JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 104, NO. E5, PAGES 11,817-11,826,MAY 25, .1999
Io's patchy SO2 atmosphere as measured
by the Galileo ultraviolet spectrometer
A. R. Hendrix, C. A. Barth, and C. W. Hord
Laboratoryfor Atmosphericand SpacePhysics,Universityof Colorado,Boulder
Abstract. The Galileo ultravioletspectrometer
hasobservedJupiter'svolcanicallyactivemoon
Io inthe2100-3200
• wavelength
range
at 13.7• resolution.
Wefindthatbothsulfurdioxide
froston the surfaceandsulfurdioxidegasin Io's atmosphere
are detectable.Io's SO2atmosphere
is not hemisphericallyhomogeneous,
but is patchy,with both thick and thin regions. During an
observationnear 120-150øWlongitude,we find that a thick SO2atmosphere(columndensity
N=l.0x10•9cm-2)covers---35%of theobserved
region,whilea thinnerSO2(N=4.0x1017
cm-2)
atmosphere
covers25% of the observedregion. The thick atmospheric
componentis significant
inthe2350-2400/it
range,
whilethethinner
component
isimportant
particularly
atwavelengths
shorter
than--2300•. Likelysources
oftheSO2atmosphere
arevolcanic
plumes
andsublimation
of SO,_frost, as well as outgassingassociatedwith the severalhotspotsin the field of view.
1.
Introduction
et al., 1979]. The infrared interferometer spectrometer(IRIS)
The primary sourcesof Io's atmosphereare sublimation of
surfaceSO2frost and volcanic plumes. In the first close-up
observations of Io since Voyager, we show that Io's
atmosphereconsists of both thick and thin components. In
this analysis of Io data, using spectra from the Galileo
ultraviolet spectrometer(UVS), we look at a larger wavelength
on Voyagermeasured
gaseous
SO2 at 1000-1500cm-• and
determinedthat the columndensity(N) of Io's SO2equilibrium
(ambient)atmosphere
was5.4x1018cm-2[Pearlet al., 1979].
(Lellouch et al. [1992] showedthat the inferred column density
could also
be
--3x1017-2x1018
cm-2 due to
non-local
thermodynamicequilibrium (LTE)effects.) The IUE was used
by Bertaux
andBelton[1979]in the 2400-3400• rangeto
range(2100-3200•) thanin earlierInternational
Ultraviolet determine an upper limit on Io's disk-averaged SO2
Explorer
(IUE)
observations;
and Hubble
furthermore,
Space
Telescope
we do not focus on the
(HST)
fine
structureof the SO2 gas, but on the broad shape of the SO2
crosssection. We usethe knowledgethat SO2 frost reflectance
and SO: gas cross section have unique shapes in this
wavelength range. We allow for both thick and thin SO2 gas
components and derive the best fit column densities of both.
We find that Io's ultraviolet albedo is explained not only by
SO2 frost, but also by an SO2 gas component overlying the
frostwith a columndensityN=4.0x10•7cm-2. Surrounding
the
frost region are several hotspots' associated with these
hotspots is a thicker SO_,gas component with a column
densityof N= 1.0x1019cm-:
atmosphericcolumn density of 4.2x10l* cm-2, and by
Butterworth
et al. [1980] in the 2900-3100fit wavelength
range to determine an upper limit for the thickness of Io's
atmosphere
at 2.lx1017cm-2. However,it wasdetermined
by
Belton [1982] that the Butterworth et al. [1980] measurements
did not have the resolution necessaryto adequatelyidentify
gaseous
SO2in the 2900-3100• wavelength
range,where
Beer's law may be inadequateto describethe transmission of
SO2 gas. Howell et al. [1989] determinedan upperlimit on the
SO, abundance
at 2.7x1018cm-2 using ground-based
spectra
near 4 gin.
More recently,Ballesteret al. [1990] usedIUE to make high
spectral
resolution
observations
of Io near3000 •.
They
found that, for both Io's leading and trailing hemispheres,the
upper limit on the average SO2 column density was
2.
2.0x1017cm-2for a homogeneous
atmosphere.Lellouchet al.
Background
Io's SO: atmosphere has been measured by many
instruments in several wavelength regions; differing results
have been obtained. Here we summarize these past results;
later we will compare the UVS resultsto thosepresentedhere.
The arrival of Voyager at Io in 1979 brought the discovery
of the existence of active volcanoes and the presence of
gaseous SO2.
Ground-based observations had already
confirmedthe presenceof SO2 frost on the surface[Cruikshank
et al., 1978; Pollack et al., 1978; Fanale et al., 1979; Smythe
[1992] observedboth the leading and trailing hemispheresof
Io usingmillimeter wavelengths. They found a typical column
densityof 6.0x1017cm-2 covering5-20% of Io's surface.
Ballester et al. [1994] usedthe HST Faint Object Spectrograph
(FOS) to make high spectral resolution observations of Io's
trailinghemisphere
in the2000-2300• range. Theyfound
that hemisphericatmospheres
have an averagecolumn density
of 6-10x10 Is cm-:, while better fits were obtained with more
confined atmospheres(--.8%of disk centeredon the subsolar
point) of N---Bx10
•7cm-2. rrafton et al. [1996] observed
Io
with the HST GoddardHigh Resolution Spectrograph (GHRS)
Copyright1999by the AmericanGeophysicalUnion.
to obtainveryhighresolution
spectra
in the2095-2135fit
range. They found that, for a homogeneous atmosphere, the
Paper number 1999JE900009.
trailinghemisphere
hada columndensityof N=6.7x10Iscm-2
0148-0227/99/1999JE900009509.00
while the column densityfor the leading hemispherewas found
11,817
11,818
HENDRIX ET AL.: IO'S PATCHY ATMOSPHERE
to be N=4.9x10t5cm'2. More confinedatmospheric
regions sublimationof SO2 frost, while millimeter wavelengthdata
(covering -35% of the disk in the center of the disk) had
detect
a thickerlocalizedcomponent
(N-5x10•6-5x10
•?cm-2),
column densities of
dueto volcanicplumes. However, it was acknowledged
that
the modeling of the millimeter wavelengthdata was still
2.1x10 •7 cm-2.
