ALMA Observations of Io Going into and Coming out of Eclipse
Imke de Pater, Statia Luszcz-Cook, Patricio Rojo, Erin Redwing, Katherine de Kleer, Arielle Moullet
DDraft version September 17, 2020
Typeset using L A TEX twocolumn style in AASTeX63
ALMA Observations of Io Going into and Coming out of Eclipse
Imke de Pater, Statia Luszcz-Cook, Patricio Rojo, Erin Redwing, Katherine de Kleer, andArielle Moullet University of California, 501 Campbell Hall, Berkeley, CA 94720, USA, and Faculty of Aerospace Engineering, Delft University ofTechnology, Delft 2629 HS, The Netherlands University of Columbia, Astronomy Department, New York, USA Universidad de Chile, Departamento de Astronomia, Casilla 36-D, Santiago, Chile University of California, 307 McCone Hall, Berkeley, CA 94720, USA California Institute of Technology, 1200 East California Boulevard, Pasadena, CA 91101, USA SOFIA/USRA, NASA Ames Building N232, Moffett Field, CA 94035, USA (Received XXX; Revised XXX; Accepted XXX)
Submitted to Planetary Science JournalABSTRACTWe present 1-mm observations constructed from ALMA [Atacama Large (sub)Millimeter Array]data of SO , SO and KCl when Io went from sunlight into eclipse (20 March 2018), and vice versa (2and 11 September 2018). There is clear evidence of volcanic plumes on 20 March and 2 September.The plumes distort the line profiles, causing high-velocity ( (cid:38)
500 m/s) wings, and red/blue-shiftedshoulders in the line profiles. During eclipse ingress, the SO flux density dropped exponentially,and the atmosphere reformed in a linear fashion when re-emerging in sunlight, with a “post-eclipsebrightening” after ∼
10 minutes. While both the in-eclipse decrease and in-sunlight increase in SO wasmore gradual than for SO , the fact that SO decreased at all is evidence that self-reactions at thesurface are important and fast, and that in-sunlight photolysis of SO is the dominant source of SO.Disk-integrated SO in-sunlight flux densities are ∼ N ≈ (1 . ± . × cm − and T ≈ −
320 K both in-sunlight and in-eclipse,while the fractional coverage of the gas is 2–3 times lower in-eclipse than in-sunlight. The low levelSO emissions present during eclipse may be sourced by stealth volcanism or be evidence of a layerof non-condensible gases preventing complete collapse of the SO atmosphere. The melt in magmachambers at different volcanoes must differ in composition to explain the absence of SO and SO , butsimultaneous presence of KCl over Ulgen Patera. Keywords:
Galilean satellites, Planetary atmospheres, Radio astronomy INTRODUCTIONJupiter’s satellite Io is unique amongst bodies in ourSolar System. Its yellow-white-orange-red coloration isproduced by SO -frost on its surface, a variety of sulfurallotropes (S –S ), and metastable polymorphs of ele-mental sulfur mixed in with other species (Moses andNash, 1991). Spectra of the numerous dark calderas,sites of intermittant volcanic activity, indicate the pres-ence of (ultra)mafic minerals such as olivine and py-roxene (Geissler et al., 1999). When Io is in eclipse(Jupiter’s shadow), or during an ionian night (visibleonly from spacecraft), visible and near-infrared imagesof the satellite reveal dozens of thermally-bright volcanic hot spots (e.g., Geissler et al., 2001; Macintosh et al.,2003; de Pater et al, 2004; Spencer et al., 2007; Rether-ford et al., 2007). This widespread volcanic activity ispowered by strong tidal heating induced by Io’s orbitaleccentricity, which is the result of the Laplace orbitalresonance between Io, Europa, and Ganymede. Somevolcanoes are associated with active plumes, which are amajor source of material into Io’s atmosphere, Jupiter’smagnetosphere, and even the interplanetary medium.The mass-loss from Io’s atmosphere is estimated at 1ton/second (Spencer and Schneider, 1996), yet the at-mosphere is consistently present, indicating an ongoingreplenishment mechanism. However, the amount of ma- a r X i v : . [ a s t r o - ph . E P ] S e p terial pumped into Io’s atmosphere by volcanism is notwell known, and it is consequently not known whetherthe dynamics in Io’s atmosphere is primarily driven bysublimation of SO frost on its surface or by volcanoes.An additional source of atmospheric gas may be sput-tering from Io’s surface.A decade after the inital detection of gaseous SO inits ν band (7.3 µ m) from the Voyager data (Pearl etal., 1979), Io’s “global” SO atmosphere was detected at222 GHz (Lellouch et al., 1990). These data revealeda surface pressure of order 4–40 nbars (2 × –2 × cm − ), covering 3–20% of the surface, at temperaturesof ∼ in the line profiles (e.g., Lellouch etal., 1992). Moreover, zonal winds would broaden the lineprofile (“competing” with temperature), while Ballesteret al (1994) noted that winds from volcanic eruptionsmay distort the lineshape, both adding complicationsto modeling efforts. Although SO has now been ob-served at mm, UV, and at thermal infrared wavelengths,its temperature and column density are still poorly con-strained.Based upon photochemical considerations alone, in aSO -dominated atmosphere one would expect at leastthe products SO, O , as well as atomic S and O (e.g.,Kumar, 1985; Summers, 1985). SO was detected at mi-crowave wavelengths at a level of a few % compared tothe SO abundance (Lellouch, 1996). While O has not(yet) been detected, S and O have been detected e.g.,in the form of auroral emissions off Io’s limb along itsequator (e.g., Geissler et al. 2004). We further notethat gaseous NaCl was first detected by Lellouch et al.(2003), and a tentative detection of KCl was reported byMoullet et al. (2013). Both NaCl and KCl were mappedwith ALMA by Moullet et al. (2015).Spatially resolved data obtained with the HubbleSpace Telescope (HST) at UV wavelengths revealed thatSO was mainly confined to latitudes within 30–40 ◦ fromthe equator, with a higher column density and latitu-dinal extent on the anti-jovian side (central meridianlongitude CML ∼ ◦ W) (e.g., Roesler et al., 1999;Feaga et al., 2004, 2009). The sub- (CML ∼ ◦ W) toanti-jovian hemisphere distribution was confirmed us-ing disk-averaged thermal infrared data of the 19- µ m ν band of SO , observed in absorption against Io with theTEXES instrument on NASA’s Infrared Telescope Fa- The fractional coverage of the gas is the fraction of the projectedsurface that is covered by the gas. cility (IRTF) in 2001–2005 (Spencer et al., 2005). Whilethese observations showed a temperature of ∼ ν + ν band at 4 µ m with the CRIRES instrument onthe Very Large Telescope (VLT) favors a temperature of ∼
170 K (Lellouch et al., 2015). Typical column densitiesin all these data vary roughly from ∼ on the sub-jovian hemisphere to ∼ cm − on the anti-jovianside.Moullet et al. (2010) used the spatial distributionderived from the HST/UV and TEXES/IRTF observa-tions to analyze SO maps at mm-wavelengths obtainedwith the Sub-Millimeter Array (SMA). By decreasingthe number of free parameters to just temperature andcolumn density, using the fractional coverage from theUV and mid-IR data, they derived a disk-averaged col-umn density of 2.3–4.6 × cm − and temperature be-tween 150–210 K on the leading (CML ∼ ◦ W) hemi-sphere, and 0.7–1.1 × cm − with 215–255 K on thetrailing (CML ∼ ◦ W) side. These temperatures andcolumn densities are considerably lower than the earliermm-wavelength measurements.As mentioned above, it is still being debated whetherthe primary source of Io’s atmosphere is volcanic ordriven by sublimation, although it is clear that bothvolcanoes and SO frost do play a role (Jessup et al.,2007; Lellouch et al., 1990, 2003, 2015; Moullet et al.,2010, 2013, 2015; Spencer et al., 2005; Tsang et al.,2012, 2016). Although much of the SO frost may ul-timately have been produced by volcanoes, the extentto which volcanoes directly affect the atmosphere is un-known; moreover, this likely varies over time. Mid-IRobservations showed an increase in the SO abundancewith decreasing heliocentric distance, which is, at leastin part, in support of the sublimation theory (Tsang etal., 2012). Further support was given by the analysis ofthe SMA maps mentioned above, which indicated thatfrost sublimation is the main source of gaseous SO , andphotolysis of SO is the main source of SO, since vol-canic activity is not sufficient to explain the SO columndensity and distribution (Moullet et al., 2010). On theother hand, SO gas is enhanced above some volcanichot spots ( McGrath et al., 2000) , and Pele’s plume con-tains the sulfur-rich gases S , S, and SO (McGrath et al.,2000, Spencer et al., 2000; Jessup et al., 2007), indicativeof volcanic contributions to Io’s atmosphere. For moreinformation on the pros and cons of the driving forces(sublimation vs volcanic) of Io’s atmospheric dynamics,see the excellent reviews of the state-of-knowledge of Io’satmosphere in the mid-2000’s by McGrath et al. (2004)and Lellouch et al. (2007). LMA Observations of Io in/out of Eclipse gas that makes up thebulk of Io’s atmosphere is expected to condense out ona similar timescale, set by the vapor pressure of thisgas, which is a steep exponential function of tempera-ture ( P vapor = 1 . × e − /T bar; Wagman, 1979).Tsang et al (2016) obtained the first direct obervationsof the SO ν band in Io’s atmosphere in eclipse withthe TEXES instrument on the Gemini telescope. Theirdisk-integrated spectra were sensitive to surface temper-ature, atmospheric temperature, and SO column abun-dance. Based on a simple model with a surface temper-ature of 127 K, they found that this value dropped to105 K within minutes after entering eclipse. A rangeof models for Io’s atmospheric cooling all showed thatthe SO column density simultaneously dropped, by afactor of 5 ±
2. They therefore concluded that the atmo-sphere must contain a large component that is driven bysublimation.Although the radical SO will not condense at thesetemperatures, it may be rapidly removed from the at-mosphere through reactions with Io’s surface (Lellouchet al. 1996). However, a bright emission band complexat 1.707 µ m, the forbidden electronic a ∆ → X Σ − transition of SO, was observed in a disk-integrated spec-trum of Io while in eclipse. Based on the linewidth, arotational temperature of ∼ vapor atdepth. These plumes do not have much dust or conden-sates, and are therefore not seen in reflected sunlight.The SO data are further suggestive of non-LTE pro-cesses, in addition to the direct ejection of excited SOfrom the volcanic vents.In order to shed more light on the core questionwhether the dynamics in Io’s atmosphere is predomi-nantly driven by sublimation of SO -ice or volcanic ac- tivity, we present spatially resolved observations of thesatellite at 880 µ m when Io moved from sunlight intoeclipse, and half a year later from eclipse into sunlight.The observations and data reduction are discussed inSection 2 with results presented in Section 3. The anal-ysis of line profiles is presented in Section 4, with a dis-cussion in Section 5. Conclusions are summarized inSection 6. OBSERVATIONS AND DATA REDUCTIONWe observed Io with the Atacama Large(sub)Millimeter Array (ALMA) on 20 March 2018 justbefore and after the satellite moved into eclipse. Simi-lar experiments were conducted when Io moved out ofeclipse on 2 and 11 September, 2018. Figure 1 showsthe viewing geometry on both occasions. All observa-tions were conducted in Band 7, the 1-mm band. Eachcontinuous observation of a source (calibrator or Io) isreferred to as a scan, and gets a scan “label”. Am-plitude and bandpass calibrations were performed onthe radio source J1517-2422 during the first ∼
15 minof each of the six ∼ and SO, andone transition of KCl. These transitions, together withthe spectral window (spw) used to observe them, the to-tal bandwidth and channel width of each spw are listedin Table 1. We typically had 3–4 beams (resolution ele-ments) across the satellite. Usually all scans on a partic-ular source are combined to create a map or a spectralline data-cube. Since we are interested in particular inhow the spatial brightness distribution and flux densitychange during eclipse ingress and egress, we imaged indi-vidual scans, and even fractions of a scan, as summarizedin Table 2.After the calibration and initial flagging was done inthe ALMA pipeline, we split off the Io data into its owndataset (referred to as a measurement set, Io.ms) andattached a new ephemeris file so that the position andvelocity got updated every minute of time. We usedthe Common Astronomy Software Applications pack-age, CASA, version 5.4.0-68 for all our data reduction.This version properly handles the tracking of Io’s motionacross the sky and velocity along the line-of-sight. Ourfinal products are centered on Io, both in space (images)and velocity (line profiles). We first created continuummaps of the satellite, used initially for additonal flag-ging and selfcalibration of the data (e.g., Cornwell andFomalont 1999). Mapping was done using tCLEAN; amodel of Io’s continuum emission served as a “start-model” in the deconvolution (“cleaning”) and selfcali-bration process. The model is a uniform limb-darkeneddisk with a disk-averaged brightness temperature thatmatches the data (typically between 65 and 80 K), anda limb-darkening coefficient q =0.3 (i.e., the brightnessfalls off towards the limb as cos θ q , with θ the emissionangle). All data were selfcalibrated twice (phase selfcalonly), although the second selfcalibration did not im-prove the data substantially over the first one.Before creating spectral line data-cubes, we split outeach spectral window (Table 1) into its own measure-ment set (Io-spwx.ms, with x=0–7), and subtractedthe continuum emission from each Io-spwx.ms (x=1–7) dataset using the CASA routine UVCONTSUB. Atthis point we have spectral image data-cubes of just theemission of each species (SO , SO, KCl).In order to create maps of the brightness distributionof each species at a high signal-to-noise (SNR), we aver-aged the data in velocity over 0.4 km/s, centered at thecenter of each line; these maps are referred to as “linecenter maps”. To evaluate line profiles, we also con-structed 3-dimensional (3D) data-cubes with RA andDEC along the x- and y-axes, and frequency (or veloc-ity) along the z-axis, where each image plane was aver-aged over 0.142 km/s, which translates roughly into afrequency resolution of ∼
160 kHz , slightly larger thanthe 122 kHz width of an individual channel in each spw.All (spectral line, line center, and continuum) maps wereconstructed using uniform weighting, and cleaned usingthe Clark or H¨ogbom algorithm with a gain of 10% inCASA’s tCLEAN routine. In essence, in this routinewe iteratively remove 10% of the peak flux density fromthat location on the map, together with the synthesized Note that the precise number depends on frequency ν , since ∆ ν = νv/c , with v the velocity and c the speed of light. beam (the telescope’s antenna pattern). This process isrepeated until essentially only noise is left in the “resid-ual” map. These socalled “clean components” form amap, the .model map in CASA. An example is shownin Figure 2. (Note that the continuum maps were de-convolved using a startmodel in tCLEAN, as describedabove). The .model map is convolved with a circulargaussian beam with a full width at half power (HPBW)that best matches the inner part of the synthesized beam(see Table 2 for the HPBW values) before being addedback to the residual map. The .model map in Fig. 2shows the clean components of the map displayed in thetop left panel in Figure 4, discussed below in Section3.2. We used a cell (or pixel) size for all maps of 0.04”,i.e., between 5.5–9 pixels/beam. RESULTS3.1.
