The Pele Plume Io Observations with the Hubble Space Telescope

GEOPHYSICALRESEARCH LETTERS, VOL. 24, NO. 20, PAGES 2471-2474, OCTOBER 15, 1997

The Pele plume (Io): Observationswith the Hubble SpaceTelescope

JohnR. Spencer,• Paola Sartore•i, 2 GildaE. Balles•er,3 Alfred S. McEwen,4 JohnT. Clarke,3 and MelissaA. McGra•h*

Abstract. In July 1996, with the Hubble SpaceTele- 1). Image resolutionwas improved by Lucy-Richardson scope(HST), we observedthe Pele plumesilhouetted deconvolution. against Jupiter at a wavelengthof 0.27pm, the first A single F255W image taken on July 8 1995 shows definitiveobservation of an Io plume from Earth. The a bright patch againstdark sky off Io's limb in (Fig- height, 420 q- 40 km, was greater than any plume ob- ure 1). The image is noisy,but the perfectalignment servedby Voyager.The plumehad significantlysmaller of the bright patch with the expectedlocation of Pele, optical depth at 0.34 and 0.41pm, where it was not and the lack of similar features off Io's limb elsewhere detected. The wavelengthdependence of the optical in this and other F255W frames, gives us some confi- depth can be matchedby a plume either of fine dust, dence that this is the Pele plume. The plume is about with minimummass of 1.2 x 100g andmaximum parti- 20 DN above backgroundsky brightness,corresponding cle sizeof 0.08pm, or of SO2 gaswith a columndensity to an I/F value (the brightnessratioed to a normally- of 3.7 x 1017cm -2 andtotal mass of 1.1x 1011g. Our illuminated Lambert surface)of 0.004. This is similar modelssuggest that early Voyagerimaging estimates of to average0.35pm Pele plumeI/F valuesseen by Voy- the minimummass of the Loki plume [Collins, 1981] ager 1 [Stromet al., 1981,Figure 2]. No plumewas seen may have'been too large by a factor of • 100. We may in 0.373 pm images taken 20 minutes earlier. have detectedthe Pele plume in reflectedsunlight, at An F255W image taken on July 7, when Pele should 0.27pm, in July 1995, but did not seeit 21 hoursearlier, have appeared on the opposite limb, does not show the so the plume may be capableof rapid changes. plume (Figure 1). Either the July 8 detection,or the July 7 non-detection, is spurious,or the plume switched 1. The Pele Plume on in the 21 hours between the frames. During the Voyager 1 encounter the Pele plume was active for at Pele was unique among the nine volcaniceruption least 9 days, but these data suggestthat more rapid plumeson Io discoveredby Voyager[Strom et al., 1981]. variations are possible. It wasthe highestplume (300 km), the most optically The Pele plume was definitely seen in three F255W thin, and was the only one not active during both en- imagestaken in Jupiter transit in July 1996 (Figure 1). counters. The Pele plume was not detected in Galileo The plume appears dark against Jupiter. Subsequent imagesin June, September,or November1996, but was F336W and F410M images do not show the plume, seenin December1996 [McEwenet al., 1997a].High- though it was further from the limb when these images temperature thermal emission from Pele has been seen weretaken (Table I). Partial saturationof Jupiterin the by Galileo'sSSI [Belton et al., 1996; McEwen et al., F336W image reducesthe precisionof measurementsin 1997b],and NIMS [Lopes-Gautieret al., 1997]instru- this filter. ments whenever observations have been made.