Denser localized
atmospheres
were measuredusingthe HST Faint Object Camera
(FOC)at2850fi,[Sartoretti
et al., 1994]andat 2325fi,,2600 uncertain, and Lellouch [1996] concludedthat
• and28503, [Sartoretti
etal., 1996]. Theymeasured
column Io's atmosphereis still somewhatunclear.
the nature of
densities
of-lx10 •8cm-2in patchescovering11-15%of the
observedregions, along with SO2frost on the surfaceof Io.
The results of these observations
are summarized
3.
Observations
in Table 1.
The Galileo UVS was built at the University of Colorado's
Lellouch [1996], in an excellent review paper, comparedthe
UV (near2100 ,•)and millimeterwavelength
resultsand Laboratory for Atmospheric and Space Physics and is
suggestedthat the UV wavelengths detect a thin localized
component (N-5x10•5-5x10•6 cm-2), associated with
describedby Hord et al. [1992]. It consists of a Cassegrain
telescopeusedto collect light and an Ebert-Fastiescanning
Table 1. Summary of Previous Io AtmosphereResults
Reference
Instrument
Wavelength
Range
Longitude
Derived
Notes
Column
Density,
cm -2
Pearl et al.
Voyager
[ 1979]
IRIS
Bertaux and
Belton
-300øW
7
IUE
2400-3400 ]k
IUE
2900-3100
]k
5.4x10•8
overLokiplume
<4.2x10•7
disk-average
<2.4x10•7
disk-average
[19791
Butterworth
et al.
80-90øW
[19801
Howell et
al. [ 1989]
Ballesteret
ground-
70øW
4
<2.7x10 •8
based
IUE
2900-3150 ]k
114øW,305øW
2.0x10•7
hemispheric
LH
- 1.6x10•7
covering~5-20%
TH
- 1.2x10•8
covering~5-20%
al. [ 1990]
Lellouchet
al. [ 1992]
ground-
mm-wave
based
of disk
of disk
Clarke et al.
HST
[ 19941
GHRS
Ballesteret
HST FOS
2300-3300
330øW
2x1019
2x1016
2000-2300
278øW, 292øW
al. [ 1994]
Sartorettiet
<4x10•
6-10x10•s
3x10 •7
HST FOC
al. [1994
]
2850
~4500
TH (270ø-
hemispheric
center of disk
-lx10 •8
covering13%of
-lx10 •8
covering0% of
284øW)
LH (67øW)
disk-average
over 10%
over 90%
disk
disk
Trafion
et
al. (1996)
HST
2095-2135
]k
GHRS
80ø-97øW
171ø-285øW
4.9x10is
6.7x10•s
LH hemispheric
TH hemispheric
2.1x10•7
covering35% of
disk, LH and TH
Sartoretti et
al.[19961
HST FOC
2325
2600
28•0]k '
LH (68øW133øW)
TH (233øW270øW)
~lx10•8
covering
-11-15%
- lx10 •8
covering~ 11-15%
of disk
LH, leadinghemisphere,
centeredon 90øW longitude;TH, trailinghemisphere,
centeredon 270øW longitude
of disk
HENDRIX ET AL.' IO'S PATCHY ATMOSPHERE
11,819
spectrometer. A diffraction grating scans the ultraviolet
spectrum,and three photomultiplier tubes record the signal.
The datareportedherewere acquiredusingthe F channel,which
roughness parameter O is not expected to vary with
wavelength,so we usedthe value determinedby McEwen et al.
[1988], O=30 ø. We also tried varying the value of the
coversthe 1620- 3231 • wavelength
regionwith a spectral opposition effect term B0, but since our smallest phase angle
resolution
of 13.7•; thesampling
is performed
every3.1 •. is 13ø, we found we could not constrainthat parameter and used
The calibration of the F channel is described by Hendrix B0=0.
To determine the best phase curve parameters to apply to
[1996]. The entire spectrumis scannedin 4.33 s, with a
0.006 s integration period for each of the 528 grating the UVS Io spectrum, we used the double-lobed Henyeypositions. The instantaneous
angularsize of the slit is 0.1ø x Greenstein phase function of (2) in the Hapke model and
0.4 ø
comparedthe resultant Hapke model to the UVS-measuredIo
The observation describedin this paper was performed on reflectancesat variousphase angles to find the best fit values
May 31, 1998. During this observation, the UVS F channel of b and c.
field of view (FOV) was held steady on Io's surface centered
(1- c)(1- b2)
c(1- b2)
near 130øW longitude. The centralportion of the FOV covered
p(c•)=
3/2+
3/2
(2)
(1- 2bcosa
+ b2)
(1+ 2bcosa
+ b2)
part of BosphorusRegio, which, as shownby Lopes-Gautier et
al. [1997], is a bright region (likely SO_,frost)surrounded
by
hot spots. The phase angle during the observation (averaged To derive the ultraviolet phase curve, we used 14 UVS Io
by Hendrixet al. [1996,
over the FOV) was 51ø, and the subsolarpoint was near 90øW. observations,previouslydescribed
1997], which coveredthe 13ø -113ø phaseangle range. The
The emission and incidence angles, averaged over the FOV,
were 43.8ø and 40.5ø, respectively. As shown in Plate 1, the observationswere taken at several different longitudes on Io's
length of the FOV was approximately one Io diameter. The leadinghemisphere
(longitudes0-180øW). By comparingthe
observation lasted 56 min, during which 784 spectra were Hapkemodel(with differentcombinationsof b and c values)
summed. Figure la displays the measuredreflectance. In with the measured reflectances, we found the best fit values of
Figure la (and other figures) the UVS-measuredspectrum is b and c to be -0.50 and 0.85, respectively.The valuesof b and
shownin 15 • bins.
c did not vary appreciablywith wavelengthanddid not affect
the derived albedos outside of the statistical
4.