Continuum Maps
The continuum maps for each of the 6 sessions, threein-sunlight and three in-eclipse, are very similar, and donot show any structure other than that the maximumtemperature is not centered on Io, but slightly displacedtowards the afternoon, as shown in Figure 3a,b. We de-termined the total flux density from such maps, since itis impossible to determine this from the uv -data, as theseare dominated by the signal from nearby Jupiter. Forthe maps in-eclipse, the sidelobe patterns from Jupiterproduce broad (similar size as Io) low-level (few % of Io’speak intensity) negative and/or positive ripples whichaffect the precise determination of Io’s flux density. Al-though ideally one would subtract Jupiter from the vis-ibility data, in practice this is not easy as Jupiter is nota uniform disk at mm-wavelengths (e.g., de Pater et al.,2019), moves with respect to Io, is mostly resolved out,and mostly on the edge or outside the ∼
20” primarybeam. We therefore opted to correct for these negativeor positive backgrounds by subtracting the average fluxdensity per pixel as determined from an annulus aroundIo in each of the 6 maps. Each map was constructedfrom all scans in the particular observing session, al-though the in-eclipse scan 6 on 11 September was notused (too much affected by nearby Jupiter).The total flux density, F J , normalized to a geocentricdistance of 5.044 AU (Io’s diameter is 1” at this distance)and averaged over all six measurements, F J = 5.43 ± T b = 93.6 ± LMA Observations of Io in/out of Eclipse ∼ ∼ T b in sunlight and in eclipse: in sunlight we find F = 5.43 ± T b = 93.6 ± F =5.25 ± T b = 90.8 ± T b may de-crease by ∼ µ m from 127 K to 105 K.At radio wavelengths we typically probe ∼ ∼ ∼ ∼ ∼ ∼ cm or so of Io’scrust is composed of two layers. For the model in Figure2c we assumed a bolometric Bond albedo, A =0.5, aninfrared emissivity (cid:15) =0.9 and a thermal inertia Γ =50 Jm − K − s − / . This number is similar to the value of70 J m − K − s − / derived by Rathbun et al. (2004)from Galileo /PPR data. At millimeter wavelengths weassumed A =0.5, (cid:15) =0.78 and Γ = 320 J m − K − s − / .These models, for a “typical” surface location at mid-latitudes, more or less match the data, and suggest thatIo’s surface is overlain with a thin (no more than a fewmm thick) low-thermal-inertia layer, such as expectedfor dust or fluffy deposits from volcanic plumes, over- lying a more compact high-thermal-inertia layer, com-posed of ice (likely coarse-grained, and/or sintered) androck. This is very similar to the model proposed by Mor-rison and Cruikshank (1973) based upon seven eclipseingress or egress measurements at a wavelength of 20 µ m, although our value for the low thermal inertia layeris ∼ µ m.They found a best fit by assuming Io to be covered byboth dark ( A =0.10) and bright ( A =0.47) areas, withΓ =5.6 and 50 J m − K − s − / , resp., where the lowthermal inertia layer is just a thin layer atop a muchhigher thermal inertia. They noted that cooling wasrapid during the first few minutes, followed by a slowerprocess which they attributed to a combination of thehigher thermal inertia, higher albedo passive componentand emission from hot spots. Although their thermal in-ertias are roughly an order of magnitude smaller thanthe values we found, the overall physical picture of athin dusty/porous layer atop a more compact high in-ertia layer is the same for all models. Our millimeterdata in particular add a strong constraint to the higherthermal inertia layer roughly a cm or so below Ios sur-face, a depth not probed at shorter wavelengths. Ourvalues for the upper dusty layer are also similar to thosereported for the other galilean satellites (e.g., Spencer,1987; Spencer et al., 1999; de Kleer et al., 2020), andthey agree with the best-fit values found in the ther-mophysical parametric study by Walker et al. (2012).Note, though, that the latter study, as well as other2-component thermal inertia studies at mid-IR wave-lengths, all refer to horizontal surface variations, whileour study refers to a vertically stacked model. In a fu-ture paper we intend to expand our 2-layer model toinclude proper dark and bright surface areas, as donefor Ganymede in de Kleer et al. (2020).3.2. Line Center Maps (averaged over 0.4 km/s)
Line center Maps on 20 March 2018SO maps: — Figure 4 (top row) shows SO maps at346.652 GHz (spw2), averaged in velocity over 0.4 km/s( ∼ ∼ ∼
15 min after entering eclipse (Table 2). Thelarge circle shows the outline of Io, as determined fromsimultaneously obtained images of the continuum emis-sion. As soon as Io enters an eclipse, the atmosphericand surface temperatures drop (Fig. 3), and SO is ex-pected to condense out on a time scale t ∼ H/c s ≈
70 sec, for a scale height H ≈
10 km and sound speedc s ≈ . × cm/s (de Pater et al., 2002), unless alayer of non-condensible gases prevents complete col-lapse (Moore et al., 2009). Once in eclipse, assumingcomplete collapse, the only SO we see should be vol-canically sourced. The letters on Figure 4 show the po-sitions of Karei Patera (K), Daedalus Patera (D), andNorth Lerna (L). Due to the excellent match between thelocation of these volcanoes and the SO emissions on thisday, these volcanoes are likely the main sources of SO gas for Io-in-eclipse. All three volcanoes have showneither plumes or changes on the surface attributed toplume activity in the past (Geissler et al., 2004; Spenceret al., 2007). SO maps: — SO can be volcanically sourced, i.e., pro-duced in thermochemical equilibrium in the vent (Zolo-tov and Fegley, 1998), or later via the reaction O + S at a column-integrated rate of 4 . × cm − s − , orwhile in sunlight it can be produced through photoly-sis of SO at a similar column-integrated rate (Moseset al., 2002). About 70% of SO is lost through pho-tolysis into S and O, but during an eclipse the onlyknown loss is through a reaction with itself: 2SO → SO + S, at a rate of 3 . × cm − s − (Moses etal., 2002). Hence, to eliminate an entire column of 10 cm − (Section 4.3) would take 8.5 hrs, or almost an hourto loose a 10 × smaller column. Hence one would not ex-pect much change in the SO column density upon eclipseingress. The data, however, clearly show a decrease inthe SO emission after eclipse ingress, though not as fastas for SO . The observed decrease suggests that SOmay be much more reactive with itself than capturedby the above reaction rate. Additional (in-between) re-actions are: 2SO → (SO) , and SO + (SO) → S O+ SO (Schenk and Steudel, 1965). At low tempera-tures (i.e., in-eclipse), both SO and S O condense out,and SO may effectively condense out through chemicalreactions in the gas-phase with the surface, producingthe above mentioned compounds (Hapke and Graham,1989). Based on our observations, it looks like suchself-reactions of SO must be very fast. Although thispossbility has been suggested in the past (e.g., Lellouch,1996), the SO “condensation” rate has never before beenobserved.As shown, the connection with volcanoes is more ten-uous for the SO emissions than for SO , except perhapsfor Daedalus Patera. However, as noted above, SO’scolumn density is not really expected to change much.With a layer of SO, and perhaps other noncondensiblespecies (e.g., O, O ), SO may indeed not completely collapse, such as modeled by e.g., Moore et al. (2009).We may also see emissions from stealth volcanoes, aspostulated by de Pater et al. (2020) to explain thewidespread spatial distribution of the 1.707 µ m SO emis-sions, which only occasionally showed a connection tovolcanoes. Disk-integrated flux densities: — We integrated the fluxdensity over Io on each map for each transition (exceptspw7, the lowest line strength, where we have no detec-tion), and plotted the results on the left panel of Fig-ure 5. For easier comparison, all flux densities in thisfigure have been normalized to a geocentric distance of5.044 AU, at which distance Io’s diameter is 1”. Assum-ing that Io’s flux density is constant while in-sunlight, itdecreases exponentially within the first few minutes af-ter the satellite enters eclipse. The dotted lines show thecollapse for each transition, modeled for the SO linesas: F i = F i ( t ) e − t j / ( H i + C i ( t j − t )) (1)where F i stands for the flux density in each transition i (for i = spw1, spw2, spw5, and spw6), t j the time(in min) from time t = 0 taken as midway during thepartial eclipse, t =8, t = 10.7, t = 19.5 min). H i shows the exponential decay constant in minutes (indi-cated on the figure). After the initial drop in intensity,further decrease is slowed down, as roughly indicated bythe term C i ( t j − t ), with C = 1 at 346.524 and 332.091GHz (spw1, spw5), C = 0 . C = 0 . ∼
20 minutes, similar tothe results shown by Tsang et al. (2016) at 19 µ m.The flux density decreases by a factor of 2 at 346.524,332.091, and 332.505 GHz (spw1, spw5, spw6), and by3.2 at the strongest line transition, 346.652 GHz (spw2).As shown by H i , the latter flux density decreases muchfaster than the others. Also the SO emission, plottedhere as the average of the two transitions, decreases bya factor of 2, although much more gradual, essentiallyfollowing a linear decay, modeled as: F i = F i ( t ) + at j (2)where a = −
15 mJy/min. This more gradual decreaseis also visible on the maps in Figure 4.3.2.2.
Line Center Maps on 2 & 11 September 20182 Sep. 2018 — Figure 6 shows the distribution of SO and SO gases on 2 September 2018, when Io moved fromeclipse into sunlight. The integrated flux densities areplotted on the right-side panel in Figure 5. As soonas sunlight hits the satellite, SO starts to sublime and LMA Observations of Io in/out of Eclipse a in mJy/min (positive sign for increasing slope) areindicated on the figure.When the satellite was in-eclipse on 2 September (scan6a in Fig. 6 ), the spatial distribution of SO gas showsvery strong emission near the SW limb, centered onP207 (91 W long., 37 S lat.), a small dark-floored pa-tera. Although thermal emission has been detected atthis site with the W. M. Keck Observatory (Marchis etal., 2005; de Kleer and de Pater, 2016), no evidenceof plume activity has ever before been recorded. Faintemissions can further been seen near Nyambe Pateraand just north of PFd1691, a dark-floored patera wherethermal emission has also been detected with the KeckObservatory (de Kleer et al., 2019). As soon as Io enterssunlight (scan 6b), the SO emission near P207 becomesmore pronounced; this is the side of Io where the Sunfirst strikes. Over the next 4–5 minutes (scans 8a, 8b)the emissions get stronger, in particular near the volca-noes. At 9 minutes (scan 12) the SO atmosphere hascompletely reformed.The bottom row in Figure 6 shows practically no SOemissions while Io is in eclipse (scan 6), except for someemission along the limb north and south of P207. Faintemissions are also seen near PFd1691, and at a few otherplaces on the disk. None of these emissions seem to bedirectly associated with known volcanoes, nor with theSO emissions on Io-in-eclipse. About 4 minutes afterentering sunlight (scan 8) strong SO emission is detectedabove P207, suggestive of formation through photodis-sociation of SO . Five minutes later we also detect emis-sions over Nyambe Patera, and another 20–30 min laterthe SO emissions track the SO emissions pretty well, asexpected if the main source of SO is photolysis of SO .