2. HST Observations 3. 1996 Plume Size and Density

We imagedIo with the Wide-Field Planetary Camera To modelthe transit images,we ignorebackscattered 2 (WFPC2) on HST in July 1995and July 1996(Table sunlightfrom the plume. This is justifiedif the plume activity level was similaron July 24 1996 and July 8 1995' the DN deficit in the plume in the July 1996 •Lowell Observatory,Flagstaff, Arizona •Institut d'Astrophysiquede Paris, Paris images,compared to the Jupiterbackground, is about adept. of AtmosphericOceanic, and SpaceScience, Uni- 320 DN, comparedto the 20 DN brightnessof the July versity of Michigan, Ann Arbor 1995 plume images. Also, modelingdiscussed below 4Lunarand PlanetaryLaboratory, University of Arizona, suggeststhat for dust or gas plume modelsthat fit the Tucson transitobservations, reflected sunlight is faint compared 5SpaceTelescope Science Institute, Baltimore,Maryland to absorbedor scatteredJupiter-light. We modeled the plume as a uniform half-disk cen- Copyright1997 by theAmerican Geophysical Union. tered on the Pele vent (18 S, 256 W), includingonly Papernumber 97GL02592. that portion that would protrudeabove Io's limb. This 0094-8534/97/97GL-02592505.00 model plume was smoothedby a 0.07 arcsecboxcar fil-

2471 2472 SPENCER ET AL.- PELE PLUME FROM HST

Table 1. Details of HST images

Sub-Earth UT HST Filter Effective Exp- Solar UT Mid Filter Wavelength Wavelength, osure W. Phase A R H Date Time Name Range,•um •um (s) Lat. Long. Angle (AU) (AU) (km)

95/07/07 14:01 F255W 0.257-0.291 0.272 600 -2.9 345.5 6.8 4.494 5.328 1.2 95/07/08 10:53 F255W 0.257-0.291 0.272 600 -2.9 162.4 7.0 4.503 5.328 2.8 96/07/24 19:12 F255W 0.257-0.291 0.272 600 -1.8 171.9 4.2 4.239 5.196 8.1 96/07/24 19:24 F255W 0.257-0.291 0.272 500 -1.8 173.7 4.2 4.239 5.196 13.9 96/07/24 20:45 F255W 0.257-0.291 0.272 500 -1.8 185.1 4.2 4.240 5.196 87.4 96/07/24 20:53 F336W 0.315-0.358 0.340 23 -1.8 186.2 4.2 4.240 5.196 97.8 96/07/24 20:59 F410M 0.400-0.417 0.409 10 -1.8 187.1 4.2 4.240 5.196 106.6

Comments' Filter wavelengthrange is between filter transmissionhalf power points, after weighting by the solar spectrum. Effective wavelength is also weighted by the solar spectrum. H is the projected distance between the Pele plume source and 1o'slimb: this much of the bottom of the plume is hidden.

95/07/O7, 1,4-01 UT 95/07/08, 10:55 UT 96/07/24, 19:18 UT ter, and was then added to each HST image. Model ..,. -:-•.. •.• ...... •k' • ' d' -- ...... ::.::,:",•'•...... '•, ... sizeand brightnesswas adjusted to exactlyremove the (negative)plume image. The modelplume brightness was then translated into an optical depth by comparison with the brightnessof the Jupiterbackground (Table 2). Althoughthe plumewas further from the limb when the F410M image was taken, it shouldhave been de-

