Global Color Variations on Io P.E. Geissler, A.S. McEwen, L. Keszthelyi Lunar and Planetary Laboratory, University of Arizona, Tucson, AZ. 85721 voice: (520) 621-2114 fax: (520) 621-4933 Email geissler@pirl.lpl.arizona.edu R. Lopes-Gautier Jet Propulsion Laboratory, 4800 Oak Grove Drive M/S: 183-601 Pasadena, CA 91109-8099 voice (818) 393-1232 Fax (818) 393-4530 Email rlopes@isaac.jpl.nasa.gov J. Granahan SETS Technology, Inc. and the University of Hawaii Science and Technology International 733 Bishop Street Suite 3100, Makai Tower Honolulu, HI 96813 voice (808) 540-4745 Fax (808) 540-4850 Email granahan@lava.net D.P. Simonelli Cornell University, 320 Space Sciences Building Ithaca, NY 14853-6801 voice (607) 255-8608 Fax (607) 255-9002 Email simonelli@cuspif.tn.cornell.edu 34 manuscript pages 18 figures 1 table Submitted to Icarus Revised 20 January 1999 Proposed running head: "Global color variations on Io" ABSTRACT Visible and near-infrared images of Io from the Galileo spacecraft reveal a surface more colorful than previously thought. Red, yellow, green, white and black hues decorate the satellite, presumably caused by a varied composition of sulfur compounds and silicates. Almost a third of Io is covered by red and orange materials, particularly at polar latitudes above +/- 30 degrees. These red regions were scarcely distinguishable in the shorter-wavelength Voyager observations. Bright red pyroclastic deposits mark the locations of many hot spots, plumes and visible surface changes, providing a prominent flag of recent volcanic activity. Io's equatorial regions are dominated by yellow materials, which occupy about 40% of the satellite's surface. White and grey materials cover about 27% of Io, primarily in equatorial areas and in localized deposits at high latitudes. These are identified with moderate-to-coarse-grained SO2 as mapped by the NIMS instrument on Galileo (Carlson et al., Geophys. Res. Lett. 24, 2479-2482, 1997). Greenish-yellow materials in small isolated spots on Io's anti-Jupiter hemisphere were recently discovered in 3 km/pixel color imaging from orbit 14. Unlike other ionian terrains, these regions have a negative near-infrared spectral slope, suggesting contamination by a non-sulfur component. Only about 1.4% of Io's surface is occupied by dark materials, which display a variety of visible colors ranging from black to red and green. Most dark spots have a shallow spectral absorption feature at 0.9 micron, suggesting magnesium-rich silicates rather than black sulfur. Little large-scale alteration in the global color and albedo pattern has occurred between the Voyager and Galileo eras; 90% of the surface appears unchanged despite the vigorous volcanic activity which must have taken place in the intervening 17 years. This suggests that over a time scale of decades, the bulk of Io's resurfacing is restricted to a few small but persistently active areas. Keywords: Io, Galileo, Spectroscopy, Multispectral imaging 1. Introduction Io is by far the most colorful satellite in the solar system. In brief glimpses provided by the Voyager missions in 1979, Io was revealed to have a variegated surface of yellow, orange, brown and white hues indicative of its violent geologic activity (Smith et al. 1979a,b). Comparison of Io's u.v.-visible colors as seen by Voyager with laboratory spectra suggested sulfur allotropes and SO2 as likely compositional candidates (Soderblom et al. 1980). Many large-scale color and albedo changes were detected by comparing images of Io acquired during the two Voyager encounters, only 4 months apart (McEwen 1988). However, the expectation that the satellite might be largely unrecognizable by the time the Galileo spacecraft arrived at Jupiter was dispelled by low resolution (160 km/pixel) imaging from the Hubble Space Telescope on the eve of Galileo's arrival (Spencer et al. 1997). Galileo multispectral imaging extends our perception of Io's color in several respects. Most important is the ability to "see" at longer wavelengths: Voyager's vidicon imagers were insensitive to wavelengths beyond 0.59 microns (orange), whereas Galileo's silicon CCD Solid State Imaging (SSI) system responds to red and near-infrared wavelengths up to 1 micron (Belton et al. 1992). Galileo has also imaged Io at a wider variety of phase angles than possible during the Voyager flybys, revealing stark differences in the photometric behavior of various surface units (Simonelli et al. 1997), and has observed the satellite over a much longer time period. At the close of Galileo's nominal mission, low resolution global color coverage has been obtained in the violet (0.413 micron center wavelength in white light; Klaasen et al. 1997), green (0.560 micron), red (0.665 micron) and near-infrared (NIR, 0.757 micron) filters, and 6-color coverage including the 889 (0.888 micron) and 968 (0.991 micron) filters extends over limited regions. The SSI observations of Io in the first 10 orbits are described in McEwen et al. (1998a). The goals of the present study are (1) to characterize the visible and near-infrared spectral reflectance of Io's diverse surface materials, (2) identify spectral units and map their global distribution, and (3) search for regional or global changes in the color and albedo of the surface in the 17 years since the Voyager encounters. Here we present an interim report on the multispectral imaging results from the start of the nominal mission up to orbit E14; the highest-resolution Io color data are yet to come at the end of Galileo's extended mission, which includes two close flybys of the satellite in late 1999. In the following sections we first examine the appearance of Io in a global mosaic of low phase-angle color images from 1996-1997 and compare it with a similar view from Voyager 1 (McEwen 1988). Next we examine the spectral properties of ionian surface terrains as seen in SSI broadband filter photometry, and use these results to define color image ratios which best discriminate between spectral units. The distributions of each of four color units are then presented, and the geologic modes of occurrence and relationships between the units are discussed. We conclude by suggesting interpretations of the new data with respect to Io's composition and resurfacing rate. 2. Global Multispectral Mosaics To begin to explore Io's color variations, we constructed global mosaics in 4 colors (violet, green, red and NIR) from the imaging data listed in Table 1. These data were acquired primarily to monitor surface changes throughout the nominal mission, and typically consisted of full resolution green filter images supplemented by coverage in the violet, red and NIR passbands at reduced resolution using SSI's 2x2 pixel summation mode, in which each pixel represents the sum of four adjacent pixels from the CCD detector (Belton et al. 1992). For the more distant observations it proved practical to conserve resources by recording up to 4 images on the CCD before reading the data out to tape. This "on-chip mosaic" strategy slightly reduced the signal to noise ratio of the data and increased the complexity of the calibration, since the dark current varied from exposure to exposure. The choice of data for global mosaics was constrained by the requirement that the images be acquired at consistent phase angles; early attempts at mosaic construction based on data from the first and second Galileo orbits were complicated by stark differences in surface contrast produced by the variations in illumination and viewing geometry (Geissler et al. 1997). In particular, the low-albedo polar regions of Io darken less than equatorial areas with increasing phase angle (Simonelli et al. 1997). Such photometric variations can be minimized by confining our attention to the lowest available phase angle images, ranging from 0.5 to 13.9 degrees (Table 1). An advantage of this approach is that the global image mosaics generated from these data can be directly compared to similar data products from Voyager 1 (2.6 to 14.7 degrees, McEwen 1988) and