11 Sep. 2018 — Figure 7 shows the spatial distribu-tions of SO and SO of Io-in-eclipse and in-sunlight on11 September 2018, but not during the transition fromeclipse into sunlight. While in-eclipse, faint volcanicallysourced SO emissions are present near P129, Karei andRa Paterae, and along the west limb near Gish Bar Pa-tera and NW of P207. The eruption at P207, so promi-nent 9 days earlier, has stopped. No SO emissions areseen above the noise level. Ten minutes later, the atmo-sphere has reformed as shown by the in-sunlight map,with most of the emissions confined to latitudes within ∼ | − | , in agreement with the latitudinal extentmeasured from UV/HST data (Feaga et al., 2009), andwith Figures 4 and 6. The SO map shows emission peaks above Karei Patera and P129. The ratio of flux densi-ties between in-sunlight and in-eclipse is about a factorof 4–5 for SO and ∼
10 for SO on this day (Fig. 5).Hence, as shown both by this large ratio and the maps,on this date there was not much volcanic activity.3.2.3.
Map of KCl on 20 March 2018
On 20 March 2018 we also detected KCl, shown inFigure 8. As shown, the distribution is completely dif-ferent from that seen in SO and SO: the south-easternspot is centered near Ulgen Patera, and emission is seenalong the limb towards the north. There may also besome emission from near Dazhbog Patera. KCl wasnot detected in September 2018, when Ulgen Paterawas out of view. Further analysis of the KCl data willbe provided in a future paper.3.3. SO Line Profiles and Image Data Cubes(Resolution 0.142 km/s ≈
160 kHz)
In addition to the spatial distribution at the peak ofthe line profiles (i.e., the line center maps) when Io goesfrom sunlight into eclipse and vice versa, the full imagedata cubes contain an additional wealth of information.3.3.1.
Image Data Cubes on 20 March 2018
In Figure 9 we show several frames of the SO im-age data cubes together with disk-integrated line pro-files from March 2018 for Io-in-sunlight (top half) andin-eclipse (bottom half). To increase the SNR, we av-eraged the data at 346.524 and 346.652 GHz (spw1 andspw2). We also averaged all scans for the in-eclipse data(Table 2, scans 7–15 in Set 2) in this view.At the peak of the line profiles (frame 3), the imageslook similar to those shown in Figure 4. Moving awayfrom the peak we see the SO distribution at a partic-ular radial velocity, v r (the velocity along the line ofsight). It is striking how similar the images are movingtowards lower or higher frequencies (positive or negative v r ), i.e., the brightness distribution is very symmetricaround the peak of the line. If there would be a hori-zontal wind of ∼
300 m/s in the prograde direction, asreported by Moullet et al. (2008) from maps when Iowas near elongation (i.e., a different viewing geometry),we would expect spatial distributions asymmetric withrespect to the line center. In the first frame, where wesee the spatial distribution of gas offset by ∼ +0.6 MHz,i.e., moving towards us (blue-shifted – B) at a speed of ∼ v r = − . gason the west (left) limb; we would see the gas on the eastlimb in frame 5 where we map the brightness distribu-tion at v r = +0 . Image Data Cubes on 2 & 11 September 2018
Figure 10 shows the image data-cube from September2 when Io moves from eclipse into sunlight. The top halfshows the image data cube when Io was in sunlight, andthe bottom half shows the results for scan 6 when Io wasin eclipse. As for the March data, the broad asymmetricwings of the line profile are clearly produced by volcanicplumes, the plume at P207 on this date.On 11 September the situation is slightly different, asshown in Figure 11. There were no detectable volcanicplumes. When Io was in-eclipse, only faint SO emis-sions were seen. At the peak of the line, emissions seemto originate near Euboea Fluctus and Ra Patera. Butoverall, if SO is volcanically sourced, most faint emis-sions may be sourced from stealth volcanism, as men-tioned in Section 3.2. On the sunlit image data cube wesee some emission from the west limb in frame 1, nearZal Patera (northern spot) and Itzamna Patera (south-ern spot), and on the east limb on frame 5 at MazdaPatera. In frame 2 the emission has shifted more to-wards the center of the disk, but is still only visible onthe western hemisphere, i.e., the side that is moving to-wards us. In frame 4 more emission is coming from theeastern hemisphere, while in frame 5 emission comes pri-marily from the eastern limb. These frames could beinterpreted as indicative of a ∼ ∼
145 K. Clearly, the spatialdistribution on this day is not as symmetric around the center of the line (frame 3) as on the other two dates.On this date most SO must have been produced bysublimation, since we do not see clear evidence of vol-canic eruptions in-sunlight nor in-eclipse. This may bethe reason that we can distinguish zonal winds such asreported before by Moullet et al. (2008). If these windsare real, they must form within 10–20 minutes after Iore-emerges in sunlight. The reason for zonal winds, if in-deed present, remains a mystery, since we would expectday-to-night winds on a body with a warm day- and coldnightside (see, e.g., Ingersoll et al., 1985; Walker et al.,2010; Gratiy et al., 2010). ANALYSIS OF SO AND SO LINE PROFILESAs shown by Lellouch et al. (1990), the SO line pro-files as observed are saturated, and the peak flux den-sity depends not only on the temperature and columndensity, but also on the fraction of the satellite coveredby the gas. With our spatially-resolved maps and fiveobserved SO transitions, we should be able to deter-mine the atmospheric temperature, column density andfractional coverage, as well as constrain the presence ofwinds. This was not possible with any of the previouslypublished observations.4.1. Fractional Coverage.
The fractional coverage of the gas on Io can be deter-mined directly from maps of the SO gas as observed inthe various transitions. However, the fractional cover-age as seen on such maps (Fig. 4) is significantly affectedby beam convolution, which makes it hard to determinethe precise fraction. A better way is to use a decon-volved map such as shown in Fig. 2 and discussed inSection 2. The total number of pixels with non-zerointensities divided by the total number of pixels on Io’sdisk gives us the fractional coverage of the gas over thedisk. This procedure works best if the SNR in themaps is high, which is certainly true for the strongesttransitions, i.e., 346.652 GHz (spw2), and likely for SOin sunlight (346.528 GHz, spw1), as shown in Figures4, 6, and 7. The .model files cannot be trusted to ac-curately represent fractional SO gas coverage in-eclipse,since the signal is so low (hardly above the noise). Ifthe brightness distribution is very flat, like the contin-uum maps of Io, this procedure underestimates the frac-tional coverage; it works best if the spatial distributionconsists of point-like sources. The fractional coverage, f r map , for SO as determined from maps in-eclipse andin-sunlight for all 3 days is summarized in Column 3of Table 3. We typically see a 30–35% coverage for Io-in-sunlight. On 20 March, ∼
15 minutes after enteringeclipse, we measured ∼ LMA Observations of Io in/out of Eclipse ∼
10% when Io had beenin Jupiter’s shadow for ∼ f r map ≈ ∼
10% in September. In-eclipse this cover-age drops to below ∼ ∼
10% on allretrieved numbers for SO , and ∼
20% for SO.4.2.
Radiative Transfer Model
To model the line profiles, we developed a radiativetransfer (RT) code analogous to that used to model COradio observations of the giant planets (Luszcz-Cook andde Pater, 2013). We assume Io’s atmosphere to be in hy-drostatic equilibrium, so the density can be calculatedas a function of altitude once a temperature is chosen(we use an isothermal atmosphere in this work). Anymolecular emissions are assumed to occur in local ther-modynamic equilibrium (LTE), as expected for these ro-tational transitions in Io’s atmosphere (Lellouch et al.,1992). We perform RT calculations across Io’s disk ata cell size of 0.01” and a frequency resolution 1/4th ofthe resolution in the observations (i.e., roughly 40 kHz).Io’s solid body rotation ( v rot = 75 m/s at the equator)is taken into account; a simple increase/decrease in v rot can account for zonal winds.In order to account for potential Doppler shifts (blue-and red-shifts) in line profiles, which might be expectedfor localized volcanic eruptions or for day-to-night windsin disk-averaged line profiles, we added a separate pa-rameter, v r , in addition to the planet’s rotation andzonal winds. With this parameter we can accuratelyfit any offset in frequency at line center. As shown be-low, we do need the freedom to shift some modeled lineprofiles to match the data; potential reasons for suchshifts are discussed below and in Section 5.We adopted a surface temperature of 110 K with anemissivity of 0.8. For the analysis of our data we ranmany models, where we varied the column density, N ,from ∼ – few × cm − for SO , a factor of 10smaller for SO, the temperature T from ∼
120 – 700 K,and the rotational and Doppler velocities, v rot and v r ,each from ∼ −
400 to +400 m/s. In the following subsec-tions we analyze line profiles for March and September.Figure 12 shows sample contribution functions for thefour SO line transitions detected in our data. The line profiles based upon the parameters in panel a) matchthe observed line profiles quite well, as shown in Sec-tions 4.3 and 4.4. Panels c and d show the changes inthe contribution functions when the temperature or col-umn density are changed. Line profiles based upon theseparameters do not match our observed line profiles, butgive an idea where one probes under different scenar-ios. The column density used in panel d) matches thatusually reported for the anti-jovian hemisphere. whilethe temperature in panel c) is similar to the atmno-spheric temperature determined at 4 µ m (Lellouch et al.,2015). In panel b we show a calculation for a temper-ature that increases with altitude, such as expected forIo based upon plasma heating from above (e.g., Strobelet al., 1994; Walker et al., 2010). Resulting line profilesagain do not match any of our data. The bottom lineof this excercise is that we typically probe the lower 10up to ∼
80 km altitudes for column densities of ∼ –10 cm − , and that different transitions are sensitive todifferent altitudes in the atmosphere. We further notethat the temperature structure in the first few tens ofkilometers above the surface is unknown, which makesinterpretation of mm-data quite challenging.4.3. SO on 20 March 2018: Sunlight → Eclipse
Disk-integrated line profiles.
We first focus on the disk-integrated line profiles ofSO for Io-in-sunlight. We have 5 transitions; althoughthere essentially is no signal in the weakest line transi-tion (333.043 GHz, in spw7), it still helps to constrainthe parameters. The free parameters in our model are N t , T t , f r t , v rot and v r , where the subscript t is usedfor disk-integrated data. We thus have to find a set ofparameters that can match the line profiles in all SO transitions. Moreover, since the fractional coverage, f r t ,should match that derived from the maps, f r map (Table3), the parameter f r t is heavily constrained for disk-integrated line profiles.While the Doppler shift, v r , in our implementationleads to a shift in frequency (i.e., velocity) of the entireline profile, both the temperature and rotation of thebody (or any zonal wind), v rot , lead to a broadening ofthe line profile. Hence high values of v rot can be compen-sated by lower atmospheric temperatures. For example,for v rot =300 m/s and T t =195 K, the lineshape matchesthe observed profiles quite well; however, for any given N t , there is not a single value for f r t that can matchthe line profiles for all transitions; moreover, f r t shouldbe equal to f r map . Based upon such comparisons wecan rule out zonal winds much larger than ∼
100 m/s,which agrees with our earlier findings where we did not0see evidence on the maps for large zonal winds, exceptperhaps for September 11 (Section 3.3). Since there isno noticeable broadening in the line profiles for zonalwinds up to ∼
100 m/s, we ignore any potential presenceof zonal winds in the rest of this section, and simply use v rot = 75 m/s.By assuming that the fractional coverage of SO on Io, f r t , should be the same for all transitions, and be equalto f r map , we get a pretty tight constraint on the columndensity and atmospheric temperature: N t = (1 . ± . × cm − and T t = 270 ± K. These numbersare summarized in Table 3, together with f r t ; the spreadin f r t between transitions is written as an uncertainty.We found that the modeled profile had to be shifted inits entirety by +20 m/s (22–23 kHz), with an estimatederror of 7 m/s. Since uncertainties in the line positionsas measured in the laboratory are of order 4 kHz , theobserved offset cannot be caused by measurement errorsin the lab. This shift is indicative of material movingaway from us. This can be caused by an asymmetricdistribution of the gas with more material on the east-ern than western hemisphere. Alternatively, it can becaused by day-to-night flows, or gas falling down ontothe surface such as expected in volcanic eruptions afterejection into the atmosphere. The rising branch of gasplumes usually occurs over a small surface area (vent),is very dense ( ∼ few 10 cm − ; see, e.g., Zhang et al.,2003), and therefore saturated. While rising, the plumecools and expands, and the return umbrella-like flow, es-sentially along ballistic trajectories, covers a much largerarea, up to 100’s km from the vent, with column densi-ties about 2 orders of magnitude lower than at the vent.Hence, since the downward flow covers a much largerarea than the rising column of gas, one can qualitativelyexplain a redshift of disk-integrated line profiles. Thisidea was used by Lellouch et al. (1994; see Lellouch,1996 for updates) to explain ∼
80 m/s redshifts in theirline profiles, which they could model if there would be oforder 50 plumes on the observed hemisphere. Althoughthis seemed quite a large number of plumes at the time,if one considers the presence of stealt plumes (Johnsonet al., 1995) and the recent publication of the spatialdistribution of 1.707 µ m SO emissions (de Pater et al.,2020), this may be a quite plausible idea.Several fits are shown in Figure 13, panels a and c. Foreach of the models shown we used the mean fractionalcoverage as derived from the line profiles in the fourspectral windows (spw1, spw2, spw5, and spw6) for thatparticular model, i.e., f r t = 0.32 for the best-fit model https://spec.jpl.nasa.gov ( N t = 1 . × ), but f r t =0.39 for N t = 1 × cm − and f r t =0.26 for N t = 2 × cm − . Whileall three model curves might match one or two spectralwindows, only one curve (red one) fits all transitions, aswell as f r map . Note, though, that none of the curvesfits the broad shoulders of the observed profiles; this isclearly caused by the relatively high velocities (Dopplershift) of the eruptions, as discussed in Section 3.3 andFigure 9.The column density and temperature hardly changefor Io-in-eclipse (scan 11+15) (Table 3). The drop influx density is mainly caused by the factor of ∼ sublimation. In this particular case thereappears to be excess emission at lower frequencies, i.e.,at velocities moving away from us.4.3.2. Line profiles for individual volcanoes.
