2209 - Moore, J. M., E. Asphaug, R. J. Sullivan, J. E. Klemaszewski ...

ICARUS

135, 127–145 (1998) IS985973

ARTICLE NO.

Large Impact Features on Europa: Results of the Galileo Nominal Mission Jeffrey M. Moore and Erik Asphaug NASA Ames Research Center, MS 245-3, Moffett Field, California 94035 E-mail: [email protected]

Robert J. Sullivan, James E. Klemaszewski, Kelly C. Bender, and Ronald Greeley Geology Department, Arizona State University, Tempe, Arizona 85287

Paul E. Geissler, Alfred S. McEwen, Elizabeth P. Turtle, Cynthia B. Phillips, and B. Randy Tufts Lunar and Planetary Laboratory, University of Arizona, Tucson, Arizona 85721

James W. Head III, Robert T. Pappalardo, and Kevin B. Jones Geological Sciences Department, Brown University, Providence, Rhode Island 02912

Clark R. Chapman Southwest Research Institute, 1051 Walnut Street, Suite 426, Boulder, Colorado 80302

Michael J. S. Belton National Optical Astronomical Observatory, Box 26732, Tucson, Arizona 85717

Randolph L. Kirk U.S. Geological Survey, 2255 N. Gemini Drive, Flagstaff, Arizona 86001

and David Morrison NASA Ames Research Center, MS 200-7, Moffett Field, California 94035 Received September 16, 1997; revised April 2, 1998

The Galileo Orbiter examined several impact features on Europa at considerably better resolution than was possible from Voyager. The new data allow us to describe the morphology and infer the geology of the largest impact features on Europa, which are probes into the crust. We observe two basic types of large impact features: (1) ‘‘classic’’ impact craters that grossly resemble well-preserved lunar craters of similar size but are more topographically subdued (e.g., Pwyll) and (2) very flat circular features that lack the basic topographic structures of impact craters such as raised rims, a central depression, or central peaks, and which largely owe their identification as impact features to the field of secondary craters radially sprayed about them (e.g., Callanish). Our interpretation is that the classic craters (all ,30 km diameter) formed entirely within a

solid target at least 5 to 10 km thick that exhibited brittle behavior on time scales of the impact events. Some of the classic craters have a more subdued topography than fresh craters of similar size on other icy bodies such as Ganymede and Callisto, probably due to the enhanced viscous relaxation produced by a steeper thermal gradient on Europa. Pedestal ejecta facies on Europa (and Ganymede) may be produced by the reliefflattening movement of plastically deforming but otherwise solid ice that was warm at the time of emplacement. Callanish and Tyre do not appear to be larger and even more viscously relaxed versions of the classic craters; rather they display totally different morphologies such as distinctive textures and a series of large concentric structural rings cutting impact-feature-related materials. Impact simulations suggest that the distinctive morphologies would not be produced by impact into a solid ice

127 0019-1035/98 $25.00 Copyright  1998 by Academic Press All rights of reproduction in any form reserved.

128

MOORE ET AL.

target, but may be explained by impact into an ice layer p10 to 15 km thick overlying a low-viscosity material such as water. The very wide (near antipodal) separation of Callanish and Tyre imply that p10–15 km may have been the global average thickness of the rigid crust of Europa when these impacts occurred. The absence of detectable craters superposed on the interior deposits of Callanish suggests that it is geologically young (,108 years). Hence, it seems likely that our preliminary conclusions about the subsurface structure of Europa apply to the current day.  1998 Academic Press

INTRODUCTION

Voyager images of the surface of Europa revealed several circular features of probable impact origin (Smith et al., 1979, Lucchitta and Soderblom 1982). Lucchitta and Soderblom (1982) classified these features as craters a few tens of kilometers in diameter that display rims, central peaks, and ejecta blankets, and large, flat, darker, and redder circular spots 100 km or larger that served as convergence points for lineaments. Tyre (348N, 1468W) was cited as the best example of an impact feature in the latter category. Malin and Pieri (1986) concurred that Tyre was probably a ‘‘relic impact crater.’’ The arrival of the Galileo Orbiter in the Jupiter system has allowed us to examine several potential impact features on Europa at considerably better resolution and greater spectral coverage than was possible by Voyager. In this paper, we use Galileo Solid State Imaging (SSI) data to evaluate the origin of large, candidate impact structures on Europa by analyzing their morphologies. We also consider what the morphologies of these features imply about the shallow interior of Europa, and we evaluate the potential role of target rheology on impact crater morphology. We make use in this paper the implications of europan crater statistics discussed by Chapman et al. (1998) and Zahnle et al. (1998). OBSERVATIONS

Galileo has observed impact features on Europa in both categories first developed by Lucchitta and Soderblom (1982): (1) impact features that have much in common with ‘‘classic’’ impact craters observed on the Moon and other dry, silicate bodies and (2) features which lack obvious continuous rims or central peaks are very flat at the scale of the whole feature (though show some high frequency relief) and owe their identification as impact features mostly to fields of secondary craters radially sprayed about them. At the time of this writing the average age of the surface of Europa is unknown, but most workers consider the low density of unambiguous impact craters to indicate a relatively youthful surface. Chapman et al. (1998) and Zahnle et al. (1998) estimate its age at 107 –108 years. There

FIG. 1. Govannan, a 10-km-diameter impact crater at 37.58S, 302.68W, exhibits a pedestal-like ejecta facies. This crater exhibits a small central peak and is p300 m deep. Image resolution is 1.2 km/pixel. Illumination is low and from the right. North is up (Galileo image PICNO E4E0001).

are very few impact features on the surface .10 km in diameter. Smaller craters seen by Voyager or Galileo are often imaged at resolutions at which these features subtend only a few pixels. Govannan A few ‘‘classic’’ impact craters (i.e., having morphologies broadly similar to lunar craters) 10 to 30 km in diameter have been observed so far in Galileo SSI images at resolutions allowing morphological analysis. The first example, named Govannan, has a diameter of 10 km and is located in rough but bright terrain at 37.58S, 302.68W (Fig. 1). This crater, observed at 1.2 km/pixel under near-terminator lighting conditions, exhibits an essentially bowl-shaped interior, a central peak about 2 km across, and no obvious terracing. A photoclinometrically derived rim-to-floor depth of p300 m yields a diameter-to-depth ratio of 30, which is approximately four times less than similar-sized craters on Ganymede and Callisto (Schenk 1991). Govannan appears to possess a discrete, scarp-bounded pedestal that extends about 3 to 4 km beyond the rim. Pedestal craters, first identified in Voyager images of Ganymede (e.g., Horner and Greeley 1982), have elevated proximal ejecta blankets with scarp-like, roughly circular outer boundaries. The pedestal-to-crater-diameter ratio (p1.4) of Govannan is close to the mean (1.5) for ganymedan pedestal craters (Horner and Greeley 1982). There are no other detectable ejecta facies associated with Govannan.

LARGE IMPACT FEATURES ON EUROPA

129

FIG 2. The p21-km-diameter ray crater Mannann’an, located at 38N, 2408W. The left image in this figure is a high-sun view of Mannann’an, acquired at a resolution of 1.6 km/pixel during orbit G1 (Galileo image PICNO G1E0003). The E11 observation on the right shows Mannann’an illuminated with the Sun 98 from the eastern horizon and at a resolution of 218 m/pixel (mosaic of Galileo images PICNO E11E0012 and 14). North is up in both pictures.

