Galileo Orbiter Ultraviolet Observations of Jupiter Aurora

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 103, NO. E9, PAGES 20,125-20,148,AUGUST 30, 1998

Galileo orbiter ultraviolet observations of Jupiter aurora

JosephAjello, • DonaldShemansky, 2 Wayne Pryor, 3 Kent Tobiska, 4 Charles Hord, 3 StuartStephens, 4 Ian Stewart, 3 John Clarke, • Karen Simmons, 3 William McClintock, 3 CharlesBarth, 3Jeremy Gebben, 3Deborah Miller, 3 and Bill Sandel6

Abstract. In 1996 duringthe first four orbitsof the satellitetour the Galileo ultraviolet spectrometer(UVS) (11304320/•) and extreme ultraviolet spectrometer (EUVS) (540-1280/•) performednear-simultaneous observations of theJupiter aurora in boththe north and south polar regions.These observations are modeledto providethe absolute surface brightness of the aurora from the H2 RydbergSystems (B, B', B", C, D, D' -->X bandsystems). The spectraldistribution andbrightness of theEUV auroraare sensitiveto H2 abundance,H2 temperature,and CI-I• abundance.Analysis of theemission spectra indicates that the EUV aurora (800-1200/l 0 are producedover a rangeof altitudescorresponding to slant column abundances of H2 from 10 •6 to 1020cm '•' or greater. The UVS spectra of thefar ultraviolet (FUV) from 1130 to 1700]i are opticallythin in H2,but highlysensitive to theCI-h columnabundance and to the secondary electronenergy distribution. The slantcolumn abundance of CI-I• absorbersfound from modelsof theFUV spectravaried in therange 0 - 10x 10•6 cm '2, indicating the presence of bothhigh altitude aurora,at or abovethe homopause,and deep aurora. The FUV spectrashow C2H•absorption bands near1520/•. The surface brightness ofthe aurora from the H2 Rydberg Systems ranged from 100 to 600 kR and of H Lyman c• was60 to 130 kR for a 2000 km wide oval. The totalpower input to theatmosphere from particle deposition is estimatedto be ~ 1 x 10•4 W.

I. Introduction energy output of the aurora and through self-absorption is Galileo began its planned l 1-orbit tour of the Jupiter sensitive to H a abundanceand temperature,as well as CHn systemin December1995. The I.JV subsystemon the Galileo abundance.The energeticsand CH4 abundancesought to be orbiter consists of two separatespectrometers. The extreme self-consistent between the two sets of data. ultravioletspectrometer (EUVS) measuresradiation from 540 These unique observations consisting of combined EUV to 1280• andis on thespinning portion of the spacecraft.and UVS spectraare an important step in understandingthe The ultraviolet spectrometer(UVS)is mounted on the scan Jupiter aurora. Voyager auroralEUV spectraobtained inside platformand operates over the range from 1130 to 4320 • the magnetospherewere partially compromisedby a high [Hord et al., 1992]. In the early orbits in 1996 at the time radiation environment.'A single preliminary model is shown the boresight of the EUVS crossedthe dawn terminator of by Broadfoot et al. [1981]. Morphological studies by Jupiter, simultaneousEUVS and UVS observations of the Herbert et al. [1987] of the northern aurora, from the aurora from the same source location were performed. The outbound leg of Voyager 2, pinpointed the System 1I[ spectra measured by each instrument provide important longitude,)•- 210øasa regionof maximumbrightness of H information about the energetics and structureof the aurora. Lytx and Ha bandsand the southernmostextent of the aurora The UVS far ultraviolet (FUV) observations furnish the CHn as the footprint of the Io toms. In addition, the Voyager column abundanceabove the aurora and energy output of the UVS spectral resolution was not sufficient to observe the Rydbergband systemsof H a, but provide no information on CHn absorptionstructure. Recently, Morrissey et al. [1997] the H a foregroundabundance. The EUVS also measuresthe obtainedsimultaneous EUV/FUV spectra (900-1650 ]•) and imaging of the Jovian aurorawith the Hopkins Ultraviolet Telescope (HUT)and Hubble Space Telescope (HST). The HUT model analysis was limited to the wavelengthsfrom •JetPropulsion Laboratory, California Institute of Technology, 1000to 1650]k usingmolecular parameters of the H 2 Lyman Pasadena. and Werner system.Significant residuals existed in the 900- 2Department of Aerospace Engineering, University of Southern California, Los Angeles. 1000 ]i wavelengthregion from the unmodeledhigher 3Laboratory for Atmospheric and Space Physics, University of Rydberg states. The Galileo EUVS observations provide Colorado, Boulder. excellent signal-to-noise (S/N) measurements inside the nFederal Data Corporation, Jet Propulsion Laboratory, Pasadena, Jovian magnetosphere. The emergent spectrum from the Califomia. •SpacePhysics Research Laboratory, University of Michigan,Ann Jupiter atmosphereis sensitive to gas temperatureand I-Ia Arbor. abundancealong the optical path. The resonancebands (v' 6 Lunarand Planetary Laboratory, University of Arizona,Tucson. 0)of the Rydberg band systems [Ajello etal., 1984, 1988i appearin the 850-1100]i wavelengthrange. If the gas temperatureis near 1000 K and the Ha columndensity above Copyright1998 by the AmericanGeophysical Union. 10•g cm a, thenthe atmosphereis thick for both the (v', 0) Papernumber 98JE00832. and (v', 1)progressions,which extendto 1160 ]k, a 0148-0227/98/98JE-00832509.00. spectral region that begins to include the UVS. Multiple

20,125 20,126 AJELLO ET AL.: EUV AND FUV OBSERVATIONS OF JUPITER AURORA

I SUN

50 Rj

00 Rj

. <15'Rj ß .:.-: RADIATION • CONSTRAINT' ENCOUNTER ...... PERIOD

-•-Z-AX!S APPROACH

EUV + I o TORUS S/C PERIOD

MAGNETOSPHERE .i.:::.?.::!:::.::!?.::.i.ii?.::!:::.i:.:::.i:.i...... P..E. RIOD i::i::i Figure 1. North ecliptic view of the Galileo trajectory during the orbital tour. The spacecraftz axis is nominally pointed in the Earth direction,and the UVS/EUVS can be boresightedto look at the toms and aurora.The orbit numbersare indicated.The white area is the time during which the bulk of the UVS/EUVS observationswere planned.Jupiter and the Io toms are available in the region called Io toms period. The 15 Rj radiation constraint is a predictedclosest approachdistance to operate the instrument. The dark stippledarea is the time when the cone angleof Jupiter system is less than '90ø and unobservableby the UVS becauseof spacecraftobstructions. scatteringin these bands is unimportant,since emissions in days per orbit for the toms and 1/4 day (6 hours) for the opticallythick bandsare fluorescedat longerwavelengths in auroralobservations by the EUVS. UVS observationsof the optically thin bands of the same progression. aurorawere planned for each orbit to accompanythe EUVS. In this paper we report the EUVS auroralobservations The total Observationtime per pole per orbit, considering obtained on the "Big Four" set of orbits. The three sets of the small duty cycle of the spinning EUVS, was 12 s. observations begin just prior to the satellite/Jupiter Excellent EUV spectra were obtained from these brief encounters:G1, C3, and E4. There were no EUVS/UVS aurora exposures.The EUVS acquiredhigh-quality spectra of the or torus spectra on G2. These first four orbits have large aurora because of the smaller than expected radiation distances from Jupiter during the toms and aurora environment and an increased thickness of radiation observationperiods, roughly definedby 90 ø Jupiter phase shieldingrelative to the Voyager UVS. Modeling techniques angle. These orbits provided the least interferencefrom developedby Shemansky[1985] wereused as a basis for the radiation and the longest and most sensitiveobservations of EUV and FUV models. the Jupiter auroraand Io toms. The geometry is shown in Previous spectral information about the U V aurora is Figure 1 with the Io toms period. The period included4 3/4 known from two decades of FUV measurements by AJELLOET AL.: EUV AND FUV OBSERVATIONSOF JUPITERAURORA 20,127

InternationalUltraviolet Explorer(IUE), HUT, and HST thedata into two wavelengthregions: the blue 1230-1300/• Goddardhigh-resolution spectrometer (GHRS). The past andthe red 1557-1619 /• asproposed by Yunget al. [1982]. historyof FUV observationsis extensive,consisting of Theratio of red/bluehas been used as a measureof the CH4 both high-resolutionspatial imaging (HST) in the Lyman (principalhydrocarbon) optical depth in front of the auroral bandsand H Lymantx andspectral measurements of varying layerwith SystemIll longitude.The two principalresults of spectral coverage and resolution (GHRS high resolution this studyshowed that the auroralbrightness and the color echellespectra of 0.07 /• full widthat half maximumratio peakedat 180ø longitude,indicating a maximum (FWHM)to IUElow resolution of 11 /• FWHM)[Clarke et columndensity of CH4 at 180ø longitude with a variationof al., 1994; Feldmanet al., 1993; Durranceet al., 1982; a factorof 2-3. This resultcould be interpretedas arising Livengoodet al., 1990;Trafton et al., 1994;Prange et al., froman atmospherethat is tmiformin longitudebut has a 1995,1997a,b; Harris et al., 1996]. longitudedependent energy in primary precipitating The mostrecent spectral model of the FUV aurorawas particles.A reconciliationof the long-termIUE resultsof publishedby Liu andDalgarno [1996]. Using a primeset of Livengood(e.g., maximumbrightness and color ratio at HSTGHRS observations from past observer cycles, Liu and 180ø longitude) with the WFPC 2 andFOC imaging is given Dalgarnoderive Ha auroral rotational temperatures of 400- byPrange et al. [1997a]and Ballester et al. [1996].Prange 600 K. The modelincludes the line transitionprobabilities et al. [1997a]give a summaryof the southernaurora from of Abgrall et al. [1993a,b], a major improvementin IUE studies.A maximumin FUV brightnessis observed at 0- modeling rotational-vibrationalcoupling. The model 40ø SystemIlI longitude. simulatedthe auroralspectra on the basis of primary In this paperwe report the first analysisof the near- electronbeams of 20, 100, and 1000 eV energiesand simultaneousFUV andEUV spectra from three of the "Big calculatedthe secondaryelectron distribution based upon Four"orbits, identified as Ganymede 1 (G1), Callisto3 (C3), cross sections for electron ionization, excitation, and and Europa4 (FA). In the next section,we describethe dissociationof Ha. The model did not includegeometry and show the data. For each orbit, we show a pair photoabsorption of the calculated Ha emissions. The of north and south observations for the UVS and EUVS secondaryelectron spectrum expected in the aurorais the instruments.We then briefly describethe modelused to basisof our currentanalysis. H a emissionintensities are analyzethe data. The modelis an extensionof earlierwork sensitiveto the presenceof low-energyelectrons. The by Shemansky[1985]. The followingtwo sectionsshow current analysis also includes effects of atmospheric modelfits to the UVS and EUVS data. A discussionsection absorption. completesthe paper and summarizesthe data/model GHRSspectral line profilesof H Lymantx in the dayglow comparison. bulge region are highly structuredand broad in the wings (-1 /•) [Eraerichet al., 1996],possibly from the collision of supersonicjetsfrom the opposite poles. Global dynamics 2. Observationsof the Jupiter Aurora may be influencedby auroral processes.High-resolution H Lyman tx line profiles in the aurora are found to be broad The instrument description for both the UVS and EUVS also, but with a self-reversedline core [Prange et al., has been providedpreviously for the Galileo Phase 1 high- 1997b]. FUV wide-field planetary camera 2 (WFPC 2) gain prime mission [Hord et al., 1992]. With the loss of the imagingof the Jupiteraurora and Io footprint in the Lyman high-gainantenna, telemetry rates were limited to a total of bandshave been accomplishedby Clarke et al. [1996] and -50 bits/s in the Galileo Phase2 mission.To compensate, Ballester et al. [1996]. Trafton et al. [1994] studiedthe the UVS/EUVS flight softwarewas rewritten to includea new northernaurora spectroscopically in the FUV withGHRS and summingcapability for low light level sourcesor bright observed the rotational lines of the Lyman system near sources that do not fill the field of view. Real-time data 1600/• andof theWerner system near i270/•. Clarkeet al. transmissionof the summeddata allows the UVSand EUVS [1994] andKim etal. [1995] with GHRSstudied the region telemetryrates to be kept at a manageable5 or 10 bits/s, of 1204to 12413,. Thereanalysis of theTrafton et al. whileat the same time allowing the instrument to count all [1994], Kim et al. [1995], andClarke et al. [1994] resultsby the photonsto build S/N. In Phase2 both instrumentssum Liu and Dalgarno[1996] gave auroraltemperatures in the spectralregions over periodsof time that are an integral rangeof 400-600K. numberof real-timeimagings (RIMs) (1 RIM - 60 2/3 s). High-resolutionspatial imaging by the faint object The unit of time for the Galileo spacecraftclock is RIMs. camera(FOC) wascarried on by Gerardet al. [1993, 1994a]. For the auroral observationsthe times sequencedwere They observeda transientbright arc in the north, indicating normally 30 or 60 RIMS (1/2 or 1 hour). The UVS was transient 3000 K or higher exosphere temperatures. In given a 1092 word x 16 bits/word data buffer in the addition,Gerard et al. [1994b] performeda high spatial CommandData Subsystem(CD$) to continuallystore and resolutionsingle rotation morphologicalstudy of the co-addthe spectralinformation at the original high-gain northernaurora in 1993, finding that the auroral oval datarate of 1 spectrumper 4 1/3 s (546 wordsx 8 bits). The correspondedmost closely to theL=30 R• field line andthat EUVSuses a portionof its ownmicroprocessor for storing this auroralstructure persists with time. 1092 words of 16 bits/wordspectral and engineering The most complete morphological study of the IUE data information. The instrument stores data as a two- wasaccomplished by Livengoodet al. [1990]. Harris et al. dimensionalarray. One dimensionrepresents wavelength, [1996] presenteda secondexhaustive analysis of the IUE andthe other dimension represents position in the sky. The data and includedthe Lytx line which Livengoodet al. wavelengthdimension is storedas superpixels(45 for the neglected.Livengood et al. analyzeda northernaurora data aurora)and is a programmableset of the 126 spectralanode setof FUVspectra obtained from 1979-1989. They binned channels(5.9 /• per channel).The programmableset of 20,128 AIELLO ET AL.: EUV AND FUV OBSERVATIONS OF JUPITER AURORA superpixelsis read into the instrumentmicroprocessor at the brightest pixels for each spectral transmission, as the slit start of the auroraobservations as the aurorafixed pattern begins to drift across the two auroral regions. The noise table (FPNT). Superpixels vary in size from two to horizontal axis is time as well as position in the sky. The four channels.The auroraFPNT was chosen to optimize the width of eachpixel is either 1/2 hour or 1 hour, depending resolution in the wavelength range for the H 2 Rydberg on the integration time between each spectral transmission. systemsin the wavelengthrange 800-1250 3,; the torus On C3 the UVS was pointed to the south pole to FPNT was designed to achieve maximum resolution on the maximize signals from the bright 0-50 ø longitudes. On EA torusion linesbetween 600 and1200 3,. Thesky position the UVS was pointedat the north pole to providea high S/N is divided into a programmablesector angle. A sector angle UVS integration on the near-180ø bright longitudes. For is the angle defined by the number of programmablescans each of theseorbits the UVS pointing was then changedat a (one for the aurora)times the spin rate (nominally 3.15 later time to the opposite pole to complete a set of near- rpm) times the scan period for the detector (21.4 ms), simultaneous observations with EUVS. The central meridian resulting in a 0.4 ø sector angle. Sectors are scanned longitude (CML) at the center time of the spectral consecutivelyfrom the starting angle (-90 ø for the aurora) as transmission is indicated for each orbit in Plate la. Black measuredfrom the north ecliptic pole until the commanded pixels or stripes indicate lost data. For the case of G1 the number of sectors have been scanned (24 for the aurora). UVS data, obtained simultaneously with the EUVS, was EUVS spatial registration was improved in Phase 2 to be flawed becauseof a grating anomaly at low spacecraft within +1 detector scan (0.4ø), whereas in Phase 1 the temperatures,a problem rectified on subsequentorbits by spatial registration was within 1.4 ø. This engineering activatingthe UVS supplementalheater. The summaryof the changeallows EUVS to obtain spatially resolvedspectra of UVS/EUVS orbitaldata to be comparedin this paper for the the poles and equator, which had an angular separation of Big Four orbits are listed in Table 1. about 0.8 ø as viewed from Galileo during a typical orbit of We show the EUVS spectraldata for each of the northern the Big Four. orbits in Figure 2 and each of the southernorbits in Figure Plate la shows typical geometry for a simultaneous 3 as an instrument count rate. The integration times in EUVS/UVS observation on the C3 southern aurora, the first secondsfor each of the spectra are indicated. The spectral time a simultaneous UVS/EUV observation was obtained. resolutionof the EUVS is 35 3, FWHM. A total of about 12 The design shows both the UVS slit and the EUVS sector, s (0.2 RIM) per orbit per pole is obtainedwith the angular which are aligned at right angles to one another. The EUVS duty cycle of 4.7 x 104 for the spinning EUVS 0.17ø boresightis mounted90 o to the spacecraftspin axis, which instantaneousfield of view. The peak counts in a channel is the spacecraftcone angle reference axis. The UVS was obtained in a single spectrum is about 10 counts. Hence commandedto thecone angle of 90o andthe clockangle of meaningful analysis can only be performed on the entire the southernaurora. Both instrumentsuse spacecraft(S/C) accumulatedspectrum for each orbit. For example, the peak motion to carry the boresightsthrough the target. For these Lyman ct signal on EA north is 80 counts in 361 RIM of orbits the drift lasts about 6 hours. We show in Plate lb the continuousspin-scan imaging (10.3 s observation time of three EUVS images from the Big Four orbits. They are target). The wavelength-integratedaccumulated spectrum of labeledJOCD, G02C, and C03C for the nameof the sequence 1232 countsrepresents an averageEUV auroraspectrum for loadsthat commandedthe spacecraftin the approachto G1, over a 1/2 rotation of Jupiter. The wavelength scale is C3, and F_A,respectively. Although simultaneousUVS/EUVS divided into superpixels for data collection; the speculumis observations occurred on some of these orbits, reconstructedwith the count information divided equally "simultaneous"is not synonymouswith looking at the same among spectralchannels contained within a superpixel.In area of the sourceregion. these data the backgroundfrom radiation environment and The EUVS imagesare presentedas a Lyman ct intensity the torus backgroundhave been subtracted. The radiation (channels100-120) in counts/sas a color brightness table environmentproduces a variable backgroundcount rate. The (yellow is brightest) for each real-time data transmission. count rate varies between -- 0.023 c/s/pixel (0.3 c/0.2 The image in the H2 Rydbergsystem (channels40-99) is RIM/pixel), a minimum value at the same level as the shown side-by-sidewith the Lyman czimage. The H Lyman radioisotope thermoelectric generator (RTG)cruise rate, to ct signal is causedby at least tttree processes,e.g., H + e, 0.29 c/s/pixel, a maximumrate at the centrifugalequator at direct electron excitation from the auroral electrons or 55 Rj and is negligible. The estimatedtorus background photoelectrons; and H 2 + e, dissociative excitation in the signal (solid curve) for the length of time of the observation auroraand H + hv solar fluorescenceprocesses. The H• is shownin Figure4, comparedto the G1 north total signal Rydbergbands are primarily excitedby electronimpact. The (dotted curve). The estimated torus backgroundis obtained images of Jupiter show both dayglow and auroral signal from the total EUVS torusspectrum from the Big Four set of ratesin H Lymanct. The EUVS auroraimage for the Rydberg orbits.The G1 torusbackground to the auroralsignal below bandsstands out as bright intense stripes on each pole as 800 /1•shows a spectralshape with a peak signal0.5 the same sector number (the vertical axis) occurs for the c/s/pixel(6 counts/0.2RIM/pixel) at 690 •. Theintegrated