We next investigate the line profiles of individual vol-canoes, Karei and Daedalus Paterae. These line pro-files were created by integrating over a circle with a di-ameter equal to the HPBW (Table 2) centered on thepeak emission of the volcano on the 346.652 GHz (spw2)map. We determined the line profile for the models inthe exact same way, so that the rotation of the satel-lite was taken into account, and the viewing geometry(i.e., pathlength through the atmosphere) was the same.Hence, the modeled line profile for a volcano on the West(East) limb is already Doppler shifted to account forthe satellite’s rotation towards (away from) us, and anyadditional shifts are intrinsic to the volcano itself. Asshown, the hydrostatic line profiles match the observedspectra for Karei Patera in sunlight very well (Fig. 13e)with a column density and temperature that are quitesimilar to the numbers we found for the integrated fluxdensities, but with a fractional coverage of almost 50%(Table 4). Hence, the column density (cm − ) of SO gas in-sunlight appears to be quite constant across Ioover areas where there is gas, i.e., over approximately30–35% of Io’s surface in-sunlight, and over about halfthe area of a volcanically active source (note that we in-tegrated here over approximately the size of the beam,so the plume itself may be unresolved). LMA Observations of Io in/out of Eclipse f r v (the subscript v stands for volcano) as listed in Table 4shows the spread in f r v between the four transitions. Ifthe spread is small, the solution is quite robust. Whenthe spread is large, the results should be taken witha grain of salt. The line center is offset by +60 m/s,indicative of material moving away from us, such asmight be expected for an umbrella-shaped plume as dis-cussed above. The in-eclipse profile (panel f) can also bematched quite well, with a similar temperature, perhapsa higher column density, but a much lower f r v .The observed profile for Daedalus Patera in-sunlight isvery different. The profiles in all four transitions have apronounced red wing (Fig. 13g). The main profile can bematched quite well with T v ≈
220 K, and N v ≈ × cm − , with a fractional coverage of 46%. The line centerappears to be Doppler shifted by −
40 m/s, i.e. mate-rial moving towards us. Note that the line offsets forthe two volcanoes are in the direction of a retrograde,rather than prograde, zonal wind; however, if such awind would prevail, we would expect the windspeed tobe largest near the limb (Daedalus Patera), i.e., oppo-site to the observations. The observed Dopplershifts aremore likely local effects, produced by the eruptions. Forthe in-eclipse profile (Fig. 13h) no good solution could befound, as indicated by the large uncertainties. This isnot too surprising, since in-eclipse most emissions arelikely volcanic in origin, since as soon as SO gas iscooled to below its condensation temperature it maycondense out. The applicability of our simple hydro-static model is therefore limited. To properly modelthese one needs to add volcanic plumes to the model,such as done by e.g., Gratiy et al. (2010). (see alsoSection 5.3).4.4. SO on 2 & 11 September 2018: Eclipse → Sunlight
Figure 14 shows several line profiles for the Septem-ber data; best fits are summarized in Tables 3 and 4.As with the March data, the disk-integrated line profilesfor both 2 and 11 September for Io-in-sunlight can bemodeled quite well with our simple hydrostatic model,in contrast to line profiles taken of Io-in-eclipse whereemissions must be volcanic in origin, and the applicabil-ity of our model is limited. From our hydrostatic mod-els we find that the SO fractional coverage on bothdays is a factor of 3 lower for the in-eclipse data thanfor Io-in-sunlight, while it was only a factor of 2 inMarch. On the latter date the satellite had only been in shadow, though, for 15 minutes, much shorter thanfor the September data. While on 11 September thecolumn density between in-sunlight and in-eclipse datais very similar, on 2 September it may be a factor of2 higher when in-eclipse, although the uncertainties arelarge enough to accommodate no-change as well.Line profiles of individual volcanoes, calculated byintegrating over a circle with a diameter equal to theHPBW, also deviate significantly from the hydrostaticmodels, although for volcanoes in-sunlight the discrep-ancies are smaller than when in-eclipse. We were ableto find a good model for P207, in particular in-sunlight,where a column density quite similar to that found forthe disk-integrated line profiles covers ∼
60% of the vol-cano. During eclipse the fractional coverage decreasesby a factor of ∼ f r v for all 4 transitions,which translates into a high uncertainty even for thein-sunlight data. 4.5. SO Line Profiles
We modeled the disk-integrated SO line profiles in Fig-ure 15 by adopting the temperature that was determinedfrom the SO profiles on the various days, since the at-mospheric temperature should not depend on the speciesconsidered. The fractional coverage as determined fromthe line center maps for Io-in-sunlight is about a factorof 2 lower for SO than for SO in March, and more likea factor of 3–4 in September. The temperature togetherwith this fractional coverage should result in a trust-worthy value for the column density, assuming againthat the atmosphere is in hydrostatic equilibrium. Withthese assumptions we find a column density of ∼ cm − on 20 March when in-sunlight, roughly a factor of10 below the SO column density. This, with the lowerfractional coverage, suggests a difference of a factor of ∼
20 between the total volumes of SO and SO gas, ingood agreement with previous observations (e.g., Lel-louch et al., 2007). On 2 September the column densityis roughly a factor of 5 lower than the SO column den-sity, which with the much lower fractional coverage alsosuggests almost a factor of 20 difference in gas volumes.On 11 September the column density is again a factor of10 below the SO value, but with a much lower fractionalcoverage this results in a difference of ∼
40 between thevolumes of SO and SO gases.As shown in Figures 4, 6, 7 and 15 we did detect SOwhen Io was in-eclipse in March and on 2 September,2but not on 11 September. The SNR in the maps, how-ever, is very low, which prevented a good estimate ofthe fractional coverage, a necessary quantity to deter-mine the column density from the data. Assuming thesame temperature as derived from the SO maps for Io-in-eclipse, we find a fractional coverage of 7% for an SOcolumn density of 10 cm − , i.e., about half the frac-tional area for the same column density as seen in thein-sunlight maps. The fractional coverage decreases fora higher value of N t , and vice versa for a lower value. On2 September the most likely scenario for Io-in-eclipse isthat both f r t and N t decrease by a factor of 2, while on11 September no SO emissions were detected in-eclipse. DISCUSSION5.1.
Summary of Observations
We observed Io with ALMA in Band 7 (880 µ m) infive SO and two SO transitions when it went from sun-light into eclipse (20 March 2018), and from eclipse intosunlight (2 and 11 September 2018). On all three dayswe obtained disk-resolved data cubes, and analyzed SO and SO line profiles for both the disk-integrated data,and for several active volcanoes on the disk-resolved datacubes. Specifics on the observations and derived param-eters are summarized in Tables 1–4.5.1.1. Disk-integrated data: SO The line-emission disk-integrated flux densities aretypically ∼ − × higher for Io-in-sunlight than in-eclipse (Fig. 5), indicative of a roughly 30–50% contri-bution of volcanic gases to the SO emissions. How-ever, there is much variability in these numbers. On11 September the SO flux density in-sunlight is 4–5times higher than in-eclipse. In March, when Io wentinto eclipse, the flux density in the strongest transition, F . , dropped exponentially by a factor of 3, in con-trast to the factor-of-2 drop in the three weaker transi-tions. F . for Io-in-sunlight was ∼ × higher than F . , about 30% above the ratio in their intrinsic linestrengths (Table 1). In contrast, the observed ratios be-tween F . with F . and with F . are 40 and60% smaller than the ratios between their intrinsic linestrengths. Since the flux density in the various tran-sitions depends also on the atmospheric temperature,which determines (in LTE) which energy levels in themolecule are populated (Boltzmann’s equation), the dif-ferences in flux density between the various transitionswas used in Section 4 to determine the column densityand atmospheric temperature. For example, for lower The disk-integrated flux densities are normalized to a geocentricdistance of 5.044 AU for intercomparison of the datasets. temperatures, the modeled F . would be too low,while the modeled F . would be too high, whichcan be qualitatively understood from the difference incontribution functions between panels a and c in Fig-ure 12. We found that neither the atmospheric temper-ature nor the column density between the in-sunlightand in-eclipse data sets did noticeably change, but thatthe differences in flux density could be accounted for bya factor of 2–3 decrease in fractional coverage. Or inother words, the column densities (cm − ) remained thesame, but there were fewer areas (2–3 times less) abovewhich SO gas was present.Tsang et al (2016) measured a factor of 5 ± ±
50 K. Thisis clearly an oversimplification, since the temperaturewill certainly vary with altitude, and also with latitude,longitude, and time of day. Walker et al. (2010) showthat the atmospheric (translational) temperature risessteeply with altitude due to plasma heating from above.Near the surface the SO gas is expected to be in equi-librium with the surface frost, rising to ∼
400 K at analtitude of 70 km during the day; at night the plasmacan reach lower altitudes so that the 400 K tempera-ture may be reached at an altitude of ∼
40 km. Theexact 3D temperature profile depends on the 3D dis-tribution of the atmospheric density, which for SO istightly coupled with the frost distribution and temper-ature. Moreover, as seen from the previous sections,volcanic plumes may affect the atmosphere and its tem-perature structure quite dramatically. We will get backto this in Section 5.3. The bottom line is that altitude-dependent changes in temperature and density affect thevarious transitions in different ways, as shown by thecontribution functions in Figure 12.In contrast to the exponential decrease in intensityduring eclipse ingress, there is a linear increase duringegress for at least about 10 minutes. Interestingly, theSO flux density in three of the four transitions ∼ ± F . , 16 ±
7% for F . and 19 ±
7% for F . .The disk-integrated flux densities for September 11are shown alongside the September 2 numbers in Figure5. The in-eclipse flux densities for SO on Sep. 11 aretypically a factor of 2 (two strongest transitions) – 3(two weakest transitions) below the Sep. 2 in-eclipse LMA Observations of Io in/out of Eclipse are well belowthe Sep. 2 values, in particular for the two weakesttransitions, which are lower by a factor of 1.18 ± ± at temperatures below its dew point(i.e., near the surface) condenses out, only gas sourcedfrom volcanic vents, or SO gas that was prevented fromcomplete collapse by a layer of non-condensible gases(e.g., Moore et al., 2009) can be present on Io-in-eclipse.These gases apparently cover ∼ µ m forbidded emissionson this hemisphere (de Pater et al., 2020). Point sourcesand glows of gases that were seen on this hemispherewith the New Horizons spacecraft, interpreted as beingcaused by plasma interactions with the (near-)surface(Spencer et al., 2007), could also be a signature of stealthvolcanism (de Pater et al., 2020).5.1.2. Disk-integrated data: SO
The SO flux density for Io-in-sunlight is highest onSep. 2 (0.66 ± ± , indicative offormation from SO through photolysis. According toMoses et al. (2002), SO is formed through photolysis ofSO at a column abundance rate of 4 . × cm − s − ;i.e., it takes about 1/2 hr to produce a full column of ∼ cm − s − , and less depending on how much SOgas is left. Above volcanoes, about 50% of SO is pro-duced this way, and another 50% may be produced ata similar rate through the reaction of O+S . The data(Figure 5) show that SO is fully restored within ∼ Disk-resolved data and line profiles