96/07/24, 20:44 UT 96/07/24, 20:59 UT (F410M) 96/07/24, 19:18 UT tected had it been as large and optically thick at this ....:':'--"...... '"'";...... "< '":•--':•:"••;•:': .... '-.x...... • :: •;;;•";:•...... ••• ...... •:•::::wavelength as in the F255Wfilter (Table2). The plume .'ad•':':':•'"•::•':'•"•'•'•:•?)•.;•'::{•?'•:d?•:k } height,420• 40 kin, makes this the largest plume ever •======•.:• •-•::•::•.•:.•;:•{•"-•"•?•,•.•.•}:•5;:•;:;;•;??•,::•:•:•:•%a...... • ?•t•?::...::.•:::' plumeseenonofIo. similar December height 1996 [McEwenGalileo etimages al., 1997a]. show aPele ':.'k;' ex. k • / ::"••:;:--:':•27 --;•5:•::'....• •:•?:'"•'•:•' ':: :•;•...... %...... •.w•:•:::•: '"-:':...•;•. 4. Models' Dust Scattering 96/07/24, 20:44 U• 96/07/24, •0:5• UT ...... -...... e..:::.:.½•:•?•'•,,•:•;,•-':-d?•sg?-•:•?•;...... :.;•;::?:.:':.'..•.•,•s:•:•?•5•-;%•;•;•:•.....:•:• Voyager 1 observationsof the Pele plume at wave- •]]::•;•:•;;•:•:•;•:•':'"'•"•]•:•s...... :•:.?•.::;•:•;?:'•";-?• "'•":"-?•?:•½•:•:.. ' ...... ,?::.:.•:•;•;: •--.,??•:% ...... lengths > 0.35•m, show aneutral color. However, the :,::".:::::•::•:•':,•...... :•:-•::•:.....:.... ß '""? :--:. :''. :???::L:•:(:•:?• "':•?•;".:-.]•'•. :..:.:...... •?'"'•:•?•"•:•Yd?•;;;::• Voyager• outerLoki plume wasblue inthe 0.35-0.6•m 'z::?;•:•:::•::':•;"•..... •:•a"•:• ...... ;•'...... •?'":"?":' ;"q• %•? ...... "::'5•'::••;•?•::::"; [Collins, 1981]. Collins attributed theblue color of

...... •.<..:-•%;•:.:.:.:-• 'a::j""'"'<.:•:::,-.•k'--,: ...... :•-•::•:.::::•?.:.:..m•::•:• •:•::::.•..•...:?'-' ...... :.•:3 • smallerthe outer thanLoki the plume wavelength toscattering of light,by andparticles this maymuch also Figure 1. Detections,probable detections, and non- explain the HST color of the Pele plume. detectionsof the Pele plume in HST WFPC2 images. All are in the F255W filter except where noted. Io's limb, equator, and the expectedlocation of the Pele source are shown in all frames. The plume is proba- Table 2. Plume optical depths, July 1996 bly seenagainst dark skyon 95/07/08, at F255W:the Pele Plume arrows note imperfectly removed cosmic rays. How- Sub-Earth ever,it is not seenin 95/07/07 F255W imagestaken 21 W. H, Optical Radius hoursearlier. The 1996images have Jupiter in the back- Filter Long. km Depth km ground,and mostare shownin both "raw"format and with plume visibility enhancedby ratioing eachimage F255W 172.8 •' 11.0 0.19-1-0.05 420-1-40 to a transposedand registeredcopy of itself: the plume F255W 185.1 87.4 0.17-1-0.07 440+80 is then visibleas both a positiveimage (dark spot off F336W 186.2 97.8 <0.15__ the limb)and negative image (bright spot off the limb). F410M 187.1 106.6 <0.06 The Pele plume is seenin silhouetteagainst Jupiter in the F255W images(the imagelabeled "19:18 UT" is an averageof the i9:i2 and 19:24UT images),but not in Comments: H is defined as in Table 1 the F336W or F410M images. •Averageof first two 97/07/24 frames. SPENCER ET AL' PELE PLUME FROM HST 2473

for the outerpart of the plume,for r _<0.01/•m. Our \ ...... Radius=O.O5/•,m.Mass=2.80x109 g Mie scatteringcode, approximating the Loki plumeas 0.25,• '•'•I Radius=O.O7/•,m,Mass=1.25x109g a uniform hemisphere with a 0.34/•mI/F of 0.03 and \ / k I Radius=O.O8/•,mMass=1.15x10 • g radiusof 200km, givesa plumemass of 9.3x 1010g for 0.20 • ' • '• r -- 0.01/•m, confirmingCollins' result. However,the Mie codecan also match the colorand brightness of the 0.• "-,I_ _'•.:.'•- ' outer Loki plume with r - 0.05/•mand m - 1.1 x 10ø g, comparable to our minimum dust mass for Pele. The 0.10 ß . Loki plume may therefore be much less massivethan deducedby Collins,if r is largerthan he assumed. 0.05