Plate 1. (a) The geometryfor the simultaneousUVS/EUVS observationson C3 (G02C-sequence)at time of crossingthe CML of Jupiter.The UVS slit is løx 0.1ø FWHM rectangularfield of view. The EUVS slit is 0.87øx0.17 ø triangular field of view. The sectorsize for the EUVS is 0.87øx 0.4 ø. (b) The EUVSimage of the aurorafor G1, C3, and E4 in H Lyct and in the H• Rydbergbands. Time is the linear x axis scale as producedby spacecraftmotion. The width of each pixel is proportional to the spectral integration time (30 or 60 RIMS), andthe heightis proportionalto the sectorsize (0.4ø ) tracedout by the spacecraftspin in 21.4 ms. GALILEO EUV/UVS OBSERVATIONS OF JUPITER AURORA

(a) C3 AUR 02 304//08:51 S/C DIST = 62 Rj

SPIN ANGLE

EUV SECTOR

uvs

S/C MOTION'

(b)

+e --- H e--- RYDBERG BANDS + hvO --.

/,/23c26C!ilL-225 de9ml, E4

G02C C3 a½//,.; 51 CML=?6

G1 ß 173 0•.22 CML-210 20,130 AJELLO ET AL.: EUV AND FUV OBSERVATIONS OF JUPITER AURORA

Table 1. Summaryof EUVS/UVS Near-SimultaneousObservations for the Big Four Orbits

EUVS

Orbit Day of Year Spacecraft StartTime End Time Duration Fill MU Latitude,Longitude 1996/Observ Distance, (RIMS) (RIMS) (RIMS) Factor, (mid- Mid-Time, ation Mid- R• % time) degrees Time (mean (DOY/hour cosine min) emission angle)

G 1 North 172/23h 44 m 67.6 3488463 3488866 403 7.7 0. 24 58-70N, 150-220W GI South 172/23h 44 m 67.6 3488463 3488866 403 5.3 0.17 76-84S,120-270W C3 North 304/11h55 m 61.7 3675730 3676179 449 4.0 0.35 65-77N,60-120W C3 South 304/09 h08 m 62.8 3675560 3675759 199 6.8 0.30 70-73S,331-60W E4 North 347/22h 27 a 55.2 3737639 3738000 361 10.2 0.47 57-74N, 100-210W E4 South 348/04h 02 m 54.0 3738001 3738300 299 --- none 77S,308W

UVS

Orbit Day of Year Space- StartTime End Time Duration Integration Fill MU Latitude,Longitude 1996/Observ craft (RIMS) (RIMS) (RIMS) Time per Factor, (mid- Mid-Time, ation Mid- Distance, channel, % time) degrees Time R• s (mean (DOY/hour cosine min) emission angle)

GI North 176/06h 08 m 43.0 3493308 3493328 20 1.44 4.1 0.25 53N,185W GI South 176/03h 36 m 44.1 3493156 3493176 20 1.44 5.8 0.34 71S,23W C3 North 304/13h 41 m 61.2 3675880 3676239 359 29.74 2.7 0.27 67N,216W C3 South 304/09h 09 m 62.2 3675700 3675879 179 9.83 2.6 0.26 69S,57W E4 North 347/21h 55 m 55.3 3737639 3737938 299 12.31 2.9 0.45 67N, 210W E4 South 348/04h 33 m 53.8 3738001 3738362 361 14.87 3.0 0.20 69S, 70W

interplanetaryH Lyman o• signal with a maximum count rate spectralresolution of theUVS is 6.5 ]i FWHM.In Figures8 of 2.4 c/s (29 c/0.2 RIM)over channels 100-120 is small and 9 we show the correspondingspectra for the south.Only compared to the auroral signal. For example, the auroral C3 south and EA north were simultaneous with the EUVS. G1 signal level at H Lyman o• on C3 north is 799 c/0.2 RIM. and EA north spectrawere able to capturethe bright aurora The interplanetary signal level is equivalent to 676 R. A from near 180 ø longitude. The observations usedin the G1 nominal FWHM of six channels with broad wings is found comparison were obtained several days later, as shown in for H Lyman o•. The total count spectrum over the G1-E4 Table 1. They actually occurredin the official G1 sequence orbits is shown in Figure 5 with the 1{; error bars from period, whereas EUVS observations were obtained in the signal statistics. The signal statistics are basedon the fact JOCDsequence period. The count rates were highest in G1 that, on the average, a detectedphoton event produces2.5 near the bright 180ø longitude, indicating a bright northern channel counts. The spectral structure of the data shows aurora.The count rate in the H2 bands in the FUV for full fundamental differences between north and south. However, spectral scans of 528 channels is determinedby the UVS both datasets show wavelength thresholds of near 880 duty cycle of either 14 (G-G photomultiplier (PMT)) spectral At wavelengthsbelow H Lyman o• the signalis much weaker scans or 7(F-G photomultipliers) scans per RIM. There are and washedout in the south. It is characteristicof deeper three separatePMTs located behind three separateexit slits penetrationby the primary particles or a differencein the in the focal plane of the spectrometer[Hord et al., 1992]. auroral atmospheresbetween north and south or limb effects. The integrationtime for eachof the 528 channels(1.5 Jd For example,the wavelengthpeaks at 1030 and1100 ,/t are channel width)in the G-PMT wavelength range of 1133 - much stronger in the north. Data from later orbits will be 1931,/t is 6 msper spectrum, summed over the integration used to further study this trend. times per channel indicatedin Figures6 and 7. The signal We turn our attention to the UVS spectra that were rate was smallest in C3 north for the opposite side of the obtained simultaneously or closest in time with the six northern auroral oval than was measured in G1. The EUVS spectra above (three in the north and three in the background count rate in the set of UVS observations is south).We show the UVS north spectrafor G1, C3, and EA small, -4 c/s/channel.The preflight laboratoryelectron and the preflight electron impact spectrumof H2 at 100 eV impact spectrumhas the lowest color ratio CR [Livengoodet discussedin the calibration publications in absolute count al., 1990], rate in Figure6 andrelative intensity, normalized to 1580 in Figure 7 [Ajello et al., 1988; Hord et al., 1992]. The CR=4K]•557_•6•9A / 4rclm0.•300x ' (1) AJELLO ET AL.: EUV AND FUV OBSERVATIONS OF JUPITER AURORA 20,131

lOO 70 1 r ß F l- -8

90 _ TORUS BACKGROUND G1NORTH FOR 11 49 s T •' 60 -5 -7 8o <

6 O 50 •o so 5 . Zm 40

_ 4 • 40 3 =, -- B --•-X o z 20 n•:Tx'• ,-' 2 • 20 • lO 1 J lO __ '7•/-X /•111 BANOS o < o o o