HST (6.8 to 9.1 degrees). No red filter data were recovered from one of the better low phase observations, during orbit E4 (Table 1), but these longitudes were adequately covered by lower resolution red filter images from orbits G2 and C9. Data processing began with radiometric calibration using procedures derived by Klaasen et al. (1997). Spatially adaptive filtering (Eliason and McEwen 1990) was then performed to partially remove isolated noise spikes due to charged particle radiation. Next, color composites or image "cubes" were assembled for each of the four observations by registering the violet, red and NIR filter images to the green image, generally the sharpest of the set. Because the observations were distant and acquired over short time intervals, this correction consists of simply translating and rescaling the images so that they overlie one another, and was performed to subpixel precision using automated procedures of the U.S.G.S. "ISIS" software system (Torson and Becker 1997). The image sets were next "navigated" (camera-pointing corrected) by manually adjusting the positions of predicted limb profiles overlain on the green filter images. This step left errors which limit the geographic accuracy of the global mosaics to an estimated +/- 1 degree latitude and longitude. The data were then reprojected to an equal-area cylindrical map projection with a scale of 5 km/pixel at the equator. A simple lunar-Lambert photometric function was employed to account for limb darkening, with values of 0.6 to 0.8, depending upon phase angle, for the limb-darkening parameter L (Table 1) from McEwen (1991). This correction, intended to normalize brightness variations across the disk of Io, was assumed independent of wavelength and applied uniformly to all of the filter bands. The images were then trimmed to exclude data too near the terminator to be useful. The moderate phase angle (4.1 - 4.9 degrees) observations from the second and ninth orbits could then be mosaicked without further adjustment, but the higher phase (13.9 degrees) E6 images and especially the low (0.5 degree) phase data from E4 required further correction to account for variation with phase angle. We chose to correct these brightness mismatches empirically, by calculating for each color a linear correction based on the correlation between the E4 and E6 images and the intermediate phase angle G2 and C9 data in the regions of mutual overlap. The resulting mosaics thus approximate the appearance of Io at 4.1 to 4.9 degrees phase. The seam-removal technique of Soderblom et al. (1978) was applied before the data were finally assembled into a global mosaic. Two versions of the resulting multispectral mosaic are shown in Plate 1. Both are false-color composites of the NIR, green and violet filter data portrayed in red, green and blue hues respectively. The top frame is linearly contrast-stretched from 0 to the maximum digital number (corresponding to a radiance factor of 1.12), and approximates the visual appearance of Io in 1996 - 1997. At bottom is an enhanced color representation, for which each of the NIR, green and violet filter mosaics have been stretched individually. Plate 2 shows the same data in orthographic projections centered on the north and south poles. Four major color units can be easily discerned in these images, particularly in the enhanced color versions: (1) red and orange materials dominate the dark polar regions of Io and are found at lower latitudes in isolated bright deposits from presently active plumes such as Pele (the bright red ring centered at 19 S, 255 W), Marduk (30 S, 209 W), Culann (20 S, 160 W) and others (see Figure 2 of McEwen et al. 1998a, for the locations of named features on Io); (2) yellow and yellow-green materials commonly coat the equatorial plains within 30 degrees of the equator, and occur in isolated patches at higher latitudes; (3) white and grey materials have the same overall distribution as the yellow unit, i.e. concentrated near the equator and in isolated polar patches; and (4) low albedo materials (black in these saturated color pictures) occur in many small spots marking the locations of active calderas and more rarely in diffuse deposits from plumes such as Pele and Babbar (39 S, 273 W). Plates 1 and 2 confirm the conclusion from groundbased (Minton 1973) and HST (Spencer et al. 1997) observations that Io tends to be yellow at the equator and red at the poles. As we shall see next, the overall distribution of these four color units has changed relatively little since 1979. An attempt to simulate a Voyager-like view of Io using the Galileo bandpasses was made by crudely synthesizing an "orange" filter mosaic by averaging the red and green filter frames. Plate 3 presents a comparison between the orange, green and violet composite from Voyager 1 (McEwen 1988), the higher resolution of the two Voyager data sets, and the synthetic "orange", green and violet composite from Galileo SSI. The results show that the Galileo simulation is still too red; the mean of the SSI green filter center wavelength (0.559 microns) and red center wavelength (0.665 microns) is 0.61 microns, significantly longer than the center wavelength of Voyager's orange filter (0.59 microns) in a spectral region where the reflectance of the red materials rises rapidly with wavelength. On the other hand, the color discrepancy actually makes it easier to identify regions in the Voyager 1 mosaic which have barely detectable reddish hues, such as the western reaches of Pele's plume deposit (20 S, 267 W). Other complications of the comparison are the small positional shifts which are generally the result of errors in image navigation, as described above. These will be rectified when the Voyager data are reprocessed and tied to an updated base map. Plate 3 demonstrates a somewhat surprising result: despite the vigorous volcanic activity and local resurfacing that has taken place over the 17 to 18 year interval between the two spacecraft observations, little alteration of the global color and albedo distributions has occurred. More than 30 small scale (tens of km) Voyager-to-Galileo surface changes have been documented (McEwen et al. 1998a), and several regional changes can be seen for the first time in Plate 3, such as the brightening of areas around Daedalus (19 N, 273 W) and Kurdalagon (50 S, 219 W). There are more similarities than differences, however, when allowance is made for imperfect data processing and genuine photometric variations. The fact that the face of Io has remained recognizable between spacecraft visits is consistent with the HST observations (Spencer et al., 1997) and the long term stability of the satellite's appearance in ground-based observations spanning several decades (Morrison et al. 1979), and suggests that most of the resurfacing which has taken place over the last two decades has been confined to a small fraction of the satellite's surface. An estimate of the area resurfaced over the last 17 years is shown in Figure 1, a map of the large-scale changes in the visible appearance of Io from the Voyager flybys to the end of Galileo's nominal mission. Only about 10% of Io's surface has markedly changed in color or albedo since 1979. New red materials, including Pele's plume deposits which altered between the two Voyager flybys, account for one third of the altered area (proportionate to the area of the red unit; Section 4). Approximately 0.8% of Io was mantled with new dark deposits, much of it during the eruption of Pillan in 1997 that took place after the acquisition of the images shown in Plates 1 and 2. The thickness of these new deposits may range from a few tens of microns, the minimum necessary to alter the optical properties of the surface, up to many meters in the case of new lava flows. 