It is the smallest candidate pedestal crater so far identified on Europa. Mannann’an Mannann’an, a rayed crater located in mottled terrain at 38N, 2408W, was observed at high sun and 1.6 km/pixel during the G1 orbit and at low sun and 218 m/pxl during the E11 orbit (Fig. 2). The G1 imaging revealed a sharp bright elliptical annulus 30 km (north–south) by 23 km, which approximates the location of the crater rim deposits. A 7 km diameter dark ring surrounds a central bright spot. Bright ray material can be identified up to p120 km from the center of the impact. Superposed on the crater rim deposits are a few discontinuous patches of dark material that may be a dark component of the proximal ejecta. The E11 image of Mannann’an revealed a very flattened crater with a rim diameter of 23 km (north–south) by 19 km. Mannann’an has no central peak but does contain within its interior several massifs, the largest of which is about p5 km across and p200 m high (from shadow measurements) and is located approximately half a crater radii from the crater center toward the east. The floor itself appears rough but level. It is difficult to associate any particular floor feature with the location of the central bright spot seen in the high sun, lower resolution G1 image. Local relief on the crater rim interior is p100 m, implying a diameter-to-depth ratio on order of p200. This diameterto-depth ratio is especially striking in light of the typical

value of p11 for similar-sized craters on Ganymede (Schenk 1991), which is a target similar to Europa in composition and surface gravity. There is no apparent terracing along the interior of the rim. A pedestal-like break in slope occurs in the continuous ejecta at p7 km beyond the rim, giving a pedestal-to-crater-diameter ratio of p1.6. The dark patches seen on the proximal ejecta in the G1 image do not appear to correspond to any material unit boundary. There appears to be little exploitation of the local tectonic fabric (as expressed on the modern surface around Mannann’an) during crater growth, implying that the crater excavated into a target which had annealed after the latest preimpact tectonic event. Pywll Crater Pwyll (268S, 2718W) was suspected to be an impact feature on the basis of low resolution Voyager coverage (McEwen 1986). Bright rays of ejecta from this feature were seen in 1.6 km/pixel images of the northern portion of Europa’s antijovian hemisphere obtained during the G1 encounter (Belton et al., 1996). These rays could be traced back to a dark spot in a brighter zone seen in Voyager images. Color global images taken with resolution of 6.9 km/pixel obtained during the G2 encounter revealed further details of an unmistakable ray pattern characteristic of a fresh impact, with some individual rays extending more than 1000 km from the center. The G2 images and a high-sun view of this location taken during the E4 flyby at

130

MOORE ET AL.

FIG. 3. A portion of the 1.2-km/pixel regional coverage of the trailing hemisphere of Europa acquired during orbit E4 showing the extent of Pwyll’s ray system seen at moderately high sun. The center of the feature in this image is a dark spot about 50 km in diameter centered on 268S, 2718W. This scene is roughly 1300 by 1700 km. North is up (mosaic of Galileo images PICNO E4E0001, 02, 03, and 04).

a resolution of 1.2 km/pixel showed a well-defined, circular, dark spot about 50 km across at the center of the ray pattern (Fig. 3). The E4 images also show that Pwyll probably formed in mottled terrain, although it is difficult to be certain because the region around the dark spot appears to be blanketed with bright ray material to distances over 100 km from the dark spot. During the E6 encounter, Pwyll and its environs were imaged under near-terminator illumination with resolutions of 240 and 56 m/pixel. The crater and its surroundings are clearly seen in the 240-m/pixel imaging. Surprisingly, the periphery of the p50-km spot does not correspond to the crater rim or floor but instead to a pedestal-like innermost portion of the continuous ejecta (Figs. 4 and 5). The crater itself is 24–25 km across (rim crest to rim crest), has a very shallow, mostly flat floor, a diameter-to-depth ratio on order of 100, and some terracing along the inner wall. Stereo observations of Pwyll by Galileo are planned for the end of 1997 which will define this value much more

precisely. Within the crater interior is an elongate (p8 by p4 km) massif, offset to the east, somewhat flat-topped to the northeast, and oriented with its long axis pN408E. Local relief on the crater floor massif ranges from p100 to p300 m, with the greatest topography seen in its rugged southwest end. Also on the floor there are three other kilometer-scale, isolated, angular blocks west of the main massif. Most of the floor is smooth with texturing in places caused by equidimensional hills whose size approaches the limit of resolution. The extent of the pedestal-like proximal ejecta closely, but not exactly, matches the area covered by the dark (and ‘‘red’’) spot seen in high-sun imaging (Fig. 4). At 240-m/ pixel resolution this pedestal (Fig. 5) resembles that of pedestal craters on Ganymede such as Achelous (Fig. 6). The pedestal scarp ranges from 30 to 35 km from the crater center, yielding a pedestal-to-crater-diameter ratio that is smaller than, and barely within the maximum measured deviation of, the mean for ganymedan pedestal craters (Horner and Greeley 1982). The pedestal-like portion of the ejecta, crater rim, and crater interior exhibits a nearly uniform low albedo, which may be the result of these surfaces all being composed of, or else coated with, the last and deepest excavated ejected material. The outer edge of Pwyll’s ejecta pedestal is defined by an outwardfacing scarp that, in places, appears convex upward. The bounding scarp is often lobate in plan. Collectively, the continuous ejecta (both within and beyond the pedestal scarp) predominantly exhibit subradially braided texture. The texture of the facies within the scarp

FIG. 4. The combination of moderate resolution (1.2 km/pixel) albedo-dominated (acquired at moderately sun) E4 image and high-resolution (240 m/pixel) low-sun E6 image of Pwyll. Note that the periphery of the p50-km dark (and from G2 color analysis ‘‘red’’) spot does not correspond to the crater rim or floor but instead to a pedestal-like proximal ejecta facies (also see Fig. 5). North is up (Albedo from Galileo image PICNO E4E0002 and shading from images E6E0030 and 31).

LARGE IMPACT FEATURES ON EUROPA

131

FIG. 5. A low-sun, 240-m/pixel view of the crater Pwyll and its surroundings. The crater is 24–25 km across (rim-to-rim), has a very shallow mostly flat floor, and, in places, some terracing along the inner wall. Within the crater interior, but offset from the center, is a roughly rectangular (p8 by p4 km) and somewhat flat-topped massif. The scarp defining the edge of pedestal-like topography within the proximal ejecta facies can be clearly seen (also see Figs. 7 and 8). North is up. Illumination is low and from the right. This is a Pwyll-centered orthographic projection (mosaic of Galileo images PICNO E6E0030 and 31).

boundary may additionally exhibit some fracture-like intersecting troughs, where seen at 56 m/pixel (Figs. 7 and 8). However, there are no obvious dune-like hummocks within this zone or elsewhere within the continuous ejecta as is so often seen in proximal lunar ejecta. Beyond the pedestal scarp the continuous ejecta facies show a few discrete flow-like features such as narrow, low-relief, radially lineated lobes (Figs. 7 and 8). However, ‘‘dry’’ continuous ejecta around very young lunar craters similar in size to Pwyll (e.g., King, Tycho) also exhibit flow-like features, such as lobate outward-facing scarps and lineations that veer in response to preexisting topography, giving the appearance of streamlining. The transition from the outer, bright continuous ejecta to a surface pitted by secondaries appears fairly abrupt (Figs. 7 and 8). Many secondaries coalesce to form chains

oriented subradially to the impact site. Closely spaced secondaries from p1 km diameter down to the limit of resolution can be seen disrupting the surface as far as some 80 km beyond the crater center, beyond which point 56-m/ pixel coverage terminates. Probable Pwyll secondaries up to p500 m in diameter are seen as far as 1000 km from the impact. These probable secondaries occur in clusters within Pwyll’s rays in high-resolution images of the Conamara Chaos region (centered 108N, 2718W) taken during the E6 encounter. Callanish On December 19, 1996, the Galileo spacecraft made its first close encounter (on orbit E4) with Europa. During this encounter three images were taken of a feature located at 168S, 3348W, identified on low-resolution (6.9 km/pixel)

132

MOORE ET AL.