-lO I • Ncu t o ff=880A I I •E O C G• I 500 600 700 800 900 1000 1100 1200 500 500 600 700 800 900 1000 1100 1200 '1300 WAVELENGTH WAVELENGTH(,A,) Figure 2. The EUVSnorth spectrum for threeof the Big Figure4. The EUVS G1 northspectrum count rate (dotted Fourorbits that obtained spectra: G1, C3, andE4. The units line) comparedto the predictedtorus backgroundsignal for signalsare shownin countsper 0.2 RIM andcounts per (solidline) for the lengthof time of the observation.The second.The signal rate magnitudesare close to the total torusbackground is obtainedfrom a summationof all the countsreceived in that orbit.The lo errorbar for the signal total torus spectrafrom the Big Four. There is a torus at any wavelengthcan be foundby dividing the count rate backgroundpresent in all aurora observationsover the by 2.5, whichgives the photonevents. The squareroot of approximately0.2 RIM of the observation.The torus the numberof photonevents is the lo errorbar. We show spectrumis subtractedfrom all EUVS auroraobservations by the lo errorbar for thepeak signal. The wavelengthregions normalizingthe estimatedtorus background signal to the of theRydberg bands are indicated. actualauroral observation. The signalrate magnitudesare close to the total counts received in that orbit. The lo error bar for the signal at any wavelength can be found by and4•lx is themonochromatic lineintensity. The color dividing the count rate by 2.5, whichgives the photon ratiois usedas a figureof meritto comparethe amount of events.The square root of thenumber of photonevents is CH 4 absorption and other hydrocarbon absorption that is the lo error bar. strongbelow 1400 ]t. We showin Figure7 the relative spectra. The presence of CH n absorption and other absorptionin the region is clear, when comparedto the The southernaurorae show a muchlarger variation in both laboratory spectrum.The laboratory spectrumis stronger at signal level and absorption. Figure 8 shows the three all wavelengthsbelow 1400 ]t. Althoughthe signallevels southernauroral observations. The signallevel in the south vary by a factorof 3 in the north,the amountof absorption was largestfor G1 dayside.The relative spectrumis shown appearsnearly the samefor the three analyzedspectra. E4 in Figure 9. The variation in color ratio is striking. E4, hasslightly less absorption below 1400 ]t thanG1 or C3. whichis the weakestspectrum in termsof countrate, is similar to the northern auroral spectra earlier. E4 shows the 1 O0 i [ [ I I I -8 smallest amount of absorption, when comparedto the • SO- G1SOUTH FOR11 49 s T laboratoryspectrum. G1 and C3, which were very strong, .< 80- CSSOUTHFO..... j_ -7 occurover similar System III longitudesand show a large O E4 SOUTH FOR 8 53 s 6 • 7o- uJ 60- • 50- r..o (a) C3 •OR•o, R,•s I • 40- 4 • • 200 E4FOR 299RIMS *' I g 3 • Oø100 ____.' umrreddoto ofG1, C3, E4 • • _ _co 3o- 2 • •< 50 •- 20- -• o D "' "' i (/')J +++ lo err• o lO- 1 Ov -50 +++o error ors 500 600 700 800 900 1000 11 O0 1200 1300 0- - o WAVELENGTH(•) ZEROBACKGROUND • -113 I I I I I • 500 600 700 800 900 1000 11O0 1200 300 • 250 61 FOR449 [•IMS I WAVELENGTH(A) o• 2OO(b) c, •OR,0• R•S "• -I Figure3 TheBUYS south spectrum forthree ofthe Big øø•oo forsignals areshown in counts per0.2 RIM andcounts per• o *+ *• second.FourorbitsThe that signalobtained rate spectra:magnitudes G1,areC3, closeand E4. to Thethe unitstotal ½,• 50 ...... countsreceived in that orbit. The lo error bar for the signal 5oo soo 7oo 8oo •oo •ooo •oo •2oo •3oo at anywavelength can be foundby dividingthe countrate WAVELENGTH by 2.5, which gives the photon events.The squareroot of Figure 5. The spectrumof the total of all countsreceived the numberof photon eventsis the lo errorbar. We show in GI+ C3+ E4 for (a) the southand (b) the north. The + lo the lo errorbar for the peaksignal. signalstatistics are indicatedas plus signs. 20,132 AIELLO ET AL.: EUV AND FUV OBSERVATIONS OF JUPITER AURORA

lOO ii ' ' G1NORTH FOR144 S/CHANNEL (43Rj,185øVV) G1SOUTH FOR 1 44S/CHANNEL (44Rj, 23øW)

ß ii --- C3NORTH FOR2974 S/CHANNEL (61Rj,?6øVV) C3SOUTH FOR 9 82S/CHANNEL (62Rj, 76øW) 80 iii ß' - E4 NORTH FOR1231S/CHANNEL (55Rj,225øW) - .-. 80 : E4SOUTH FOR 14 87 S/CHANNEL (55R j, 225øW) '•' • .!!iil li!i ' ' ' e+ H2---•'- UV(LAB) - • 60 t,.i• •! r• 60 •'•

!?,' 20 :••' ':'". [.t.. . ' - 20 .•i i r;,,,,'4ii'q:. b:. t•.'.-•i.•,'- ':,..:.-."'"'•

o 11 O0 1200 1300 1400 1500 1600 1700 800 11 oo 1200 1300 1400 1500 1600 1700 1800 WAVELENGTH(,,•) WAVELENGTH(,,•) Figure 6. The UVS spectrummeasured simultaneous to or Figure 8. The UVS spectnu•measured simultaneous to or closest in time to the EUVS observations for the north closest in time to the EUVS observations for the south aurorafor G1, C3, andFA. The preflight UVS • + e (100 aurorafor G1, C3, andFA. The l c•error bar for the signal eV) electronimpact calibration spectrum is alsoshown. The levelin any channelcan be calculatedas the squareroot of l c• error bar for the signal level in any channelcan be the plotted count rate times the length of time of the calculatedas the squareroot of the plottedcount rate times measurement,given in secondsin theplot. the length of time of the measurement,given in secondsin the plot.

oval [Clarke et al., 1996] to capturethe bright auroral amountof absorption.G1 and C3 representrelatively deep longitude regions (185ø in the north and 23ø in the south for aurora.The color ratio changeis a factor of 5. G1) establishedfrom IUE and HST diagnosticsof the aurora Supportimaging by the Galileo solid state imaging (SSI) [Prange et al., 1997a]. The figures show the latitude- systemdid not occurearly in the mission. A cycle 5 HST longitude grid system of Jupiter at the time of the program was in place to obtain aurora images at the same observation. time as the G1 (G1JUFIXTMD, sequenceobservation name) north and G1 (G1JUAURMAP) south observations. Auroral 3. Data Analysis images were obtained using the WFPC 2 Na Wood's filter F160WB [Clarkeet al., this issue].The southaurora image, The signalthat is measuredby the UVS or EUVS, summed projectedback in time, for the CML of the UVS observation over all channels for a monochromatic source at is shownin Plate 2. The north auroraimage, projectedback wavelength)• that partially fills the field of view can be in time, for the CML of the UVS observation is shown in expressed Plate 3. The bright aurorain the north near the limb was observedby the UVS in G1. A few hoursearlier, the bright (c/s)= 4nz (c/s/R) (2) aurora shown in the south was in the UVS field of view. Both observationswere well placedon the WFPC 2 reference where4•.r x is the total surfacebrightness in rayleighs(R)

Ii G1NORTH FOR144 S/CHANNEL (43Rj,185øW) 5 6 - • G1SOUTH FOR144 S/CHANNEL (44Rj,23øW) I I C3NORTH FOR29 74 S/CHANNEL (61Rj,76øW) - II C3SOUTH FOR 9 82S/CHANNEL (62Rj, 76øW) • 5 I I E4SOUTH FOR 14 87 S/CHANNEL (55Rj225øW) •"4 .1 :1i'I II't. e+H2'•"'UV(LAB)E4NORTH FOR1231S/CHANNEL (55Rj 225øW) • I• t e+ H2 • UV(LAB)

,..3

• 2 i,

1 ['"";'"''• ' ' . . "..i • I ,, ,,,,,,, '• " t,".•,. • , - "• / t

• • oo • •oo • 3oo • 4oo • soo • •oo • 7oo • soo • 100 • 200 • 300 • 400 • •00 • 600 • 700 • 800 WAVELENGTS{•) WAVELENGTH(A) Figure 7, The UVS relative intensity variation as a Figure 9. The UVS relative intensity variation as a function of wavelength,normalized to 1580 /!, for the functionof wavelength,normalized to 1580 /• for the northernaurora G1, C3, E4. The preflightelectron impact southernaurora G1, C3, and E4. The preflight electron spectrum is also shown. impact spectrumis also shown. AJELLO ET AL.: EUV AND FUV OBSERVATIONS OF JUPITER AURORA 20,133

SUN

// t/ / /

Plate 2. Hubble SpaceTelescope WFPC 2 image of the G1 southaurora as would be observedby Galileo UVS at time of G1 southobservation. The HST observationoccurred at day 176 15• 07m with a CML of 6'.

(unitsof 106 photons/cm2 s)at the top of the atmosphere,build up a spectrum.For example, for a G-G PMT scan the k•=' is the instrumentcalibration for an extendedsource and dutycycle is 1.385 x 10'3 per channelor 6.155 x 10'3 per containsthe factors relating to the instrument field of view, monochromaticline. The kx'xt calibration factors that were transmissioncharacteristics, and detector quantum efficiency, usedare given in the instrumentdescription paper [Hord et fd,,tyis the duty cycle for the spectral bandpass(4.44 al., 1992] and represent the current performance of the channels), and f• is the fill factor for the observation. The instrumentbased on in-flight calibration at H Lyman a from fill factors have been given in Table 1 for all the observationsof the interplanetaryhydrogen glow [Ajello et observations.The fill factor is defined as the percentageof al., 1994; Pryor et al., 1992]. The sensitivityat H Lyman a the field of view that is filled by the auroral source. This for the UVS is 170 c/s/kR and for the EUVS is 3.35 c/s/kR. factor is based on an auroral oval that is 2000 km wide and The Lyman a intensitiesmeasured by the two instrumentsare is uniformly bright over the circumferenceof the L = 30 Rj within a factor of 2 for the C3 south and E4 north footprint from the referenceoval [Clarke et al., 1996]. (A simultaneousauroral observations.The factor of 2 agreement recent analysisof an observationof the auroraloval by SSI is attributed to field of view differences. These intensities on orbit G7 indicates a much smaller width of- 200 km (A. vary between 145 kR and 106 kR for the EUVS and 65 and Ingersoll et al., Imaging Jupiter's aurora at visible 103 kR for the UVS on C3 and E4, respectively and will wavelengths,submitted to Icarus, 1998; hereinafterreferred scaleupward by a factor of 10 for a 200 km wide auroral arc. to as submitted paper); a 10 times smaller width would The Rydberg band contribution to the brightness in the require a factor of 10 increase in the intensities derived channels of H Lyman a has been removed. This here). The longitude-latitudedependence of reference auroral correspondencebetween the instrumentsis reasonable,given ovals has been establishedby WFPC 2 [Clarke et al., 1996], the different fields of view of the two instruments. The which are close to the L = 30 Rj footprint in updated intensity of the Rydberg bands of H2 should be about 10 magnetic field models. The duty cycle for the EUVS has times brighter than H Lyman a, since the total electron alreadybeen given as 4.722 x10'n arising from the spin-scan cross section for the entire Rydberg system is about 10 imaging mode for the array detector; the duty cycle for the times the H Lyman a dissociative cross section [Liu et al., UVS is basedon the single pixel wavelengthscan requiredto 1998]. 20,134 AJELLOET AL.: EUV AND FUV OBSERVATIONSOF JUPITERAURORA

WFPC II LYMAN BAND IMAGE DOY 176//18:19 PROJECTED FOR GALILEO G1 NORTH DARKSIDE OBSERVATION 176//06:20

90 60 120 SUN

150

180 + sic

UTC_TIMEß 1996--176//06:19:48 TargetRa/Dec' 206.33/-7.20 Deg S/C CLK ß03493328:00:0:0 S/Cto Body Center:( 42.919114Rj ) Z-axisPointing ( Ra / Dec)' 102.80/ 25.00 Deg

Plate 3. Hubble SpaceTelescope WFPC 2 image of the G1 north aurora as would be observed by GalileoUVS at time of G1 north observation.The HST observationoccurred at day 176 18h 19m with a CML of 119'. The terminatoris shownnear 60'W longitude.

The model monochromaticline intensity, 4•r./'•.mod, H a ionization,and Tr[x x (q (oo:CHn))]and Tr[x x (q (oo:C2H2))] emergingfrom an auroral atmosphereof H 2, CH n, C2H•, and are continuum absorption terms for CH n and C2H2, other hydrocarbons penetrated to altitude z by a primary respectively. The notation •1 (oo:H2),•1 (oo:CHn),and q particle, ion, or electron is given by (oo:C2H•)are the column densitiesof H 2, CHn, and C2H2 absorberspresent to the top of the atmosphereas measured 4n:Z•.mød(oo) = gxq(z+Az: H2) Tr[xx(q(oo:H2))] from a layer at altitude z. In this paper we assume two analytic forms for hydrocarbonabsorption in the region of the homopause.First, we assumethat for CH4 and C2H2, the Tr[xx(n(oo:CH4))]Tr[xx(n(oo:C2H•))] (3) product of transmissionfunctions is given by

Where g• is the emission rate per molecule per second in a Tr[xx(q(oo:CH•))] Tr[xx(q (oo:C2H2))] particular rotational line J'-• J' '; q(z+Az:H•) is the column density of H• emitters in a layer, Az tlfick = exp[-xx (CHn)] exp[-% (C2H•)] (4) (nominally several scale heights in this analysis), at altitude z for the lower edge of the layer and Tr[x• (q (oo:H•))] This equation is an approximationof extinction by absorber is a transmissionfunction through H2 gas based on a curve foreground gases separated from the emission region of growth formulation of Voigt line profiles and molecular [Shemansky,1985]. Second,we use the analytic solution for AIELLO ET AL.- EUV AND FUV OBSERVATIONSOF JUPITERAURORA 20,135