3. Spectral Measurements Measurements of the spectral reflectance of various surface units were obtained from 6-color imaging observations during Galileo's 10th and 14th orbits (Table 1). These data were acquired at phase angles from 36 to 75 degrees, much higher than those used to construct the global mosaics. In addition to the previously described calibrations, these images were corrected for instrumental scattered light via a wavelength-dependent sharpening procedure (Gaddis et al. 1996). Next, in the same manner as described above for the global mosaics, we corrected the data for limb darkening and adjusted the brightness to estimate near-normal reflectance, i.e. reflectance at 4 to 5 degrees phase. The limb darkening function of Io at high phase is nearly Lambertian (McEwen 1991), and the corresponding L values used for photometric normalization are listed in Table 1. We next scaled the spectra by an empirical phase angle correction factor to match the global mosaic; the scale factors used are also listed in Table 1. These corrections are of course approximate and do not include any opposition effects. As applied here, they are simple multiplicative factors that affect only the estimated albedos, not the shapes of the spectra. A useful check is provided by the HST data, which were acquired with similar filters at nearly normal geometry (6.8 to 9.1 degrees phase; Spencer et al. 1997). The comparison suggests that the spectral slopes reported here are generally accurate and the near-normal albedos are correct to within about 10%. Measurements were made at a sufficient number of surface locations to define the differences between, and the variability within, each of the 4 color classes. The sites sampled are shown in Figure 2; in each case, measurements were obtained from the largest possible rectangular regions within homogenous areas. This proved problematic in the case of the dark materials, frequently represented by small spots only a few pixels in diameter, and resulted in larger variance for the dark materials' spectra. Figure 3 shows the spectra of the red materials. They are characterized by steep positive slopes from the violet to the red filters, slightly less steep from red to the NIR, and shallow positive slopes at longer wavelengths. Reflectances at 4 to 5 degrees phase range from 0.1 - 0.2 in violet, increasing to > 1.0 at 1 micron. The spectral feature that distinguishes red materials from the other color classes is their steep slope between green and NIR wavelengths. Figure 4 shows the spectra of selected yellow materials. They also display steep positive slopes from the violet to the green, but have a pronounced inflexion to much shallower slopes between green and red, grading to still shallower slopes at wavelengths beyond red. Near-normal reflectances range from 0.1 - 0.4 in violet and reach a maximum of 0.7 - 1.2 at 1 micron. The yellow materials are distinguished by the sharp change in spectral slope at green wavelengths. Many yellow-unit materials appear somewhat greenish, perhaps indicating a distinct composition. In high resolution 6-color imaging obtained in orbit E14, a few small areas less than 100 km in length can be seen to have a distinct green hue in contrast to the more typically yellow plains materials nearby (Figure 5). Spectra from three such spots are shown in Figure 6. These "green" materials are also characterized by a sharp change in spectral slope at green wavelengths, but unlike the yellow unit they reach maximum reflectance in the green to red filter bandpasses and display a negative spectral slope at longer wavelengths. This is not a characteristic of common sulfur allotropes, but may be due to iron contamination (Kargel et al. 1998) or perhaps silicates (Section 5). These spots are too small to appear distinct in the global color mosaics, and their global distribution is as yet unknown. Figure 7 shows the spectra of the white materials, which are actually pale pink or yellow at visible wavelengths. Their distinguishing features are their relatively shallow slopes at short wavelengths and their high albedos in the violet, ranging from 0.8 - 0.9. They display shallow positive slopes in the infrared, similar to those of the yellow and red units, and reach maximum near-normal reflectances of 1.2 - 1.3 at 1 micron. Again, the higher resolution color imaging obtained in orbit E14 reveals previously unknown color variations within and between bright deposits, such as subtle pink hues at the core and margins of the bright deposit Bactria Regio centered at 44 S, 127 W. Dark materials appear to be the most variable of the four classes (Figure 8), but some of this variability might be due to not fully resolving the dark units. Distinguished by their low reflectance relative to their surrounds, dark materials show a diversity of colors and an array of albedos ranging from 0.1 to 0.5. Among the darkest is the active hotspot Amaterasu (38 N, 307 W), with a maximum near-normal reflectance of only 14% at 0.76 microns. Amaterasu was even darker (~5% reflectance) during the Voyager era (McEwen et al., 1985). Most of the dark spots are generally red at visible wavelengths, but several of the brighter spots such as Kanehekili (16 S, 38 W; spectrum number 8 in Figure 8) have steep slopes from violet to green, and sharp inflexions at green wavelengths. Perhaps these are dark materials which are masked by or mixed with other units. Kanehekili is also a hot spot, but the flux contribution due to its thermal emission (measured while Io was eclipsed by Jupiter during orbits G7 and E15) is less than 2% of its reflected radiance in the 968 filter, so no correction of the spectra for thermal emission was performed. Particularly bright and active hot spots such as Pele and Pillan can have much greater thermal flux contributions (McEwen et al., 1998b), but the emitting regions are so small that these would appear indistinguishable from noise spikes in these images and they would be removed by spatial filtering during the early steps of data processing. An especially important feature is found in the long wavelength spectra of the dark materials: an apparent absorption can be seen at 0.89 microns which is absent from the red, yellow and white units. Figure 8 shows this previously unknown spectral feature in data from two separate C10 observations. The absorption can also be seen in the long-wavelength data from orbits E4 and E14. 4. Spectral Unit Mapping Each of the four color units described above can be characterized by a distinctive spectral property which sets it apart from other ionian terrains. A simple means of separating the classes and mapping the spatial distributions of the individual units is to create color ratio images from the global 4-color data set. Important albedo information is lost by dividing one spectral band by another, information necessary for example to distinguish the dark spots from their surrounds. On the other hand, color ratios have the advantage of relative insensitivity to errors in the limb darkening correction, and are also unaffected by wavelength-independent brightness variations due to other factors such as unresolved shadows. Red, yellow and white materials are best distinguished by their spectral slopes and can be easily mapped in color ratio images. Dark materials, because of their variable colors, can be more readily identified by their local albedo contrasts. In this section we derive the global spatial distributions of the red, yellow, white and dark materials, point out prominent changes that have taken place since the Voyager flybys, and discuss the relationships between the units and their geologic modes of occurrence. Red Materials The most important consequence of SSI's improved long wavelength sensitivity is the ability to distinguish bright red, diffuse deposits such as produced by the presently active plume Pele. These are probably pyroclastic volcanic deposits, based on their association with active plumes (McEwen et al. 1998a). Because of their spectral similarity to other units at green and shorter wavelengths, these materials were poorly discriminated in Voyager imaging observations, appearing similar in color to, but slightly darker than neighboring terrains (Plate 3). However, their