On 20 March the SO emission is dominated by thevolcanically active Karei and Daedalus Paterae, while some low-level emission is seen near North Lerna. On2 September the emission is dominated by P207, whilewe also see emission near PFd1691 and Nyambe Patera.The situation is less clear on 11 September: P207 Paterawas no longer active, while emissions on Io-in-sunlightwere seen above Karei and Nyambe Paterae. In-eclipsevery low-level activity was seen over Karei, Gish Bar, Raand P129 Paterae, north of P207, and near Euboea Fluc-tus. SO emissions in March tracked the SO emissionsreasonably well, i.e., both Karei and Daedalus Pateraeshowed activity. On 2 September no clear SO emis-sions were seen during eclipse, but ∼ emissions.This all suggests that the main source of SO is photolysisof SO . As mentioned above, at volcanic eruption sitesa full column of ∼ cm − s − will be produced in 15min. The SO peak intensity levels above P207 changedfrom (cid:46) ∼ ∼ ∼ and SO emis-sions is most clearly seen in the lineshapes. While disk-integrated Io-in-sunlight data can usually be matchedquite well with hydrostatic models, the in-eclipse profilesdeviate considerably from such gaussian-shaped profiles.Both in-sunlight and in-eclipse disk-integrated line pro-files show broad low-level wings out to ∼ line pro-file of Karei Patera showed a +60 m/s shift, and a − Post-eclipse Brightening
As mentioned above, the flux density on 2 September ∼
10 minutes after emerging from eclipse was consider-ably higher, up to ∼
20% in some transitions, than 1/2hr later. This appears to be an anomalous post-eclipsebrightening effect. A ∼
10% brightening of the satellitefor about 10–20 min after emerging from eclipse was firstreported by Binder and Cruikshank (1964) at a wave-length of 450 nm, i.e., they observed the satellite’s sur-face in reflected sunlight. The authors noted that Io was ∼
10% brighter when it emerged from eclipse, which de-creased over the next ∼
15 minutes. They suggested thatthe brightening might be caused by an atmospheric com-ponent that condenses on Io’s surface during the eclipse.This makes the satellite bright; the ice should evaporateonly minutes after receiving sunlight again, resulting ina slow darkening, back to its original reflected-sunlightintensity. We note that these observations were obtainedbefore Io’s atmosphere and its volcanic activity were de-tected – in fact, based on their data, the authors sug-gested Io to have an atmosphere. During subsequentyears, both detections and non-detections (e.g., Cruik-shank et al., 2010; Tsang et al., 2015) of this “post-eclipse brightening” effect have been reported at wave-lengths from the UV to the mid-IR. Explanations of theeffect range from condensation with subsequent sublima-tion of SO -frost (Binder and Cruikshank, 1964; Fanaleet al., 1981; Belluci et al., 2004), to changes in Io’s re-flectivity due to sulfur allotropes as a result of changesin surface temperature (Hammel et al., 1985), to inter-actions of amospheric molecules with Jupiter’s magneto-spheric plasma (Saur and Strobel, 2004). Some authorsconcentrate on phenomenae causing a brightening of thesurface, and others of the atmosphere. No clear expla-nation has been provided yet; and as shown by the data, the effect has not always been detected, which has beeninterpreted by possible differences in frost coverage atdifferent longitudes.The gradual increase in flux density in our data dur-ing the first ∼
10 minutes after emerging from eclipse intosunlight is exactly how Binder and Cruikshank (1964)explained their observed post-eclipse brightening of Io’ssurface: The surface was bright since the SO -ice cover-age had increased due to condensation while in eclipse;as soon as the surface warmed, SO sublimed, the satel-lite’s surface darkened, and the atmosphere reformed.The situation, as we observed it, is a bit more complexin that we see the SO flux density to “overshoot” after10 minutes (the end of our observing session 1 on Sep.2), before reaching a steady state (in observing set 2on Sep. 2). In the next section we show that this mayresult from the interaction of volcanic plumes with thereforming atmosphere.5.3. Comparison of Data with Atmospheric Models
Summary of Published Plume Simulations
In the following we compare the above results withsimulations of volcanic plumes by Zhang et al. (2003)and McDoniel et al. (2017). These simulations includea full treatment of gas dynamics, radiation (heating andcooling through rotational and vibrational radiation),sublimation and condensation. McDoniel et al. (2017)coupled Zhang et al.’s (2003, 2004) original plume modelto a model of a sublimation-driven atmosphere, devel-oped over the years by Moore et al. (2009) and Walkeret al. (2010, 2012). Simulations with this coupled modelshow how a volcanic plume on the dayside expands ina sublimating atmosphere. The authors present mod-els for a Pele-type plume both on the night and dayside. They assumed a night side surface temperature of90 K, and 116-118 K during the day. The gas eruptsfrom the vent at a temperature of ∼
600 K and a sourcerate of ∼ SO molecules/second at hypersonic veloc-ities of close to 1 km/s. It then expands and cools. Atan altitude of ∼
300 km a canopy-shaped shock forms(due to Io’s gravity field) where the radially expand-ing molecules turn back down to the surface. Mostof the gas falls down ∼ gas condenses and forms a ringaround the volcano which matches the red ring observedaround Pele. Due to plume expansion and vibrationalcooling the gas temperature above the vent decreases tovery low ( ∼
50 K) temperatures, while the temperaturein the canopy-shock is of order 300–400 K.On the night side the model shows an average columndensity N v = 1 . × cm − over a region up to ∼ LMA Observations of Io in/out of Eclipse N v ≈ cm − , and drops by an order of magnitudeover a 30-km distance.A Pele-type plume on the day-side is different becausethere is also SO sublimation from Io’s surface, andhence the plume expands in a background atmosphere.The extent to which a dayside sublimation atmosphereis affected by plumes depends on the size, density andejection velocity of the plume, as well as on the densityof the sublimation atmosphere, which is set by the tem-perature of the surface frost (for details, see McDoniel etal. 2017). Plumes that do not rise up above the exobase(i.e., the altitude at which the mean free pathlength be-tween collisions is equal to one atmospheric scaleheight,which is typically at an altitude of ∼ Comparing ALMA Data with AtmosphericSimulations Sublimation atmosphere: — During eclipse ingress inMarch the SO flux density decreased exponentially,caused by a decrease in the volume of SO molecules(assuming a hydrostatic atmosphere we showed that thecolumn density and temperature did not change much;only the fractional area decreased). With such a tenuousatmosphere, one would expect the surface temperatureto drop instantaneously when entering eclipse, as shownto be true by Tsang et al (2016). Given a diffusion timeof 70 sec (Section 3.2.1; de Pater et al., 2002), the SO molecules are expected to rapidly condense onto the sur-face, which means that the number density of moleculesjust above the surface decreases, resulting in a down-ward motion of gas above it. Moore et al. (2009) showthat changes occur primarily in the bottom 10–20 km.They further show that even a small amount of non-condensible gases will form a diffusion layer near the sur-face. Once this layer is several mean-free path-lengthsthick, it will prevent or at least slow down further col-lapse of the SO atmosphere. They predict this to hap-pen after about 20 min. They also predict that in thiscase the gas column density and the atmospheric tem-perature remain essentially the same. In their calcula-tions they assumed, though, that SO is non-condensible,while our data show that SO in essence rapidly con-denses through self-reactions on the surface (the fluxdensity or volume of SO molecules decreases linearly ata rate of 15 mJy/min). However, since our observa-tions show essentially no change in column density andtemperature, and some SO gas is always present, evenwhen volcanic activity is low (Fig. 11), atmospheric col-lapse may indeed be retarded by a diffusive layer of non-condensible gases near the surface. We cannot excludethe possibility of SO emissions due to stealth volcan-ism, however.During eclipse egress in September both the SO andSO emissions increase linearly, though SO is clearly de-layed compared to SO , which we attributed to forma-tion through photochemistry (Section 5.1.2). The atmo-sphere is restored within about 10 min after re-emergingin sunlight. This suggests that the surface heats upessentially instantaneously, causing SO -ice to start tosublime immediately. This is very different from thecalculations by Moore et al. (2009), who show the at-mosphere to reform on a much ( (cid:38) × ) longer timescale. Volcanic plumes: — The beamsize in our data is ∼ ∼ v r of order 700 m/s, which agrees wellwith the wings in our disk-integrated line profiles, whichare caused by the plumes (Figs. 9,10).Within 1–2 minutes after emerging in sunlight, theP207 plume increased in intensity, and continued to in-crease for the next several minutes (Fig. 6). During thisperiod, the plume transitions from a night-side plume toa day-side plume, when SO sublimation from SO frostbecomes important. McDoniel et al. (2017) show cal-culations of a plume transitioning from the night to thedayside, and back into the night, a process that takesalmost a full Io day (42.5 hrs). The ALMA observa-tions, in contrast, show a very accelerated process sinceeclipse egress only takes a few minutes. During thesefew minutes, SO frost starts to sublime and the at-mosphere reforms, while the volcano continues to ejectgases. The plume starts to interact with the atmospherewhile it is forming. A re-entry shock forms where theplume material hits the atmosphere. The resultinghigh temperature (Section 5.3.1) accelerates SO subli-mation, which gets entrained in the plume flow, causingthe plume area to grow. Hence, the observed bright-ening and expansion of the SO emissions near volcanicvents, i.e., near regions where we see some (though some-times faint) SO emissions during eclipse is consistentwith McDoniel et al.’s simulations. It may also causethe post-eclipse brightening effect we see about 10 minafter eclipse egress, where the sudden change from nightto day and the interaction of the plume with the re-forming atmosphere may lead to a temporary “excess”in SO emissions, likely due to an altitude-dependenttemporary increase in atmospheric temperature.The authors further show that the average columndensity over the vent at night is ∼ cm − , andthat during the day the column density over the plumematches that over the dayside hemisphere at distances (cid:38)
150 km. This essentially agrees with our observations,where column densities over the plume and backgroundatmosphere on the dayside are very similar. The differ-ences in brightness we see between the day- and night-(eclipse) side, both disk-averaged and over volcanoes, are mostly explained by changes in the fractional areacovered by the gas, but columns of gas over these ar-eas are very similar. Given our relatively low spatialresolution this may well be consistent with the models.The temperature that best matches our line pro-files, ∼ ∼
50 K abovethe vent up to 300–400 K at the canopy shock. Forcomparison, when Moullet et al. (2008) parameterizedthe Zhang et al. (2003) night-side plumes at a location ∼ ◦ away from disk center, they found that the modelscould be mimicked well with an isothermal temperatureof ∼
190 K. However, given how complex the plume–atmosphere interaction is (McDoniel et al., 2017), we donot think that the atmosphere can be modeled correctlyusing a simple isostatic atmosphere.In March we detected vigorous eruptions at Karei andDaedalus Paterae. At Karei Patera the fractional cover-age in sunlight was roughly 3 times larger than in eclipse,with an atmospheric temperature of 270 ±
50 K both insunlight and in eclipse. As mentioned before, the entireprofile was redshifted by 60 m/s, while in eclipse the pro-file was slightly skewed, peaking more at the blue sideof the spectrum. The overall shift towards the red isindicative of plume material falling back down onto thesurface, away from us; since the umbrella-shaped plumeis much larger in extent than the rising column of gas,this can qualitatively explain the line profiles.Daedalus Patera, in contrast, shows a strong red-shifted shoulder, somewhat similar to the blue shiftedshoulder for P207 in September. The entire profile wasslightly blueshifted, presumably because the umbrella-shaped plume material from a volcanic ejection near thelimb has a large component of material moving towardsus (i.e., similar to the material that explains the blue-shifted wing of the line for P207). The red-shifted wing,though, shows that a large component of plume materialis also moving away from us.When Moullet et al. (2008) modeled the Zhang etal. plumes for comparison with their radio data, theydid not see such red-shifted shoulders in the models.This, together with our observations of line profiles thatare very asymmetric, in particular in-eclipse, shows thatthe volcanic eruptions are much more complex than theZhang et al. (2003) and McDoniel et al. (2017) mod-els predict. This is not too surprising; volcanic erup-tions are likely not axisymmetric, and may fluctuate inejection speed, direction, and gas content on timescalesmuch shorter than we can capture in our observations.Yet, it is re-assuring that our observations do qualita-tively match many features in the model.