0.00 .... I .... • , , 5. Models' SO• Gas Absorption 0.20 0.25 0.30 0.35 0.40 0.45 0.50 W•velen9th, microns The extinctionof Jupiter light by the Pele plumein the F255W filter imagemay alsobe explainedby the presenceof SO2 gas, which would make Pele a "stealth op[•c•] •ep[•, comp•ed [o •us• •c•[[e•g mo•e]• •e- •cdbea • [•e levi. •e o•c•]]•fio•s • [•e •e cu•es plume" [Johnsonet al., 1995]. SO2 gas absorption •e •ue [o ou• •ump[•o• o[ • •g]e pA•[•c]e•e • e•c• [Mannattand Lane, 1993]is muchmore efficient in the mo•e]. •o•o•[•] bA• •ow [•e w•e]e•g[• •ge o• F255W filter than at longerwavelengths. The SO• ab- e•c• fi]•e• (•ble H), •a •e•fic•] e•o• b• •ow [•e sorptioncross-section •r in the F255W filter bandpass, u•ce•[•[]• • [•e ob•e•ea optics] aep[•. •e [wo weightedby the spectrumof reflectedlight from Jupiter (Mark Vincent,pets. comm.), is • 5 x 10-lø cm•. As- sumingSO• absorptionis in thelinear regime [Ballester et al., 1994],the observedoptical depth of 0.19 givesa column abundance N • 3.7 x 1017cm -•. This value is We modeled the plume as a 420 km radius cloud approximate,because the Manatt and Lane SO2 spec- of spherical particles with uniform projected optical trum is at 300 K, and has relativelylow spectralresolu- depth, using Mie theory [Bobtenand Huffman, 1983, tion. This amount of gaswould not producedetectable Appendix A], and assumingthe sulfur real refractive absorption in the F336W and F410M filters. index of 1.957. For relevant particle sizes, the value of Our inferred SO• abundanceis comparableto other the complexrefractive index is unimportant. We de- estimates of local SO• abundance on Io. The darkness termined the amount of direct Jupiter light scattered of Io's disk near Pele in 0.2-0.3•umHST imagessug- out of the beam, the amount of Jupiter light scattered gestedN = 10ls cm-• [Sartorettiet al., 1996],depen- into the beamfrom otherparts of the planet(which was denton assumptionsabout the surfacealbedo. Voyager negligible),and the amountof light addedto the plume infrared spectra gave N - 5 x 10ls cm-• over Loki image by backscatteredsunlight. [Pearlet al., 1979]. HST spectroscopydetected SO2 The HST plume color requires particles with radius absorptionlocalized above Pele in early August 1996, r _• 0.08/•m (Figure 2). The plumemass m neededto shortlyafter our Jupiter transit observations (McGrath match the F255W optical depth increaseswith decreas- et al., pers.comm.), supporting the SO• modelfor our ing particle size, becausesmaller particles are muchless observationsbut not ruling out the presenceof dust in efficientscatterers. For r - 0.07/•m,m - 1.25+0.3x10 ø addition. g. Particles with r - 0.05- 0.08/•m are poor backscat- The total massof gasover Pele derivedfrom our 1996 terers, and the backscattered sunlight added to the observations,if the exinctionis due to SO•, is 1.1 x 1011 plume is only 10-20% of the scatteredJupiter light g, about 100 times the minimummass of dust required lost from it in the F255W filter, consistent with the to explainthe sameobservations. Rayleigh scattering plume's dark appearanceand justifying our neglectof is negiligiblefor thesegas abundances [Collins, 1981], as is refraction. backscatteredsunlight in deriving optical depths. Par- ticles smaller than 0.02/•m scatter too isotropically to 6. Resurfacing Rates be consistentwith the observeddarkness of the plume. Application of the Mie scattering model to the ten- Using a ballisticapproximation, a 420 km plume al- tativeJuly 8 1995observation ofPele in backscatteredtitude requiresa launchspeed of 1.1 km/s (43%of Io's sunlight, assumingthe same plume radius as the 1996 escapevelocity, but physicallyplausible [Keiffer, 1982]), observations,gives m - 6 x 108g for r - 0.07/•m,about and a flight time of 1630sec, including the reductionin half the July 1996 mass. gravity with altitude. With a plume dust massof 10o Collins [1981] used a Rayleigh-likeparticle scatter- g, a 420 km radiusfallout area, and a particledensity ing law, valid only for r _<0.01/•m, to model the Voy- of 1 g/cm3, this givesa minimumresurfacing rate of agerimages of the Lokiplume. He derivedm _>i0 il g 3.5x 10-3 cm/yr,equivalent to a globalrate of 4.6 x 10-5 2474 SPENCER ET AL.- PELE PLUME FROM HST