steep spectral slopes at wavelengths from 0.56 to 0.76 microns are distinctive, and they are much brighter than other materials in a NIR/green ratio image (Figure 9). Red-orange materials dominate the polar regions of Io, with distinct boundaries at latitudes higher than +/- 30 degrees. The darkest of these red polar regions appear to have shallow spectral absorptions longwards of 1 micron (Carlson et al. 1997). Red materials are also found in discrete deposits at lower latitudes, usually associated with high temperature, low albedo volcanic edifices which show significant change between the Voyager and Galileo eras. The most prominent example is Pele itself: a ring of bright red pyroclastic deposits encircles the vent, with a minimum diameter of over 500 km reaching outwards to at least 1100 km. Diffuse dark deposits are found within this bright red ring, extending up to 200 km from the central caldera. Pele's red deposits are large enough to be resolved in Hubble Space Telescope images of Io (Spencer et al. 1997) which first showed their distinct spectral slope in the green to near-IR wavelength range. Globally, many smaller deposits of red materials up to 100 km in length are found adjacent to active volcanic centers at low to mid latitudes. Examples of bright red deposits emplaced since Voyager are found near Marduk (30 S, 209 W), Zamama (18 N, 174 W), Culann (20 S, 160 W) and Prometheus (2 S, 152 W), all sites of Voyager-to-Galileo surface morphological changes and of temperature anomalies detected by the NIMS instrument on Galileo or by SSI imaging during Io eclipse, or both (Lopes-Gautier et al. 1999). With the exception of Pele, these new deposits are strikingly asymmetric with respect to their source vents, unlike the white deposits from more quiescent, presumably SO2-rich plumes. The red pyroclastic deposits may result from overpressurized (underexpanded) plumes, whereas the white rings are from balanced or umbrella-shaped plumes (Kieffer, 1982). A number of other distinctly red deposits are located near hot spots detected by Galileo, including an extended region of northern Colchis Regio near Lei-kung Fluctus (33 N, 206 W); Tupan Patera (19 S, 140 W); Malik (34 S, 129 W); Zal (35 N, 76 W); Mithra (59 S, 267 W); Sethlaus (52 S, 194 W); Kurdalagon (50 S, 218 W); Rata (35 S, 200 W); Fuchi (28 N, 330 W); and Amirani (18 N, 117 W), although the hot spot detected by NIMS along Amirani is about 9 degrees to the north of the red deposit (Lopes-Gautier et al. 1999). Four prominent red deposits are in areas in which no hot spots have yet been detected. These are Tohil (26 S, 157 W); a small patera west of Culann (23 S, 166 W); an un-named patera west of Ulgen (39 S, 290 W); and Euboea Fluctus (47 S, 350 W), where the bright red materials exhibit an unique bilateral symmetry around the source vent. Finally, several faded or otherwise indistinct red deposits may mark sites of recent, but presently inactive volcanism: Dazbog (55 N, 300 W), darker and perhaps active during the Voyager era; a nearby un-named patera (48 N, 310 W); Surt (43 N, 340 W); an un-named spot at 52 S, 202 W; the area east of Volund (near 28 N, 168 W); and an un-named region centered at 31 N, 146 W which is the site of a Voyager-to-Galileo surface morphological change (McEwen et al. 1998a). This is clearly not an exhaustive list, as shown by the recent E14 data: new red deposits (as well as smaller hot spots) can be expected to be found as we continue to image Io at higher resolution. Yellow Materials Yellow and greenish-yellow materials are characterized by an inflexion in their spectra at green wavelengths (Figures 4 and 6). They are therefore bright in a color ratio comprised of the green filter image divided by the sum of the violet and NIR filter images (Figure 10). Yellow-green materials dominate the equatorial region of Io, up to latitudes of +/- 30 degrees, and occur in localized concentrations at higher latitudes (especially within or adjacent to white, presumably SO2-rich patches). This unit appears to be mixed with, or perhaps compositionally or genetically related to the red and white materials; contacts between yellow and red units are diffuse and poorly defined, as are the contacts between yellow and white materials. Moreover, deposits which appear white or pale yellow at low phase angles take on more saturated yellow hues at higher phase, in a manner analogous to lunar phase reddening (e.g., Gradie and Veverka 1986). A distinct and extensive deposit of yellow materials mantles the topography of Lei-Kung Fluctus to the north of Colchis Regio (centered at 47 N, 214 W). This deposit appears to have enlarged relative to its appearance in Voyager images; either new yellow deposits have been emplaced, or formerly red materials have altered to yellow. Another yellow deposit appears to have recently formed along the eastern margin of Loki. The global distribution of the more greenish materials is as yet unknown, since the color imaging observations to date largely lack adequate spatial resolution to discern them. Regions with particularly low NIR/green ratios, possibly indicative of negative near-infrared spectral slopes, include the areas near Heiseb (30 N, 235 W), Ra (10 S, 325 W), Babbar (38 S, 277 W), Karei (2 S, 12 W) and along the eastern margin of Bosphorus Regio (16 S, 103 W). White Materials In agreement with previous observations (e.g., McEwen et al. 1988), grey and white spectral units on Io correspond to areas of abundant, coarse to moderate grained SO2, as determined by the Near-Infrared Mapping Spectrometer (NIMS) instrument on Galileo (Carlson et al. 1997). This does not appear to be a one-for-one correspondence, however, as deep SO2 bands are seen by NIMS in yellow regions as well. White and grey materials are easily distinguished in both Voyager and Galileo imaging observations using either the violet filter albedo or the violet/green color ratio (shown in Figure 11). Because the bandpasses of these two filters are similar for both spacecraft, comparisons between the data sets can be done on a consistent basis. Even so, the sharp dependence of this unit's brightness on phase angle (Simonelli et al. 1997) requires that we restrict the present discussion to gross changes in spectral unit boundaries. To first order, then, remarkably little change has occurred in the global distribution of white materials over the last 17 years. New white deposits have formed principally on the trailing hemisphere: in the region surrounding Daedalus (19 N, 273 W); to the northwest and southeast of the detached linear dark feature north of Loki (18 N, 303 W); and to the south of Pele (from 33 S, 230 W to 43 S, 260 W). An unusually bright deposit is located in a basin north of Colchis Regio (35 N, 190 W). A diffuse bright ring outlines the deposit, presumably due to fine-grained SO2 frost of unknown origin. Dark Materials Figure 12 shows the distribution of dark materials, defined here as isolated spots and patches locally darker than their surrounds and mapped on the basis of albedo contrast rather than absolute albedo. Most dark materials on Io are actually quite red at this resolution and are brighter than much of the Earth's Moon; they appear black only in saturated color pictures. Low albedo materials occur both as diffuse pyroclastic deposits and as continuous units confined to caldera interiors and flows. The green filter near-normal albedos of dark materials range from 12% (at Amaterasu Patera; 37 N, 305 W) up to almost 40%, while their colors vary considerably between deposits. The colors of active calderas also appear to vary across their surfaces; Loki appears much redder near its margins than at its center, for example. Globally, most dark calderas are located near the equator; few are found at latitudes higher than +/- 60 degrees, perhaps partly because of poorer resolution. Low albedo spots are the most changeable of ionian surface features, with more than 20 Voyager-to-Galileo shape, color or albedo differences documented (McEwen et al. 