Photometric
error bars.
We
Corrections
also attemptedto use the single-lobed Henyey-Greenstein
phase
functionbutfoundthatit inadequately
fit the high-phase
To convert from a measuredreflectance to an albedo, we
angleobservation(c•=113ø);Io is moreforwardscatteringat
correctedfor observational geometry using the Hapke
high phase angles than the single-lobedphase function
photometric
function[Hapke, 1993]:
predicted,likely due to volcanicparticles.
In Figurelb we showthe Hapkemodelplottedalong with
rm(f,
f0,0t,
X)-- 0)(•) ]'tO[(l+B(c•,X))p(00]R(•
,•) (1) UVS-measuredreflectances of
•+•0
Although we are investigating the amountof SO,.gas in Io's
atmosphere, and Hapke's model applies to surfaces,we use
this
function
because we detect both
Io's leading hemisphere at
various
phase
angles.Thedatain Figure
lb arefor2900•,
the
surface
and the
andwe find that the photometricparametersmaking up this
Hapkemodel(with the exceptionof the single-scatteralbedo)
arereasonable
forall wavelengths
in the2100-3200
• region.
To comparethe datawith atmospheric
models,we converted
atmosphericcomponents. In (1), rm(f, f0, 0•, X) is the from the derived single-scatteralbedo to to the geometric
measured
reflectance,to(K)is the single-scatteralbedo, g and albedousing(3) from Hapke [1993, p. 353]:
go are the cosines of the emission and incidence angles,
respectively, B(c•, X) accountsfor the opposition surge,
R(©, c•) correctsfor surfaceroughness,and p(c•) is the phase
8
2
3
correction. (All terms are discussedby Hapke [1993].) We
used(1) to solvefor the single-scatteralbedo as it varies with
ga(X)
=--{[l+B0]p(0)-I
}+U(
©) rø(1+rø)
(3)
wavelength.
In (3), ga(X) is the geometricalbedo,U is an empirical term
The measuredreflectancerm(f, fro, c•, X) is determinedby
accounting for macroscopic roughness (given by Hapke
calibratingthe measuredspectraby dividing by the calibration
[1993, p. 353]) and r0 is the diffusive reflectance [Hapke,
curvefor the F channel[Hendrix, 1996] and removing reflected
1993, p. 196]. Using the photometric correctionsof (1)-(3),
solar featuresby dividing by the solar spectrum. We useda
we obtain a geometric albedo comparable to the HST
solar spectrummeasuredby the UARS Solar Stellar Irradiance
ComparisonExperiment (SOLSTICE)instrument [Rottman et measurementof Clarke et al. [1994]. The derived geometric
al., 1993]. Thesolarflux in the2000-3200• regionvaries albedo is shown in Figure l c. The geometric albedo is
by less than 1.5% over the solar cycle, so our results are not providedin tabularform in Table 2.
significantly affected by using a solar spectrummeasuredon a
different day than the UVS Io observation.
The solar
spectrum,originally at 0.1 nm resolution, was double-boxcar
smoothedto match the resolution of the UVS and interpolated
to the UVS wavelength scale.
Io's ultraviolet photometric parameters have not been
previously determined, although visible photometric
parameters have been derived by Simonelli and Veverka
[1984, 1986] and McEwen et al. [1988].
The surface
5.
Models
In this study,we fit three differentmodelsto the derivedIo
albedo. The three models are outlined in Table 3 and described
below.
The results of the models are shown in Table 4.
We
quantitatively analyze the fit of the model to the data by
minimizing
Z2 overthe2100-3200
• wavelength
range:
11,820
HENDRIX ET AL.: IO'S PATCHY ATMOSPHERE
b
o
ß
0.15
ß
ß
i
,
ß
,
i
ß
ß
ß
i
ß
ß
ß
i
ß
ß
,
i
ß
ß
ß
i
ß
ß
ß
E 0.10
0.12
c
o
0.09
0.06
0.06
0.04
0.03
0.02
ß
200
..I,.ll...I.,.I,..I...111
220
0.00
I
240
260
280
300
320
ß
ß
i
ß
ß
ß
i
ß
ß
ß
i
ß
ß
ß
i
,
ß
ß
i
ß
ß
ß
ß
.
.
i
.
.
.
i
.
.
.
i
.
.
.
i
,
,
,
i
,
,
i
0.08
o
0.00
ß
340
0
20
wovelength(nm)
40
60
80
100
120
phase angle (degrees)
c
0.08
c)
,•
0.06
,•
0.04
Figure
•
Shown
0.00
220
240
260
280
300
320
340
wovelencjth(nm)
Z:= E
abs(model- ga)
(a) Measured reflectance of Io (calibrated
are data from several
UVS
observations
of Io on the
leading hemisphere (0-180øW). Also plotted is the Hapke
model usingderivedphotometricparameters(c0=0.11, b=-0.5,
c=-0.85, B0=0, 0=30ø). The observation discussedin this
paper is at 51ø phase angle. (c) Geometric albedo of Io
determined by applying Hapke photometric correction to
...........................
200
1.
data/solarspectrum)
at 13.7 ,• resolution;(b) several
observations
demonstrating
Io's phasecurveat 2900 ,•.
measured
(4)
ga
reflectance.
narrowto be accountedfor by the SO2 frost, which has a much
broaderlocal maximumin reflectance(Figure 2a). In contrast,
the feature
is consistent
with
the narrow
local
maximum
in
transmission
of a thick SO: column(Figure 2c).