FIG. 6. Achelous Crater, Ganymede (628N, 128W). This is a 34-km-diameter pedestal crater imaged at low sun and high resolution (172 m/pixel). Note that like Pwyll, the texture of the continuous ejecta grossly resembles that of lunar craters. The pedestal is simply a convex-outward break in the slope within this deposit. The inset was processed to accentuate the topography of the pedestal scarp. Lighting is from the right. North is up (Galileo image PICNO G7G0020).

images taken earlier in the mission (orbit G2) as a p100 km diameter low albedo spot upon which several lineaments converged. The low altitude of the E4 encounter permitted high-resolution (120 m/pixel), low-sun imaging of the southern portion of the feature (Fig. 9). These images revealed that the low albedo spot, named Callanish, was a genetic but heterogeneous suite of landforms and deposits. Morphologically, Callanish can be divided into two

zones and several other associated features (see map, Fig. 10). The inner zone is approximately 50 km in diameter and is characterized by a rugged-textured surface (rough inner unit on map). Near the center of this zone is a 10- to 15-km, slightly higher albedo feature composed of radially arrayed lobes about a central depressed annulus (bright central lobate unit on map). Nested across the boundary of the two zones are a number of elongate concentric

FIG. 7. The 240-m/pixel coverage of the region south of the center of Pwyll and its environs with a strip of high-resolution (56 m/pixel) images covering a portion of this scene, the eastern half of which is shown in Fig. 8. Note the large, tilted, tabular blocks and their associated detachment moats in the far western part (240 m/pixel coverage) of the figure. North is up. Illumination is low and from the right. This is a Pwyll-centered orthographic projection (moderate-resolution Galileo images PICNO E6E0030 and 31; high-resolution images E6E0060, 61, 62, 63, 64, and 65).

LARGE IMPACT FEATURES ON EUROPA

133

FIG. 8. High-resolution (56 m/pixel) images of the proximal ejecta facies of Pwyll seen just south to southwest of the crater. Collectively, the continuous ejecta (both within and beyond the pedestal scarp) predominantly exhibits subradially braided texture. The texture of the facies within the scarp boundary may additionally exhibit some fracture-like intersecting troughs. Beyond the pedestal scarp, the continuous ejecta facies shows a few discrete flow-like features such as thin, narrow, radially lineated lobes. This continuous ejecta facies fairly abruptly ends exposing a surface gouged by numerous secondary impacts from Pwyll. North is up. Illumination is low and from the right (mosaic of Galileo images PICNO E6E0063, 64, and 65).

massifs. The zone encircling the central rough zone (smoother outer flow unit on map) is characterized by very small, finely textured equidimensional hills, where not disrupted by subsequent tectonics. The outer zone is distinctly darker and redder than the inner zone. The material of the outer zone appears to pond in low areas and appears to have been emplaced in a fluidized state, perhaps as a slurry. This zone is broken by several large, concentric, continuous troughs (typically p800 m wide), additional smaller troughs, and numerous fractures. The simple troughs coalesce into a complex multiple-terrace trough up to p4 km wide along the southeast periphery of Callanish. The troughs are interpreted to be tectonic in origin and may be graben. Crosscutting/superposition relationships between the tectonic features and the deposits indicates that most of the deposits were in place prior to tectonization (e.g., graben scarps cut most superjacent deposits). There is some indication, however, that a small amount of the material of the outer zone (smoother outer flow unit) was still mobile following the formation of the troughs. This implies that the troughs did not form at the time of impact but probably soon afterward. Just outside Callanish to the southwest are two undulatory 10-km depressions. Two prominent ridges intersect Callanish, one from the WNW, the other from the ESE. Both ridges are modified in the portions that cross the outer zone of Callanish and disappear at its inner boundary, indicating these ridges either predate the formation of Callanish or that their formation was affected by the presence of Callanish. Additionally, there are numerous small pits, often with raised

rims, found in the area surrounding Callanish, some of which form pit chains oriented radially to the center of the structure. These pits are interpreted to be secondary impact craters. It is the presence of these secondaries that permits an unambiguous interpretation of Callanish as an impact feature. Photoclinometrically derived topography (verified with shadow measurements) of Callanish shows the feature to be regionally flat (Fig. 11). The low-sun geometry of these images minimized the effects of varying photometric properties of the surface, and so shape strongly controlled shading. The photoclinometric technique used was that of Kirk (1987), which iterates between the shaded model and the actual scene to converge on a solution. There is no evidence for a present-day depression associated with this impact feature. Maximum relief across the high spatial frequency components of Callanish is p100 m, save the large concentrically oriented massifs which show relief approaching 200 m. The depths of the concentric troughs are p100 m. The surfaces beyond the rims of these troughs appear to ramp up gradually (starting several kilometers outside the troughs) to reliefs of more than 50 m at the rims. These raised-trough rims may be the complement to the downwarping of the surfaces along the flanks of large, steepsided ridges seen in the vicinity of Callanish and elsewhere on Europa (Greenberg et al., 1998). Tyre Seen at Voyager resolution (p2 km/pixel) and very low sun, Tyre is a circular low-albedo patch with little morpho-

134

MOORE ET AL.

FIG. 9. Southern portion of Callanish (168S, 3348W) imaged at 120 m/pixel. North is up. Illumination is low and from the right. Compare it with the geologic map (Fig. 10) and see text for discussion (mosaic of Galileo images PICNO E4E0016 and 17).

logic character (Fig. 12). Several concentric rings can be seen to have little relief in comparison to the N–S trending ridges crosscutting them. As Fig. 12 illustrates, Tyre and Callanish resemble one-another at comparable resolutions and lighting. Tyre, like Callanish, serves as the nexus of converging lineaments and ridges. It was a target for high-

sun (phase angle 178), three-filter (404, 559, 986 nm) imaging at 570 m/pixel by SSI during the G7 orbit of Galileo. The Galileo G7 images reveal that the appearance of Tyre is dominated (at this resolution and lighting) by approximately a dozen concentric narrow rings (Fig. 13). The outermost ring diameter is p125 km and the innermost

FIG. 10. Geologic map of the materials related to the Callanish shown in Fig. 9. Tectonic structure is not mapped. North is up. Details of the material units are discussed in the text.

LARGE IMPACT FEATURES ON EUROPA

135

directions. The dots, typically p0.5–1 km in diameter, become less numerous with distance from Tyre and often form chains that are radial to the center of Tyre. These dots are interpreted to be the sites of secondary craters created by the Tyre-forming impact. Several of the narrow rings seen in Galileo G7 images of Tyre correspond to ridges seen in low-sun p2-km/pixel Voyager 2 views (Fig. 12). Low-sun Galileo images of Callanish resampled to .1 km/pixel produce a view of its circumferential features that can be compared with the Voyager views of Tyre. Some of Callanish’s troughs now have a ridge-like appearance. This is due both to the pixeldominating brightness of the sun-catching walls and the rise of trough-flanking ramps. Callanish’s inner annulus of massifs (interior to its troughs) produces the most prominent ridges on that feature. The Voyager images of Tyre indicate that a similar inner annulus of massifs must be less prominent than that of Callanish. Tyre appears to have little regional relief at Voyager resolutions, indicating that this feature, like Callanish, has no great central basin. Highresolution (p160 m/pixel), low-sun imaging of Tyre is planned for early 1998, which should greatly improve our knowledge of its morphology and topography. FIG. 11. Topography of the southwestern portion Callanish region derived from photoclinometry of the 120-m/pixel coverage using the method of Kirk (1987). The top figure is a gray-scale Digital Terrain Model (DTM) of this area; elevations .75 m above the mean are saturated, and those .60 m below the mean are black. The appearance of a shallow central p10-km-diameter depression (upper-right of figure) is suspect, as this area has a different intrinsic albedo which can cause shape-fromshading algorithms to produce erroneous results. The original image used to make this DTM contained data drop-out that resulted in the east–west bands of low resolution running across the center of the scene. The middle plot is an east–west topographic profile across the center to the western edge of the DTM. The lower figure shows the location of the profile on the original image.

recognizable ring has a diameter of p30 km. The ring spacing is typically between 5 and 7 km. In the center of Tyre there is a small (p14 km across) discrete deposit of relatively higher albedo material. The individual rings are usually very narrow (p1 km) single strips, though they are sometimes doubled (e.g., in the northwest of Tyre). The rings are similar in color to, but darker than, the surrounding plains. Unlike the younger lineae that crosscut them, the rings do not appear compositionally distinct from the ice they intrude. These rings are interpreted to be troughs, similar to those surrounding Callanish. The spaces between the rings are often filled with discrete deposits of darker, redder (compositionally distinct) material. There is a particularly large deposit of this darker, redder intraring material forming a belt p5 km wide, and extending some 30 to 40 km from the center of Tyre. Beyond this main belt of dark red material is a spray of small dots in all

FIG. 12. Tyre (top) as seen by Voyager 2. The left view has a resolution of p1.9 km/pixel and the right has a resolution of 2.2 km/pixel. In both images lighting is low and from the left. Callanish (bottom), shown for comparison, was resampled at 1.2 km/pixel. Lighting is low and from the right. North is up (Voyager 2 images FDS 20649.37 and 20652.11; Galileo images E4E0015, 16, and 17).