12 I ...... ANALYTICFORMULA 1 /(1 + All of the UVS observations consideredin this analysis ANALYTICFORMULA EXP(-B) occurredon the darksideor within a few degreesof the dawn GLADSTONEAND SKINNER, 1989 10 - terminator. Since there is no measurable solar albedo in the FUV, there is no added constant of solar reflectance, unlike HUT or IUE auroral dayside spectra. The H• emission rate gx-factor for an emission line (J' -->J'', v'-->v'', Ryd-->X) following electron impact 0.6 excitation from a rotational level (J•-}J', v• -}v', X-->Ryd) in the ground state to rotational level J'is given by

ii 0.4 g• (J' -->J' ', v' -->v' ' )=F , (œ,z)rox(J'-->J' ', v' -->v' ')x

i 0.2 q(z,H:) Y•f(vi,Ji,r(z))Q(s, vi,v' ;Ji,J') (6) vi ,Ji=J',J':l: 1 l l

0 ,c'i, I'r J, • .... , I l,'•,::::::l 200 1300 1400 1500 1600 1700 where F, (œ,z) is the secondary electron distribution at WAVELENGTH altitude z as a function of energy œ; Q(œ) is the excitation Figure 10. Transmission function of hydrocarbon cross section for rotational level v', J' from initial level absorbersin the FUV for two analytic models and the model v•,J,; p is the fraction of the population of H: molecules in atmosphereof Gladstone and Skinner [1989]. level v•, J• at temperature T(z); and co,is the emission branching ratio. At the resolution of the Galileo UVS and EUVS, where tens of rotational lines are summed, we have a Chapmanlayer aurorain a homogeneousH2, CHn.,and C2H• verified that the branchingratios are adequatelydescribed by atmosphere. In this case it can be shown (G.R. Gladstone, using the Allison and Dalgarno [1970] oscillator strengths private communication,1997) that the product and Honl-London factors for the calculation of transition probabilities. The more accurate rotational line oscillator Tr[xx(q(*o:CHn))] Tr[xx(q strengthsfrom Abgrall et al. [1993a,b] are necessaryfor the high-resolutioncapability of HST, where perturbedrotational = 1./[1. + q;x(CHn)+xx (C:H:)] (5) lines are resolved [Liu and Dalgarno, 1996]. However, the where q;x(CH4)-t- '1;x (C:H:) is the total hydrocarbon slant more accurate excitation cross sections of the Rydberg optical depth above the peak of the Chapman layer from bands, recently measuredat high resolution, are requiredto both CHn and C:H:. We show in Figure 10 the transmission produce synthetic spectra [Liu et al., 1998]. These excitation function wavelength dependencies for the analytic cross sections are given in Table 2 for the B, B', C, D' approximations,(4)and (5), which are comparedto the statesof H2. The B state cascadecross section value at 100 model atmosphereresult with a Chapmanlayer peak at 370 eV [Ajello et al., 1988] is stated for the combined cascade km [Gladstone and Skinner, 1989]. The difference between cross sections of the E, F --> B and G,K --> B andH--> B the analytic approximation equation (5) and the model transitions [James et al., this issue]. However, the cascade atmosphereca•lculation is not significantand is causedby crosssection at low energy (œ-20 eV) is reducedby 1/2 for the grid spacing.A displacementof the Chapmanlayer peak the modeling process of this paper from earlier estimates by a few kilometers can bring the transmission functions [Shemanskyet al., 1985; Ajello et al., 1982, 1988]. The for the analyticmodel and the model atmospherecalculation recent modeling of high-resolution FUV laboratory spectra of Gladstoneand Skinner into close agreement.The two at 20 eV indicates the cascadecross section is roughly 1/2 analytic models differ by nearly a factor of 2 in the of the earlier estimates (X. Liu, private communication, estimatedCH n column densities encounteredin a deep 1998). This cross section is extremely important in Galileo aurora.The modelshave enoughfree parametersthat interpreting FUV spectra, since the model spectral it is not possibleto choosebetween the two for simulating distribution and color ratio (1) is affected by this cross the data. From this point on, we chooseto show only the section. Since the EUVS spectra are optically thick with a representationfrom (4), except for one case for the UVS to measuredwavelength threshold of 880 •, weneglect in the illustratethe effects of C2H:. modelthe emissionsof the n=4 Rydbergseries, B' ' ly.+.and D' 1H., to the groundstate, whichradiate primarily from 800-900./t [Ajelloet al., 1984;Liu et al., 1995]. Table2. ExcitationCross Sections from the X ly.g+ GroundState at 100 eV for H: In order to calculate the g factor, it is important to quantify the flux of secondary electrons from electron State CrossSection at 300øK, Rangeof Vibrational impact ionization of H:. We show in Figure 11 the 10'17 cm: Levels Considered secondaryelectron distribution, measuredby Opal et al. [1971]. The secondaryelectron energy spectrumis largely 8:5?. (Direct) 2.62 0-34 C :l-I, 2.41 0-13 independentof primary impact energy. This secondary B' :5'.+, 0.16 0-7 electronflux was usedin the model calculationsfor exciting D :Fl. 0.30 0-2(FI+) the H: aurora. The secondary electron flux can be 0-14 (Fl-) parameterizedby the sum of four Maxwellians: B ' Y•+,,(Cascade) 0.29' 4 'Cross sectionsat low energy are reducedby 1/2 from earlier F• = Z Cn•n(Tn) (7) estimates [Ajello et al., 1982, 1988]. 20,136 AJELLO ET AL.: EUV AND FUV OBSERVATIONS OF JUP1TERAURORA

' ' ' ' I I lOo model.The differencesbetween the Maxwellian secondary electronsand the monenergeticgenerated electron impact

10-1 inducedfluorescence spectrum are as large as 50% in the FUV for wavelengthsabove 1270 •. Thesetwo distributions represent the extremes of a "cold" and "hot" electron • 0-2 distributionfunction, respectively. The spectrain Figure12 ß ADTAOPALe, t al. (1971) •_ representthe extremeof effectson the FUV spectrumby the x nMODEL,• 4 '• secondaryelectron distribution for the samegas temperature. . 10'3 The enhancementof low-energyelectrons in the Maxwellian f(E): n•__Cn 1 Cn* (E/En)En* EXP(eV) (-E/En) Tn (øK) '• 2 206756 2 84 3 25 x 104 distributionserves to pump the Lyman bandsrelative to the 0 4056933 9 7 1 13 x 105 10-4 Wernerbands. The intensityof the v'' progressionwith v' 2 84163x 10'2 39 5 4 58 x 105 9 84 x 10'4 18036 2 09 x 106 = 0 of theB stateis increasedrelative to progressionsfrom higher vibrational levels. The use of the 100 eV

10-52 ...... 10• ...... 100, ...... 1000 monoenergetic"hot" electron distributionin analyzing E(eV) Jupiteraurora spectra will lead to smallerCH n columll Figure 1 1. Secondaryelectron flux from e + H2 primary abundancesthan the coldMaxwellian distribution. The actual ionization.The dataof Opal et al., [1971]and model are Jupiterelectron distribution at anylongitude in the auroral shown,along with the model parameters asdescribed in the ovalmay be variablewith altitude and time between these text.f(E) is the secondaryelectron distribution. two extremes. The final consideration in the modeling is an understandingof the atmosphericdistribution of H2 andCH n, whereC, are constantsand (I), are Maxwelliansat electron the dominanthydrocarbon in the nonauroralatmosphere. temperatureT,. The constantsC, and correspondingC2H 2, thenext most abundanthydrocarbon, has a mixing temperaturesare given in Figure11. We have comparedthe ratio in the peak auroralregion altitudes(250-400 km) of Opalelectron production spectrum in the energy range 10- 10-2 to 10-2 of CHn, albeit for thenonauroral atmosphere 1000 eV againstthe steadystate distributionfor electron [Gladstoneet al., 1996]. The actualmixing ratios of the precipitationin the Jupiter atmosphere[Singhal et al., hydrocarbonsin the auroral atmosphereare not known. 1992]. The agreementin this rangeis excellent. Livengoodet al. [1990] give argumentsfor both the net By consideringall electrons from 10 to 1000 eV, we upward and downward motion of the semi-infinite accuratelymodel the effect upon the g factor of cross photochemicallayer of hydrocarbonsby consideringthe sectionvariation with energy.The preponderanceof low- addedeffects of vertical transportand electronimpact energyelectrons in the distributionfunction enhances the dissociation.They estimatean upwardor downwardmotion effect of the (E,F), (G,K) (double minimum potential of the homopauseby 1 or 2 scaleheights, H= 30 km for functions)and H---) B cascadecross section (optically 200 K. With this consideration,we showin Figure13 the forbiddenexcitation processes)contribution in the 1270- large scale, estimated distribution of the important 1600• spectralregion, relative to a monoenergetic100eV atmosphericconstituents in the auroral atmosphere, based electrondistribution. We showin Figure12 the importance on the Galileoprobe results of the equatorialregion for Ha of the electron energy distributionto the model FUV [Seiff et al., 1996, 1997] and CHn from photochemical spectrum.In addition, we overplot the 100 eV prefiight modelsfrom beforethe Galileo probe data [Gladstoneet al., electronbeam calibrationspectrum, together with a 100 eV 1996]. The effect of the increase of the accelerationof gravityat 60øN is the primaryeffect that hasbeen used to modify the Galileo probe results.We have not modifiedthe 'l temperaturedistribution of the Galileo probe. Beneaththe

I 0 100 eV PRE-FLIGHTEBEAM LAB DATA

i tl lOOev MONOENERGETICDISIRIBUIION MODEL

"• 0 8 MAXWELLIAN01STRIBUTION MOOEL A '•' 15oo

z 0.6 > [•016 C•V•2] (800-- 900 A) EUVLAYER3

_ooo

0.2 • H2,SEIFFelo1,1997 •,',• ',,,, ...... , 0.0 ...... • .... 500t/CH,, EQUATORIAL, GLADSTONE....'x•1996 'N• [10 •eCM -2](900-1000 A)EUVLAYER2 -- • • 00 • 2OO • 3OO • 4OO • 5OO • 6OO • 7OO / [1020 CM -2] (1000-1200 A) œUVLAYœR1 WAVELENGTH t CHCH4'POLARPOLAR'• HOMOPAUSE, • •' .••['102ø-21 CM-2]FUV LAYER u • ....L:.:''::...... tL • (12oo-1700 A) Figure 12. The Maxwellianmodel of secondaryelectrons 1014 1016 1018 1020 1022 1024 of Figure 11 are used for the productionof an electron COLUMNABUNDANCE (cm -2) impactinduced FUV spectrum and a monoenergeticelectron Figure 13. Thelow-latitude nonauroral atmosphere based impactmodel spectrum at 100 eV. In addition,we show the on theGalileo probe results [Seiff et al., 1996;1997] and preflightlaboratory spectrum at 100 eV spectrum. thephotochemical models for CHn [Gladstone et al., 1996]. AJELLOET AL.: EUV AND FUV OBSERVATIONS OFJUPITER AURORA 20,137

homopausethe mixingratio of CHn is 2 x 10'3. The lowest to determine a model signal intensity in c/s/channel. The layerin themodel for theEUV andthe highestlayer for the FWHM variesslightly over the spectrumby 10%. We then FUV, near the homopause,are the only layers whose comparethe model signal intensity with the data directly. emissions may be affected by CHn abgorption in the This type of calculationis necessaryfor the EUVS, rather modelingprocess. We mark in Figure 13 four important than trying to model the calibrated data. The instrument altitudesabove the 1 bar referencelevel, correspondingto sensitivity is not constant over the FWHM of 35 /•. the columndensity locations of 10'6 cm'2 (0.0001 I.tbar) at Shemansky [1985] and Broadfoot et al. [1981] have 1240km, 10'8 cm'2 (0.006 I.tbar) at 660 km, 1020cm '2 (0.4 successfullyused this approachfor the Jupiterdayglow for I.tbar) at 340 km, and 102• cm'2 at 265 km (3 I.tbar) the VoyagerUVS. The Galileo EUVS, which is the flight (pressures).The pressuresand altitudesfor each layer can spareVoyager UVS, has been modifiedto have improved vary significantly,depending on the temperaturestructure in radiation shielding, a holographic grating with reduced the auroralzone. The auroraltemperatures are very uncertain scatter,a smallergrating constant,and a larger field of view and time variable with effectsfrom joule heating and H3* [Hord et al., 1992]. cooling [Kirn et al., 1992]. We addressthe modelingproblem, realizing that more 4. Modeling the UVS observations than one atmosphericlayer will contributeto the emergent UV spectrum.We indicateschematically in Figure 13 the We have modeled the UVS observations for G1, C3, and layer that contributesthe most emissionto fourwavelength E4. The resultsare shown in Figures 14, 15 and 16. The intervalsin the EUV andFUV foundfrom this study with the FUV model spectrahave very little dependenceon H2 understandingthat the FUV layer may be very extendedin foregroundgas abundance.Electric field modificationsof the the homosphere.The FUV layer can be extended,since slant electron energy spectrumare not consideredhere. However, optical depth unity varies by a factor of 10 or more from the spectraare very sensitiveto CHn absorption [Gladstone 1250/•, whereCH n opacitydominates to 1600/•, where and Skinner, 1989; Durrance et al., 1982]. The models have C2H2and other hydrocarbon/aerosolopacity is important. varying amountsof CHnslant column densitiesfrom 0to 2 x The temperaturesfor theselayers in the equatorialregion 10'7 cm'2 with the C2H 2 abundancemaintained at 104to 10'2 havebeen measured by the Galileoprobe. In the modeling of the CH4 abundance[Gladstone and Skinner, 1989; process,we use a temperatureof 1000 K for the 10'6 cm-2 Gladstone et al., 1996]. The best model fits to the data for layer,900 K for the 10'8 cm -2 layer, and600 K for the 102ø the threenorth observations occur at a columndensity of 0 and 102' cm-2 layers, basedon the mean line-of-sight - 1.0 x 10•6 cm'2 of CH4for both sets of transmission rotational temperaturefrom HST analysisof high-resolution function models. The best model fits to the data for the FUV spectra[Liu and Dalgarno, 1996]. In general,we find southobservations occur at a columndensity of 7.5 _+2.5 x the sensitivityto the H2 gas temperatureof the modelsto be 1016cm '2 for G1 andC3 andnear 0 cm'2 for EA.The sampled very weak in the 200-600 K range, the extremes of latitudes and longitudes for G1, C3, and EA north were temperatures measured by Galileo probe and HST, within 35 ø of one another. The FOV of the UVS is 0.1øx respectively,in this region of the stratosphereabove 250 1.0ø. On the sky the arc length of the aurora that is km altitude where the auroral FUV emission probably observedis about 6000 km and ensuresa range of distinct occurs.We will be discussingthe variouslayers in the next regions are observedin each case. Ballester et al. [1996] sectionas we modelthe FUV andEUV spectra. have measured small-scale aurora structures of 3300 km arc AtomicH is an importantabsorber in the atmospherefor length. The UVS north aurora were all observedin the IUE wavelengthsbelow 911 /•. Weuse H/H2 mixing ratios given enhancedemission region fixed around180-210 ø [Ballester by Shemansky[1985]. The mixing ratios are 0.021, 0.13, et al., 1996; Gladstone and Skinner, 1989]. We have and 0.67 for verticalH2 columndensities of 1 x 1018cm '2, 3 sampledtoo few points in theseobservations to make any X 1016cm '2, and 1 x 10'5 cm-2, respectively. The atomic H statement on morphology. We consider in more detail the compositionat the presenttime is highly speculative,since energeticsof the auroraand the locationof the homopause modelsof the auroral atmosphereare not well established by comparisonwith the EUVS. [Waite et al., 1983]. Most of the UV emissionthat escapesthe atmosphere The spectralanalysis scales model spectrafor a least takesplace in the Lyman bandsof H 2. The Lymanbands (B squaresfit to the data. The requirementof the model is to --> X) are spreadout over 35 upper state vibrational levels minimizethe differenceC• ød(c/s) - C•"' (c/s) in the least connected to 14 lower state vibrational levels as well as the squaressense, when summedover the numberof channels,n: Lyman continuum.The Lyman band system constitutes ~10,000 rotational lines at-600 K. At an H2 column R(4KI u•n, 4• I • ,4• I uyca) densityof 102øcm '2 the line centeroptical depths of the Rydbergresonance bands (v' ,0) below1110 /• are saturated. 1In[ •[C•ø•t(i)-C•ata(i)]2} 1/2 (8) At this columndensity, self-absorption from v'' =1 is also i=l significant. High auroraltemperatures of 1000 K can even bring about self-absorption from v' '=2. The upper The functional dependenceof the residualR is def'medin wavelengthlimit for self-absorptionby v''=l is 1160 J• termsof the emissionfrom eachlayer. The dataanalysis is andby v'' =2 is 1230 ]k. For the lattercase, Lyman tx performedfor both the UVS and EUVSobservations by a fluorescencewould affect the spectraconsiderably. The self- convolution of the model line intensities with the absorptionmanifests itself in the EUVS spectrttmonly. The instrument transmission function, taken to have a three strong Rydberg transitions (C, B', and D) have all trapezoidalline profile with a peak half-width of 10% of the their emissionin the EUV below1300 /l, andare greatly instrumentFWHM, multiplied by the wavelengthsensitivity attenuatedin the radiative transport through the Jupiter 20,138 AJELLOET AL.: EUV AND FUV OBSERVATIONSOF JUPITERAURORA