1998a). Among these, spots which appear newly formed or have grown darker tend to correspond to detectable hot spots indicative of current activity, such as Altjira (35 S, 109 W); an un-named patera at 1 S, 217 W; Kurdalagon (51 S, 218 W); and Culann, Marduk and Zamama as previously mentioned. On the other hand, very little alteration is evident among existing dark diffuse deposits: the prominent dark pyroclastic deposits of Babbar, Pele and Isum (30 N, 208 W) appear largely unchanged from their appearance in Voyager images. Dark diffuse materials apparently deposited by airborne plumes have disappeared from the area around 30 S, 303 W, but these appear to be mantled with new grey-white deposits. Dark deposits from Loki may also have been buried, but the surface was obscured by active plumes during the Voyager era. A large new dark deposit was emplaced during the eruption of Pillan (12 S, 242 W) in June, 1997, after the acquisition of the images used in Figure 12. The eruption blanketed a 500,000 km2 area of formerly red Pele plume ejecta with a diffuse dark deposit of unknown composition. Many low-albedo spots appear slightly darker at 0.889 microns than in the NIR (0.756) and 1-mm (0.968) filters (Figure 8). The correlation of this apparent infrared absorption with the dark spots was verified by creating an 0.889 micron "band-depth map", computed by dividing the sum of the NIR and 1-mm filter images by the 0.889 micron filter image. The result for the Kanehekili region, observed in orbit C10, is compared with the corresponding NIR image in Figure 13. Regions with relatively deep 0.889 micron absorptions appear bright in the ratio, and can be seen to correspond to dark spots at Kanehekili, Hi'iaka, Shamshu and several dark areas in the vicinity of Masubi. The apparent band depth reaches a maximum of 30%, although this might be reduced to perhaps 20% when noise is considered, and may be even deeper at higher spatial and spectral resolution. Coverage of Io in the 0.889 micron filter is so far quite limited (Table 1), but 0.889 micron absorptions have also been detected at Shamash, Zal, Gish-Bar, Fjorgynn, Loki, Amaterasu and Babbar. All of these are active hot spots (Lopes-Gautier et al. 1999). Global Color Unit Map An estimate of the proportion of Io covered by each of the four color units was made by combining the individual distributions described above and producing the composite global color unit map shown in Figure 14. A crude classification was achieved by thresholding the NIR/green, violet/green and green/(NIR + violet) color ratio images to identify red, white and yellow units respectively. Threshold values were chosen to simultaneously minimize overlap between units (i.e., areas classified as more than one type of material) and gaps at unit boundaries (not belonging to any unit). Areas of overlap were assigned to units on the basis of confidence of detection: regions disputed between red and yellow were assigned to the red unit, while pixels intermediate between yellow and white were included with the white unit. (In reality, much of Io may be covered by materials which are mixtures of the end-member spectral units; McEwen et al. 1988.) The distribution of dark materials was superimposed last. About 3.7% of the surface was left unclassified, mostly because of no coverage (at the poles) or indeterminate unit membership, such as small gaps along unit boundaries. These regions were excluded from subsequent pixel counts and surface area computations. From Figure 14 we estimate that approximately 40% of Io's surface is covered by yellow materials. Red materials make up about 32%, primarily confined to latitudes above +/- 30 degrees. 27% of the area is covered by the white spectral unit, and little more than 1% is occupied by dark materials. These figures are accurate only to perhaps 5%, based on the proportion of unclassified and overlapping pixels to the total, except for the fraction of dark materials (believed to be correct to within a factor of 2). 5. Discussion At least two issues can be partially addressed by studying Io's global color variations. First, what chemical compounds are responsible for Io's red-orange, yellow-green, white and dark hues? Second, where and how is the satellite being resurfaced? Of course, Io's composition can not be uniquely identified on the basis of broad-band 6-filter photometry. We can only offer interpretations based upon the observed distributions of and relationships between the color units, and upon comparisons with laboratory spectral measurements. Figures 15 to 18 show the spectral reflectance of a selection of candidate surface constituents, convolved to the filter bandpasses of Galileo SSI. These can be compared (with caution!) to the Io spectra of Figures 3 to 8. The most interesting new compositional information from SSI is the observation of an apparent 0.9 micron absorption among dark materials. A search of over 650 minerals and rocks from the digital spectral libraries of Clark (1993), Grove et al. (1992), Gaffey (1976), and R. Singer (unpublished) revealed that the deepest 0.9 micron absorptions bands, calculated in a manner comparable to Figure 13, are possessed by orthopyroxenes (Mg-rich silicate minerals common in terrestrial mafic and ultramafic rocks) such as enstatite (MgSiO3) and bronzite-hypersthene ((Mg,Fe)SiO3) (Figure 17). These orthopyroxenes must contain at least a small amount of iron to produce the 0.9 micron Fe2+ absorption band (e.g., Adams, 1974). Clinopyroxenes, in contrast, have absorptions at longer wavelengths and would not produce the observed spectral behavior. There are reasons to expect magnesium-rich silicates in Io's active volcanic centers: the high temperatures exhibited by hot spots (1700 K or hotter) are inconsistent with sulfur compounds or even iron-rich silicates such as basalts unless heated well beyond their melting points (McEwen et al. 1998b), and magnesian silicates are predicted to comprise Io's mantle as a result of extensive and repeated magmatic differentiation (Keszthelyi and McEwen 1997). The interpretation of the dark, caldera-filling materials as orthopyroxene-bearing assemblages thus seems reasonable, but is not unique. Other compounds with 0.9 micron absorptions include iron oxides and hydroxides such as hematite (Fe203) and goethite (FeOOH), sulfates such as jarosite ((Na,K)Fe3(SO4)2(OH)6) and even carbonates such as siderite (FeCO3) (Figure 18). Some of these candidates can be dismissed on geochemical grounds: water and hydroxyl are unlikely to be abundant on Io, making phases such as goethite and jarosite improbable, and the availability of carbonates to form siderite is likewise questionable. Moreover, all of these compounds have melting points even lower than that of basalt, making their identification as the source of the high temperature hot spots untenable. Given the variability of the dark materials' visible color and the likelihood of contamination by sulfur compounds, however, we can not yet make an unique interpretation. Further SSI observations are expected during Galileo's extended mission, including coverage at higher spatial resolution, but a definitive identification will probably require measurements at longer wavelengths (in the 2 micron spectral region diagnostic of orthopyroxenes for example), like those planned for NIMS during the close flybys of Io in late 1999. Sulfur and its compounds remain the most viable compositional candidates for the variegated colors of Io, as concluded by Voyager investigators (Soderblom et al., 1980). So far, only SO2 has been definitively identified on Io by its diagnostic near-infrared absorption features (Fanale et al., 1979; Smythe et al., 1979; Carlson et al., 1997). Even the brightest regions on Io are tinged with pale pink and yellow hues, however, suggesting contamination with sulfur. Because NIMS sees SO2 in both white and yellow regions, the yellow unit may in places be a dilute mixture of sulfur with SO2, perhaps a thin veneer which is colored at visible wavelengths