Ingersoll,A. P., Io meteorology:How atmosphericpressure cm/yr.Johnson etal. [1979]estimated a global plume is controlledlocally by volcanoesand surfacefrost, Icarus resurfacingrate of 3.5 x 10-4 cm/yr).The much larger 81, 298-313, 1989. globalplume resurfacing rates of ,-• 10 -2 - 1 cm/yresti- JohnSon,T. V., A. F. Cook,C. Sagan,and L. A. Soderblom, matedby [Johnsonand Soderblom, 1982] resulted from Volcanicresurfacing rates and implications for volatileson thehigh Loki plume mass determined by [Collins,1981], Io. Nature 280, 746-750, 1979. and may thus be too large. JohnsonT. V., and L. A. Soderblom,Volcanic eruptions on Our minimum July 1996 Pele dust kesurfacingrate Io: Implicationsfor surfaceevolution and massloss. in Satellitesof Jupite?,editedby D. Morrison,pp. 634-646, givesonly 10 micronsof depositionin the fourmonths U. Arizona Press, Tucson, 1982. betweenVoyager i and Voyager2, probablyinsufficient JohnsonT. V., D. L. Matson, D. L. Blaney,G. J. Veeder, to producethe major albedochanges seen at Pele in and A. G. Davies, Stealth plumeson Io, Geophys.Res. that interval. Either the plume is much more massive Left. 22, 23, 1995. than our lower limit, due to a large populationof par- Keiffer S. W., Dynamicsand thermodynamicsof volcanic eruptions:implications for the plumesof Io. In Satellites ticleslarger or smallerthan 0.07/•m,or to a largegas of Jupiter,edited by D. Morrison,pp. 647-723,U. Arizona content,or it wasmuch more massive during the Voy- Press,,Tucson, 1982, ager era. Lellouch,E., M. J. S. Belton, I. DePater, G. Paubert, S. If the observedplume is due to SO2gas, simple scal- Gulkis, and T. Encrenaz,The structure,stability, and globaldistribution of Io's atmosphere,Icarus 98, 271-95, ing from the dust calculationgives a largelocal SO2 1992. depositionrate of 0.39cm/yr. The Peleejecta deposits, R. Lopes-Gautier,A. G. Davies,R. Carlson,W. Sfnythe,L. however,appear to be poorin SO2frost [McEwen et al., Camp,L. Soderblom,F.E. Leader,R. Mehlman,and the i988]. If the plumewe see is dueto SO2,the ultimate Galileo NIMS team, Hot spotson Io: Initial resultsfrom fate of that SO2 is an interestingquestion: perhaps it Galileo'sNear I.nfraredMapping Spectrometer, Geophys. flowshorizontally to high-albedocold traps elsewhere Res. L•ett.,this issue,1997. Manatt, S. L. and A. L. Lane,A compilationof SO2absorp- [Ingersoll,1989]. tion cross-sectionsfrom 106-403 nm, J. Quant. $pectrosc. Radiat. Trans. 50, 267-276, 1993. 