5.1
Model Components
In the modelsdescribedhere, we useda medium-grain(~ 100Figure 2 displaysthe important componentsof our models, 200 [tm)SO,_ frost reflectance (Figure 2a). The SO,_frost
SO: frostreflectanceand SO: gas cross section. The frost and reflectancewas from Wagneret al. [1987], scaledto agreewith
gas are spectrallydistinct in this wavelengthrange. As shown the medium-grain SO: frost data from Nash et al [1980].
in Figure 2a, SO: frost increases sharply in reflectance Carlson et al. [1997] found that the size of the SO: frost grains
between3000 and3200 •. Between3000 and2000 •, the in Io's equatorialregions is of the order 300 ILtm,and because
reflectance increases slightly with decreasing wavelength. most of the reflectance from Io's surface during the UVS
There is a broad local maximum
in reflectance
near 2350observationswas from the equatorialregions,this comparison
2400•. ForSO,_
gas,Figure2c shows
thatcolumn
densities shouldbe adequate. The SO2 gas cross section usedin this
near 5x1016cm-2are bestdetectedshortwardof 2300 • and are study (at 293+10K) was from Manart and Lane [1993]' it is
practically undetectableat longer wavelengths. As the column shownin Figure2b degradedto the UVS spectralresolution. It
density increases, the transmission of the gas decreases is unclearhow much the use of room-temperatureSO2 cross
overall,and the maximumin transmissionnear2350-2400 ,• sections in our Io atmospheremodel affects our results. The
becomesmore narrow. Also, as column density increases,the temperatureof Io's atmosphere is uncertain; Ballester et al.
fine band structure near 2100 ,• increases and the bands [1994] obtained temperatures between 110 and 250 K for
SO: atmospheres,
while Lellouchet al. [1992]
becomestronger.At columndensities
near5x10Iscm-2,these constrained
bands saturate out and the reflectance in that wavelength
region is black.
Comparingthe spectrain Figures 2a and 2c to the albedo in
Figure l c, we see that Io displays a sharp increase in albedo
foundtemperatures
as high as 500-600 K.
5.2
Model
1'
Surface
Reflectance
Only
Model 1 accountsonly for surfacereflectanceand ignores
between
3000and3200,&,consistent
withSO:frost.At other any gaseoussulfurdioxide. In Model 1, we fixed theamountof
wavelengths, Io's albedo is generally flat, except for a
SO: frost on the surface(X• in Table 3) at 35%. This amount
maximum
near2350-2400,•. Thisfeature
appears
to be too was basedon comparingthe amountof the FOV filled with the
HENDRIX ET AL.: IO'S PATCHY ATMOSPHERE
Table
2.
Geometric
Wavelength,
•
Table
11,821
Albedo
Albedo
Wavelength,
,•
Albedo
Wavelength,
,•
Albedo
2115.83
0.0119
2491.58
0.0203
2867.33
0.0184
2130.86
0.0156
2506.61
0.0134
2882.36
0.0151
2145.89
0.0140
2521.64
0.0185
2897.39
0.0156
2160.92
0.0165
2536.67
0.0221
2912.42
0.0131
2175.95
0.0139
2551.70
0.0165
2927.45
0.0154
2190.98
0.0133
2566.73
0.0139
2942.48
0.0177
2206.01
0.0138
2581.76
0.0150
2957.51
0.0174
2221.04
0.0157
2596.79
0.0184
2972.54
0.0153
2236.07
0.0151
2611.82
0.0194
2987.57
0.0163
2251.10
0.0151
2626.85
0.0167
3002.60
0.0199
2266.13
0.0176
2641.88
0.0140
3017.63
0.0222
2281.16
0.0212
2656.91
0.0134
3032.66
0.0221
2296.19
0.0206
2671.94
0.0132
3047.69
0.0212
2311.22
0.0250
2686.97
0.0144
3062.72
0.0219
2326.25
0.0253
2702.00
0.0144
3077.75
0.0235
2341.28
0.0303
2717.03
0.0147
3092.78
0.0238
2356.31
0.0297
2732.06
0.0156
3107.81
0.0256
2371.34
0.0289
2747.09
0.0148
3122.84
0.0305
2386.37
0.0246
2762.12
0.0155
3137.87
0.0381
2401.40
0.0290
2777.15
0.0148
3152.90
0.0472
2416.43
0.0222
2792.18
0.0200
3167.93
0.0587
2431.46
0.0199
2807.21
0.0173
3182.96
0.0706
2446.49
0.0187
2822.24
0.0125
3197.99
0.0814
2461.52
0.0205
2837.27
0.0149
3213.02
0.0906
2476.55
0.0246
2852.30
0.0176
3.
Models
Model
1: Surface reflectanceonly
Description
albedo=XiRso2 + X2Rothe
,
X•= percentcoverageby SO2 frost
Rso2= reflectance of SO2 frost
X:= percentcoverageby other component(X2= 100% - Xt)
Rother-reflectanceof other component
2: Hemisphericatmosphere
albedo=(XiRso2 + X2Rothe,)S
S= transmissionof SO2 gas (see text)
3: Patchyatmosphere
albedo=XtRso2S
• + X2Rothe,
S2 + X-•Rothe
,
St= transmission
by atmospheric
componentoverlyingSO2 frost
S2=transmission
by atmospheric
componentoverlyingothersurface
material
X•= 100%-X•-X2= regionof surfacecoverageby othersurface
material and no gas= 40%
11,822
Table
HENDRIX ET AL.: IO'S ['ATCHY ATMOSPHERE
4.