but transparent in the infrared. Ordinary orthorhombic sulfur (S8) could comprise the equatorial yellow unit, although other materials such as polysulfur oxides, sodium sulfides and sulfates can not be ruled out. A possible clue to the makeup of both the yellow and red units is the transition from yellow to red-orange hues at latitudes near +/- 30 degrees. Large S8 molecules may be rapidly destroyed at high latitudes on Io by charged particle irradiation (Johnson 1997), leaving short-chain S3 and S4 polymers which may account for the red color of Io's poles. Thermal equilibration to stable cyclo-octal sulfur should proceed more slowly at the poles than at the equator because of the lower surface temperatures (McEwen et al. 1998a), favoring the formation of colored sulfur diradicals by charged particles, ultraviolet (Steudel et al., 1986) and X-radiation (Nelson et al. 1990). Similar short-chain allotropes of sulfur, quenched in their high temperature forms, could also comprise the red pyroclastic materials seen at lower latitudes and perhaps even some of the diffuse black deposits. Alternatively, the dark diffuse deposits at Pele, Babbar and Pillan could simply represent silicate pyroclastic eruptions familiar from the terrestrial planets. The question may be resolved by long wavelength SSI and NIMS observations of these areas later in the mission. On the other hand, the greenish materials probably can not be accounted for by sulfur allotropes alone; uncontaminated sulfur specimens commonly have flat or positive spectral slopes in the visible to 1 micron wavelength range. The greenish materials may be made up of sulfur contaminated by iron (Kargel et al. 1998) or perhaps silicates such as olivine or pyroxene (Figure 16). Basic questions about the emplacement and alteration of Io's spectral units remain to be answered. Perhaps the most puzzling is the lack of change in the global color and albedo pattern between the Voyager and Galileo eras, despite abundant evidence for vigorous volcanic activity in the intervening 17 years. This is particularly surprising in view of the fact that over million-year time scales, the globally averaged resurfacing rate must be on the order of mm/year, based on the absence of impact craters and the estimated flux of impactors (e.g. Johnson and Soderblom, 1982; Zahnle et al., 1998), while several cm/year may be needed to account for Io's present global heat flow (e.g., Spencer and Schneider 1996). Yet apparently as much as 90% of Io has not been resurfaced at all in the past two decades, unless the new deposits have subsequently been removed or are identical in color and albedo to the materials they bury. This requires that the resurfacing rate in the active areas amounts to ten times the global average. In fact, the degree of concentration may be much greater than an order of magnitude, since many of the newly resurfaced areas are probably quite thin: deposits only a few tens of microns deep are sufficient to alter the optical properties of the surface, and there is evidence that some deposits, such as those from the high-latitude plumes Surt and Aten, have eroded away or otherwise faded over the past two decades. If the bulk of Io's resurfacing is restricted to just 1% of the surface, the approximate area of new or altered dark materials, then meters to tens of meters thickness must have accumulated over the last two decades in these persistently active areas to account for the global average. Galileo may provide a refined estimate of time scales for resurfacing by plume deposits through continued observations of the Pillan area; Pillan's dark deposits should soon be obscured by fresh red materials from Pele, and the clock started ticking in June 1997, at the time of Pillan's eruption. 6. Conclusions Both the acquisition and analysis of Galileo multispectral images of Io are ongoing at this writing, with the highest resolution observations yet to come. Nonetheless, we can draw the following conclusions from this interim study of Io's global color variations: 1. The global pattern of color and albedo on Io is essentially unchanged from its appearance in 1979. This comes as something of a surprise, given the rapid changes in the satellite's appearance between Voyager 1 and Voyager 2, just 4 months apart. The implication is that, on a time scale of decades, Io's resurfacing is largely confined to many small, persistently active areas of the satellite. 2. Red materials, practically unseen by Voyager, dominate Io's polar regions at latitudes above 30 degrees. They are also found in isolated deposits at lower latitudes that closely correspond to sites of recent or ongoing activity. These red deposits provide a prominent "flag" indicating volcanically active areas along Io's equator. 3. Yellow materials cover much of Io's lower latitudes, particularly on the trailing hemisphere. Contacts between yellow and white regions are diffuse and gradational, and the two units have similar overall distributions. Along the equator, the yellow coloration could simply be a sulfurous stain masking bright, generally SO2-rich materials. 4. Small greenish-yellow deposits have been identified in recent high resolution color observations with negative spectral slopes from 0.56 to 1 micron. The global distribution of these unique materials is not yet known, but they may commonly contribute to the greenish tinge of the yellow unit. 5. White materials occupy the equatorial regions of Io below 30 degrees and are found in isolated deposits at higher latitudes, frequently adjacent to yellow deposits. White areas appear to indicate relatively uncontaminated SO2, perhaps freshly deposited or recently reworked. 6. Dark spots and associated low albedo deposits occupy only about 1% of Io's surface, and most are found within 60 degrees of the equator. Dark materials on Io are distinguished by an apparent infrared absorption feature interpreted to be due to Mg-rich silicates. Acknowledgments We thank Fraser Fanale and Jim Crowley for constructive reviews, and Julianne Moses for helpful discussion. We are grateful to the Galileo Project, the SSI Team, and especially the Photoscience Group at JPL for their efforts in acquiring these data. References Adams, J. 1974. Visible and near-infrared diffuse reflectance spectra of pyroxenes as applied to remote sensing of solid objects in the solar system. Journ. Geophys. Res. 79, 4829-4835. Belton, M.J.S., K.P. Klaasen, M.C. Clary, J.L. Anderson, C.D. Anger, M.H. Carr, C.R. Chapman, M.E. Davies, R. Greeley, D. Anderson, L.K. Bolef, T.E. Townsend, R. Greenberg, J.W. Head III, G. Neukum, C.B. Pilcher, J. Veverka, P.J. Gierasch, F.P. Fanale, A. P. Ingersoll, H. Masursky, D. Morrison, and J.B. Pollack 1992. The Galileo Solid-State Imaging Experiment. Space Science Reviews 60, 413-455. Carlson, R.W., W. Smythe, R. Lopes-Gautier and 8 others 1997. The distribution of sulfur dioxide and other infrared absorbers on the surface of Io. Geophys. Res. Lett. 24, 2479-2482. Clark, R.N., G. Swayze, A. Gallagher, T. King, and W. Calvin 1993. The U. S. Geological Survey, Digital Spectral Library. U.S.G.S. Open File Report 93-592. Gaffey, M., 1976. Spectral reflectance characteristics of the meteorite classes, Journ. Geophys. Res. 81, 905-920. Eliason, E.M., and A.S. McEwen 1990. Adaptive box filters for removal of random noise from digital images. Photogrammetric Engineering and Remote Sensing 56, 453-458. Fanale, F. P., R. H. Brown, D. P. Cruikshank, and R. N. Clarke 1979. Significance of absorption features in Io's IR reflectance spectrum. Nature, 280, 761-763. Gaddis, L., A. McEwen and T. Becker 1995. Compositional variations on the Moon: Recalibration of Galileo solid-state imaging data for the Orientale region and farside. Journ. Geophys. Res. 100, 26345-26355. Geissler, P.E., A.S. McEwen, D.P. Simonelli, R. Lopes-Gautier, A.G. Davies, J. Granahan, T. Denk, and the Galileo SSI team 1997. Global color variations on Io. Lunar and Planet. Sci. 28, 403-404. Gradie, J. and J. Veverka 1986. The wavelength dependence