LMA Observations of Io in/out of Eclipse CONCLUSIONSWe used ALMA in Band 7 (880 µ m) to observe Io infive SO , two SO, and one KCl transitions when it wentfrom sunlight into eclipse (20 March 2018), and fromeclipse into sunlight (2 and 11 September 2018). Wesummarize the main findings as follows: • The disk-averaged brightness temperature at 0.9 mmis 93.6 ± ∼ µ m (Tsang et al., 2016), suggests that Io’ssurface is composed of a thin low-thermal-inertia (50 Jm − K − s − / ) layer, overlying a more compact high-thermal-inertia (320 J m − K − s − / ) layer, indica-tive of a thin ( (cid:46) few mm) layer of dust or fine-grainedvolcanic deposits overlying more compact layers of rockand/or coarse-grained/sintered ice. • The SO and SO disk-integrated flux densities aretypically about 2–3 times higher on Io-in-sunlight thanin-eclipse, indicative of a 30–50% volcanic contributionto the emissions. • During eclipse ingress, the SO flux densitydropped exponentially, with the 346.652 GHz transition(strongest line intensity) faster and more than the othertransitions. Following eclipse egress, the SO flux densi-ties increased linearly, with the 346.652 GHz transitionfaster than the others. • Eclipse egress observations show that the atmo-sphere is re-instated on a timescale of 10 minutes, con-sistent with the interpretation of the post-eclipse bright-ening effect reported for observations of Io’s surface re-flectivity. An atmospheric post-eclipse brightening wasseen in several SO transitions, where the flux densitywas up to ∼
20% higher 10 min after re-emerging in sun-light compared to 1/2 hour later. • We attribute the variations in emissions and differ-ences between line transitions during eclipse ingress andegress, as well as the atmospheric post-eclipse brighten-ing effect to altitude-dependent changes in temperature,likely caused in/by volcanic plumes and their interactonwith the atmosphere, such as simulated by McDoniel etal. (2017). • The SO flux density dropped/increased linearly afterentering/re-emerging from eclipse, in both cases clearlydelayed compared to SO . This provides confirmationthat SO may be rapidly removed through reactions withIo’s surface once in eclipse, and that photolysis of SO is a major source of SO. • Spectral image data cubes reveal bright volcanicplumes on 20 March and 2 September; no plumes weredetected on 11 September. Plumes on the limb createhigh-velocity wings in the disk-integrated line profiles (at (cid:38)
600 kHz, or (cid:38)
500 m/s). Such high velocities matchthose predicted in plume simulations by Zhang et al.(2003) and McDoniel et al. (2017). • In addition to the few obvious volcanic plumes in ourspectral image data-cubes, the low level SO emissionspresent during eclipse may be sourced by stealth vol-canic plumes or be evidence of a layer of non-condensiblegases preventing complete collapse of SO , as modeledby Moore et al. (2009). • Based upon hydrostatic model calculations, typicaldisk-integrated SO column densities and temperaturesare N t ≈ (1 . ± . × cm − and T t ≈ −
320 Kboth for Io-in-sunlight and in-eclipse. SO column den-sities are roughly a factor of 5–10 lower. The main dif-ferences between in-sunlight and in-eclipse flux densitiesappear to be caused by a factor of 2–3 smaller fractionalcoverage in-eclipse (i.e., down from 30–35% SO and ∼
12% SO in-sunlight). • The active volcanoes on 20 March and 2 Septembershow similar SO column densities and temperatures asfor the disk-integrated profiles, but with a very highfractional coverage ( ∼ • Line profiles of in-eclipse data are very asymmetric,both for disk-integrated profiles and individual volca-noes. Some volcanoes show red-shifted, and others blue-shifted shoulders both in-sunlight and in-eclipse. Some-times the entire profile is slightly red- or blue-shifted.The line profiles must be strongly affected by intrinsicproperties of volcanic plumes (e.g., ejection speed, di-rection, density, and variations therein), in addition totheir viewing geometry. • The data are suggestive of a 300–400 m/s horizontalprograde wind on 11 September, when no volcanic activ-ity was reported; however such a wind is not supportedby disk-integrated line profiles. No zonal winds weredetected on 20 March and 2 September, when volcanicplumes were seen. • KCl gas has only been detected on 20 March, sourcedmainly from near Ulgen Patera. No SO or SO gaswas detected at this location. Hence the magma in thechambers that power volcanoes must have different meltcompositions, and/or the magma has access to differentsurface/subsurface volatile reservoirs. • Our data can be qualitatively explained by thenight-side plume simulations of Zhang et al. (2003)and day-side simulations by McDoniel et al. (2017),although it is also clear that the data are much morecomplex than the models can capture.8Our observations begin to clarify the role of volcanismin forming Io’s atmosphere. However, many questionsstill remain, including, e.g., Io’s overall atmospherictemperature profile, in particular in the first 10-20 kmabove the surface; longitudinal variations in column den-sity; winds; volcanic sources; magma composition. Al-though it is clear that low-level emissions are presentduring eclipse, we do not yet understand the cause ofthese: perhaps stealth volcanism, or a layer of non-condensible gases preventing complete collapse of theSO atmosphere. To further address these questions weplan to obtain ALMA data at a higher spatial resolutionwhen the satellite is at eastern and western elongation.Future work will also need to include realistic plumemodels in addition to the hydrostatic models employedhere. ACKNOWLEDGEMENTSWe are grateful for in-depth reviews by David Gold-stein and one anonymous referee, which helped im-prove the manuscript substantially. This paper makesuse of ALMA data ADS/JAO.ALMA REFERENCESBallester, G. E., M. A. McGrath, D. F. Stobel, X.Zhu, P. D. Feldman, and H. W. Moos, Detection of theSO atmosphere on Io with the Hubble Space Telescope,Icarus 111, 2-17, 1994.Bellucci, G., Aversa, E.D., Formisano, V., et al., 2004.Cassini/VIMS observation of an Io post-eclipse bright-ening event. Icarus, 172, 141-148.Binder, A.P., Cruikshank, D.P., 1964. Evidence foran atmosphere on Io. Icarus 3, 299-305.Cornwell, T. J., and E. B. Fomalont 1999. Self-calibration. In Synthesis Imaging in Radio AstronomyII (G. B. Taylor, C. L. Carilli, and R. A. Perley, Eds.),pp. 187-199, ASP Conf. Series. Astron. Soc. of thePacific, San Francisco.Cruikshank, D.P., Emery, J.P., Korney, K.A., Bel-lucci, G., Aversa, E., 2010. Eclipse reappearances of Io: Time resolved spectroscopy (1.9–4.2 µ m). Icarus,205, 516-527.de Kleer, K., I. de Pater, 2016. Time Variability of Io’sVolcanic Activity from Near-IR Adaptive Optics Obser-vations on 100 Nights in 2013-2015. Icarus, 280, 378-404.de Kleer, K., de Pater, I., Molter, E., Banks, E,Davies, A., Alvarez, C., Campbell, R., et al., 2019. Io’svolcanic activity from Time-domain Adaptive OpticsObservations: 2013–2018. Astron. J., 158, 129 (14pp).https://doi.org/10.3847/1538-3881/ab2380de Kleer, K., Butler, B., de Pater, I., Gurwell, M.,Moullet, A., Trumbo, S., Spencer, J., 2020. Thermalproperties of Ganymedes surface from millimeter andinfrared emission. PSJ ??, in prep.de Pater, I., H.G. Roe, J.R. Graham, D.F. Strobel,and P. Bernath, 2002. Detection of the Forbidden SO a ∆ → X Σ − Rovibronic Transition on Io at 1.7 µ m. Icarus Note , 296-301.de Pater, I., F. Marchis, B.A. Macintosh, H.G. Roe,D. Le Mignant, J.R. Graham, and A.G. Davies, 2004.Keck AO observations of Io in and out of eclipse. Icarus,169, 250-263.de Pater, I., C. Laver, F. Marchis, H.G. Roe, and B.A.Macintosh, 2007. Spatially Resolved Observations of theForbidden SO a ∆ → X Σ − Rovibronic Transition onIo during an Eclipse. Icarus, 191, 172-182.de Pater, I., Sault, R.J., Moeckel, C., Moullet, A.,Wong, M.H., Goullaud,C., DeBoer,D., Butler, B., Bjo-raker, B., ´Ad´amkovics, M., Cosentino, R., Donnelly,P.T., Fletcher, L.N., Kasaba, Y., Orton, G., Rogers, J.,Sinclair, J., Villard, E., 2019. First ALMA millimeterwavelength maps of Jupiter, with a multi-wavelengthstudy of convection.
Astron. J. , , 139 (17 pp).https://doi.org/10.3847/1538-3881/ab3643de Pater, I., de Kleer, K., ´Ad´amkovics, M., 2020. HighSpatial and Spectral Resolution Observations of the For-bidden 1.707 µ m Rovibronic SO Emissions on Io. Plan-etary Science Journal , submitted.Fanale, F.P., Banerdt, W.B., Cruikshank, D.P., 1981.IO: Could SO condensation/sublimation cause thesometimes reported post-eclipse brightening? GRL, 8,625-628.Feaga, L.M., McGrath, M., Feldman, P.D., 2009. Io’sdayside SO atmosphere. Icarus 201, 570-584.Geissler, P.E., McEwen, A.S., Keszthelyi, L., Lopes-Gautier, R., Granahan, J., Simonelli, P., 1999. Globalcolor variations on Io, Icarus, 140, 265-282.Geissler, P.E., Smyth, W.H., McEwen, A.S., et al.,2001. Morphology and time variability of Io’s visibleaurora. J. Geophys. Res., 106, 26,137-26,146. LMA Observations of Io in/out of Eclipse . Icarus 64, 375-390.Jessup, K.L., Spencer, J.R., Yelle, R., 2007. Sulfurvolcanism on Io. Icarus 192, 24-40.Johnson, T. V., Matson, D. L., Blaney, D.L., Veeder,G.J., Davies, A., 1995. Stealth plumes on Io, Geophys.Res. Lett. 22, 3293-3296.Kieffer, S.W., Lopes-Gautier, R., McEwen, A.,Smythe, W., Keszthelyi, L., Carlson, R., 2000.Prometheus: Io’s wandering plume. Science 288, 1204-1208.Kumar, S. 1985. The SO atmosphere and ionosphereof Io: Ion chemistry, atmospheric escape, and modelscorresponding to the Pioneer 10 radio occultation mea-surements. Icarus, 61, 101-123.Lellouch, E., M.J.S. Belton, I. de Pater, S. Gulkis,and T. Encrenaz, 1990, Io’s atmosphere from microwavedetection of SO , Nature, 346, 639-641.Lellouch, E., M. Belton, I. de Pater, G. Paubert, S.Gulkis, and Th. Encrenaz, 1992, The Structure, Stabil-ity, and Global Distribution of Io’s Atmosphere, Icarus,98, 271-295.Lellouch, E., 1996. Urey Prize Lecture. Io’s Atmo-sphere: Not Yet Understood. Icarus, 124, 1-21.Lellouch, E., Paubert, G., Moses, J. I., Schneider,N. M., and Strobel, D. F., 2003. Volcanically-emittedsodium chloride as a source for Io’s neutral clouds andplasma torus. Nature 421, 45-47.Lellouch, E., Strobel, D. Belton, M., Paubert, G.,Ballester, G., de Pater, I., 1994, Millimeter wave ober-vations of Io’s atmosphere: new data and new models,BAAS, 26, 1136.Lellouch, E., Paubert, G., Moses, J. I., Schneider,N. M., and Strobel, D. F. 2003. Volcanically-emitted sodium chloride as a source for Io’s neutral clouds andplasma torus. Nature, 421, 45-47.Lellouch, E., McGrath, M. A., Jessup, K. L., 2007.Io’s atmosphere, in Io after Galileo: A New View ofJupiter’s Volcanic Moon . Eds. Lopes, R.M., & Spencer,J. R. Springer. ISBN 3-540-34681-3.Lellouch, E., Ali-Dib, M., Jessup, K.-L., Smette, A.,K¨aufl, H.