Results
% SO2
Rother
Frost (Set)
Hemispheric Percent
Thick
ColumnDensity
N, cm-2
model 1
35%
0.018
model 2
35%
0.020
model 3
35%
0.032
SO,,Gas
Coverage
ThickSO2
Percent
Thin
ThinSO2
ColumnDensity
SO, Gas
ColumnDensity
N, cm-2
Cov-erage
Z2
N, cm-2
13.1
5.0e 16
10.6
25%
1.0e19
35%*
4.0e17
9.1
* overlyingSO2 frost
to SO,_frost) but doesnot simulatethe other
bright material in solid state imaging (SSI)camera images, (attributable
which covers approximately +15ø latitude. This is also in featuresin the measuredalbedo, which we find from Model 3
gasof varyingthicknesses.
Theimportance
of
agreementwith the resultsof McEwenet al. [1988] for this areduetoSO,_
longituderegion. We thensolvedfor the bestfit reflectanceof this modelis to demonstratethat SO, gas is indeeddetectable
another surfacematerial, covering the regions void of SO2 in this wavelength range.
frost, which we assumedhad no variation in reflectance with
wavelength (R,,,he..in
Table 3).
The resultof this modelis shownin Figure 3a. The best fit
value of Rothe
,. was 0.018. Figure 3a shows that this model
accounts
for the steepincrease
in albedonear 3100 •
5.3
Caveats
Models 2 and 3 include transmission by SO2 gas, which
requiresan introductionbefore we explain those modelsand
Plate 1 Imageof the UVS Io observation
indicatingthe regionobserved.The FOV washeld in this position for 56 min; the
784 spectraacquiredduringtheobservation
wereaveragedandusedin theanalysisto determinethealbedoof the regionobserved.
The observedregionincludesSO2 frost-richBosphorus
Regio (---15øS-15øN)
and severalhotspotsin the 15øS-30øSand 15øN-30øN
regions.
HENDRIXET AL.' IO'S PATCHYATMOSPHERE
11,823
b
O
0.12
8x10
ßßß! ßßßi ßßß! ßßßi ß, , i ßß! ß' '
0.10
18
6x10
0.08
E 4x10-18
0.06
0.04
2x10
0.02
-18
_
0.00
ß
200
.
.
i
220
,
,
.
i
240
,
,
.
i
260
.
,
,
i
280
,
.
,
i
,
300
,
,
i
.
.
320
,
340
200
220
240
wavelength(nm)
260
280
300
320
340
wavelength(nm)
1.0
Figure 2. (a) Reflectanceof SO: frost [Wagner et al., 1987]
at two grain sizes. The thin line (upper) is the Wagner et al.
reflectancescaledto agree with the small grain size (<50 gm)
of Nash et al. [1980]' the thick line is the Wagner et al.
reflectancescaledto correspondwith the mediumgrain size
(-100-200 gm)of Nash et al. [1980]. (b) SO•_gas cross
section from Manart and Lane [1993]. The laboratory-
• 04
Z 02
O0
200
measured cross section
220
240
260
280
300
320
340
wavelength(nm)
results. Belton [1982] pointed out that, due to the densely
packednatureof the spectrallines in the SO•_cross section,
Beer's law may not be an appropriate approximation for the
transmissionof this gas.
Ballesteret al. [1994] provide an in-depth investigation of
methods to use to account for curve-of-growth effects due to
Beer's law, they applied a Malkmus model with a given k
distributionto determinecolumn densities of SO: measuredat
Io in high-resolution HST spectra. This method solves the
0.08
0.08
0.06
0.06
0.04
0.04
0.02
0.02
0.00
0.00
240
260
280
300
320
340
to match
thecrowded
SO:linesin the2100 .• region.Instead
of using
0.10
220
smoothed
differentcolumndensities'(top) N=5.0x10lø cm-2, (middle)
N=5.0x10•7cm-:,and (bottom)N=5.0x10•8cm-2
0.10
200
was double-boxcar
the UVS resolution. (c) SO2 gas transmission for three
200
220
240
wovelength(nm)
260
280
500
520
540
wavelength(nm)
Figure3. Ultravioletalbedoof Io asmeasured
by theUVS. Measured
albedois shownastrianglesin 15 .•
bins, while model is shown as a solid line.
Error bars are due to statistical
error in the measurement.
Comparisonwith (a) Model 1 (surfacereflectanceonly), where the best fit model included35% SO• frost and
_
65% of a fiat 0.018reflectance;
and(b) Model2 (surfacereflectance
plushemisphericSO: atmosphere),where
the bestfit modelincluded35% SO: frostplus65% flat reflectance
of 0.020 and a hemisphericSO2 atmosphere
of columndensityN=5.0x10•øcm-:.
11,824
HENDRIX ET AL.' IO'S PATCHY ATMOSPHERE
problemof not knowing the cross section exactly in regions
where the line spacing is small. Clarke et al. [1994], in
allowed the rest of the FOV to be covered by a surface
componentof reflectanceRothe
r. We allowedtwo atmospheric
analyzinga HSTGHRSspectrum
of Io in the 2300-3300
components
to exist: a "thick" component
(N>lx10•8 cm-'-)
anda "thin"component
(N<lx10•8cm-:). We trieddifferent
region effectively ignored the problem by stating that the
crosssection had not been measuredwell enough at near-UV
wavelengths,and they usedBeer's law in their approximations
of the thicknessof Io's atmosphere.
amountsof coverageby the thick and thin componentswhile
keeping X•=35%, corresponding to Bosphorous Regio
between-15øS and 15øN. We did not allow the gaseousSO'- to
We used a combination
of these two methodsß
For
extend beyond 30ø latitude; higher latitude regions were
wavelengths
between
2100 and2300 • weusedtheMalkmus assumedcoveredby the nonfrost surfacematerial and had no
gas coverage. In this way, gas was allowedto cover 60% of
model describedby Ballester et al. [1994]:
the observedregion. For instance,we tried allowing the thick
gascomponentto cover the entire 35% of the frost portion of
S= exp(- -- y[(1 + •
- 1])
(5)
the FOV, while the thin gas component covered 25% of the
2
FOV over the nonfrostmaterial, and the remaining40% had no
wherewe use y=1.85 over this wavelength range, as obtained gascoverage. We tried many combinations of thick and thin
by Ballesteret al. [1994] near2100 • for a particular
k gas component coverage, while keeping the frost coverage
distribution.