Table 3. UVS Modelin• Results

Orbit Model Observed Observed Model EnergyInput SlantCH n SlantCH n SecondaryElectron Emission Emission Rate Emission Emission Rate at 2000 Model Model Distribution at Rateat Top from Galileo Rate from Rate at km Wide Abundance Abundance Level of 1 x 1020 of Atmos- 1160-1725,1.,, Galileo Sourcein Source from(4), b from(5), b cm-2, phere kR Lyman-ct, Atmosl>here Regionfor 10•6 cm -2 10•6 cm '2 10• e/crn2s Rydberg kR Rydberg 10% UV Bands, Bands, Efficiency, kR kR ergs/cm2s

G 1 North 529 290 70 941 1510 1 1 8.2 G1 South 593 541 130 1920 3072 5 7 17.0

EB North 338 284 73 601 961 1 1 5.3 C3 South 292 130 65 945 1512 5 10 8.3

E4 North 460 226 103 818 1312 0 0 7.2 E4 South 176 226 86 312 499 0 0 2.7

EUVS Modelinl•Results for LowestLayer Orbit EmissionRate at Top of EmissionRate at Top of Atmospherewith 1 x 1017cm 2 Atmospherewith 2 x 1017cm 2 of CH4,e of CH4,e

G1 North 129 1025 G 1 South 162 1238

C3 South 128 1057

E4 North 120 1140

Without Lyman at C2H2column abundances are 10'1 - 10'2 of CI-h. Most of the emissionbrighmess from the Rydberg band systemsoccurs in the FUV.

atmosphere, both by self-absorption and hydrocarbon at the sourceregion is 941 kR from the Rydberg bands. We absorption. The spectral data from the EUV and FUV assumea 10% conversionefficiency of input energyto UV observations will be comparedfor the first time to models radiatedpower. The energy input deposition rate is 151 to demonstrateself-consistency can be achieved. ergs/cm2/s.The auroralsource is assumedto be 2000 km The resultsfor the UVS analysisare given in Table 3. In wide and to fill the 0.1 ø width of the UVS slit. The observed detail, the model for the G1 observations for the north countrate for G1 north requiresa secondaryelectron flux of indicatean emergentRydberg band (800-1700 /•) surface 8.2 x 108electrons/cm2/s for a single-layercolumn density brightness of 529 kR. The modeled surfacebrightness can of 1 x 102ocm 2 (-350 km altitudereferenced to 1 bar). The be comparedto the measuredsurface brightness of 290 kR, radiated power observed in the aurora is similar to HST integratedover the wavelengthrange of 1140-1625/•. In measurementsof an intense FUV aurora [Ballester et al., addition,70 kR is attributedto H Lyman at, not includedin 1996]. The root-mean-square(RMS) uncertaintyof the 1 x the H 2 band emission rate. The models for the G1 10• crn-2 CH n columndensity model to the datais 38% for observationsin the north are shown in Figures 14a and 14c Figure 14a. The channels containing Lyman at were not for the transmissionfunction models describedby (4) and consideredin the RMS calculation.The largest discrepancy (5). The best fit models for both cases indicate a column occursin the 1300 - 1400 /• region and can be densityof CHn of about1 x 10• cm-2 The two modelsdiffer accommodatedby the simple one-layer model of (4) or (5) in the amountof C2H2, where we have useda mixing ratio of by includinga "warmer"electron distribution than the Opal 10'2 of CHn for Figure14a and 10'• of CHn for Figure14c. distribution. The effects of the differing amountsof C2H2 to the FUV The comparison of the model and data for G1 south is spectrum occur at wavelengths of strong absorption at shownin Figure 14c. The model resultsfor the G1 aurora for 1340,1440, 1480,and 1520 /•. A valueof 10-1 of theCH n the south listed in Table 3 show a radiated surface abundancefor the C2H2 abundanceseems to causetoo much brightnessof 593 kR in the Rydbergbands at the top of the absorptionin the deepC2H 2 bands,whereas the 10'2 mixing atmosphere.The measuredsurface brightness is 541 kR at ratio shownin Figure 14a is not enough.The sensitivity to the spacecraftin the UVS wavelength range and 150 kR in measuringCH n beginsat a slant columndensity of 1 x 10• H Lyman at. The secondaryelectron flux requiredfor this cm'2. At thiscolumn density the optical depth is about0.15 brightnessis 1.7 x 109electrons/cm2/sfor a columndensity near 1250 /•. of 1 x 102ocm '2. The southaurora is a factorof 5 deeperin Additionalauroral modeling parameters are given in Table the CHn slant column density than the north. We find Figure 3. Table 3 indicatesthat prior to transmissionthrough the 14b, which uses the transmission function of (4), indicates overlying atmosphere,the surfacebrightness for G1 north a bestfit of 5 x 10•6 cm '2 for theCH n columndensity, while AJELLO ET AL.: EUV AND FUV OBSERVATIONS OF JUPITER AURORA 20,139

, • lOO ,, data (a) G1 Nodh Jupiter Atmosphere TransmissionFunction _j 80 SLANTCt'I,• = 0 cm-2 1.0 SLANTCH,• = 1 X 10•6cm2 z • o.• • 60 /' ,'"":• SLANTCH4= 2 X 1016Cm2 , Ii/ / •' SLANTCH,•= 5 X10•6cm -2 • 0.6 40 •'i',i" ' S,ANT CH, = •X

J • 0.4 o 20 • 0.2 0 . / "•, • 0.0 1100 1200 1.300 1400 1500 1600 1700 1800 1600 WAVELENGTH 1300 1400 1500 WAVELENGTH

120 I! .! /• ...... Galileo UVS 01 South FUV Aurora I 11 _ SLANTC., = 2 X 300 ,, , lOO / II SLANTCH, = 250i.... , (9) 8o , • s• Hzand Maxwellion •condoryElectrons Model 60 200i• • ,, , RMSfit=1.44 • Model*Transmission 4o o o 100 2o 50 0 L o• 1100 200 1500 1400 1500 1600 1700 1800 WAVELENGTH(•) 1 •00 1400 1 •00 1600 WAVELENGTH lOO

60 / I I•1 • s• c•, = s x •o"• -•

20 0 1100 1200 1300 1400 1500 1600 1700 1800 WAVELENGTH(,•)

Hydrocarbon Absorber Cross-Sections Jupiter Model Atmosphere :••..,.:,•_,,•• 1 X [ X • ' NS=1'2 x 1ø1ncm-3 X X (h) lO-17 •o.•_-•.,x,o,•cm-• • • ¾ 10-18/1010-15-16 _]•:...... ]_]]]]/•.].] • , ...... • ...... ß._ ' •=':'z :..-_ ,C4H,o (d)_ 500 600 10-19 L C•H4N• C•H,• . c•,, ....-.,• = x _ 1300 1400 1500 1600 o 400 WAVELENGTH(A) ...... ?H••ol Totol Density •nsityZs•• Hydrocarbon Absorber Cross-Sections '-• • 1010-15-16 • A C2H4(e) • < 300 10 17 200 10 18 19 C•HC••'X.,._ C•H...... ,,,,, ..... •0 100 ß ';-., , , ...... , 12 14 20 1.300 1400 1500 1600 10 6 108 1010 10 10 1016 1018 10 WAVELENGTH(/•.) Density(cm -•) Figure 14. The G1 UVS FUV observationson day of year (DAY) 176 and model (a)north and (b) south for the hydrocarbontransmission function of (4) and model (c)north for the hydrocarbontransmission function of (5), 1./(1. + x). In Figures 14d and 14e are the absorption cross section of the six absorbers, CH4 [Mountet al., 1977],C2H2 [Nakayama and Watanabe,1964; Suto and Lee, 1984; Wu et al., 1989], CaI-I6 [Mount et al., 1978], C•H•, C•H•, and C•H]o (see Gladstone et al. [1996] for remaining absorption cross sectionreferences) to model the G1 south aurora. In Figure 14f is the transmission function derived from the Gladstone et al. [1996] model atmosphere.The numberdensity at the Chapman layer peak is 1.2 x 10TM cm -3. We plot the best fit modelin Figure14g of the G1 southobservation compared with the (31 southdata and the sourceregion H 2 emission spectrum.We plot in Figure 14h the altitude profile of the modelatmosphere [Gladstone et al., 1996] andthe volumeemission rate, V(cmas").The Chapmanlayer vertical peak column abundancesare listed along with the altitudez, and numberdensity N(z,). the transmissionfunction of (5) givesa bestfit of 7 x 10'6 theobservation in the 1300-1500• wavelengthrange. The crn-• for the CH4 column density. It is not possible to Ix factors for the observationare given in Table 1 and are computea meaningful RMS uncertainty,since it is not not much different from those for the north. The major possibleto model the Jupiterspecmnn in the wavelength resultof the analysisof G1 southis the failure of the simple rangeof 1300-1500]k. The uncertainties in the E,F andG,K one-layermodel to describedeep aurora.We can interpret and H cascade cross sections or the electron energy this effect to indicate the presence of additional absorbers distribution could not account for this large deficit in the and additionallayers. A more realistic atmosphericmodel of one-layermodel. As we have noted, the model in the same the primary absorbers altitude distribution is requiredto spectralregion in the north is also somewhatstronger than explain the FUV spectrum. 20,140 AIELLO ET AL.: EUV AND FUV OBSERVATIONSOF JUPITERAURORA

To further explore the "extra" absorbersin the G1 south legitimate means of obtaining the CH n abundancesfrom the spectrum,we make use of the auroral atmospheretreatment color ratio of (1). A similar analysis for G1 north gives a of Gladstone and Skinner (G & S) [1989], including Chapman layer peak of 405 km. In order to model deep Chapman profiles of the auroral emission rate in the H: aurora, the atmosphere is better approximatedby at least bands, V(z). V(z) can be written two layers or multi-layers, as above, with (1) a topside diffusive separation layer above the homopause, where the V(z) = k N(z) exp[-N(z)/N(z,)] (9) only emittersand absorbersare H and H:, and (2) a layer of mixed H:and hydrocarbons. wherek is a constantof proportionality, N(z) is the total A major result of this analysis is confirmation of the numberdensity at altitude z, and z, is the altitude of the IUE result that C:H: is present above the auroral emissions. Chapmanlayer peak volume emissionrate. The original G This is particularly clear in the fit to the strong absorption & S model atmosphereincluded equatorial photochemical bandnear 1520/•. Areadditional absorbers actually present altitude profiles of just two hydrocarbonabsorbers, C:H: and in the aurora?Photochemical models of Jupiterpredict CHn. Their work found IUE FUV spectracould be fit with to be presentin a similar amount to C:H:. We concludethat these two absorbers. However, the recent work by the data are also consistentwith the presenceof C:H6 above Gladstone et al. [1996] plots altitude profiles at 60ø polar the aurora. The longer wavelength data absorption edge of latitudesfor six absorbers:CH 4, C:H2, C2H4,C:H 6, Calla, and C:H6(above 1550 /•)comparedto CH4helps to minimize C4H•0. The altitude distributions derived by Gladstone et systematic offsets between data and model from 1400-1500 a1.[1996] did not make use of the Galileo probe results, and /•. TheCall6 absorption edge below about 1550 /• servesto altitudes in this discussionbased on these profiles will need improve model fits to Jupiter's reflection spectrum from revision. 1500to 1740/• [Gladstoneand Yung, 1983). As shownin The G & S transmissionfunction (Figure 10) differs from Figure 14g, the absorption edge occurs at even longer the analytic functions in the sense that the FUV wavelengths in C•Ha and Chill0' Use of a modem transmission is increasedin spectral regions (1100 - 1300 photochemical model with six hydrocarbon absorbers does /!,) wherethe absorptionis largest.This is the effectthat is improve the fits to the spectrum, although the neededto explain the deep aurora,like G1 south. The mean identifications may not be unique. Other aerosols or transmission function is defined hydrocarbonsmay also producecontinuous absorption that affects the FUV spectrum. The major remaining Tr.•()O = JV(z)exp[-%x""(z)/!.t] dz/ JV(z) dz (10) discrepanciesbetween the G1 south data and model remain in the 1300-1400/• range.At this point, the use of a "warmer" electron distribution and a more accurate E,F cross wherexx"" is the total vertical optical depth and Ix is the sectionvalue would probably reduce theseremaining residual cosine of the emission angle. For this analysis, we use the differences. Opal et al. [1971] secondaryelectron function to generate The model for the C3 observations for the north indicates the FUV fluorescence spectrum, without any overlying a radiatedsurface brightness of 338 kR. We show the model hydrocarbon absorption, as the estimate of the relative fits in Figures 15a and 15b. In the north for C3 the best fit spectral intensity of the source emission region. A typical occursfor a columndensity of 1 x 10•6 cma for theCH n thin emission spectrumis shown in Figure 12. For the G1 column density for both transmission function models south aurora data, the use of the new transmission function considered.The secondary electron flux requiredfor this with the six absorbers(CH•, C:H:, C:H•, C2H6, C3H8, and brightnessis 5.3 x 108electrons/cm:/s for a columndensity CnH•0)provides improved agreementwith the data, compared of 1 x 102øcm -2. The RMS uncertaintyis about 65% in to models using the analytic expressions of (4) and (5) Figure 15a. The models in Figure 15b show a best fit for the shown in Figure 10. We varied z, in the Chapman profile, which in turn, varied the mean transmission function, until we m'mirnized the RMS deviation between the G1 south data 50 _ _ SLANTCH. - • X •0 %• ' (a)C3 North and the G1 south model, which is the productof the mean 40 SLANTCH,, - 2 X 10'•cm • transmissionfunction and the sourcespectrum. We show in 30 jq SLANTCH,= 5 X10'•cm 2 / • .... S•NTCH.= 1 X 10'/cm • Figures 14d-14h the results of the analysis. Figures 14d and 20 •v ':'v. ... 14e show the absorption cross sections for the six lO ... •.:• • absorbersused to calculate the total optical depth. Figure •o 14f shows the mean transmission. Figure 14g comparesthe 11oo 12oo 1300 1400 1500 1600 1700 ] 800 WAVELEN6TH(A) model and the data and gives the RMS fit. For G1 south, we find an RMS fit of 1.44%. We show in Figure 14h the altitude of the volume emission rate and the altitude profile of the principal constituents. The peak in the Chapman • 40 ] I SLANTCHa= 5 X 10•'cm -2 •Z• 30 i•III I ...... SLANTCH,•=1 X 10"crn-' functionoccurs at 358 km. We also list the Chapmanlayer 20 tx• column densities for each atmospheric component and the total atmospheric density. The slant column densities of .H? :'.1 .. CHn andC:H: for G1 southby this techniqueare 8 x 10•6 11 O0 1200 1 300 1400 1500 1600 1700 1800 cma and5 x 10•s cm':. The slantcolumn densities are nearly WAVELENGTH the same values found from the analytic transmission Figure 15. The C3 UVS FUV observations on DOY 304 functions of (4) and (5). Although a single-layer model and model (a)north and (b) south for the hydrocarbon cannot fit the spectra from deep aurora, it appears to be a transmission function of (4). AJELLO ET AL.: EUV AND FUV OBSERVATIONS OF JUPITER AURORA 20,141

