents fg accordingto informationprovided by the magnetometer team (N. F. Ness, private communication, 1986). It is clear that several different types of wave modes are present during this interval. The bursty emissions abovef• correspondto the line at 9 kHz in Figure7 and are '1919:46 likely plasma waves associatedwith fun- The intense emissions with limited bandwidth in the region from perhaps 0.25 to 0.5 jANUARY' 24, 1986 f• are likely to be whistlermodes. Emissions in the range below a few hundred hertz could be Doppler-shifted ion Fig. 7. This wideband spectrogrammay be representativeof the acoustic waves. sourceregion of the bursty Uranian radio emission.The narrowband emissionat 9 kHz is likely to be an electrostaticband nearfuH which With the presenceof the bursty plasma wave emissionsat could couple into the escaping electromagneticmode. These data very low frequenciesit is possible to conceive of three-wave were obtained in the vicinity of the Miranda L shell. interactions involving one of the low-frequencyemissions, the intense narrow band near 9 kHz, and an electromagnetic mode which we detect as a freely propagating radio wave as radio emission.We will discussthe possiblesignificance of the Voyager leaves Uranus. narrowbandemission in Figure 7 in the next section. The above-describedthree-wave processis similar to one of the favored mechanismsfor the generation of continuum radi- 3. DISCUSSION AND CONCLUSIONS ation at the earth, Jupiter, and Saturn. That is, the continuum The observationspresented above show the characteristics radiation is thought to be a nonlinear process involving an of a new radio emission discovered in the region around intense band near fun and some low-frequency plasma wave Uranus. The emission exhibits extremely narrowband bursts which producesan electromagneticwave (the continuum radi- with bandwidths of a few percent or less and time scalesof ation) at nearly the same frequency as the upper hybrid band seconds.The bursts are accompaniedat times by a diffuse (see,for example, Melrose [1981]). An alternative hypothesisis underlying structureof more temporally continuousbands. It a linear conversionfrom the upper hybrid band to an electro- is conceivable that the diffuse component is the result of the magnetic mode (see,for example, Jones [1976]). superposition of several weak bursts. The emission is only The theories for generation of continuum radiation, then, weakly presentduring the close-indayside trajectory of Voy- are perhaps applicable to the Uranian emission near 5 kHz ager 2 and is prominent in the radial distancerange of about and, if so, point to the region near Miranda's orbit, because 20-300 Rt• as the spacecraftleaves Uranus in the predawn intense narrow bands near fun are detected there. (There is direction. little evidence to help us choose between the linear and non- It is tempting to associatethe sourcelocation for the bursty linear generation mechanisms,however.) But the major prob- KURTH ET AL.' SPORADICRADIO BURSTS FROM URANUS 11,963

VOYAGER 2 JANUARY 24, 1986 -2 56.2: 31.1 : 17.8 _: io.o = .6a - 3.11 - 1.78 I.oo - .562 - .311 - 78 _-- .IOO • .O56 • .031 :_ .018 _: .010 •

SCET 1900 1920 1940 2000 R(Ru) 4.80 5.22 5.71 6.25 Xrn 14.1':' 12.7':' 12.4':' 13.2':' Fig. 8. Spectrum analyzer data from the region around the outbound Miranda L shell crossing,which is a possible sourceregion for the 5-kHz Uranian radio emission.

lem with this generation mechanism is that the continuum have to be converted into electromagneticwaves in the pro- radiation usually has a very smooth temporal behavior, as cessof soliton collapseto account for the observedamplitude shown in Figure 5, strikingly different from that demonstrated of the radio burst at > 300 Rv. Sincesoliton collapsegenerally in Figures 1, 2, and 4. Very little has been mentioned in the results in plasma heating, this efficiencyfor electromagnetic literature about the temporal variability afforded by the wave generation is high by several orders of magnitude. Fur- theories for continuum radiation, so it is not possible to im- ther, the frequencyof the pump (Langmuir) wave would vary mediately assesswhether the linear and/or nonlinear theories throughout the collapse process,and the resulting electro- can account for smooth temporal variations at earth, Jupiter, magneticwave would almost certainly not be monochromatic, and Saturn and the very bursty behavior at Uranus. as observed. There is another mechanism which comes to mind which The energy problem could be solved if many solitonswere might explain the very bursty temporal variations of the Ura- radiating' however,the effectof superimposingradiation from nian 5-kHz radio emission. So!iton collapse has long been severalcollapsing solitons would be to spreadboth the tempo- considereda viable mechanismfor solar type III radio bursts ral and spectral profile of the radio emission.Type III solar generatedin the interplanetarymedium (see,for example,Gold- bursts