In the2300-3250
• range,
we used
Beer'slawto constantat 35% and the total gas coverageconstantat 60%.
determinethe best fit column densities. We also tried using
Figure 4a displaysour best fit model comparedto the UVSthe Malkmus model in this wavelength range; however, the y
measured
albedo.
The best
fit
model
included
a thin
value is unknown here due to a lack of high-resolution
(N=4.0x10•7cm-'-)SO:gascomponent
overlyingthe SO:frost
laboratory measurementsof the SO_,cross section in this
region, a thick (N=l.0x1019 cm-'-) SO2 gas component
range. When we simply used y=1.85 over the entire overlying material of reflectance Rothe
r = 0.032. The higher
wavelength range, we obtained the same best fit column
latitudes observed in the FOV were assumedto only contain
•r
40'N)
densites as when we used Beer's law. As will be shown, the
rangeof valuesof column densities derived in this analysis is
larger than the possible error due to using Beer's law versus a
random
5.4
band model.
Model 2:
the surfacematerial of reflectance Rother,
and no atmospheric
component. In Figure 4b we show the three individual
componentsof the model. The thick atmospheric component
is important
near2350-2400•, whereit addsa narrowlocal
maximum to the model. As first pointed out by Clarke et al.
[1994], the thick component is saturated out at most
wavelengths in this range and provides only an emission
Homogeneous SO2 Gas
In Model 2 we allowed for surfacereflectanceby SO,frost
and another component (as in Model 1), as well as
transmittance by SO: gas distributed homogeneously across
the observed region. In Model 2, we fixed the percent
coverageby SO_,frostat 35% and solvedfor the best fit values
of the reflectanceof the other surfacecomponent and the best
fit column density (N) of the SO: gas. We accountedfor the
average slant angle to the observed region by incorporating
the averagert and g0 (cosineof emissionand incidenceangles,
respectively) values over the FOV into the transmission part
of Model 2, where the two-way absorptionZ=(1/g+l/g0)N.
feature
near2400 •. Figure4c displays
the measured
albedo
and the model not including the thick gas component to
demonstrateits effect near 2400 •.
Uncertainties
in
the
derived
column
densities
were
determinedbased on the goodness of fit of the model to the
data. We found that the thick atmospheric component column
densitycould vary by a factor of 2 before the quality of the fit
was unsatisfactory' the range of best fit values is thus
N=5.0x101s-2.x
101ø cm-'
For the thin component,
satisfactoryfits to the data were achievedfor N=3.x10•vß
9.0x 1017 cm--'
Thebestfit model(N=5.0x1016cm-2)shownin Figure3b
shows that a homogeneousatmosphereimproves the fit to the
data over Model
1 but still
does not account
for the maximum
in albedo
near2350•. Addinggaseous
SO,alsomeans
thata
brighter surfacereflectance(other than SO'-frost) is required:
a fiat 0.020 reflectance fits the data best. It is not expected
that SO,-gas would be distributed homogeneously across the
observed
frost,
hotspots,
region with
and nearby
its varying
volcanic
sources
plumes.
of gasHowever,
such as SO:
we
include
the results of this model here to demonstrate
the effects
6.
Discussion
Using Model 3, which incorporates a localized thick SO:
atmosphere, SO_,frost and an additional surface component
with a flat reflectance, a good fit was obtained over the
measured
wavelength
range(2100-3200fk). In particular,
the
fit wasimproved
in the2250-2400• region,where
thickSO'atmosphereaccountsfor an increasein the measuredalbedo.
of including patchy regions of thick and thin atmospheric
components(Model 3).
6.1
5.5
Model 3:
Patchy Atmosphere
Model 3 allows for surfacereflectanceby SO, frost and
another component (as in Model 1), as well as transmittance
by both a localizedrelativelydenseSO: atmosphereand a thin
SO, atmosphere
component.We fixed thepercentcoverageby
SO: frost at 35% covering the central part of the FOV. We
Comparisons
With
Previous
Results
The results presented here are consistent with those
obtainedby Clarke et al. [1994] using a HST GHRS spectrum
in the same wavelength range; they found that a model
including10%coverage
by N=2x1019cm-'-and90% coverage
by N=2x10•6cm-2wasconsistent
withtheirdata. Atmospheric
thicknesses
on the orderof N-lx1018 cm-'-wereobtainedby
Sartoretti et al. [1994, 1996] using HST FOC images also in
HENDRIX ET AL.: IO'S PATCttY ATMOSI)HERE
0.10
11,825
0.06
0.08
0.04
0.06
0.04
0.02
0.02
0.00
0.00
200
220
240
260
280
300
320
340
200
220
wavelength(nm)
240
260
280
300
320
340
wavelength(nm)
C
ß ß ß i
0.10
ß ß ß i r.
,
.
!
,
,
,
i
ß ß ß i
ß ß ß i
ß ß ß
0.08
0.06
Figure 4. (a) Ultraviolet albedo of Io as measuredby the
UVS, compared with best fit Model 3. Measured albedo is
0.04
shownastrianglesin 15 ,• bins,whilemodelis shownas a
0.02
solid
line.
Error
bars
are due to
statistical
error
in
the
measurement. (b) Individual model components. Thin line:
.