136

MOORE ET AL.

morphology of Pwyll appears to offer conflicting evidence for the nature of the target in which it formed. We note two hypotheses for Pwyll: (1) Pwyll formed in effectively solid ice and (2) it formed in ‘‘thin’’ ice overlying a deep ocean. Our approach has not been to interpret these interior models in any detailed way, other than to evaluate to what degree they are consistent with morphological, gravitational, and multispectral observations.

FIG. 13. Tyre (centered 348N, 146.58W) This high-sun view was acquired in the 404-, 559-, and 986-nm filters at a resolution of 570 m/pixel, which were combined to make this image. The feature is dominated by p12 narrow concentric rings and several broader bands of darker (and ‘‘redder’’) material. The diameter of the main band of darker material is p70 km. The outermost ring has a diameter of p125 km. Beyond this main belt of dark red material is a spray of small dots in all directions. The dots are typically p0.5–1 km in diameter, become less numerous with distance from Tyre, and often form chains that are radial to the center of Tyre. These dots are interpreted to be the sites of secondary craters created by the Tyre-forming impact. North is up (Galileo images G7E0040, 42, and 44).

DISCUSSION, MODELING, AND SPECULATION

The detailed views provided by Galileo of five impact features $10 km in diameter reveal that these features exhibit surprising variation. These variations, moreover, are not merely among features of very different sizes but even among those whose sizes are similar. Impact features on Europa change with increasing size from a more-orless ‘‘classic’’ crater morphology, such as those of Pwyll and Mannann’an, to something which can only be interpreted as an impact feature by the presence of associated secondaries, such as Callanish and Tyre. We begin by proposing hypotheses for the origins of these features and related landforms, then we test them against available data, physical modeling, and/or compare them with related hypotheses concerning the crustal and subcrustal structure of Europa. Pwyll Formation and Evolution Hypotheses Of the larger ‘‘classic crater’’ features, Pwyll was best observed and will be used to represent its class. The geo-

Pwyll formation in solid ice. Impact into a completely solid, brittle target is the simplest explanation for ‘‘classic’’ crater morphology (e.g., a central massif, distinct and roughly circular raised rim, bright rays, and copious secondaries). However, this explanation presents some problems. If Pwyll formed in a lunar-like fashion in solid ice, how did it acquire a diameter–depth ratio nearly ten times flatter than similar-sized craters on Ganymede? In this scenario, the dark, ‘‘red’’ material associated with the pedestal was emplaced as a solid, even though material with these photometric properties is often associated with endogenic landforms and fluid-rich deposits (Geissler et al. 1998). Large, tilted, tabular blocks (and their associated detachment moats) are seen within 80 km of Pwyll (Fig. 7). The appearance of such blocks is interpreted elsewhere (see Carr et al. 1998, Greeley et al. 1998) as evidence that the solid ice layer at these locations was, at the time of block mobilization, not more than 1 or 2 km thick. This is thinner than the presumed excavation depth (p3 km) of craters the size of Pwyll (e.g., Melosh 1989, p. 78). Pywll formation in a thin ice crust over liquid. Pwyll’s low relief and some of its associated landforms might be explained if the impactor punched a hole through the ice crust, sending ejecta far and wide yet not resulting in significant topographic relief due to the lack of support from the ‘‘thin’’ ice layer. In this hypothesis, the crater floor is a frozen pond and the dark annulus of the pedestal was created by dark fluid excavated from depth mixing with bright, solid ejecta. If this hypothesis were valid, it would have to explain the following: (a) the lack of any circumferential structures around Pwyll, which are expected from impacts into thin-ice layers above subsurface liquids (see subsequent discussion) and (b) the existence of a central massif. A simultaneous solution to (a) and (b) is to postulate that the original crater was actually significantly smaller than 26 km and that the circumferentially fragmented pieces of crust that did form were rafted toward the center of the crater. In this case, the unusually shaped and off-centered central massif at the center of Pwyll would not be a central peak at all, but a rafted portion of the rim. A third hypothesis, that Pywll formed in a thin H2O layer overlying near-surface silicates, appears to be ruled out by E4 and E6 gravity results (Anderson et al. 1997) which indicate that the H2O layer is $100 km thick. We prefer the ‘‘formation in solid ice’’ hypothesis, both on the

LARGE IMPACT FEATURES ON EUROPA

basis of the observations and subsequently discussed modeling. Hypotheses for Callanish and Tyre Formation Although rather less likely (given their association with secondary craters), the possibility that Callanish and Tyre are not impact features will be briefly reviewed. Callanish and Tyre both exhibit some of the features of venusian coronae and terrestrial ice cauldrons. Coronae are a class of features on Venus typified by a concentric annulus of tectonic features (e.g., Pronin and Stofan 1990). Generally circular in planform, coronae on Venus range in size from 60 to 2600 km in diameter. The tectonic annulus may comprise extensional features, compressional features, or a mix of the two. The annulus width varies greatly, but generally widens with increased overall corona size. The interiors of most coronae on Venus are typically smooth plains (volcanic deposits), which represent either the preexisting surface or new flows associated with corona formation. Volcanic flows originating at these tectonic features, such as annular fractures, are associated with many coronae. The topographic expression of coronae includes domes to plateaus, plateaus with interior lows, and rimmed depressions. Most workers agree that coronae on Venus are the surface manifestations of mantle plumes or diapirs (Stofan et al. 1992, Squyres et al. 1992, Janes et al. 1992). An alternative nonimpact hypothesis was proposed by Wood (1981) and Thomas (1997); both suggested that multiringed features like Tyre were formed analogously to ice cauldrons. Ice cauldrons on Earth form under ice sheets when localized subglacial heating forms a domeshaped reservoir of liquid water melted at the base of the ice. Eventually, the water pressure exceeds the ice overburden pressure and the water drains from the reservoir causing the surface above it to sag, creating concentric fractures, and forming the ice cauldron. While Callanish and Tyre both exhibit some of the features of venusian coronae and ice cauldrons, these nonimpact processes would not produce the fields of radially arrayed pits or dots around Callanish and Tyre interpreted by us to be secondary craters. Also, both nonimpact processes produce large-scale topography (e.g., broad raised rims, central rises, or broad central depressions), whereas Callanish and Tyre exhibit no detectable regional topography. If Callanish and Tyre were formed by impact as we very strongly suspect, we must then reconcile the landforms of these features with that process. The nearest analogous impact feature on an icy satellite may be the palimpsests seen on Ganymede. Initial Voyager-based work characterized palimpsests on Ganymede as roughly circular, highalbedo patches or spots with little or no relief (Smith et al. 1979). Their centers are usually smooth but may become texturally rough around their peripheries. Palimpsests are