are long lived and have relatively broad bandwidths man [1983]). Type III bursts do not show a rapidly varying because of these collective effects. We must conclude, then, component, but the model assumesthe collective action of that soliton collapse is not a viable mechanism for the ob- many solitons and electromagneticradiation from the "con- servedwaves, and we are led to considerother possibilities. densate"radiation left by soliton collapseand burnout [Gold- It is possibleto produce a short, monochromatic pulse re- man, 1984]. sponsein a detector simply by sweeping a narrow beam of There is some speculationthat electromagneticradiation monochromaticwaves over the detector.The pulselength is could be generatedfrom the collapseof a single soliton if ion determined by the beam width and angular velocity of the acoustic waves were present to take up momentum and serve beam. The very short pulse durations reported herein imply a as the third wave in wave vector and frequency matching very narrow beam or very large angular velocities.The mono- conditions. The electron plasma oscillationsresponsible for chromatic spectrumis inherent in the source. the ponderomotive force which collapsesthe soliton is the One excellent method of producing a monochromatic, pump wave in this scenario.This mechanismis suggestedby highly collimated beam is via a laser mechanism,since these the time scalesof the burstsreported herein and presupposes are two basic features of a laser. We suggestthat the 5-kHz the existenceof energetic streams of electrons in the source Uranian bursts could be the result of some lasing process region to excite the plasma oscillations' however, we have no operating in the magnetosphereof Uranus. Calvert [1982] has evidencefor suchstreaming. suggestedthat several properties of auroral kilometric radi- For the sakeof completenesswe have examinedthe energet- ation at the earth can be explained by coupling the wave ics implied by generationof individual bursts via soliton col- amplification mechanism of Wu and Lee [1979] with a posi, lapse.If we assumethe burst illustratedin the bottom panel of tive feedbackmechanism. High resolution spectrogramsof au- Figure 3 was beamedinto a solid angle of only 1 sr (prompted roral kilometric radiation do not show the type of temporal by the restricted observability of the emissionillustrated in and spectral behavior seen in Figure 2, however. Instead the Figure 6) and the sourceis a plasma consistingof 1-keV elec- terrestrialemission is composedof narrowbandtones which tronswith a densityof 1 cm-3 (veryoptimistic in viewof the are nearly continuous in amplitude and which vary in fre- plasma observationsin the region near Miranda's L shell (R. quency very rapidly. Also, the auroral kilometric radiation L. McNutt, personal communication, 1986)), then 100% of the mechanismoperates in regionswhere fp <f•. 11,964 KURTH ET AL.: SPORADICRADIO BURSTSFROM URANUS

If we use current ideas on the generation of continuum that considerabletheoretical effort will be required to firmly radiation instead of the Wu and Lee mechanism for auroral establishany mechanism. kilometric radiation, we might begin to fashiona viable mechanism for the Uranian bursty emission.This choiceis suggest- Acknowledgments.The authorswould like to acknowledgeD. N. ed by the similarity in bandwidth and slowly varying spectral Nicholson, J.P. Sheerin, and C. K. Goertz for useful discussions.We behavior of the continuum radiation and the Uranian emis- are also gratefulto N. F. Ness and the Voyager magnetometerteam for information on the magneticfield strengthand J. W. Belcherand sion coupled with the apparent low magnetic latitude of the R. L. McNutt of the Voyager plasma scienceteam for information on source at Uranus. The upper hybrid bands at the earth associ- the plasma densitiesand temperaturesused herein. The researchat ated with continuum radiation source regions are usually lo- the Universityof Iowa was supportedby the National Aeronautics cated near the magneticequator. and SpaceAdministration through contract954013 through the Jet PropulsionLaboratory and the Office of Naval Researchthrough The model is complicated by the reliance of both the linear grant N00014-85-K-0404.The researchat TRW was supportedby and nonlinear theories for the generation of continuum radi- NASA throughcontract 954012 with the Jet PropulsionLaboratory. ation on coupling from the electrostaticBernstein mode into The Editor thanks M. D. Desch and another referee for their assist- the electromagneticmode. For a simple laser model one has ance in evaluating this paper. an amplification processwhich operates directly on the elec- REFERENCES tromagneticmode, which executesmultiple reflectionsthrough Bridge, H. S., J. W. Belcher,B. Coppi, A. J. Lazarus, R. L. McNutt, the amplification region before being emitted. In the con- Jr., S. Olbert, J. D. Richardson,M. R. Sands,R. S. Selesnick,J. D. tinuum radiation mechanism the plasma