0.00
,•,1,,,I,,,I,,,I,,,I,,11111
200
220
240
260
280
300
320
340
wavelength(nm)
X•Rso:S•=(35%)(Rso:)(S
• whereN=4.0x10•7cm-:);thick line:
X:Ro,he,
S:=(25%)(0.032)(S
2 whereN=l.0x1019 cm-2);dashed
line: X•Ro,he,=(40%)(0.032
). (C) Albedo compared to model
excluding thick component(thick line in Figure 4b).
thiswavelength
range(2325•, 2600 •, 2850 •). Themodel the thick component cover part of the frost and the thin
presented here that is consistent with UVS data includes 35%
coverageby an atmosphericcomponentof N=4.x10 •7 cm :.
This "thin" component agrees with the column densities
measured
near2100 .• in high-resolution
HST spectaby
Ballester et al. [1994] and Trafton et al. [1996].
These
comparisons indicate that the near-UV wavelengths
(particularly
near2400.,•)areimportant
in detecting
column
densities of N-•lx10•a-lx1019
cm-:.
Measurements near
2100• arevital in detecting
column
densities
of N-lx10 •6lx10 •7 cm-2. These results are also consistent with the
component cover the other part of the frost and part of the
nonfrost region. We tried letting the thick component cover
the nonfrost material and the thin componentcover the frost.
We found that the best fit had the thicker gas component
over the nonfrost material and the thinner gas component over
the frost. This indicates that a thicker column density of SO"gas is associatedwith nonfrost regions. Several hot spots
were included in the FOV in the nonfrost regions, and the
Amirani plume was also included in the FOV. These results
suggest that sublimation of surface frost produces a thin
suggestion by Lellouch [1996] that different wavelength (N-4x10•7 cm-2) atmospheric
component,while outgassing
ranges detect SO2 column densitiesof varying thicknesses. from hot spots and volcanic plumes producesa thicker gas
Near2100•, a "thin"component
is detectable,
likelydueto component(N--.1.0x10
•9cm-2).Thismayindicatethat"stealth
sublimationof surfacefrost.
In millimeter wavelength plumes" [Johnson et al., 1995] are associatedwith these hot
observations, thicker gas components are detected, spots, contributing to the thick SO"-gas component, where
presumablydue to volcanicplumes. We suggestthat the near- dustparticulates are not included in the plumes, so they are
UV wavelengths
detectstill a thicker SO._gas component,due undetectable at visible wavelengths.
to volcanic plumes and outgassingat hotspots (discussed
below).
7.
6.2
Implications
In our model we varied the thicknessof coverageof the
thickandthin atmosphericcomponents.We triedletting the
thickcomponent
coverthe entireSO._frost regionandhaving
the thin component
coverthe othermaterial. We triedletting
Conclusions
The conclusionsof this work may be summarizedin the
three following points.
1. Sulfurdioxidefrostalonedoesnot accountfor the shape
of Io'salbedo
in the2100-3200
• wavelength
range.
11,826
HENDRIX ET AL.' IO'S PATCItY ATMOSPHERE
2. Sulfur dioxide gas must be included in the model of Io's
near-ultraviolet albedo, and the gas must be in a thick
localized region, not hemisphericallyhomogeneous.
3. Our best fit model includesa thick atmosphericregion
covering 25% of the observedregion with a column density
N--,1.0x10
•9cm-2. Also importantin fitting the near-UVdata
is a thinneratmospheric
regionof columndensityN-4.0x10 •7
cm-2. We findthatthethinnerSO2gascomponent
is likely due
to sublimation of SO2 frost, while the thicker component is
associatedwith volcanicplumesand outgassingat hotspots.
Hendrix, A. R., The Galileo ultraviolet spectrometer: In-flight
calibration and ultraviolet albedos of the Moon, Gaspra, Ida and
Europa. Ph.D. thesis,Univ. of Colo., Boulder,1996.
Hendrix, A. R., et al., Galileo ultraviolet spectrometerobservationsof Io
(abstract),Eos Trans.AGU, 77, Fall. Meet. Suppl.,F437, 1996.
Hendrix, A. R., C. A. Barth, C. W. Hord, A. I. F. Stewart, K. E.
Simmons, W. E. McClintock, J. M. Ajello, and A. L. Lane, !o:
Surface and atmosphere observationsby the Galileo ultraviolet
spectrometer(abstract),Eos Trans.AGU, 78 (46), Fall Meet. Suppl.,
F418, 1997.
}lord, C. W., et al., Galileo ultraviolet spectrometerexperiment,Space
Sci. Rev., 60, 503-530, 1992.
Howell, R. R., D. B. Nash, T. R. Geballe, and D. P. Cruikshank,Highresolution infrared spectroscopy of Io and possible surface
materials, Icarus, 78, 27-37, 1989.
These results are consistent with previous high spectral
resolution observations near 2100 ,• which indicated column
densitiesof N-3x10 •7cm-2[Ballesteret al., 1994; Traftonet
Johnson,T. V., D. L. Matson, D. L. Blaney, and G. J. Veeder, Stealth
plumeson Io, Geophys.Res.Lett., 22, 3293-3296, 1995.
Lellouch, E., Io's atmosphere: Not yet understood,Icarus, 124, 1-21,
al., 1996]; they are also consistent with earlier near-UV
Lellouch, E., M. Belton, I. DePater, G. Paubert, S. Gulkis, and T.
observations
which
indicated
column
densities
of
N-•l.5x10 •9cm 2 in localizedregions[Clarkeet al., 1994].
Put together, these observations all indicate that different
regionsof the ultraviolet detectdifferent SO2 gas thicknesses
dueto the strong variation in wavelength that the SO2 cross
section
exhibits
for different
column
densities.
This
is in
agreement with the suggestion by Lellouch [1996] that
observations performedin varying wavelength ranges detect
different parts of Io's atmosphere.
1996.
Encrenaz, The structure, stability, and global distributionof Io's
atmosphere,Icarus, 98, 271-295, 1992.