137

generally thought to represent impact ‘‘scars’’ whose morphology either (a) developed by viscous relaxation of an original impact crater with initially ‘‘classic’’ morphology (e.g., Passey and Shoemaker 1982) or (b) formed as is, due to unusual (nonbrittle) target properties, with little change since the impact event (e.g., Greeley et al. 1982). Alternatively, it has been proposed that they could be of endogenic origin (e.g., Squyres 1980), but the predominant interpretation of these features remains exogenic. Work by Greeley et al. (1982), Fink et al. (1984), Croft (1984), and Schenk (1996) using Voyager images and Jones et al. (1997) examining images of palimpsests taken on Galileo orbits G1 and G2 favor the present appearance of palimpsests as representing the original morphology in which a fluidized slurry or a slurry with large solid chunks was ejected at the time of impact. In this hypothesis, palimpsests form having no rim or else a highly modified and subtle one, an inner fill, and a peripheral unit of continuous ejecta, beyond which only secondary craters can be seen. The outer edge of this peripheral unit coincides with the edge of the high-albedo spot that delineates a palimpsest viewed in high sun (Fig. 14, top). A comparison of Callanish with Buto Facula on Ganymede (centered 138N, 2038W), a palimpsest imaged during orbit G8 at 182 m/pixel and at low sun, shows that both features are regionally very flat, do not exhibit classic impact crater morphology, and owe their interpretation as impact features to their association with probable fields of secondary craters. However, some significant differences between the principal deposits of the two impact features can be discerned. The main interior deposit of Buto is very flat save for scattered sub-kilometer-scale blocks embedded within it (Fig. 14, top, see arrows), whereas the corresponding deposit within Callanish (rough inner unit in Fig. 10) is very rough at the subkilometer scale. Buto’s central smooth deposit is contained within a shallow scarp-enclosed depression, whereas the outer contact of Callanish’s rough inner unit is recognized by an abrupt change in texture (and crosscutting of preexisting ridges) with no change in general elevation. Buto’s smooth peripheral unit of ejecta mantles, or drapes, preexisting topography in such a way as to allow that topography to still be discerned. This unit develops a ropy texture composed of quasi-concentric outward-facing lobes as it approaches its well-defined outer contact (Fig. 14, bottom). In contrast, Callanish’s corresponding unit (smoother outer flow unit in Fig. 10) embays preexisting topography, ponding in local topographic lows. The differences between the two features may be due to the relative amount of liquid and overall crustal strength of the respective target areas at the time of impact. The presence of a central, scarp-enclosed depression in Buto implies that the crust was strong enough to support this topography immediately after impact, as this depression was in place in time to contain a deposit which was em-

138

MOORE ET AL.

placed in a mobile state. Buto’s proximal ejecta appears to have incorporated enough solid material to mantle preexisting topography, rather than merely ponding in local lows, and to form and retain a lobe-like texture as it slowed to a halt. Callanish’s proximal ejecta appears to have been composed of enough liquid to drain into local lows, unable to retain any texture other than that imparted by the few blocks embedded within it. On Callanish, the termination of the preexisting ridges at the contact with the rough inner unit is interpreted to be the perimeter within which all preexisting landforms within the target were destroyed. The diameter of the rough inner unit (p50 km) is taken to represent the equivalent of the crater rim. The transient crater may have been p30 km across (see scaling in McKinnon and Schenk 1995). The overall flatness of this interior unit may be due to the total lack of crustal strength within the target. A possible explanation for this unit’s texture and topography is that it was initially a pond covered with floating debris. It has been observed that craters bring to the surface material from a depth equal to approximately one-third their transient diameter (Melosh 1989, p. 78). If this applies to Callanish, then the maximum depth at which the target became mostly liquid would have been on the order of 10 km. Models of Impacts into Europa Two impact simulations were performed for the study of cratering on Europa in order to explore the differences between impacts into a solid-ice target versus liquid targets covered by an ice crust. We used a 2D Lagrangian shock physics hydrocode which models brittle solids using Hooke’s law together with a fragmentation algorithm (Benz and Asphaug 1995) for strains beyond the elastic limit. This ‘‘early time’’ impact simulation code is described by Asphaug et al. (1996) and Asphaug (1997), where it was applied to cratering and disruption of rocky targets. For ice, the parameters governing fracture have not been established, and estimates for dynamic tensile strength vary by an order of magnitude. We have derived dynamic tensile strength constants for use in our code from published data by Lange and Ahrens (1983); the quantitative interpretation of these data remain the subject of some uncertainty (B. Ivanov and T. J. Ahrens, personal communication), another uncertainty involves the scaling of these strength

constants to size scales well beyond those attainable in the laboratory (Asphaug et al. 1996). We are interested mainly in the hydrodynamical evolution in either scenario, rather than the specifics of fragmentation. Because our simulations are well inside the gravity regime (where the particular failure model is not so important when considering excavation and ejection), the emplacement of velocities by the shock (bowl evolution) is better constrained by our modeling than the mode and locality of cracks. To gain some insight into the impact process for two candidate scenarios, we modeled a solid-ice target and a solid-ice layer 6 km thick overlying liquid water. This thickness is not an a priori constraint on the depth of the europan lithosphere, but rather a value which allows for bolide penetration so that we can explore two uniquely different scenarios, the hope being to understand whether the Pwyll/Callanish distinctions can be simply understood in terms of impactors totally penetrating or partially penetrating the europan crust. In both cases, the grid is a 2D axisymmetric cylinder 50 km in radius and 50 km deep. These bottom and side boundaries are artificial, since Europa probably ought to be represented as a half-space for impacts of this scale; for this reason, we terminate our calculations before reflection of the impact shock wave from these boundaries begins to influence the flow. A 1.14km diameter ice sphere impacts each target at 10 km/sec; in the gravity regime this produces a transient cavity p14 km diameter, which is a reasonable estimate for the diameter of the transient cavity of Pwyll. This impact speed is slower than expected on Europa, but leads to a more stable calculation; specifics of the impact speed are lost during the early-time coupling of projectile to target (Holsapple and Schmidt 1987). Figure 15a shows both of these initial states; the impactor is represented as a single vertex in the hydrocode grid. The second frame (Fig. 15b) shows pressures (dyne/cm2) in cross-section 16 sec after impact, gray-scale-coded logarithmically. The faster sound speed in the solid-ice layer propagates the stress wave more rapidly than in the liquid-water layer, resulting in a very different shock morphology in the two target structures. In particular, the sound speed is significantly greater in ice than in water. The acceleration behind the shock is also greater in ice, in part because of the sound speed difference, and in part because of its greater nonlinearity due to phase

FIG. 14. The palimpsest Buto Facula, Ganymede (centered 138N, 1038W). The top half of this figure is a 180-m/pixel Galileo mosaic of this palimpsest illustrating the relatively smooth deposit contained within an inward-facing scarp (arrows) and the extensive continuous ejecta deposit, which corresponds to the sharply bounded high-albedo annulus seen in the underlying high-sun 1.3-km/pixel Voyager context image. In the Galileo images, lighting is low and from the right. North is up (Galileo images PICNO G8G0006 and 7; Voyager 2 image FDS 20635.45). The bottom half of this figure illustrates details of Buto ejecta. Buto’s smooth peripheral unit of ejecta mantles or drapes preexisting topography in such a way as to allow buried topography to still be discerned. It developed a ropy texture composed of quasi-concentric outward-facing lobes as it approached its outer margin (see sketch map at lower right for location of these features). In contrast, Callanish’s corresponding unit (see Fig. 9, and find smoother outer flow unit in Fig. 10) embays preexisting topography with ponding in local topographic lows.

LARGE IMPACT FEATURES ON EUROPA

139

140

MOORE ET AL.

transformation and fracture damage. The third frame (Fig. 15c) shows particle speeds at the end of the simulation (256 sec). By this time the shock has reflected off the back of the target boundary (an artifact of the simulation space). The difference in speed (Fig. 15c) in the two layers constitutes a shear across the ice/water boundary, which may lead to disruptive effects far from the crater bowl, consistent with what one sees around impact structures such as Callanish. The simulation for solid ice (left) shows no such surface shear, and we might therefore expect a lack of circumferential morphology. The fact that Pwyll lacks circumferential/radial fracture, and the fact that it maintains a central peak, supports our conjecture that it formed in an ice layer which was thick compared with the excavation depth of the crater (that is to say, deeper than the p6 km assumed in our simulation). The discontinuity in speeds (the scale bar is in log cm/sec) at the contact boundary in the layered target will almost certainly result in surface rupture away from the crater. Conversely, if Pwyll was created in solid ice (left) we would expect a lack of circumferential morphology (rings). This is consistent with the data, not to mention the existence of a probable central peak. Among the more interesting results of this modeling is that identical impacts produce fewer and collectively slower moving ballistic fragments when the target is ice over water than in a purely solid-ice target. To compare this result with observations, a count was made of the secondaries around Pwyll and Callanish. The area counted was limited to a zone ranging between 60 and 80 km from the center of each impact feature. Only secondaries $1 km across were counted. This large size was chosen because the Pwyll images have a resolution of 240 m/pixel. The Callanish images were degraded to the same resolution. Both impacts produced the same area density of secondaries (Pwyll 22.8 6 2.1 3 103 km2, Callanish 24.0 6 3.5 3 103 km2), given such uncertainties as potential deficiencies in secondary crater recognition due to variations in countsurface textures and the (small) differences in lighting. In light of the hydrocode modeling, this observation can be interpreted as the result of Callanish, the presumably larger impact, having excavated through an ice layer and into a liquid substrate, producing less solid ejecta than would have been derived from a completely solid target. The alternative explanation for the secondary crater densities