absorption, and (3)absorption by the CH 4 column. We •60•" 50 AA ....SLANTON. - 0cm ' (a)E4 North begin the analysis using the result for the CH n column z 40 abundancesfrom the previous section for UVS. The model • •0 •/t• • SLANTCH,=5 X 10'•cm -• analysis of the EUVS observations is performed using a

o multiple linear regression technique, involving emission from two or three atmospheric layers [Bevington, 1969]. • • 00 1200 1300 • 400 • 500 • 800 1700 • 800 The emergent radiation was fit to the observed data. The WAVElENgTH(•) amount of contribution from each layer was the fitting parameter. The final model spectrumis a linear sum of the 40• acre (b)E4 South emission from each layer. 3o• .... qlANT CH4 = o cm-2 SLANTCH,• = 1 X 1016cm-2 We show in Figure 17 the first attempt at modeling the 20: A/• t•I SLANTCH.=2 X 10'6cm -2 G1 aurora. The figure shows the southern aurora modeled 71./•t• .....CH4 =.5 X 10'6½m -2 with a 5 x 10'6 cm'• CHnslant columndensity, which is the mean column abundanceindicated by the UVS. The model in •o0 .:.::.• ._• 1100 1200 1300 1400 1500 1600 1700 1800 Figure 17 uses a two-layer model with a temperatureof 600 WAVELENGTH(•) K for the 1 x 10•ø cm'• layerand 900 K for the 1 x 10'• cm'• layer. The interplanetary H Lyman ot line profile serves as Figure 16. The E4 UVS FUV observationson DOY 347 the emission model for analyzing the Jupiter H Lyman and model (a)north and (b) south for the hydrocarbon transmission function of (4). ot line in the EUVS spectra but is not used to yield any physical parametersof the aurora. This observed emission

C3 southaurora of 5 x 10'6 Clll'2 for theCHn column density andfrom the modelof (5) (not shown)of 1 x 10'7 cm'2 for 8 the CHn colunto density. The models for the south show a (a) G1 South radiated surface brightness of 293 kR. The electron flux DATA requiredfor this brightnessis 8.3 x 10• electrons/cm•/sfor a MODEL columndensity of 1 x 10:ø cm '•. Onceagain the modelin • SLANTH2: 1020cm '2, CH 4: 5x 1016 cm '2 the 1300-1500]k wavelengthrange seriously overestimates g SLANTH2= 1018 cm'2 / IP Lc• PROFILE the auroralintensity by a factorof 3ß The samekind of L9

800 850 900 950 1000 1050 1100 11.50 1200 1250 1300 5. Modeling the EUVS observations WAVELENGTH(A) Figure 17. The G1 EUVS data for the south aurora with a The EUVS data are of sufficient quality to perform the two-layermodel atmosphere of 1 x 10'• cm': of H: in layer 1 first extensive wavelengthmodeling in the spectral range of and5 x 10'6 cm':of CHnand1 x 10:• cm':H: in layer 2, 800-1200 /•. The models of the EUVS data are sensitive to showing (a) the amount of radiation from each layer at the (1) the H: gas temperaturewhich controls the number of top of the atmosphere,contributing to the measuredsignal, levels participating in the absorptionprocess, (2) the H• basedon a linear regressionanalysis to the data, and (b) the column abundance which determines the amount of self- resultantcomparison of data and model. 20,142 AJELLO ET AL.: EUV AND FUV OBSERVATIONS OF JUPITER AURORA

8 I I I I I I I I I large to explain the rotational line structure observed by (a) 03 South HST. HST has measuredthe auroral temperaturesmany times DATA to be near 500 K [Liu and Dalgarno, 1996]. This MODEL temperatureresult from HST appears to be stable over the .... SLANTH2= 1021 cm'2 severalFUV spectrasurveyed. CH4= 5 x 1016 cm-2,T = 3000K The one remaining variable is the hydrocarbon abundance. We can simulate an increase in the hydrocarbon abundanceby allowing the CH4 column density to rise be- yond the values measuredby the UVS in the lowest layer (He - 1 x 1020cm':). The justificationfor this increasefollows from the probable UVS identification of additional hydrocarbonabsorption in the 1300-1500][ rangefrom additional layers. In addition,Pryor [1989] has shown the presence of high-altitude complex hydrocarbon absorbersin Jo- o vian auroral zones from Voyager photopolarimeter observa- 8oo 850 900 950 '•000 '•050 '•'•00 '•'•,50 '•200 '•250 300 WAVELENGTH(,,•) tions in the UV, which exhibited increased darkening at the poles, with greatest darkening contrast in the south. We i show in Figures 19a and 19b the resultsfor G1 for the north (b) with2 x 10'7 cm':CHn slant columndensity for the transmission function model of (4). A nearly identical fit is DATA 6 foundfor 5x 10'• cm': CH• slant columndensity for the transmission function model of (5). We summarize the model Rydberg band surfacebrightness in Table 3 for the z _(2 two columnabundances of CHnof lx10 '* cm-: and2 x 10'* u• 4 cm':.The changeof a factorof 2 in columnabundance leads to a factor of 8 variation in surfacebrightness. Most of the O emissionis in the FUV and is unobservedby the EUVS. The best fit surfacebrightness for the EUVS to match the UVS occurs about halfway between the two column abundance values at about 1.5 x 10 '7 cm':. The EUVS models are not dependenton the C:H: slant column density,since the CHn 0 absorption cross sections are about twice as large in the 800 8`50 900 9`50 1000 10`50 1100 11`50 1200 12,50 1300 800-1000J[ wavelengthrange. We haveinput to themodel WAVELENGTH(A) codesthe same C:H: slant column densities as in the UVS Figure 18, The C3 EUVS data for the south aurorawith a two-layermode] atmosphere of 1 x 10'9 cm':of H: in layer1 modeling, relative to CH4 column density. and5 x 10'6 cm'2 of CH4 and1 x 10:' cm':of H: at a The quality of the RMS fit shown in Figure 19 is temperatureof 3000 K for H: in layer 2, showing (a) the excellent in G1 north for both sets of transmission function output of each layer at the top of the atmosphere models. The layer contributing 98.3% of the radiated contributing to the measured signal, based on a linear intensity is the third layer (1 x 10:ø cm': slant column regression analysis to the data, and (b) the resultant density),since most of the emissiontakes place in the FUV comparisonof data and model. in the Lymanbands. Yet the secondlayer (1 x 10'8 cm-: slant column density) contributes most of the short- wavelengthEUV 880-1100• radiation.In total, the second line is simply used in the linear regression analysis along layer contributes 1.4% of the emergent intensity. The with the model H: bands. There are two problems with the resultsfor the G1 southaurora are similar (Figure 20). There two-layer model. First, the output power from the Rydberg is also a measurablecontribution from the first layer (1 x bands and the Lyman bands is a factor of 10 below the 10'• cm': slantcolumn density) in the vicinity of 900 •; correspondingintensities from the UVS. Second,•the model althoughit contributesless than 0.3% of the outputpower structures do not match the data structures at 1030 and 1 120 in the H: bands. The shortest wavelength E[N radiation ][. TheRMS fit is not verysatisfactory. In orderto match arisesfrom very high in the atmospherefrom an altitude of the UVS power from the aurorafor the Rydberg bands the near 1300 km above the 1 bar level. The various observed temperatureof the gas was increasedin order to augmentthe spectralstructures are adequatelyexplained by the individual Lyman band output relative to the Werner bands and the spectra from each layer. Self-absorption and hydrocarbon high (B', D) Rydberg bands. As the gas temperature absorptionare equallyimportant in understandingthe EUV increases,the self-absorption in the Rydberg bands becomes aurora spectra from Jupiter. The deep layer (layer 3) is more significant. At 3000 K it is possible to match the responsiblefor thepeak at 1150 •, andlayer 2 produces UVS and EUVS power outputsfor all the orbits. We show in mostof thebroad peak near 1020 •. Mostof thepeak near Figure 18 the C3 south observation, which occurredat the 1020• arisesfrom H: bandsrather than H Lyman[5. The same time as the C3 UVS observation. The two instruments emission cross section of H Lyman [5 from dissociative are boresightedas shown in Plate l a. The match with the excitationof H: is about 4% of the emission cross section UVS intensity of 300 kR is perfect. However, the RMS fit of H Lyman a from dissociativeexcitation of H: [Ajello et is not satisfactory, and the requiredgas temperatureis too al., 1996). AJELLO ET AL.' EUV AND FUV OBSERVATIONS OF JUPITER AURORA 20,143

i i 6. Comparisonof UVS observationsto HST (a) G1 North The HST GHRS obtaineda low-resolution spectrumof the southern Jupiter aurora at the same time as the G2 6 SLANTH2: 1020cm 2, CH4 =2 X1017 cm '2 observations. We show in Figure 24 a calibrated GHRS SLANTH2: 1018cm 2 specmunand two calibratedUVS spectranormalized at 1610 SLANTH2 = 1016cm 2 ]t. The two UVS auroraspectra are G1 and E4 south.We IP Lu PROFILE have modeled these two spectra to obtain CH4 column densitiesof 5 to 7 x 1016cm -2 and 0 cm'2, respectively.The HST spectrumand the UVS spectra appear to be well calibrated to one another. Indeed, the intensity for this / x particularHST spectrumis about 1 MR for a 2000 km wide source. Analysis of the HST spectrum by the same techniques employed here would yield a CH4 column

o abundancemidway betweenthe abundancesfound for the two 800 8.50 900 9.50 1 OOO 10.50 11 DO 11.50 1200 12.50 1300 UVS spectra. WAVELENGTH(/•,)

i i 1

(a) G1 South DATA

MODEL 6 __ _ _ SLANTH2:1020cm2,CH4=2X1017cm '2 ..... SLANTH2: 1018cm2 SLANTH2 = 1016cm 2

_J o /

o o 8oo 8..50 900 950 1000 1050 1100 11..50 1200 1250 1300 800 850 900 950 1000 1050 1100 1150 1200 1250 1.500 WAVELENGTH(A) WAVELENGTH(,•,) Figure 19. The G1 EUVS data for the north aurorawith a three-layermodel atmosphere of 1 x 10•6 cm'2 of H• at T = 1000 K in layer 1, of 1 x 10•8 cm• of H• at T=900 K in (b) -- DATA layer2 and2 x 10•7 cm '2of CH• and1 x 102ocm -• H 2 at 600 MODEL

K in layer 3, showing (a) the count rate producedby each 6 layer contributingto the measuredsignal, basedon a linear regressionanalysis to the data with (4) as the transmission function, and (b) the resultantcomparison of data and model. z u) 4