7. Conclusions McEwenA. S., T. V. Johnson,D. L. Matson, and L. A. SoderblOm,The globaldistribution, abundance, and sta- Theseobservations show that the largestof Io's plumes bility of SO2 on Io, Icarus 75, 450-478 1988. can be detected from Earth, opening up a new avenue McEwen, A. S., J. Spencer,D. Simonelli,T. Johnson,L. Keszthelyi,and the GalileoSSI Team, Io plumeobserva- for studyof theseremarkable phenomena. The 1996 tions from Galileo and HST, Lunar Planet. Sci. XXVIII, transit observationsmay be due to dust or gas,but the 911, 199")a. tentative 1995 observationsof the plume in backscat- McEwen,A. S., D. P. Simonelli,D. Senske,K. P. Klaasen, teredsunlight would require a dustcomponent. If the L. P. Keszthelyi,T. V. Johnson,P. Geissler,M. H. Carr, plumeis due to gas,the F255W Jupitertransit images and M. J. S. Belton, High-temperaturehot spotson Io as seenby the Galileo:Solid State Imaging (SSI) experiment. in Figurei providedramatic visual confirmation of the Geophys.Res. LetL, this issue,1997b. patcry-atmospheremodel derived from theoretical and Pearl,J. (•., R. Hanel, V. Kunde,W. Maguire,K. Fox, S. observationalconsiderations [Ingersoll, 1989; Lellouch Gupta,C. Ponnamperuma,and F. Raulin,Identification et al, 1992;Ballester et al., 1994;Sartoretti et al., 1996], of gaseousSO2 and new upper limits for othergases, Na- by directlyimaging a large,volcanically-derived atmo- ture280, 757-758, i979. Sartoretti, P., M. J. S. Belton, and M. A. McGrath, SO2 spheric"patch" over Pele. If the 1995observations in distributionson io, Icarus 122, 273-287, 1996. backscatteredsunlight are real, they provideevidence StromRG, SchneiderNM, TerrileRJ, CookAF, and Hansen for very rapid time variationsin the Pele plume. C., Volcaniceruptions on Io, J. Geophys.Res. 86, 8593- 8620, 1981. Acknowledgments. This work wasfunded by Space TelescopeScience Institute grants GO-6028 and GO-6774. J. R. Spencer,Lowell Observatory,1400 W. Mars Hill Rd., FlagstaffAZ 86001.(e-mail: [email protected]) References P. Sartoretti, Institut d'Astrophysique de Paris, 98 bis Boulevard Arago, 75015 Paris, France. Belton, M. J. S. and the Galileo SSI team, Galileo'sfirst G. E. Ballester, J. T. Clarke, Dept. of Atmospheric imagesof Jupiterand the Galileansatelhtes, Science, 27.i, Oceanic, and Space Science,University of Michigan, 2455 377-385, 1996. Hayward, Ann Arbor, MI 48109. Ballester,G. E., McGrath M. A., StrobelD. F., Zhu X., A. S. McEwen, Lunar and Planetary Laboratory, FeldmanP. D., and Moos H. W., Detectionof the SO2 University of Arizona, Tucson, AZ 85721. atmosphereof Io with the HubbleSpace Telescope, Icarus, M. McGrath, SpaceTelescope Science Institute, 3700 San 111, 2-17, 1994. Martin Dr., Baltimore, MD 21218. Bohren,C. F., and D. R. Huffman,Absorption and scattering of light by smallparticles, Wiley, New York, 1983. Colhns,S. A., Spatialcolor variations in thevolcanic plume (ReceivedApril 7, 1997; revisedJuly 21, 1997; at Loki,on Io, J. Geophys.Res. 86, 8621-8626,1981.. acceptedJuly 28, 1997.)