is that the impact energy that formed Callanish was very similar in magnitude to that which formed Pwyll and the hydrocode modeling is incorrect in predicting less ejecta from ice-over-water targets. If so, the morphology of Callanish implies that the solid layer at this target was much thinner than that of the Pwyll target, whereas the first explanation is consistent with ice-layer thicknesses at both targets being similar. In a separate numerical experiment, Turtle and colleagues (Turtle et al. 1998) have used the finite-element code TEKTON (see Melosh and Raefsky 1980) to model the collapse of craters on Europa using a technique similar to that described in Turtle and Pierazzo (1998). The finiteelement mesh consisted of a rigid (on the time scale of crater collapse) layer of ice Ih, with a power law rheology (Kirby et al. 1985), over a layer which has a very lowviscosity Newtonian rheology. The formation of a transient crater was approximated by applying Maxwell’s Z model (Maxwell and Seifert 1975) to the mesh to determine what material was ejected in the impact and to calculate the displacement of unejected material. An approximate disrupted zone around the transient crater was incorporated to match the observation that material within the final crater (1.5–2 times the diameter of the transient crater (Melosh 1989, p. 138)) collapses quite quickly, within a few minutes for a 25-km diameter crater on Europa (Melosh 1989, p. 126). Collapse simulations were performed for transient craters 13 and 26 km in diameter (final craters roughly 25 and 50 km in diameter, respectively) with brittle ice layers at the surface ranging from 3 to 51 km thick. For each simulation, radial stresses at the surface and at the bottom of the brittle layer were tracked to determine where extensional stresses exceeded the fracture stress of water ice (0.5 MPa from Parmerter and Coon 1972). Preliminary results indicate that relaxation of an impact crater formed in a layered surface, with a thin icy crust overlaying a deeper layer of liquid water (or any Newtonian material with a viscosity lower than the effective viscosity of ice Ih), induces sufficient stress to cause fracturing of the overlying ice. In the case of the larger crater (final diameter of 52 km), brittle layer thicknesses between 6 and 15 km result in extensional stresses exceeding the fracture stress of ice at the bottom of the brittle layer in a broad region (.60 km radius) around the final crater. Such a region is consistent with the formation of multiple

FIG. 15. Results of a 2D Lagrangian shock physics hydrocode experiment. Two models were studied at high resolution: a solid-ice target and a solid-ice layer, 6 km thick, overlying liquid water. In both cases, the 2D axisymmetric target grid is 50 km in radius and 50 km deep, and the calculation ends before boundary reflections become a problem. A 1.14-km diameter ice sphere impacts both targets at 10 km/sec, which in the gravity regime produces a p14-km diameter transient cavity. See text for details. (a) This frame shows both of these initial states; the impactor is represented as a single vertex. (b) The second frame shows pressures (dyne/cm2) in cross section 16 sec after impact in logarithmic scale. The faster sound speed in the solid-ice layer propagates the stress wave more rapidly than in the liquid-water layer, resulting in a very different shock morphology in the two target structures. (c) The third frame shows particle speeds at the end of the simulation (256 sec).

LARGE IMPACT FEATURES ON EUROPA

141

142

MOORE ET AL.

rings around the crater, although fracture propagation has not been modeled in detail. For ice layers both thinner (3 km) and thicker (24 km) than this, the region of extensional stress is much narrower. In the models with the thickest ice layers (.42 km), the disrupted material around the transient crater does not intersect the underlying water layer, so only a very narrow region of extensional stress above the fracture criterion exists. This suggests that models with ice layers between p6 and p15 km thick are consistent with the formation of multiple rings around Europan craters of sizes comparable to Callanish and Tyre. For the smaller transient crater (final diameter #25 km), only the thinnest ice layers (3–6 km) result in extensional stresses which approach or exceed the fracture criterion in a sizable region around the final crater. This region narrows to models with thicker ice layers, and in models with ice layers .15 km thick there is no region of extensional stress at all. Thus, ice thicker than 10–15 km is consistent with the lack of rings around the crater Pwyll. In order to form ringed craters of this size, the local crust would need to be quite thin (i.e., 3 to 6 km). These results indicate that the absence of rings around Pwyll and their presence around Callanish are consistent with a p10- to 15-km thick layer of ice over a lower viscosity material (such as water) at the time of crater formation. However, other crustal configurations, for example, a horizontal channel of low-viscosity material between two layers, may also generate stress fields consistent with the observations. Furthermore, this model only requires that the low-viscosity material has a Maxwell time comparable to the time scale of crater collapse. This is consistent with but not limited to liquid water. Analyses of Landforms Related to Impact Features Turning from impact event modeling, we will now consider the implications of landform geometries for the properties of the upper few-hundred meters of Europa’s crust. The three concentric rings surrounding Callanish in its southwest quadrant are relatively flat-floored troughs p800 m wide and spaced p6 km apart. Their flat-floored morphologies indicate that they may be graben. Most extensional structures on Europa are consistent with formation by means of tensile failure of the lithosphere (Golombek and Banerdt 1990, Pappalardo 1994). Graben, however, are formed by means of normal faulting, and they are apparently rare on Europa. If the Callanish troughs were indeed graben, this would have important implications for the properties of Europa’s lithosphere in the vicinity of Callanish when these troughs were created, as it would mean that the lithosphere apparently experienced shear failure at a relatively shallow depth. The depth of intersection of graben faults is commonly associated with the depth to a mechanical discontinuity

from which the faults have propagated, such as the brittle/ ductile transition depth (e.g., Golombek 1982). This depth d can be estimated from the width w of a graben and dip d of graben faults as d 5 0.5 w tan d. If we assume the typical terrestrial value of p508 as a likely fault dip, then the Callanish graben faults would meet at only p500 m depth, which may indicate an extremely shallow brittle/ ductile transition. For a fault dip as high as 808, this depth would still be just p2 km. Note that these calculations assume that the graben formed after cessation of movement associated with the collapse of the transient crater. Shear failure of the lithosphere at shallow depth implies that the (brittle) lithosphere must be even shallower. Examining relationships between the extensional failure style of the lithosphere and its strength (Golombek and Banerdt 1990, Pappalardo 1994), normal faulting on Europa at a depth of just p500 m to 2 km implies a tensile strength of #0.1 MPa (1 bar) for d 5 508 and #1 MPa (10 bar) for d 5 808. This lithospheric strength is an order of magnitude less than the strength of laboratory ice as previously adopted to describe Europa (cf. Golombek and Banerdt 1990) and is more appropriate to fractured ice. Low ice strength in the vicinity of Callanish may have been a local phenomenon associated with impact and shock of the region immediately surrounding the impact feature shortly after the event which formed it. This would imply that the graben ringing Callanish formed shortly enough after the impact event that the adjacent lithosphere had not annealed significantly. Now we address the origin of pedestals around europan (and ganymedan) craters. We propose that they form as a consequence of plastic deformation of initially warm ice in the rim region of the continuous ejecta that is ultimately halted by the cooling and stiffening of this material. We prefer this hypothesis over those involving liquidized ejecta or ejecta composed of slurries because the textures of the pedestals seen at high resolution, such as those of Pwyll and Achelous (on Ganymede), show little if any indication of the presence of liquid during emplacement (Figs. 6 and 8). The pedestals manifest themselves as convex-upward breaks in slope upon which the surface texture is apparently unaffected. The surface texture of the ejecta appears to antedate pedestal formation. For the ‘‘plastic deformation’’ hypothesis to be viable, several assumptions must be accepted. The differential stress to induce strain within the near-rim ejecta must come from an initial rim of significant relief. Craters like Pwyll and Govannan may owe their present relief to the topography-flattening process of viscous relaxation, such as that described by Passy and Shoemaker (1982). These craters may have had more Ganymede-like or even Moon-like crater relief immediately after formation. If so, then the rims of these craters may have been as much as 900 m above the original surface (Melosh 1989, p. 88, Pike 1980).