The C3 south observation depicted in Plate l a is shown in Figure 21 and is simultaneouswith UVS observationof C3 south.The model employsa column density of CHn of 2 2 x 1017 cm-2. This particularmodel producesa surface brightnessof about1 MR and•s listedin Table3. The actual Rydbergband surfacebrightness should be near 292 kR, the UVS value. Table 3 shows this value can be attained 800 850 900 950 1000 1050 1100 1150 1200 1250 1.500 for a columndensity of CH• between1 x 10•7 cm '• and2 x WAVELENGTH(,&,) 10•7 cm'2. We show in Figure 22 the results for the F_,4 Figure 20. The G1 EUVS data for the south aurora with a north, which is simultaneous with the UVS F_,4north. The three-layermodel atmosphere of 1 x 10• cm'• of H2 at T = modelingresults for surfacebrightnesses are given in Table 1000 K in layer 1, of 1 x 10•8 cm'2 of H• at T=900 K in 3. layer2, and2 x 1017cm '2 of CHn arid1 x 1020cm '2 of H2 at The statisticsof the EUVS observations can be improved 600 K in layer 3, showing (a) the count rate producedby each layer at the top of the atmosphere,contributing to the by summingall three observationsin the north and the measuredsignal, based on a linear regressionanalysis to the south. We plot in Figure 23 the model fit to the summed data, and (b) the resultantcomparison of data and model. The spectrumfor a columnabundance of CHn of 2 x 10•7 cm'•. transmissionfunction model used in these plots was (4). The The point is to clearly identify the real spectralfeatures in regression analysis results for transmission function (5) the data and show the good agreementin spectral structure were identical with about twice the amount of CHn required between the data and model. for best fit. 20,144 AJELLO ET AL.: EUV AND FUV OBSERVATIONS OF JUPITER AURORA

I two layers and hydrocarbonsor aerosolsin addition to CHn: (a) C3 South (1) an upper layer of emissionfrom an atmosphereof H 2 and DATA H located above the homopause and (2) a lower layer MODEL containinghydrocarbons. The details of the FUV spectracan 6 SLANTH2 = 1020 cm '2 , CH4 =2 X 1017 cm '2 SLANTH2 = 1018crn '2 be fit with an electron energy distribution that is variable with time and altitude. SLANTH2 = 1016 crn '2 IP L(• PROFILE The Opal et al. [1971] distribution generates an electron impact inducedfluorescence spectrum that fits most of the spectralstructure in the FUV. We show in Figure 25 an FUV model based on a warm 27 eV electron energy distribution exciting an H2gas with an elevated vibrational temperature of 1400 K. The model provides a better fit of some of the electron impact induced fluorescence spectra in the wavelengthrange from 1270 to 1400/• for shallowaurorae, . - _ o such as occurred on E4. Electric field accelerations could 8oo 850 900 950 1000 1050 1100 1150 1200 1250 1300 WAVELENGTH(/•) modify the Opal distribution. Note the different structurein

(b) (a) E4 North DATA -- DATAI I I 6 _ _ _ SLANTH2:1020cm'2.CH4=2X1017cm'2 SLANTH2 = 1018cm '2 SLANTH2--1016 cm-2 w

o 0 I 8oo 850 900 950 1000 1050 1100 1150 1200 1250 1500 800 850 900 950 1000 1050 1100 1150 1200 1250 1500 WAVELENGTH(/•) WAVELENGTH Figure 21. The C3 EUVS data for the south with a three- layermodel atmosphere of 1 x 10•6 cm-2of H 2 at T = 1000 K in layer 1, of 1 x 10•s cm'2of H2at T=900 K in layer2, and (b) 2 x 1017cm '2Of CH n and1 x 1020cm '2ofH 2 at 600 K in layer DATA 3, showing (a) the count rate producedby each layer at the MODEL top of the atmosphere,contributing to the measuredsignal, 6 based on a linear regressionanalysis to the data, and (b) the resultant comparison of data and model. The transmission function model used in these plots was (4). The regression z analysis results for transmissionfunction (5) were identical • 4 with about twice the amountof CH n requiredfor best fit. O •-"

7. Discussion 2 We have presented an analysis of near-simultaneous EUV/FUV observationsof the Jupiter auroraeby the Galileo

EUVS and UVS. Consistency was achieved between the 0 l analysisof the two data sets in terms of the Rydberg band 800 850 900 950 1000 1050 1100 1150 1200 1250 1300 energy outputsat the top of the atmosphere.Auroral sources WAVELENGTH(/•) need to be modeledover a range of a altitudes. For the UVS Figure 22. The E4 EUVS data for the north with a three- observations presented here the FUV emission from layermodel atmosphere of 1 x 10•6 cm'2of H 2 at T = 1000 K wavelengthsbelow 1400 /• couldbe a componentof the in layer1, of 1 x l0 is cm'2of H2at T=900 K in layer2, and aurora from above the homopause(-300 - 400 km altitude). 2 x 1017cm '2 of CHn and1 x 1020cm '2 of H2 at 600 K in layer 3, showing (a) the count rate producedby each layer at the For auroraearising deep in the atmosphere,the wavelengths top of the atmosphere,contributing to the measuredsignal, near1600 /• canarise partially from regions up to one or basedon a linear regressionanalysis to the data, and (b) the two scale heights deeperin the stratosphere[Gladstone et resultant comparison of data and model. The transmission al., 1996; Gladstone and Skinner, 1989] near 250 - 300 km function model used in these plots was (4). The regression altitude. The scale height of the auroral atmosphereat 600 K analysis results for transmission function (5) were identical is 90 km. For deep aurorathe model must include at least with about twice the amountof CHn requiredfor best fit. AIELLO ET AL.: EUV AND FUV OBSERVATIONS OF JUPITER AURORA 20,145

(a) North 9- G1 SOUTHFOR 1 44 S/CHANNEL(62R j, 23øW)

DATA ...... E4 SOUTHFOR 14 87 S/CHANNEL(55R j, 225øW) 8 MODEL HST GHRS SPECTRUM SEPTEMBER 1996

,•- 7

< 6- • 0.8 •; !• : 5 L

4- • o.• ,i;',,", "1'• z ' •' I• :Jf' - 3 o.4•,!• !1',',,', "

1 - O.0F , •'• , •' , 0 1200 1500 1400 1500 1600 1700 800 850 900 950 lOOO lO5O 1 lOO 1150 1200 1250 1300 WAVELENGTH WAVELENGTH(A)

, I Eigure 24. Comparisonof sou• auroral• s•c•a from HST G•S from September 1996 •d O•ileo •S spectra (b) South from O l •d E4 of •e southernaurora. •e GHRS spec•m DATA wasat 4.8 A ••, comp•edto •e UVS 6.5A •HM. MODEL between0 cm-a and 10 x 10'6 cm'a (Table3). The cosineof the emission angle for these observations is close to 0.3 (Table 1). The vertical column densitiesof CH4 are in the range 0 cm-a to 3 x 10•6 cm'a. The EUVS and UVS observationsin the wavelengthrange of 1000-1300]k are consistentwith vertical columndensities of Ha of 3 x l019 cm-a to 1 x 10a• cm'a. The lower limit is based on the smallest coluntodensity required to fit the EUVS dataand the high- altitude auroraemeasured by the UVS, which requiresthe emissionto take placenear the homopause.The upperlimit 800 850 900 950 1000 1050 1100 11,50 1200 12,50 1 300 is estimated, based on the location of the visible aurora at WAVELENGTH(,&,) an altitudeof 240 km (A. Ingersoll et al., submittedpaper, 1998), since the EUVS models are not sensitive to column Figure 23. The linear regression fit to the summedEUVS data for G1 + C3 + E4: (a)north and (b) south. The densitiesin excessof 1 x 10aø cm -a. The lowestlayer in the transmissionfunction model used in these plots was (4). The EUVS modelingoccurs at an altituderange of 300 to 400 regression analysis results for transmission function (5) km above the 1 bar level. Higher atmosphericlayers were identical with about twice the amount of CH 4 required producethe observedEUV radiationfrom 880 to 1000 •. for best fit. Theseabundances provide the best comparisonof the auroralratio of Ha/CH4 column densitieswhich can be used as a basis for volumetric altitude distributions. Earlier works the wavelengthregion longward of 1270,/k for the warmand [Gladstoneand Skinner, 1989; Yung et al., 1982] have cold distributions.Until there is a high-resolutionstudy of quantifiedthe CH4 abundancefrom IUE work. The ratio of the E,F -->B cascadecross-section dependence at low impact energies (12-50 eV), the spectral modeling in the range 1oo 1270-1400J[ will remainuncertain. In addition,the high- 90 resolutionstructure of the C2Ha absorption6ross section is -- 3 x 1020cm-2; 1400 K; Te = 27eV critical in building an accuratetransmission model for the 8o • 3 x 1019cm-2; 500 K; Opal et al. (1971) Ha Lyman band rotationalline spectrum.This study shows a 70 simple one-layer miz/ed atmospheredoes not work as a basis for a model for either the Galileo UVS or EUVS 60 observations. A statement can now be made of the Jupiter 5o auroralparameters that influence the multilayer modeling of 40 spectra.The parametersto be consideredfor future work are 30

(1) location of homopause, (2) altitude distribution of 20 hydrocarbons and aerosols, (3) secondary electron 10 distribution with altitude, (4) I-[ vibrational temperatures, 0 and (5)multienergy primary particle deposition into the 1100 1200 1300 1400 1500 1600 1700 aurora, which can deposit significant amounts of energy X (A) above the homopauseas well as deep in the stratosphere, Figure 25. A comparisonof the Ha electron impact causing multi-Chapman layers. inducedfluorescence spectrum for an Opal et al. [1971] The slant column densitiesof CH4, the principal absorber, secondaryelectron distribution for a 500 K gas and a warm required to match the six UVS auroral observations lie 27 eV secondaryelectron distribution for a 1400 K gas. 20,146 AJELLOET AL.: EUV AND FUV OBSERVATIONSOF JUPITERAURORA

10-10 -- CRAVENS et al. (1975) There may be multiple ways to produce a given • GLADSTONE & transmission function from models and match the energy 10-9 SKINNER (1989) •• -{- WAITE et al. (1988) output with the EUV. A variety of transmission functions 10-8 have been studiedin this paper (Figures 10 and 14). Future work will include extendingour EUV/FUV models to include 10-7 -400km [H2] •,= 3 x 10 19cm-• ,•,,,/10--- keVKORAL &COHEN (1965)many-layer atmospheres with more realistic chemical 0.1Izb HOMOPAUSE •. distributions and auroral energy deposition, to producenew 10'6 -270km[H2] •,= 1 x 1021 cm-2 '% - transmission functions that may clarify the role of ' 3 IJ.b additionalabsorbers, which may be particularlyimportant in 10-5 :-'.,•,48 keV the EUV. '•, - We summarize the results from this analysisfor the EUVS 10-4 in Section 7.1, and similarly, we summarizeresults for the UVS in Section 7.2. This analysis of the combined 10-3 EUVS/UVS data sets will be examined in more detail to

10-2 ...... ,1 I , , • •l • ...... I understandthe model variables of temperature, secondary 01 102 103 104 105 106 electron distribution functions,and high-resolutioneffects of ENERGY OF PRIMARY (eV) absorptionby other hydrocarbons. Figure 26. Depth of penetration of primary electrons based on cross sectionsof Northcliffe and Schilling [1970]. 7.1. Major EUVS Results Data Results. Three EUV spectral images of Jupiter H2/CH4 column densities at the homopausein the aurora is auroraand dayglow; Spectra are highly variable with cutoff uncertainand may be variablewith valuesbetween 103 to near 880 106. This rangeincludes the equatorialmixing ratio of 5 x Modeling Results. Model developmentwith Iaq self- IOn[Gladstoneet al., 1996]. Furthermodeling will determine absorption,latest H2excitationcross sections, hydrocarbon the auroral ratio more exactly. absorption and Opal et al. [1971] secondary electron The combined EUVS and UVS observations show that the distribution (0- 1 keV); Peak energy deposition occurs at aurora is producedover a wide range of altitudes. For a H• slantcolumn density - 3 x 10•9 to 10•2 cm-2; EUV aurora single mean primary energy the emission layer resemblesa areproduced over a rangeof altitudesfrom 1 x 10•6 cm ': to 1 Chapmanlayer, whose emission profile is distributedwith x 10:• cm-: H: slant columndensity (250-1300 lcmaltitude) an e-folding distance of an atmospheric scale height with the bulk of the integratedintensity (>99%) occurringat [Gladstone and Skinner, 1989]. The mean primary electron altitudesfrom 1 x l02gcm '2 to Ix 1022cm ': H: slant column energiesrequired to penetrateto the deep EUV/FUV layers density (250-700 km altitude); Primary electron energy 10- are shown in Figure 26 with a comparison to previous 50 keV with energy input to Rydberg bands of -30-300 ergs/cm2sfor a 2000 km wide aurora. estimates from Cravens et al. [1975], Gladstone and Skinner [1989] and Koral and Cohen [1965]. The cross sections for 7.2 Major UVS Results electron energy loss are from Northcliffe and Schilling [1970]. The primary energies requiredto penetrate to the Data Results. Simultaneous (C3 south and F.A north) deep1029- 1022cm '2 layersare in the range10 - 50 keV. EUV/UVS auroralimaging and near-simultaneousimaging on However, a more general model of the emission layers is other observations;Surface brightnessof auroraof Rydberg requiredwith a knowledgeof the distribution function of bands are 100-500 kR with Lyman ct surfacebrightnesses primaries[Gerard and Singh, 1982]. Perhapsthe joint fields from 50 to 150 kR. and particlesand remotesensing observations of the -30 Rj Modeling Results. Aurora are modeled with peak aurora in the C3 and G8 orbits will help quantify the depositionat a slantcolumn density of CHnof0-1xl0 26cm-• primary particle distributionentering the loss cone. in the northand 0-10 x 10•6 cm'2 in the south; Secondary We can estimatethe energy input to the aurora requiredto electron distribution function - 10*9 e/cm:/s at 250 - 400 produce the observations described in this paper. The km altitude. modelingresults are summarizedin Table 3. We indicate the emission rate at the top of the atmospherein the Rydberg bands for both instrumentsas predictedby the model. For Acknowledgments. The researchdescribed in this text was carried out the UVS the surfacebrightness emission rates at the top of at theJet PropulsionLaboratory, California Institute of Technology,the Universityof Colorado,University of SouthernCalifornia, University of the atmospherefrom both the model and the data are in Michigan, and University of Arizona. The work was supportedby the close agreement,since most of the output flux is in the Galileo Project and Space Telescope Program Office. J.C. Lymanbands. The surfacebrightnesses described in Table 3 acknowledgesgrant 5828-01-94A from ST ScI to the University of for the 2000 km wide model source exceed 100 kR for all Michigan. We appreciatecareful review of the manuscriptby G. R. Gladstone. observations.We use the 10% efficiency[Waite et al., 1983] for converting energy deposition from a primary electron flux above 1 keV into the source region to emission from the Rydbergbands. The energy deposition fluxes are shown References in Table 3. For the brightest auroraon G1 souththe energy flux to excitethe Rydbergbands is about30 ergs/cm:s.The Abgrall,H., E. Roueff,F. Launay,J.-Y. Roncin,and J.-L. Subtil,Table total energy flux to all processesat the sourceregion is 307 of the Lymanband system of molecularhydrogen, Astron. Astrophys. ergs/cm:s. Suppl.,101,273-321, 1993a. AIELLO ET AL.: EUV AND FUV OBSERVATIONS OF JUP1TERAURORA 20,147