-U., Marchis, F., 2015. Detection and charac-terization of Io’s atmosphere from high-resolution 4- µ mspectroscopy. Icarus, 253, 99-114.Macintosh, B., D. Gavel, S.G. Gibbard, C.E. Max, I.de Pater, A. Ghez, and J. Spencer. 2003. Speckle imag-ing of volcanic hot spots on Io with the Keck telescope.Icarus, 165, 137-143.Marchis, F., D. Le Mignant, F. Chaffee, A.G. Davies,T. Fusco, R. Pranage, I. de Pater and the Keck Scienceteam, 2005. Keck AO survey of Io’s global volcanic ac-tivity between 2 and 5 µ m. Icarus, 176, 96-122.McDoniel, W.J., Goldstein, D.B., Varghese, P.L., andTrafton, L.M., 2017. The interaction of Io’s plumes andsublimation atmosphere. Icarus, 294, 81-97.McEwen, A.S., Soderblom, L.A., 1983. Two classes ofvolcanic plumes on Io. Icarus 58, 197-226.McGrath, M.A., Lellouch, E., Strobel, D.F., Feldman,P.D., Johnson, R.E., 2004. Satellite atmospheres. In Jupiter: Planet, Satellites & Magnetosphere , Eds. F.Bagenal, T. E. Dowling, and W. McKinnon, pp. 457-483. Cambridge, UK: Cambridge University Press.McGrath, M.A., Belton, M.J.S., Spencer, J.R., Sar-toretti, P., 2000. Spatially resolved spectroscopy of Io’sPele plume and SO atmosphere. Icarus 146, 476-493.Moore, C., Goldstein, D. B., Varghese, P., Trafton,L. and Stewart, B. 2009. 1-D DSMC simulation of Io’satmospheric collapse in eclipse. Icarus, 201, 585-597.Morrison, D., Cruikshank, D. P. 1973. Thermal prop-erties of the Galilean satellites. Icarus, 18, 224-236.Moses, J. I., Nash, D.B., 1991. Phase transforma-tions and the spectral reflectance of solid sulfur - Canmetastable sulfur allotropes exist on Io?, Icarus 89, 277-304.Moses, J.I., Zolotov, M.Y., Fegley, B. Jr., 2002. Pho-tochemistry of a volcanically driven atmosphere on Io:Sulfur and oxygen species from a Pele-type eruption.Icarus, 156, 76-106.Moullet, A., Lellouch, E., Moreno, R., Gurwell, M.A.,Moore, C., 2008. First disk- resolved millimeter obser-vations of Io’s surface and SO atmosphere. Astrophys.Astron. 482, 279-292. doi:10.1051/0004-6361:20078699.Moullet, A., Gurwell, M.A., Lellouch, E., Moreno, R.,2010. Simultaneous mapping of SO , SO, NaCl in Io’satmosphere with the Submillimeter Array. Icarus, 208,353-365.0Moullet, A. et al., 2013. Exploring Ios atmosphericcomposition with APEX: First measurement of 34SO2and tentative detection of KCl. Astrophys. J. 776, 32.http://dx.doi.org/10.1088/0004-637X/776/1/32, 9 pp.Moullet, A., 2015. Exploring the Solar Systemwith ALMA. 2015. Revolution in Astronomy withALMA: The third year . ASP Conference Series, Vol.499. Daisuke Iono, Ken’ichi Tatematsu, Al Wootten,Leonardo Testi, eds. Astronomical Society of the Pa-cificMoullet, A, Lellouch, E, Gurwell, M, Moreno, R,Black, J, Butler, B. 2015. AAS DPS Meeting and new upperlimits for other gases on Io. Nature 280, 755-758.doi:10.1038/280755a0.Rathbun, J.A., Spencer, J.R., Tamppari, L.K., Mar-tin, T.Z., Barnard, L., Travis, L.D., 2004. Mapping ofIo’s thermal radiation by the Galileo photopolarimeter-radiometer (PPR) instrument. Icarus, 169, 127-139.Retherford, K.D., Spencer, J.R., Stern, S.A., et al.,2007. Io’s atmospheric response to eclipse: UV auroraeobservations. Science 318, 237-241.Roesler, F.L., Moos, H.W., Oliversen, R.J., Wood-ward, R.C., Retherford, K.D., Scherb, F., McGrath,M.A., Smyth, W.H., Feldman, P.D.F., Stro bel, D.,1999. Far-ultraviolet imaging spectroscopy of Io’s at-mosphere with HST/STIS. Science 283, 353-357.Saur, J., Strobel, D.F., 2004. Relative contributionsof sublimation and volcanoes to Io’s atmosphere inferredfrom its plasma interaction during solar eclipse. Icarus,171, 411-420.Schenk, P. W., Steudel, R., 1965. New findings inthe chemistry of the lower oxides of sulfur. AngewandteChem. Intern. Ed. Engl. 4, 402-409.Sinton, W.M., Kaminsky, C., 1988. Infrared obser-vations of eclipses of Io, its thermophysical parameters,and the thermal radiation of the Loki volcano and envi-rons. Icarus, 75, 207-232.Spencer, J. R., 1987. The Surfaces of Europa,Ganymede, and Callisto: An Investigation Using Voy-ager IRIS Thermal Infrared Spectra, Ph.D. thesis, Uni-versity of Arizona. Spencer, J.R., N.M. Schneider, 1996. Io on the eve ofthe Galileo mission. Ann. Rev. Earth and Plan. Sci.,24, 125-190.Spencer, J. R., L. K. Tamppari, T. Z. Martin, andL. D. Travis, 1999. Temperatures on Europa fromGalileo PPR: Nighttime thermal anomalies, Science 284,15141516, 1999.Spencer, J.R., Lellouch, E., Richter, M.J., L´opez-Valverde, M.A., Lea Jessup, K., Greathouse, T.K.,Flaud, J.-M., 2005. Mid-infrared detection of large lon-gitudinal asymmetries in Io’s SO atmosphere. Icarus176, 283-304.Spencer, J.R., Stern, S.A., Cheng, A.F., et al., 2007.Io volcanism seen by New Horizons: A major eruptionof the Tvashtar volcano. Science, 318, 240-243.Strobel, D.F., Zhu, X., Summers, M.E., 1994. On thevertical structure of Io’s atmosphere. Icarus 111, 18-30.Summers, M. E. 1985. Theoretical studies of Io’s at-mosphere. Ph.D. thesis, California Institute of Technol-ogy, Pasadena.Tsang, C.C.C., Spencer, J.R., Jessup, K.L., 2015.Non-detection of post-eclipse changes in Io’s Jupiter-facing atmosphere: Evidence for volcanic support?Icarus, 248, 243-253.Tsang, C.C.C., Spencer, J.R., Lellouch, E., Lopes-Valverde, M.A., Richter, J.J., 2016. The collapse ofIo’s primary atmosphere in Jupiter eclipse. JGR 121,1400-1410.Wagman, D. D. 1979. Sublimation Pressure and En-thalpy of S0 . Chem. Thermodynamics Data Center,Nat. Bureau of Standards, Washington, DC.Walker, A.C., Gratiy, S.L., Goldstein, D.B., Moore,C.H., Varghese, P.L., Trafton, L.M., Levin, D.A., Stew-art, B.D., 2010. A comprehensive numerical simulationof Io’s sublimation driven atmosphere. Icarus 207, 409-432.Walker, A., Moore, C., Goldstein, D., Varghese, P.,Trafton, L. 2012. A parametric study of Io’s thermo-physical surface parameters and subsequent numericalatmospheric simulations based on the best fit parame-ters. Icarus, 220, 225-253.Zhang, J., Goldstein, D. B., Varghese, P. L.,Gimelshein, N. E., Gimelshein, S. F., and Levin, D. A.2003. Simulation of gas dynamics and radiation in vol-canic plumes of Io. Icarus, 163, 182-187.Zhang, J., Goldstein, D., Varghese, P., Trafton,L., Moore, C., Miki, K., 2004. Numerical modelingof Ionian volcanic plumes with entrained particulates.Icarus, 172 , 479-502. LMA Observations of Io in/out of Eclipse Table 1: ALMA data: Species and frequencies
Species Frequency wavelength Line strength E (low) Spectral window Bandwidth Channel width(GHz) mm (cm − /mol/cm ) (cm − ) spw total (MHz) (kHz)Continuum 334.100 0.897 0 2000 15625SO Table 2: Time table of observations in 2018
Date Start time End time Sub-long. Sub-lat. Scans Diameter Io Array conf. HPBW HPBW Commentsmonth-day hr:m:s hr:m:s deg (W) deg combined arcsec arcsec km03-20 10:02:29 10:21:41 337.2 -3.40 7,11,15 in set 1 1.058 C43-4 0.35 1205 sunlight03-20 10:54:43 11:01:18 343.7 -3.40 7 in set 2 1.058 C43-4 0.35 1205 eclipse03-20 10:54:43 10:57:40 343.4 -3.40 7a in set 2 1.058 C43-4 0.35 1205 eclipse03-20 10:57:40 11:01:18 343.8 -3.40 7b in set 2 1.058 C43-4 0.35 1205 eclipse03-20 11:03:19 11:13:54 345.1 -3.40 11,15 in set 2 1.058 C43-4 0.35 1205 eclipse09-02 21:46:21 21:53:00 19.5 -2.96 6 in set 1 0.885 C43-3 0.30 1235 (partial) eclipse09-02 21:46:21 21:49:40 19.2 -2.96 6a in set 1 0.885 C43-3 0.30 1235 eclipse09-02 21:49:40 21:53:00 19.7 -2.96 6b in set 1 0.885 C43-3 0.30 1235 partial eclipse09-02 21:54:01 22:00:36 20.5 -2.96 8 in set 1 0.885 C43-3 0.30 1235 sunlight09-02 21:54:01 21:57:20 20.3 -2.96 8a in set 1 0.885 C43-3 0.30 1235 sunlight09-02 21:57:20 22:00:36 20.7 -2.96 8b in set 1 0.885 C43-3 0.30 1235 sunlight09-02 22:01:22 22:04:28 21.3 -2.96 10,12 in set 1 0.885 C43-3 0.30 1235 sunlight09-02 22:22:00 22:40:08 25.3 -2.96 6,8,10,12 in set 2 0.885 C43-3 0.30 1235 sunlight09-11 17:36:12 17:54:03 14.8 -2.95 6,8,10,12 in set 1 0.867 C43-5 0.22 924 eclipse09-11 18:24:01 18:41:56 21.5 -2.95 6,8,10,12 in set 2 0.867 C43-5 0.22 924 sunlightIo’s diameter is 3642 km.Sub-long, sub-lat are Observers’ sub-longitude and sub-latitude.On each day, 2 sets of data were taken, typically one when Io was in eclipse and onewhen it was in sunlight.Sans 6, 7, 8 and 11 are 6–7 min long; scans 10, 12 and 15 are typically 1-2 min long.March 20: partial eclipse started at 10:46:40; full eclipse started at 10:50:22.Sep. 02: partial eclipse started 21:49:45, and ended 21:53:29.Sep. 11: partial eclipse started 18:13:52, and ended: 18:17:35. There was no differencebetween sunlight scans 6, 8, 10, and 12; nor between first and last half of scan 6. Table 3: Analysis of SO and SO line profiles from disk-integrated spectra Date species fr fr t N t T t v r,t comments% % × cm − K m/s03-20 SO ± ± ± ± +50 − +20 ± ± ± ± ± +50 − +20 ± ± ± ± ± ± ± ± ± ± ±
25 0 sunlight, set 209-02 SO ± ± ± +2 − ±
50 0 eclipse, scan 609-02 SO 10 ± ± ± − ±
10 sunlight, set 209-02 SO 1 ± ± ± . . − ±
10 eclipse, scan 609-11 SO ± ± ± ±
25 0 sunlight09-11 SO ± ± ± ±
50 0 eclipse09-11 SO 8 ± ± ± +0 . − . ± : Fractional coverage fr map as determined from the spw2 maps for SO , and spw1maps for SO. : Fractional coverage fr t , Column density N t , temperature T t , and radial velocity v r,t for the global (disk-integrated) atmosphere. : We set the temperature equal to that determined from the SO line profiles. Table 4: Analysis of SO line profiles for individual volcanoes Date Volcano fr v N v T v v r,v comments% × cm − K m/s03-20 Karei P. 48 ± ± ±
40 +60 ± ± ± ±
50 +60 ±
10 eclipse, scan 11+1503-20 Daedalus P. 46 ± ± ± +50 − − ± ±
10 1.5 ± ± +50 − − ±
20 eclipse, scan 11+1509-02 P207 61 ± ± ±
25 0 sunlight, set 209-02 P207 25 ±
10 2 ± ±
50 0 eclipse, scan 609-02 Nyambe P. 55 ±
10 1 ± ±
50 0 sunlight, set 209-02 Nyambe P. 20 ± ± ±
50 0 eclipse, scan 6 : Column density N v , temperature T v , fractional coverage fr v , and radial velocity v r,v for individual volcanoes. However, note that the models are hydrostatic models,i.e., not particularly well suited for active volcanoes. LMA Observations of Io in/out of Eclipse Offset Right Ascension (arcsec) O ff s e t D e c l i n a t i o n ( a r c s e c ) Marchmoving into eclipseSeptembermoving out of eclipse
Eclipse geometry during March and September 2018
Io Io
Figure 1.
The geometries of Io moving into eclipse (March 2018) and coming out of eclipse (September 2018). (Adapted fromthe Planetary Ring Node: http://pds-rings.seti.org/tools/). m J y / p i x e l Figure 2.
This .model map shows the sum of all CLEAN components per pixel as obtained from CASA’s tCLEAN routinewhen deconvolving the original Io-in-sunlight map at 346.652 GHz. After convolution with the HPBW, and restoration to theresidual map, this particular .model map results in the map displayed in the top left panel of Fig. 4.
LMA Observations of Io in/out of Eclipse x xMarch 20, sunlight September 11, eclipsea) D b)c) Figure 3.