of phase coefficients. Icarus, 66, 455-467. Grove, C.I., S. J. Hook, and E. D. Paylor II, 1992. Laboratory Reflectance Spectra of 160 minerals 0.4 - 2.5 micrometers JPL Pub. 92-2. Johnson, R.E., 1997. NOTE: Polar ``Caps" on Ganymede and Io Revisited. Icarus, 128, 469-471. Johnson, T. V. and L. A. Soderblom 1982. Volcanic eruptions on Io - Implications for surface evolution and mass loss. Chapter 17 in: Morrison, D. (ed.), Satellites of Jupiter. Tucson, AZ, University of Arizona Press, p. 634-646. Kargel, J.S., P. Delmelle, and D.B. Nash 1998. Volcanogenic sulfur on Earth and Io: Composition and Spectroscopy. Submitted to Icarus. Keszthelyi, L. and A. McEwen, 1997. Magmatic Differentiation of Io, Icarus, 130, 437-448. Kieffer, S.W. 1982. Dynamics and thermodynamics of volcanic eruptions: Implications for the plumes of Io. In Satellites of Jupiter (D. Morrison, Ed.), pp. 647-723. Univ. of Arizona Press. Tucson. Klaasen, K., M. Belton, H. Breneman, A. McEwen, M. Davies, R. Sullivan, C. Chapman, G. Neukum, C. Heffernan, A. Harch, J. Kaufman, W. Merline, L. Gaddis, W. Cunningham, P. Helfenstein and T. Colvin 1997. In-flight Performance Characteristics, Calibration, and Utilization Of The Galileo SSI Camera. Opt. Eng. 36, 3001-3027. Lopes-Gautier, R., A. McEwen, W. Smythe, P. Geissler, L. Kamp, A.G. Davies, J. Spencer, R. Carlson, F.E. Leader, R. Mehlmen, L. Soderblom, and the Galileo NIMS and SSI Teams 1999. Hot spots on Io: Global Distribution and Variations in Activity. Icarus, in press. McEwen, A.S. 1988. Global color and albedo variations on Io. Icarus 73, 385-426. McEwen, A.S. 1991. Photometric Functions for Photoclinometry and Other Applications. Icarus 92, 298-311. McEwen, A.S., D.L. Matson, T.V. Johnson and L.A. Soderblom 1985. Volcanic hot spots on Io: Correlation with low-albedo calderas. Journ. Geophys. Res. 90, 12345-12379. McEwen, A. S., L. A. Soderblom, T. V. Johnson, and D. L. Matson 1988 The global distribution, abundance, and stability of SO2 on Io. Icarus, 75, 450-478. McEwen, A.S., L. Keszthelyi, P. Geissler, D.P. Simonelli, M.H. Carr, T.V. Johnson, K.P. Klaasen, H.H. Breneman, T.L. Jones, J.M. Kaufman, K.P. Magee, D.A. Senske, M.J.S. Belton, and G. Schubert, 1998a, Active volcanism on Io as seen by Galileo SSI. Icarus, 135, 181-219. McEwen, A.S., L. Keszthelyi, J.R. Spencer, G. Schubert, D.L. Matson, R. Lopes-Gautier, K.P. Klaasen, T.V. Johnson, J.W. Head, P E. Geissler, S. Fagents, A.G. Davies, M.H. Carr, H. H. Breneman and M.J.S. Belton 1998b. High-temperature silicate volcanism on Jupiter's moon Io. Science, 281, 87-90. Minton, R., 1973. The red polar caps of Io. Comm. L.P.L., 10, no. 188, p. 35-39. Morrison, D., D. Pieri, J. Veverka, and T.V. Johnson 1979. Photometric evidence of long-term stability of albedo and colour markings on Io. Nature, 280, 753-755. Nash, D.B., F.P. Fanale and R.M. Nelson 1980. SO2 frost: uv-visible reflectivity and limits on Io surface coverage. Geophys. Res. Lett. 7, 665-668. Nash, D.B. 1993. A case for Na2S on Io's surface: sulfide volcanism? (abstract) in Io: An International Conference, San Juan Capistrano Research Institute, San Juan Capistrano, CA., 75-76. Nelson, R.M., W.D. Smythe, B.W. Hapke and A.J. Cohen 1990. On the effect of X-rays on the color of elemental sulfur: Implications for Jupiter's satellite Io. Icarus 85, 326-334, Simonelli, D.P., J. Veverka, and A.S. McEwen 1997. Io: Galileo evidence for major variations in regolith properties. Geophys. Res. Lett. 24, 2475-2478. Smith, B.A., L.A. Soderblom, T.V. Johnson, A. Ingersoll, S.A. Collins,. E.M. Shoemaker, G.E. Hunt, H. Masursky, M. Carr, M.E. Davies, A.F. Cook, J. Boyce, G.E. Danielson, T. Owen, C. Sagan, R. Beebe, J. Veverka, R. Strom, J. McCauley, D. Morrison, G. Briggs, and V.E. Soumi 1979a. The Jupiter system through the eyes of Voyager 1. Science 204, 951-972. Smith, B.A., L.A. Soderblom, R. Beebe, J. Boyce, G. Briggs, M. Carr, S.A. Collins,. A.F. Cook, G.E. Danielson, M.E. Davies, G.E. Hunt, A. Ingersoll, T.V. Johnson, H. Masursky, J. McCauley, D. Morrison, T. Owen, C. Sagan, E.M. Shoemaker, R. Strom, V.E. Soumi, and J. Veverka 1979b. The Galilean satellites and Jupiter: Voyager 2 imaging results. Science 206, 927-950. Smythe, W. D., R. M. Nelson, and D. B. Nash, 1979. Spectral evidence for SO2 frost or adsorbate on Io's surface. Nature, 280, 766. Soderblom, L.A., K. Edwards, E. Eliason, E. Sanchez, and M. Charette 1978. Global Color Variations on the Martian Surface. Icarus, 34, 446-464. Soderblom, L., T. Johnson, P. Kupferman, D. Pieri, D. Morrison, E. Danielson, B. Smith, J. Veverka, C. Sagan, and A. Cook 1980. Spectrophotometry of Io - Preliminary Voyager 1 results. Geophysical Research Letters, 7, 963-966. Spencer, J.R. and N.M. Schneider 1996. Io on the eve of the Galileo mission. Annu. Rev. Earth. Planet. Sci., 24, 125-190. Spencer, J.R., A.S. McEwen, M.A. McGrath, P. Sartoretti, D.B. Nash, K.S. Noll, and D. Gilmore 1997. Volcanic resurfacing of Io: Post-repair HST imaging. Icarus 127, 221-237. Steudel, R., G. Holdt and A.T. Young 1986. On the colors of Jupiter's satellite Io: Irradiation of solid sulfur at 77 K. J. Geophys. Res. 91, 4971-4977. Torson, J.M. and K.J. Becker 1997. ISIS - A Software Architecture for Processing Planetary Images (abstract). Lunar and Planet. Sci. 28, 1443. (See also http://wwwflag.wr.usgs.gov/isis-bin/isis.cgi). Zahnle, K., L. Dones, and H. F. Levison 1999. Cratering Rates on the Galilean Satellites. Icarus, in press. FIGURE CAPTIONS Plate 1. Global color of Io in an equal-area cylindrical projection. This mosaic of low phase angle images from Galileo's 2nd, 4th, 6th and 9th orbits is a false-color composite made up of NIR (0.76 micron), green and violet filter images. The top frame is linearly contrast-stretched from 0 to the maximum digital number, and approximates the visual appearance of Io in 1996 - 1997. At bottom is an enhanced color representation, for which each of the NIR, green and violet filter mosaics have been stretched individually. Plate 2. North and South polar projections of Io's color. Plate 3. Comparison between Voyager 1 and Galileo views of Io. (Top) The face of Io in 1979 is shown in this orange, green and violet mosaic of low phase angle Voyager 1 images from McEwen (1988). (Bottom) Io in 1996 - 1997. Synthetic "orange", green and violet composite from Galileo SSI red, green and violet filter images. Despite the vigorous volcanic activity and local resurfacing that has taken place over the 17 year interval between the two sets of spacecraft observations, little alteration of the global color and albedo distributions has occurred. Figure 1. Map of the large-scale changes in the visible appearance of Io from the Voyager flybys to the end of Galileo's nominal mission. Only about 10% of Io's surface has markedly changed in color or albedo since 1979. Pele's changes mostly occurred between the two Voyager flybys. Figure 2. Sites sampled for the spectral measurements shown in Figures 3, 4, 7 and 8. These data were extracted from high phase angle images acquired during orbit 10. In each case, measurements were obtained from the largest possible rectangular regions within homogenous areas. Figure 3. Spectra of the red materials at 4 to 5 degrees phase. Red and orange materials on Io are characterized by steep positive slopes from the violet to the red filters, slightly less steep from red to the NIR, and shallow positive slopes at longer wavelengths. The spectral feature that distinguishes red materials from the other color classes is their relatively steep slope between green and NIR wavelengths. Figure 4. Spectra of selected yellow materials. Yellow materials display steep positive slopes from the violet to the green, but have a pronounced inflexion to much shallower slopes between green and red, grading to still shallower slopes at wavelengths beyond red. The yellow materials are distinguished by the sharp change in spectral slope at green wavelengths. Figure 5. Greenish-yellow regions. Among the surprises in this high resolution 968-green-violet false