Lopes-Gautier,R., A. G. Davies, R. Carlson, W. Smythe, L. Kamp, L.
Soderblom, F. E. Leader, R. Mehlman, and Galileo NIMS team. Hot
spotson Io: Initial results from Galileo's near infrared mapping
spectrometer,Geophys.Res. Lett.. 24, 2439-2442, 1997.
Manatt, S. L., and A. L. Lane, A compilationof SO2 absorptioncross
sectionsfrom 106-403 nm, J. Quant. Spectrosc.Radiat. Transfer,50,
267-276, 1993.
McEwen, A. S., T. V. Johnson, D. L. Matson, and L. A. Soderblom, The
globaldistribution,abundance,andstabilityof SO2 on Io, lcarua, 75,
450-478, 1988.
Nash, D. B., F. P. Fanale, and R. M. Nelson, SO, frost: UV-visible
_
Acknowledgments.
M. Beltonandan anonymous
reviewerprovided
reflectivity and Io surface coverage, Geophys. Re•s.Lett., 7, 665-
constructivecomments. Many thanksto W. K. Tobiska, K. E. Simmons,
668, 1980.
W. Sweet, and J. Gebben for assistancewith observationsand data; A.
Pearl, J. C., R. Hanel, V. Kunde, W. Maguire, D. Fox, S. Gupta, C.
McEwen for informationon SSI results;B. Knapp for the SOLSTICE
Ponnamperuma,
and F. Raulin, Identification of gaseousSO2 and
solar spectrum. A.R.H. is grateful for helpful discussionswith M.
new upperlimits for othergaseson Io, Nature, 280, 755-758, 1979.
McGrath
and G. Ballester.
Pollack, J. B., F. C. Witreborn, E. F. Erickson, D. W. Strecker, B. J.
Baldwin, and T. E. Bunch, Near-infrared spectra of the Galilean
satellites' Observationsand compositionalimplications,Icarus, 36,
271-303, 1978.
References
Ballester, G. E., D. F. Strobel, H. W. Moos, and P. D. Feldman, The
atmospheric
abundance
of SO•_on Io, lcarlts,88, 1-23, 1990.
Ballester,G. E., M. A. McGrath, D. F. Strobel, X. Zhu, P. D. Feldman,
and H. W. Moos,Detectionof the SO2 atmosphere
on Io with the
Hubble SpaceTelescope,Icarus, 111, 2-17, 1994.
Rottman,G. J., T. N. Woods, and T. P. Sparn, Solar-StellarIrradiance
Comparison Experiment, 1, Instrument design and operation, J.
Geophys.Re•., 98, 10667-10677, 1993.
Sartoretti,P., M. A. McGrath, and F. Paresce, Disk-resolved imagingof
Io with the Hubble SpaceTelescope,Icarus, 108, 272-284, 1994.
Sartoretti, P., M. J. S. Belton, and M. A. McGrath, SO, distributionson
Io, Icarus, 122, 273-287, 1996.
_
Belton,M. J. S., An interpretationof the near-ultravioletabsorption SimoneIll,D. P., andJ. Vcverka, Voyager disk-integratedphotometryof
Io, Icarus, 59, 406-425, 1984.
spectrumof SO2: Implicationsfor Venus, Io, and laboratory
measurements, Icarus, 52, 149-165, 1982.
Simonelli, D. P., and J. Veverka,
Bertaux, J. L., and M. J. S. Belton, Evidence of SO, on Io fi-om UV
_
observations, Nature, 282, 813-815, 1979.
Butterworth,P.S., J. Caldwell, V. Moore, T. Owen, A. R. Rivolo, and A.
L. Lane,An upperlimit to the globalSO2 abundanceon Io, N(llllre,
285, 308-309, 1980.
Carlson, R. W., et al., The distribution of sulfur dioxide and other
infrared absorberson the surface of Io, Geophy,•.Res. Lelt., 24,
2479-2482, 1997.
Clarke,J. T., J. Ajello, J. Luhmann,N. Schneider,and I. Kanik, Hubble
SpaceTelescope
UV spectralobservations
of Io passinginto eclipse,
J. Geophys.Res., 99, 8387-8402, 1994.
Phase curves of materials on Io'
Interpretation in terms of Hapke's function, Icarus, 68, 503-521,
1986.
Smythe, W. D., R. M. Nelson, and D. B. Nash, Spectral evidence for
SO2 fi'ostor adsorbate
on Io's surface,Nature, 280, 766, 1979.
Trafton, L. M., J. J. Caldwell, C. Barnet, and C. C. Cunningham,The
gaseoussulfur dioxide abundance over Io's leading and trailing
hemispheres:
HSTspectra
of Io'sC•B,-X•A•bandof SO,near
2100 A, A•trophys.J., 456, 384-392, 1996.
Wagner, J. K., B. W. Hapke, and E. N. Wells, Atlas of reflectance
spectraof terrestrial,lunar and meteoriticpowders and frostsfrom
92 to 1800 nm, Icarus, 69, 14-28, 1987.
Cruikshank,
D. P., T. J. Jones,andC. B. Pilcher,Absorptionbandsin the
spectrumof Io, Astrophys.J., 225, L89-L92, 1978.
Fanale, F. P., R. H. Brown, D. P. Cruikshank, and R. N. Clark,
A. R. Hendrix, C. A. Barth, and C. W. Hord, Laboratory for
Atmosphericand Space Physics,University of Colorado, CampusBox
Significance
of absorption
featuresin Io's IR reflectancespectrum, 590, Boulder,CO 80309-0590. (hendrix@lasp.colorado.edu)
Nature, 280, 761-763, 1979.
Hapke, B., Theory of Reflectance and E•nittance Spectroscopy, (ReceivedJuly 15, 1998; revisedJanuary29, 1999'
CambridgeUniv. Press,New York, 1993.
acceptedFebruary 10, 1999.)