Abgrall, H., E. Roueff, F. Launay, J.-Y. Roncin,and J.-L Subtil,Table Herbert, F., B. R. Sandel, and A. L. Broadfoot, Observations of the of the Werner band systemof molecularhydrogen, Astron. Astrophys. Jovian UV aurora by Voyager, J. Geophys.Res., 92, 3141-3154, Suppl.,101,323-362, 1993b. 1987. Ajello, J. M., S. K. Srivastava,and Y. L. Yung, Laboratorystudies of uv Hord, C. W., et al., Galileo ultravioletspectrometer experiment, Space emissionsof H 2 by electronimpact: The Werner and Lyman band Sci. Rev., 60, 503-530, 1992. systems,Phys. Rev. A, 29, 636-653, 1982. James,G. K., J. M. Ajello, andW. Pryor, The MUV-visible spectrumof Ajello, J. M., D. E. Shemansky,T. L. Kwok, and Y. L. Yung, Studiesof H• excitedby electronimpact, J. Geophys.Res., this issue. extreme-ultravioletemission from Rydbergseries of H 2 by electron Kim, Y. H., J. L. Fox, and H. S. Porter, Densities and vibrational impact,Phys. Rev. A, 29,636-653, 1984. distributionof H•+ in theJovian auroral ionosphere,. J. Geophys.Res., Ajello, J. M., et al., A simple ultraviolet calibration source with 97, 6093-6101, 1992. reference spectraand applicationto the Galileo orbiter ultraviolet Kim, Y. H., J. J. Caldwell, and J. L. Fox, High-resolutionultraviolet spectrometer,Appl. Opt.,27, 890-914,1988. spectroscopyof Jupiter'saurora with the Hubble space telescope, Ajello, J. M., W. R. Pryor,C. A. Barth,C. W. Hord, A. I. F. Stewart, K. Astrophys.J., 447, 906-914, 1995. E. Simmons,and D. T. Hall, Observationsof interplanetaryLyman 0t Koral, K. F., and A. J. Cohen, Empirical equationsfor electron with the Galileo ultraviolet spectrometerexperiment: Multiple backscatteringcoefficients, NASA Tech.Note, TN D-2909, 1965. scatteringeffects at solarmaximum, Astron. Astrophys., 289, 283-303, Liu, W., andA. Dalgamo,The ultravioletspectra of the Jovianaurora, 1994. Astrophys.J., 467, 446-453, 1996. Ajello,J. M., S. M. Ahmed,and X. Liu, Lineprofile of H Lyman[• from Liu, X., S. M. Ahmed, R. A. Multari, G. K. James,and J. M. Ajello, dissociativeexcitation of H2,Phys. Rev. A, 53, 2303-2308, 1996. High-resolutionelectron impact studyof the far ultraviolet emission Allison,A. C., and A. Dalgamo, Band oscillatorstrengths and transition spectrumof molecularhydrogen, Astrophys. J. Suppl.,101,375-399, probabilitiesof the Lymanand Werner systems of H2, HD andD•, At. 1995. Data, 1,289-304,1970. Liu, X., D. E. Shemansky,S. M. Ahmed, G. K. James,and J. M. Ajello, Ballester, G., et al., Time-resolved observationsof Jupiter's far- Electronimpact excitation and emissioncross sections of the H• ultraviolet aurora, Science,274,409-412, 1996. Lyman and Werner systems,J. Geophys.Res., in press,1998. Berington,P. R., Data Reductionand Error Analysisfor the Physical Livengood,T. A., D. F. Strobel,and H. W. Moos, Long-termstudy of Sciences,pp. 164-186,McGraw-Hill, New York, 1969. longitudinaldependence in primary particle precipitationin the north Broadfoot,A. L., B. R. Sandel,D. E. Shemansky,J. C. McConnell, G. R. Jovianaurora, J. Geophys.Res., 95, 10,375-10,388,1990. Smith,J. B. Holberg,S. K. Atreya,T. M. Donahue, D. F. Strobel,and Morrissey,P. F., P. D. Feldman,J. T. Clarke,B.C. Wolven, D. F. Strobel, J. L. Bertaux, Overview of the Voyager ultraviolet spectrometry S. T. Durrance,and J. T. Trauger, Simultaneousspectroscopy and resultsthrough Jupiter encounter, J. Geophys.Res., 86, 8259-8284, imagingof the Jovianaurora with the Hopkinsultraviolet telescope 1981. andthe Hubblespace telescope, Astrophys. J., 476, 918-923, 1997. Clarke, J., T., L. B. Jaffel, A. Vidal-Madjar, G. }L Gladstone,J. H. Mount, G., E. S. Warden, and H. W. Moos, Photoabsorptioncross Waite, R. Prange,J.-C Gerard,J. Ajello, andG. James,Hubble Space sectionsof methanefrom 1400 to 1850/•,Astrophys. J., 214, L47- Telescope Goddard high-resolutionspectrograph H 2 rotational L49, 1977. spectraof Jupiter'saurora, Astrophys.. J., 430, L73-L76, 1994. Mount,G., E. S. Ward, andH. W. Moos, Photoabsorptioncross sections ClarkeJ. T., et al., Far-ultravioletimaging of Jupiter'saurora and the Io of methaneand ethane from 1380-1600/• at T = 295 K and T = 200 "footprint",Science, 274,404-409, 1996. K, Astrophys.J., 224, L35-L38, 1978. Clarke,J. T., et al.., HubbleSpace Telescope imaging of Jupiter'sUV Nakayama, T. and K. Watanabe, Absorption and photoionization auroraduring the Galileoorbiter mission, J. Geophys.Res., this issue. coefficientsof acetylene,propyne and 1-butyne,J. Chem.Phys., 40, Cravens,T. E., G. A. Victor, and A. Dalgamo, The absorptionof 558-561, 1964. energeticelectrons by molecularhydrogen gas, Planet. Space Sci., Northcliffe,L. C., andR. F. Schilling,Range and stoppingpower tables 23, 1059-1070, 1975. for heavyions, Nucl. Data Tables,A7, 233-463, 1970. Durrance,S. T., P. D. Feldman,and H. W. Moos, The spectrumof the Opal, C. B., W. K. Petersen,and E. C. Beaty, Measurementsof Jovianaurora 1150-1700/t, Geophys. Res. Lett., 9, 652-655,1982. secondary-electronspectra by electronimpact ionization of a number Emerich, C., L. B. Jaffel, J. T. Clarke, R. Prange,G. R. Gladstone,J. of simplegases, J. Chem.Phys., 55, 4100-4106,1971. Sommeria,and G. Ballester,Evidence for supersonicturbulence in Prange,R., D. Rego,and J. C. Gerard,Auroral Lyman 0t andH 2 from the upperatmosphere of Jupiter,Science, 273, 1085-1087,1996. the giant planets,2, Effect of the anisotropyof the precipitating Feldman, P. D., M. A. McGrath, H. W. Moos, S. T. Durrance, D. F. particleson the interpretationof "colorratio", J. Geophys.Res., 100, Strobel,and A. F. Davidsen,The spectrumof the Joviandayglow at 7513-7521, 1995. 3/• resolutionwith the Hopkinsultraviolet telescope, Astrophys. J., Prange, R., S. Maurice, W. M. Harris, D. Rego, and T. Livengood, 406, 279-284, 1993. Comparisonof IUE and HST diagnosticsof the JovianAurorae, J. Gerard, J. C., and V. Singh,A model of energy depositionand EUV Geophys.Res., 102, 9289-9301, 1997a. emissionin the Jovianand Saturnianatmosphere and implications,J. Prange,R., D. Rego,L. Pallier, L. B. Jarel, C. Emerich, J. Ajello, J. T. Geophys.Res., 87, 4525-4532, 1982. Clarke, and G. Ballester, Detection of self-reversed Lyman alines Gerard,J. C., V. Dols,F. Paresce,and R. Prange,Morphology and time from the Jovianaurorae with the Hubble spacetelescope, Astrophys. variationof the Jovianfar UV spectra:Hubble Space Telescope J., 484, L169-L173, 1997b. observations,J. Geophys.Res., 98, 18,793-18,801,1993. Pryor, W., Auroral ultravioletdarkening on the outer planets,Ph.D. Gerard,J. C., D. Grodent, R. Prange,J. H. Waite, G. R. Gladstone,V. thesis,chap. IV, Universityof Colo.,Boulder, 1989. Dols, F. Paresce, A. Storrs, L. B. Jarel, and K. A. Franke, A Pryor,W., J. M. Ajello, C. A. Barth,C. W. Hord, A. I. F. Stewart,K. E. remarkable event on Jupiter observed in the ultraviolet with the Simmons,W. E. McClintock,B. R. Sandel,and D. E. Shemansky,The Hubblespace telescope, Science, 266, 1675-1678,1994a. Galileo and Pioneer Venus ultravioletspectrometer experiments: Gerard,J.C., V. Dols,R. Prange,and F. Paresce, The morphology of the Solar Lyman 0t latitude variation at solar maximum from north Jupiter ultraviolet aurora observedwith the Hubble space interplanetaryLyman 0t experiments,Astrophys. J., 394, 363-377, telescope,Ptanet. Space Sci., 42,905-917, 1994b. 1992. Gladstone,G. R., andT. E. Skinner,Spectral analysis of Jovianauroral Serif,A., D. B. Kirk, T. C. D. Knight,J. D. Mihalov,R. C. Blanchard,1L emissions,in TimeVariable Phenomena in the JovianSystem, edited E. Young,G. Schubert,U. vonZahn, G. Lehmacher,F. S. Milos,and by M. J. S. Belton,R. A. West and J. Rahe,NASA Spec.Publ., SP- J. Wang,Structure of the atmosphereof Jupiter,Science, 272, 844 - 494,221-228, 1989. 845, 1996. Gladstone,G. R. andY, •L.Yung, •An analysis of thereflection spectrum Seiff, A., D. B. Kirk, T. C. D. Knight,L. A. Young,F. S. Milos,E. of Jupiterfrom 1550A to 1740A, Astrophys.J.,266, 415-424,1983. Ventatapathy,J. D. Mihalov,R. C. Blanchard,R. E. Young,and G. Gladstone, G. R., M. Allen, and Y. L. Yung, Hydrocarbon Schubert,Thermal structureof Jupiter'supper atmospherederived photochemistryin the upperatmosphere of Jupiter,Icarus, 119, 1-52, from the Galileoprobe, Science, 276, 102-104,1997. 1996. Shemansky,D. E., An explanationfor the H Ly0tlongitudinal asymmetry Harris, W. M., J. T. Clarke,M. A. McGrath, and G. E. Ballester, in the equatorialspectrum of Jupiter:An outcropof paradoxical Analysisof Jovianauroral H Ly0t emission(1981-1991), Icarus, 123, energydistribution in the exosphere,J. Geophys.Res., 90, 2673-2694, 350-365, 1996. 1985. 20,148 AJELLO ET AL.: EUV AND FUV OBSERVATIONSOF JUPITERAURORA

Shemansky,D. E., J. M. Ajello, and D. T. Hall, Electron impact electronimpact: Laboratory study and applicationto Jovianaurora, excitationof Ha: Rydbergband systemsand the benchmarkcross Astrophys.J. Lett.,254, L65-L69, 1982. sectionfor H Lyman•, Astrophys.J., 296, 765-773,1985. Singhal,R. P., S.C. Chakravarty,A. Bhardwaj,and B. Prasad,Energetic electronprecipitation in Jupiter'supper atmosphere,J. Geophys. J. Ajello, Jet Propulsion Laboratory, Califomia Institute of Res., 97, 18,245-18,256, 1992. Technology,MS 183-601,4800 Oak Grove Drive, Pasadena,CA 91109. Suto,M., and L. C. Lee, Quantitativephotoexcitation and fluorescence (e-mail:[email protected].nasa.gov) studiesof C•Ha in vacuumultraviolet, J. Chem.Phys., 80, 4824-4831, C. Barth, J. Gebben,C. Hord, W. McClintock, D. Miller, W. Pryor,K. 1984. Simmons,and I. Stewart, Laboratory for Amaosphericand Space Trafton, L. M., J. C. Gerard, G. Munhoven, and J. H. Waite, High Physics,University of Colorado,Boulder, CO 80303. resolutionspectra of Jupiter'snorthern auroral ultravioletemission B. Sandel,Lunar and PlanetaryLaboratory, Universityof Arizona, withthe Hubble space telescope, Astrophys. J., 421,816-827,1994. Tucson,AZ 85721. Waite, J. H., T. E. Cravens,J. Kozyra, A. F. Nagy, S. K. Atreya, and R. J. Clarke, SpacePhysics Laboratory, University of Michigan, Ann H. Chen,Electron precipitation and relatedaeronomy of the Jovian Arbor, MI 48109. thermosphereand ionosphere, J. Geophys.Res., 88, 6143-6163,1983. D. Shemansky,Department of AerospaceEngineering, University of Wu, C., W. Robert, T. S. Chien, G. S. Liu, D. L. Judge, and J. J. SouthernCalifornia, Los Angeles,CA 90089. Caldwell, Photoabsorptionand direct dissociationcross sections of S. Stephensand K. Tobiska, FDC/Jet PropulsionLaboratory, Callain the 1530-1930 • region:A temperaturedependent study, J. Pasadena,CA 91109. Chem.Phys., 91,272-280, 1989. Yung, Y. L., G. R. Gladstone,K. M. Chang,J. M. Ajello, and S. K. (ReceivedSeptember 10, 1997;revised March 4, 1998; Srivastava,Ha fluorescencespectrum from 1200 to 1700• by acceptedMarch 9, 1998