color composite of the Prometheus region of Io are several small, greenish spots up to 100 km in length. Numbers refer to the spectra shown in Figure 6. The area pictured is 1780 km by 1450 km, centered on 6 S, 174 W. Figure 6. Spectra of small greenish spots imaged at 3 km/pixel during orbit 14. These previously unknown features reach maximum reflectance in the green or red filter passbands, and have negative spectral slopes in the near-infrared. Figure 7. Spectra of the white materials. Io's brightest materials are actually pale pink or yellow at visible wavelengths. The distinguishing spectral features of these SO2-rich regions are their relatively shallow slopes at short wavelengths and their high albedos in the violet filter. They display shallow positive slopes in the infrared, similar to those of the yellow and red units, and reach maximum normal reflectances at 1 micron. Figure 8. Spectra of the dark materials. Distinguished by their low reflectance relative to their surrounds, dark materials display a diversity of colors and an array of albedos ranging from 0.1 to 0.5. An apparent absorption can be seen at 0.89 microns which is absent from the red, yellow and white units. Figure 9. NIR/green color ratio image showing distribution of red-orange materials (bright regions). Red-orange materials dominate the polar regions of Io and are also found in discrete deposits at lower latitudes, usually associated with high temperature, low albedo volcanic centers which show significant change between the Voyager and Galileo eras. The most prominent example is Pele: a ring of bright red pyroclastic deposits encircles the vent, with a minimum diameter of over 500 km reaching outwards to at least 1100 km. Figure 10. Color ratio image showing the distribution of yellow-green materials (bright regions), made up of the green filter image divided by the sum of the violet and NIR filter images. Yellow-green materials dominate the equatorial region of Io, up to latitudes of +/- 30 degrees, and occur in localized concentrations at higher latitudes (especially within or adjacent to bright, presumably SO2-rich basins). Figure 11. Violet/green ratio image showing the distribution of grey and white materials (bright regions). Grey-white spectral units on Io correspond to areas of abundant, coarse to moderate grained SO2, as mapped by the Near-Infrared Mapping Spectrometer (NIMS) instrument on Galileo (Carlson et al. 1997). Figure 12. Distribution of the dark materials, mapped on the basis of local albedo contrasts with their surrounds. Figure 13. 0.889 micron "band-depth map", computed by dividing the sum of the NIR and 1-mm filter images by the 0.889 micron filter image. The result for the Kanehekili region (right), observed in orbit C10, is compared with the corresponding NIR image (left). Regions with relatively deep 0.889 micron absorptions appear bright in the ratio, and can be seen to correspond to dark spots at Kanehekili, Hi'iaka, Shamshu and several dark areas in the vicinity of Masubi. The area pictured is 1710 km by 1950 km, centered on 20 S, 60 W. Figure 14. Composite global color unit map. Approximately 40% of the mapped surface is covered by yellow materials (light gray), slightly concentrated on the trailing hemisphere. Red materials (dark gray) make up about 32%, primarily confined to latitudes above +/- 30 degrees. 27% of the area is covered by the white spectral unit (shown as white), and little more than 1% is occupied by dark materials (black). About 3.7% of the surface was left unclassified, mostly because of no coverage (at the poles) or indeterminate unit membership. Figure 15. Laboratory spectra of some sulfur compounds suggested as possible compositional analogs of the red, yellow and white materials. Sources: Nash et al., 1980 (SO2, extrapolated to 1 micron); Nash, 1993 (Na2S, extrapolated to 1 micron); Clark et al., 1993 (sulfur). Figure 16. Laboratory spectra of silicates (olivines and pyroxenes) with negative near-infrared spectral slopes similar to the green materials. Source: Clark et al. (1993). Figure 17. Laboratory measurements of magnesium-rich silicates similar in spectral reflectance to Io's dark spots. Source: Clark et al. (1993). Figure 18. Other compositional candidates having deep absorptions at 0.9 microns. Sources: Grove et al. 1992 (siderite); Clark et al. 1993 (others). Image Picture Filter Phase Latitude Longitude Resolution La Scale Number Number Angle Range Range (km/pixel) Factorb global mosaic 359986578 G2I0073 GREEN 4.08 -89.37 90.00 84.56 264.56 4.928 0.74 359986600 G2I0074 RED 4.07 -89.37 90.00 84.58 264.58 9.856 0.74 359986604 G2I0075 VIOLET 4.08 -89.37 90.00 84.59 264.59 9.855 0.74 359986607 G2I0076 7560 4.07 -89.37 90.00 84.59 264.59 9.855 0.74 374575800 E4I0015 GREEN 0.50 -90.00 89.30 45.00 225.00 5.889 0.78 374575845 E4I0016 GREEN 0.47 -90.00 89.30 45.06 225.06 5.884 0.78 374575922 E4I0018 VIOLET 0.48 -90.00 89.30 45.15 225.15 11.754 0.78 374575945 E4I0019 7560 0.47 -90.00 89.30 45.18 225.18 11.749 0.78 374575968 E4I0020 9680 0.47 -90.00 89.30 45.21 225.21 11.745 0.78 374576000 E4I0021 8890 0.47 -90.00 89.30 45.24 225.24 11.740 0.78 383758500 E6I0060 GREEN 13.89 -89.97 89.97 -174.23 5.77 11.251 0.62 383758504 E6I0061 RED 13.89 -89.97 89.97 -174.23 5.77 11.251 0.62 383758507 E6I0062 VIOLET 13.90 -89.97 89.97 -174.23 5.77 11.252 0.62 383758511 E6I0063 7560 13.90 -89.97 89.97 -174.23 5.77 11.252 0.62 401740700 C9I0005 GREEN 4.50 -89.90 90.00 -42.47 137.53 20.987 0.74 401740704 C9I0006 RED 4.50 -89.90 90.00 -42.47 137.53 20.986 0.74 401740707 C9I0007 VIOLET 4.51 -89.90 90.00 -42.47 137.53 20.986 0.74 401740711 C9I0008 7560 4.51 -89.90 90.00 -42.46 137.54 20.985 0.74 C10 6 color 413570400 10I0010 GREEN 36.42 -12.83 89.75 -26.34 153.66 10.688 0.40 1.3 413570407 10I0011 RED 36.41 -10.85 89.75 -26.33 153.67 10.687 0.40 1.3 413570900 10I0012 VIOLET 36.04 -16.14 89.75 -26.00 154.00 10.634 0.40 1.3 413570907 10I0013 7560 36.03 -21.45 89.75 -25.99 154.01 10.632 0.40 1.3 413571400 10I0014 9680 35.66 -12.36 89.75 -25.67 154.33 10.579 0.40 1.3 413571407 10I0015 8890 35.65 -9.29 89.75 -22.00 154.34 10.577 0.40 1.3 413744178 10I0025 GREEN 60.97 -89.69 90.00 -221.08 -55.47 5.138 0.15 1.4 413744200 10I0027 RED 60.94 -89.69 90.00 -235.05 -55.45 10.277 0.15 1.4 413744204 10I0028 VIOLET 60.94 -89.69 90.00 -235.18 -55.44 10.278 0.15 1.4 413744207 10I0029 7560 60.94 -89.69 90.00 -235.18 -55.44 10.279 0.15 1.4 413744645 10I0030 9680 60.97 -89.69 90.00 -234.84 -54.84 10.353 0.15 1.4 413791001 10I0035 VIOLET 74.32 -89.73 1.97 -182.02 -2.03 10.382 0.00 1.6 413791046 10I0036 9680 74.34 -89.73 -2.28 -181.90 -2.42 10.389 0.00 1.6 413791069 10I0037 8890 74.37 -89.73 -1.62 -181.89 -15.70 10.395 0.00 1.6 413791545 10I0038 GREEN 74.56 -89.73 -6.69 -181.38 -66.74 10.454 0.00 1.6 413791600 10I0039 RED 74.58 -89.73 -3.97 -181.39 -1.90 10.461 0.00 1.6 413791623 10I0040 7560 74.60 -89.73 -2.35 -181.30 -1.81 10.467 0.00 1.6 E14 6 Color 440873539 14I0001 GREEN 35.99 -90.00 12.98 100.90 215.23 2.990 0.40 1.3 440873552 14I0002 RED 35.99 -90.00 12.96 100.94 215.30 2.989 0.40 1.3 440873565 14I0003 VIOLET 35.99 -90.00 12.94 100.98 215.36 2.989 0.40 1.3 440873578 14I0004 7560 35.99 -90.00 12.93 101.01 215.42 2.988 0.40 1.3 440873600 14I0005 9680 35.99 -90.00 12.91 101.05 215.49 2.988 0.40 1.3 440873613 14I0006 8890 35.99 -90.00 12.90 101.08 215.56 2.987 0.40 1.3 440873626 14I0007 GREEN 35.99 -7.04 88.66 109.76 225.22 2.988 0.40 1.3 440873639 14I0008 RED 35.99 -7.05 88.66 109.81 225.28 2.987 0.40 1.3 440873652 14I0009 VIOLET 36.00 -6.86 88.66 108.61 223.50 2.987 0.40 1.3 440873665 14I0010 7560 36.00 -6.86 88.66 108.65 224.96 2.986 0.40 1.3 440873678 14I0011 9680 36.00 -6.87 88.66 108.70 225.02 2.986 0.40 1.3 440873700 14I0012 8890 36.00 -6.88 88.66 108.74 225.09 2.985 0.40 1.3 a Lunar-Lambert limb-darkening correction parameter from McEwen (1991). b The spectra in Figures 3,4,6,7 and 8 were normalized by these multiplicative factors to match the 4 to 5 degree phase global observations. Table 1. Galileo imaging data used in this study.