instability servesto Sullivan,R. E. Hartle, K. W. Ogilvie, E. C. Sittier, Jr., F. Bagenal, amplify the electrostaticwave, which in turn couplesinto an R. S. Wolff, V. M. Vasyliunas,G. L. Siscoe,C. K. Goertz, and A. ordinary mode electromagneticwave. The radio wave propa- Eviatar, Plasma observations near Uranus: Initial results from Voyager 2, Science,233, 89, 1986. gates away from the sourceinto regionsof lower plasma den- Calvert, W., A feedback model for the source of auroral kilometric sity and is subject to no further amplification. In the coupling radiation, J. Geophys.Res., 87, 8199, 1982. mechanismdiscussed by Lembegeand Jones [1982], for exam- Goldman, M. V., Progressand problemsin the theory of Type III ple, there is an intermediate wave between the electrostatic solar radio emission,Solar Phys.,89, 403, 1983. and ordinary modes called the Z mode, which propagates Goldman, M. V., Strong turbulence of plasma waves, Rev. Mod. Phys., 56, 709, 1984. betweenfw and the L = 0 cutoff [Stix, 1962]. Perhapsmulti- Gurnett, D. A., W. S. Kurth, and F. L. Scarf, Narrowband electro- ple in-phasereflections of the Z mode can reinforcethe ampli- magneticemissions from Saturn'smagnetosphere, Nature, 292, 733, fication of the electrostaticmode at fw under special con- 1981. ditions which could result in enhanced emission of the ordi- Gurnett, D. A., W. S. Kurth, and F. L. Scarf, Narrowband electro- magneticemissions from Jupiter's magnetosphere, Nature, 302, 385, nary electromagneticmode (see Figure 12 of Lembegeand 1983. Jones [1982]). In this scenarioit would be the Z mode which Gurnett,D. A., W. S. Kurth, F. L. Scarf,and R. L. Poynter,First lases with the photon being emitted through the Z mode to plasmawave observations at Uranus,Science, 233, 106, 1986. ordinary mode couplingprocess. Jones,D., Sourceof terrestrialnonthermal radiation, Nature, 260, In any case, angular motion of the beam seemsto be as- 686, 1976. Jones,D., Latitudinal beamingof planetaryradio emissions,Nature, sured by the dramatic tumbling motion of the magneticfield 288, 225, 1980. of Uranus, which is due to the 60ø tilt with respect to the Kurth, W. S., D. A. Gurnett,and R. R. Anderson,Escaping nonther- angular momentum axis of the planet [Ness et al., 1986]. In mal continuumradiation, d. Geophys.Res., 86, 5519, 1981. fact, it could be that the large dipole tilt is responsiblefor the Lembege,B., andD. Jones,Propagation of electrostaticupper-hybrid dramatic difference between the smoothly varying nar- emissionand Z modewaves at the geomagneticequatorial plasma- pause, d. Geophys.Res., 87, 6187, 1982. rowband emissionsfrom Jupiter, Saturn, and earth compared Melrose, D. B., A theory for the nonthermalradio continua in the to the very bursty nature of the Uranian emission.While both terrestrialand Jovian magnetospheres,J. Geophys.Res., 86, 30, Jupiter and Saturn rotate more rapidly than Uranus, the ap- 1981. parent motion of a spacecraftin magnetic latitude at Uranus Ness,N. F., M. H. Acufia,K. W. Behannon,L. F. Burlaga,J. E. P. Connerney,R. P. Lepping,and F. M. Neubauer,Magnetic fields at is much greater than at the other planets. This distinction Uranus, Science,233, 85, 1986. would imply a beam which is broad in azimuth but narrow in Scarf,F. L., and D. A. Gurnett,A plasmawave investigation for the latitude. Jones [1980] has predicted strong latitudinal beam- Voyagermission, Space Sci. Rev.,21, 289, 1977. ing as a consequenceof the linear conversion of the elec- Stix, T. H., The Theory of PlasmaWaves, p. 13, McGraw Hill, New trostatic mode into the electromagneticmode at earth and York, 1962. Warwick,J. W., D. R. Evans,J. H. Romig,C. B. Sawyer,M. D. Jupiter; however,Jones's model would predict rapidly drifting Desch,M. L. Kaiser,J. K. Alexander,T. D. Carr, D. H. Staelin,S. frequenciesas the observerchanged magnetic latitude. For the Gulkis,R. L. Poynter,M. Aubier,A. Boischot,Y. Leblanc,A. Leca- short time intervals for which we have observations this drift cheux,B. M. Pedersen,and P. Zarka, Voyager2 radioobservations is not apparent. of Uranus, Science,233, 102, 1986. Wu, C. S., and L. C. Lee, A theory of the terrestrialkilometric radi- It is not within the scopeof this paper to developin detail a ation, Astrophys.d., 230, 621, 1979. generation mechanismfor the new radio emissiondescribed herein. The main purposehas been to fully describethe obser- D. A. Gurnett and W. S. Kurth, Departmentof Physicsand vations and draw somegeneral conclusions about the require- Astronomy,The Universityof Iowa,Iowa City,IA 52242. ments on the source mechanism. These requirements lead F. L. Scarf,TRW Spaceand TechnologyGroup, One SpacePark, Redondo Beach,CA 90024. quite naturally to a number of possibilitiessuch as solitoh collapse and lasing mechanisms.We suggestthat the lasing (ReceivedMay 28, 1986; mechanism holds the greatest potential for successbased on revisedJuly 16, 1986; the rather simple-mindedapproach taken above. It is obvious acceptedJuly 21, 1986.)