Continuum image of Io at 334.1 GHz taken on 20 March 2018 while Io was in sunlight (panel a), and on 11 Septemberwhile Io was in eclipse (panel b). Io North is up in these images. The white circle shows the approximate size of Io’s disk.The X indicates the approximate sub-solar location, and the approximate beam size is indicated in the lower left corner. Thetemperature scale is from 0 to ∼
90 K, but not quite linear to bring out the slight asymmetry in the emission. c) Simplethermal conduction model at mid-latitudes that can explain the differences in brightness temperature between the infrared andmillimeter data when entering an eclipse. (see text for details). SO2 at 346.652 GHz
Io in sunlight Io entering eclipse (sc 7a) Io in eclipse (sc 11+15) mJy/bm K K K SO at ~345.5 GHz mJy/bm K Io in sunlight Io entering eclipse (sc 7a) Io in eclipse (sc 11+15)
K K
20 March 2018: Io going into eclipse
D DLD DD D
Figure 4.
Top row: Maps of the spw2 data of the SO distribution on Io-in-sunlight, and ∼ ∼
15 (scans 11+15)min after entering eclipse. Bottom row: maps of the averaged spw1 & spw3 SO data taken at the same times as the SO maps.All maps were averaged over 0.4km/s ( ∼ LMA Observations of Io in/out of Eclipse -40 -20 0 20Time in min. after eclipse ingress
20 March 2018 F l u x d e n s i t y ( J y ) -20 0 20 40Time in min. after eclipse egress
02 September 2018 P a r t i a l e c l i p s e P a r t i a l e c l i p s e F l u x d e n s i t y ( J y ) S e p t e m b e r a) b) S e p t e m b e r Figure 5.
Flux densities integrated over individual maps (as in Figs. 4, 6, and 7) as a function of time (filled circles for March20 and September 2 ; crosses (x) for Sep. 11). The colors refer to different spectral windows. The data for SO were averagedover spw1 and spw3 to increase the SNR. The dotted lines superposed on the data in panel a show the exponential decrease(equ. 1) or the linear slope (equ. 2) after entering eclipse, whichever is appropriate. In panel b the dotted lines show the linearincrease after emerging from eclipse on September 2. All data are normalized to a geocentric distance of 5.044 AU. SO2 at 346.652 GHz
Io in eclipse (sc 6a) ~2 min into sunlight (sc 8a) Io in sunlight (Set 2) mJy/bm partial eclipse (sc 6b)~5 min into sunlight (sc 8b) ~9 min into sunlight (sc 12)
Pf NPf Pf PfPf PfN N NNN
Io in eclipse (sc 6) mJy/bm ~4 min into sunlight (sc 8)
NN PfPf
SO at ~345.5 GHz
Pf N
Io in sunlight (Set 2)
PP P PP P PPP
Figure 6.
Top 2 rows: Maps of the spw2 data of the SO distribution on Io in eclipse (scan 6a), and emerging into sunlight on2 Sep. 2018, starting with a partial eclipse (scan 6b), as indicated. Bottom row: maps of the averaged spw1 & spw3 SO data.All maps were averaged over 0.4km/s ( ∼ LMA Observations of Io in/out of Eclipse mJy/bm Io in sunlightIo in eclipse
SO2 at 346.652 GHz
NK RRK N
Io in sunlight mJy/bm
Io in eclipse
NK R
SO at ~345.5 GHz
11 September 2018: Io coming out of eclipse
NK RP1 P1P1P1 GPPGGP G P EE EE
Figure 7.
Top row: Maps of the spw2 data of the SO distribution on Io in eclipse and in sunlight on 11 Sep. 2018. Bottomrow: Maps of the SO distribution on Io in sunlight, and in eclipse on 11 Sep. 2018. The maps from spw1 and spw3 wereaveraged to increase the SNR. Io North is up in these frames. All maps were averaged over 0.4km/s ( ∼ DaU f
20 March 2018: KCl at 344.82 GHz mJy/bm
Figure 8.
Map of the spatial distribution of KCl on 20 March 2018. The map was averaged over 0.4km/s ( ∼ LMA Observations of Io in/out of Eclipse
20 March 2018
SunlightEclipse mJy/bm
B RRB F l u x d e n s i t y ( m J y ) F l u x d e n s i t y ( m J y ) K L D
Figure 9.
Individual frames at a few different frequencies (or velocities) from our March sunlight → eclipse data for thecombined SO spw1 & spw2 data. All scans 7–15 were averaged for the in-eclipse (Set 2; Table 2) and separately for thein-sunlight data. Each frame is averaged over 0.142 km/s or ∼ SunlightEclipse mJy/bm
BB RR Offset Frequency (MHz)3210 F l u x d e n s i t y ( m J y ) Offset Frequency (MHz) Offset Frequency (MHz) Offset Frequency (MHz) Offset Frequency (MHz)Offset Frequency (MHz) Offset Frequency (MHz) Offset Frequency (MHz) F l u x d e n s i t y ( m J y ) PfP N
111 22 33 44 5 5Velocity (km/s) Velocity (km/s) Velocity (km/s) Velocity (km/s) Velocity (km/s)Velocity (km/s) Velocity (km/s) Velocity (km/s) Velocity (km/s) Velocity (km/s)0 0.5-0.5 1-1 1.5-1.5 0 0.5-0.5 1-1 1.5-1.5 0 0.5-0.5 1-1 1.5-1.5 0 0.5-0.5 1-1 1.5-1.5 0 0.5-0.5 1-1 1.5-1.5 0 0.5-0.5 1-1 1.5-1.5 0 0.5-0.5 1-1 1.5-1.5 0 0.5-0.5 1-1 1.5-1.50 0.5-0.5 1-1 1.5-1.5 0 0.5-0.5 1-1 1.5-1.5 2 1 0 -1 -2 2 1 0 -1 -2 2 1 0 -1 -2 2 1 0 -1 -2 2 1 0 -1 -2 2 1 0 -1 -2 2 1 0 -1 -2 2 1 0 -1 -2 2 1 0 -1 -2 2 1 0 -1 -2
Figure 10.
Individual frames at a few different frequencies (or velocities) from our 2 September eclipse → sunlight data for thecombined SO spw1 & spw2 data. For the eclipse data we show results for scan 6 only (6a+6b); the sunlight scans are for set 2(see Table 2). Each frame is averaged over 0.142 km/s or ∼ LMA Observations of Io in/out of Eclipse
11 September 2018
SunlightEclipse mJy/bm
B RRBOffset Frequency (MHz) Offset Frequency (MHz) Offset Frequency (MHz) Offset Frequency (MHz)Offset Frequency (MHz) Offset Frequency (MHz) Offset Frequency (MHz) 2 1 0 -1 -2Offset Frequency (MHz) F l u x d e n s i t y ( m J y ) F l u x d e n s i t y ( m J y ) Figure 11.
Individual frames at a few different frequencies (or velocities) from our 11 September eclipse → sunlight data forthe combined SO spw1 and spw2 data. Each frame is averaged over 0.142 km/s or ∼ Contribution functionContribution function A l t i t u d e ( k m ) A l t i t u d e ( k m ) Temperature (K)a) b)c) d)
Figure 12.
Sample disk-averaged contribution functions for the four SO transitions detected in our ALMA data, based uponour uniform, hydrostatic model atmosphere. a) Contribution functions for a column density ( N t = 1 . × cm − ) andisothermal temperature ( T = 270 K) that match most of our data (Sections 4.3, 4.4). b) A temperature profiles as indicated bythe dashed line. This profile is inspired by profiles affected by plasma heating from above, such as shown by Gratiy et al. (2010).Resulting line profiles do not match our data. c) Contribution functions from panel a for a much colder isothermal atmosphere( T = 170 K). Resulting line profiles do not match our data. d) Contribution functions from panel a for a much higher columnabundance ( N t = 10 cm − ), such as expected on the anti-jovian hemisphere. Resulting line profiles do not match our data. LMA Observations of Io in/out of Eclipse N=1.35E16
T=220KT=270KT=320K
T=270
N=1E16N=1.35E16N=2E16 a) b)c) d) T=270 K
N=7E15N=1.2E16N=2E16 e) Karei Patera
T=220 K T=270 K
N=2E16, f=0.41N=1.5E16 N=2E16 g) f ) h) SO : March 20
Offset frequency (MHz) Offset frequency (MHz) F l u x d e n s i t y ( J y ) F l u x d e n s i t y ( J y ) F l u x d e n s i t y ( J y ) F l u x d e n s i t y ( J y ) T=270 K
N=1E16N=1.35E16N=2E16
T=270 K
T=170 K T=270 K
N=2E16, f=0.22N=1.5E16, f=0.29 N=2E16
T=270 K T=170 KT=145 K
N=5.5E16 N=1.35E16N=1E16
Daedalus Patera
Sunlight Eclipse
Total flux density f=0.26f=0.32f=0.39 f=0.20f=0.17f=0.13 K N=1E16N=1.35E16N=2E16 f=0.20f=0.17f=0.13f=0.34f=0.32f=0.33f=0.68f=0.48f=0.36 f=0.15f=0.36f=0.46 f=0.60f=0.43f=0.19 Figure 13. SO line profiles (in black) with superposed various models. The red lines show the best fit models. All panels showdata and models at 346.652 GHz (spw2), except for panel d. a) Disk-integrated flux density for Io-in-sunlight, with superposedthe best fit ( N t = 1 . × cm − ) model at the best fit temperature T t = 270 K, and a fractional coverage fr t = 0 .
32 (inred). Several models are shown to provide a sense on the accuracy of the numbers; the fractional coverage for these modelsis indicated on the right side of the line profile. b) Disk-integrated flux density for Io-in-eclipse, with superposed the best fit( N t = 1 . × cm − ), T t = 270 K, fr t = 0 .
17 (in red). c) Same as panel a to show the sensitivity on the temperature. d)Same as panel b, but at 332.505 GHz (spw6). e–h): data for Karei and Deadalus Paterae, integrated over 1 beam diameter (inblack). Various hydrostatic models are superposed, as indicated. T=270 K
N=1E16N=1.5E16N=2E16
Total flux density
Sunlight
T=220 K
N=1E16N=1.2E16N=1.5E16
T=220 K T=270 K
N=1E16N=7E15 N=1.2E16
Nyambe PateraP207 Patera a) b)c) d)e) f )g) h)
T=270 K
Total flux density
SO : September 11SO : September 02
Offset frequency (MHz) Offset frequency (MHz) F l u x d e n s i t y ( J y ) F l u x d e n s i t y ( J y ) F l u x d e n s i t y ( J y ) F l u x d e n s i t y ( J y ) T=270 K
N=1.5E16N=2.5E16N=3E16
Eclipse
T=220 K T=270 K
N=2E16, f=0.26N=1.5E16 N=3E16
T=270 K T=320 K
N=1E16N=7E15 N=1E16
T=270 K f=0.49f=0.38f=0.33 f=0.17f=0.13f=0.12N=1E16N=1.5E16N=2E16 N=1E16N=1.5E16N=2E16f=0.46f=0.35f=0.30 f=0.16f=0.12f=0.11f=0.68f=0.61f=0.51 f=0.18f=0.29 f=0.21f=0.21f=0.25f=0.53f=0.62 f=0.45 Figure 14. SO line profiles (in black) with superposed various models. The red lines show the best fit models. All dataand models are at 346.652 GHz (spw2). The temperature ( T ), column density ( N ), and fractional coverage ( fr ) are indicatedfor each model. Panels a–f) are for 2 September, g–h) for 11 September. a) Disk-integrated flux density for Io-in-sunlight. b)Disk-integrated flux density for Io-in-eclipse. c– d) Line profiles for P207 Patera in-sunlight and in-eclipse. e– f) Line profilesfor Nyambe Patera in-sunlight and in-eclipse. g–h) Line profiles for the total flux density for 11 September in-sunlight andin-eclipse, respectively. LMA Observations of Io in/out of Eclipse T=270 K
N=1E15N=2.7E15N=7E15
T=270 K
SO: 20 March 2018 b)c) d)
SO: 2 September 2018 e) SO: 11 September 2018 f )
Eclipse
T=270 K
N=6E14N=1E15N=4.5E15
T=270 K
N=1E15N=1.7E15N=2.7E15
T=270 K
N=1E15N=1.7E15N=2.7E15
T=270 K
N=6E14N=1E15N=1.7E15N=1E15N=1.7E15N=2.7E15 f=0.11f=0.08f=0.06 f=0.07f=0.05f=0.04f=0.09f=0.07f=0.05f=0.19f=0.10f=0.08f=0.21f=0.14f=0.09 f=0.11f=0.08f=0.04 a) Sunlight
Figure 15.
SO line profiles (in black) with superposed various hydrostatic models. The red lines show the best fits; all modelsshown are at 346.528 GHz (spw1). The temperature ( T ), column density ( N ), and fractional coverage ( frfr