143

LARGE IMPACT FEATURES ON EUROPA

The cooling rate of warm ice at the middle layer of the rim deposit would only be p0.18K yr21. This value was derived from the heat conduction equation solved for an initially isothermal cooling slab (see Carslaw and Jaeger 1959, pp. 93–99), using a value of 1026 m2 sec21 (taken from Hobbs 1974, p. 361) for thermal diffusivity and a slab thickness equal to the presumed height of the initial rim, inferred from lunar data (Pike 1980). If the initial temperature of this layer was p250 K, then total strain on the order of 1 would occur over the first 100 years, based on the equations of strain by Goldsby and Kohlstedt (1997) using differential stresses derived from the weight of the overlying rim and a grain size of 1 mm. Strain of this scale is probably sufficient to cause an outward bulge to the topography of the near-rim ejecta deposit. The conclusion of this analysis is very preliminary and general, given the use of arbitrary (but geologically reasonable) values and simple analytical models, but it lends credence to the ‘‘plastic deformation’’ hypothesis for pedestal formation. The principal challenge for this hypothesis is the need for initially warm ice (.200 K) in the near-rim ejecta deposit. Currently, it is not known how warm this material could be immediately after impact. It is worth noting that pedestal craters are only observed on icy satellites thought to have extended histories of extensive heating (e.g., Europa and Ganymede). Perhaps a combination of warm material brought up from a target with a steep thermal gradient and heating by the impact event itself of material included in this deposit can fulfill this requirement. An alternative to this hypothesis is that the process of general crater-wide viscous relaxation may deform the region just beyond the crater rim to form a pedestal. However, modeling of the viscous relaxation of crater topography (e.g., Thomas and Schubert 1988, Hillgren and Melosh 1989) has not indicated this possibility, though it must be said that finescale phenomena such as pedestal formation have not been explicitly considered in these models. Finally, we consider the implications of the absence of superposed craters on Callanish. Some 3200 km2 of Callanish interior deposits were imaged at 120 m/pixel. In these low-sun images, the interior of Callanish appears topographically rugged at the kilometer scale. Although some of the features on this surface occasionally form shapes similar to degraded impact craters, we find no examples of features larger than our recognition limit of 0.5 km diameter that are likely to be real impact craters. Inasmuch as impact features as large as Callanish are thought to form on Europa much less often than craters 0.5 km diameter, the possibility that the impact event is old and that Callanish’s interior has been resurfaced more recently cannot be completely ruled out. However, our previously discussed geomorphological analysis leads us to conclude that Callanish’s interior deposits probably formed very soon after the impact. The upper limit to the density of impact

craters superposed on Callanish is similar to the very low density of nonsecondary impact craters observed on the surface of the ice-raft-like disrupted terrain (Conamara Chaos, centered at p98N, 2748W) imaged on orbit E6. Using the Shoemaker (1996) cratering flux on Europa, extrapolated to smaller sizes with a lunar-like size distribution, the interior deposits of Callanish would be younger than 5 3 105 years. Like all estimates of crater ages on Europa, this estimate is subject to large uncertainty (based on small-number statistics, in this case zero) and to the uncertainties in models and observations of the cratering rate on Europa during its history. However, even the most conservative cratering rate estimates (e.g., Zahnle et al. 1998) would place the age of Callanish’s interior deposits at not greater than p108 years. If the age range implied by the absence of craters on the surface of Callanish were correct, and if our interpretation of the formation of Callanish and Tyre as impacts into a p10- to 15-km thick ice shell overlying liquid were also correct, then it would seem likely that Europa today has an underlying liquid water layer probably of global extent, for it is unlikely that such an ‘‘ocean’’ would cease to exist in the last few percent of Solar System history. CONCLUSIONS

1. Callanish and Tyre are probably impact features. This conclusion is based on the presence of pits or dots we interpret to be secondary craters surrounding both features. We interpret Callanish and Tyre to have morphologies that are the consequence of impacts into liquid-rich target materials. Observational evidence for this conclusion is the liquid behavior at the time of emplacement of the proximal ejecta around Callanish, the lack of any classic crater-rim facies, and the lack of any regional relief (i.e., Callanish is regionally flat). Tyre was seen at less favorable viewing but enough similarity between it and Callanish allows us to extend to Tyre the conclusions we reached concerning Callanish. 2. Craters ,30 km in diameter, such as Mannann’an and Pwyll, formed entirely within a solid target that exhibited brittle behavior on time scales of the impact event. We suspect that Mannann’an and Pwyll’s very large diameterto-depth ratios are due to the viscous relaxation of largescale topography. The presence of a well-defined darker, redder annulus beyond Pwyll’s rim and the absence of an equally prominent annulus around similarly sized Mannann’an is due to the rapid fading of this material (on the assumption that Pwyll is younger). 3. Modeling of the ejecta distribution from both impacts into solid ice and impacts into ice over water indicates that similar-sized impact events produce more and faster ejecta if excavation occurs exclusively in solid ice. The comparative abundance of secondary craters around Pwyll versus

144

MOORE ET AL.

Callanish imply that Callanish may have underproduced ejecta as a consequence of its excavation into a liquid layer, consistent with the results of this modeling. 4. The assumed depth of excavation for Callanish, along with the modeling of ring formation, leads us to conclude that this target was an ice layer p10–15 km thick overlying liquid. If Tyre were to represent a similar situation, then the very wide (near antipodal) separation of these features would lend support to the possibility that the solid crust of Europa was globally p10–15 km thick when these impacts occurred. An ice layer of this thickness is consistent with the appearance of more-or-less ‘‘classic’’ (and ringless) impact-feature morphologies for craters ,30 km in diameter. Our preliminary finite element modeling is consistent with these conclusions. 5. Pedestal ejecta facies, such as that of Pwyll, may be produced by the relief-flattening movement of plastically deforming but otherwise solid ice within the rim region of the continuous ejecta that was warm at the time of emplacement. The texture of Pwyll’s proximal ejecta shows little if any evidence for liquid associated with its emplacement. The pedestal, in detail, is manifested only as a convex-upward scarp across which the continuous ejecta texture is unchanged. The same observations apply to Achelous, a well-imaged ganymedan pedestal crater. If this hypothesis were valid, it would imply that ice composing the initial rims of pedestal craters is very (.200 K) warm. 6. If the age (#108 years) implied by the absence of craters on the surface of Callanish, and our interpretation of the formation of Callanish and Tyre as impacts into a p10- to 15-km thick ice shell overlying liquid were correct, then it would seem likely that Europa today has an underlying liquid water layer probably of global extent, for it is unlikely that such an ‘‘ocean’’ would cease to exist in the last few percent of the Solar System’s history. ACKNOWLEDGMENTS We thank W. B. Banerdt, K. Zahnle, and especially M. J. Cintala for their careful reviews of this manuscript. This investigation was funded by NASA’s Galileo Project.

REFERENCE Anderson, J. D., E. L. Lau, W. L. Sjogren, G. Schubert, and W. B. Moore 1997. Europa’s differentiated internal structure: Inferences from two Galileo encounters. Science 276, 1236–1239. Asphaug, E. 1997. Impact origin of the Vesta family. Meteor. Planet. Sci. 32, 965–980. Asphaug, E., J. M. Moore, D. Morrison, W. Benz, M. C. Nolan, and R. J. Sullivan 1996. Mechanical and geological effects of impact cratering on Ida. Icarus 120, 158–184. Belton, M. J. S., J. W. Head III, A. P. Ingersoll, R. Greeley, A. S. McEwen, K. P. Klaasen, D. Senske, R. Pappalardo, G. Collins, A. R. Vasavada, R. Sullivan, D. Simonelli, P. Geissler, M. H. Carr, M. E. Davies, J. Veverka, P. J. Gierasch, D. Banfield, M. Bell, C. R. Chapman, C.

Anger, R. Greenberg, G. Neukum, C. B. Pilcher, R. F. Beebe, J. A. Burns, F. Fanale, W. Ip, T. V. Johnson, D. Morrison, J. M. Moore, G. S. Orton, P. Thomas, and R. A. West 1996. Galileo’s first images of Jupiter and the Galilean satellites. Science 274, 377–385. Benz, W., and E. Asphaug 1995. Simulations of brittle solids using smooth particle hydrodynamics. Comput. Phys. Comm. 87, 253–265. Carr, M. H., M. J. S. Belton, C. R. Chapman, M. E. Davies, P. Geissler, R. Greenberg, A. S. McEwen, R. Greeley, R. Sullivan, J. W. Head III, R. T. Pappalardo, K. P. Klaasen, T. V. Johnson, J. Kaufman, D. Senske, J. M. Moore, G. Neukum, G. Schubert, J. A. Burns, P. Thomas, and J. Veverka 1997. Evidence for a subsurface ocean on Europa. Nature 391, 263–365. Carslaw, H. S., and J. C. Jaeger 1959. Conduction of Heat in Solids, 2nd ed. Oxford University Press, Oxford. Chapman, C. R., W. J. Merline, B. Bierhaus, S. Brooks 1998. Cratering in the jovian system: Intersatellite comparisons. Lunar Planet. Sci. XXIX, No. 1927. Lunar and Planetary Institute, Houston. [Abstract] [CD-ROM] Croft, S. K. 1984. A proposed origin for palimpsests and anomalous pit craters on Ganymede and Callisto. Proc. Lunar Planet. Sci., J. Geophys. Res. 88, B71–B89. Fink, J. H., D. E. Gault, and R. Greeley 1984. The effect of viscosity on impact cratering and possible application to the icy satellites of Saturn and Jupiter. J. Geophys. Res. 89, 417–423. Geissler, P. E., R. Greenberg, G. Hoppa, A. McEwen, R. Tufts, C. Phillips, B. Clark, M. Ockert-Bell, P. Helfenstein, J. Burns, J. Veverka, R. Sullivan, R. Greeley, R. T. Pappalardo, J. W. Head III, M. J. S. Belton, and T. Denk 1998. Evolution of lineaments on Europa: Clues from Galileo multispectral imaging observations. Icarus 135, 107–126. Goldsby, D. L., and D. L. Kohlstedt. 1997. Flow of Ice I by dislocation, grain boundary sliding, and diffusion processes. Lunar Planet. Sci. XXVIII, 429–430. Golombek, M. P. 1982. Constraints on the expansion of Ganymede and the thickness of the lithosphere. J. Geophys. Res. 87, A77–A83. Golombek, M. P., and W. B. Banerdt 1990. Constraints on the subsurface structure of Europa. Icarus 83, 441–452. Greeley, R., R. Sullivan, J. Klemaszewski, K. Homan, J. W. Head III, R. T. Pappalardo, J. Veverka, B. E. Clark, T. V. Johnson, K. P. Klaasen, M. Belton, J. M. Moore, E. Asphaug, M. H. Carr, G. Neukum, T. Denk, C. R. Chapman, C. B. Pilcher, P. E. Geissler, R. Greenberg, and R. Tufts 1998. Europa: Initial Galileo geological observations. Icarus 135, 4–24. Greeley, R., J. H. Fink, D. E. Gault, and J. E. Guest 1982. Experimental simulations of impact cratering on icy satellites. In Satellites of Jupiter (D. Morrison, Ed.), pp. 340–378. University of Arizona Press, Tucson, AZ. Greenberg, R., P. Geissler, G. Hoppa, B. R. Tufts, D. D. Durda, R. Pappalardo, J. W. Head, R. Greeley, R. Sullivan, and M. H. Carr 1998. Tectonic processes on Europa: Tidal stresses, mechanical response, and visible features. Icarus 135, 64–78. Hillgren, V. A., and H. J. Melosh 1989. Crater relaxation on Ganymede: Implications for ice rheology. Geophys. Res. Lett. 16, 1339–1342. Hobbs, P. V. 1974. Ice Physics. Oxford University Press, Oxford. Holsapple, K. A., and R. M. Schmidt 1987. Point source solutions and coupling parameters in cratering mechanics. J. Geophys. Res. 92, 6350– 6376. Horner, V. M., and R. Greeley 1982. Pedestal craters on Ganymede. Icarus 51, 549–562. Janes, D. M., S. W. Squyres, D. L. Bindschadler, G. Baer, G. Schubert, V. L. Sharpton, and E. R. Stofan 1992. Geophysical models for the

LARGE IMPACT FEATURES ON EUROPA formation and evolution of coronae on Venus. J. Geophys. Res. 97, 16,055–16,067. Jones, K. B., J. W. Head III, C. R. Chapman, R. Greeley, J. M. Moore, G. Neukum, and R. T. Pappalardo 1997. Morphology of palimpsests on Ganymede from Galileo Observations. Lunar Planet. Sci. XXVIII, 679–680. Kirby, S. H., Durham, W. B., and Heard, H. C. 1985. Rheologies of H2O ices I, II, and III at high pressures: A progress report. In Ices in the Solar System (J. Klinger, D. Benest, A. Dollfus, and R. Smoluchowski, Eds.), pp. 89–107. Kirk, R. L. 1987. A Fast Finite-Element Algorithm for Two-Dimensional Photoclinometry. Ph.D. dissertation, pp. 185–257. California Institute of Technology, Pasadena. Lange, M. A., and T. J. Ahrens 1983. The dynamic tensile strength of ice and ice–silicate mixtures. J. Geophys. Res. 88, 1197–1208. Lucchitta, B. K., and L. A. Soderblom 1982. The geology of Europa. In Satellites of Jupiter (D. Morrison, Ed.), pp. 521–555. University of Arizona Press, Tucson. Malin, M. C., and D. C. Pieri 1986. Europa. In Satellites (J. A. Burns and M. S. Matthews, Eds.), pp. 689–717. University of Arizona Press, Tucson. Maxwell, D., and Seifert, K. 1975. Modeling of cratering, close-in displacements and ejecta. Defense Nuclear Agency Report, No. DNA36287, Washington, DC. McEwen, A. S. 1986. Exogenic and endogenic albedo and color patterns on Europa. J. Geophys. Res. 91, 8077–8097. McKinnon, W. B., and H. J. Melosh 1980. Evolution of planetary lithospheres: Evidence from multiring basins on Ganymede and Callisto. Icarus 44, 454–471. McKinnon, W. B., and P. M. Schenk 1995. Estimates of comet fragment masses from impact crater chains on Callisto and Ganymede. Geophys. Res. Lett. 22, 1829–1832. Melosh, H. J. 1989. Impact Cratering: A Geologic Process. Oxford University Press, Oxford. Melosh, H. J., and A. Raefsky 1980. The dynamical origin of subduction zone topography. Geophys. J. Roy. Astron. Soc. 60, 333–354. Pappalardo, R. T. 1994. The Origin and Evolution of Ridge and Trough Terrain and the Geological History of Miranda. Ph.D. dissertation. Arizona State Univ., Tempe. Parmerter, R. R., and M. D. Coon 1972. Model of pressure ridge formation in sea ice. J. Geophys. Res. 77, 6565. Passy, Q. R., and E. M. Shoemaker 1982. Craters and basins on Ganymede and Callisto: Morphological indicators of crustal evolution. In Satellites

145

of Jupiter (D. Morrison, Ed.), pp. 379–434. University of Arizona Press, Tucson. Pike, R. J. 1980. Geometric interpretation of lunar craters. U.S. Geol. Survey Prof. Paper No. 1046, pp. C1–C77. Pronin, A. A., and E. R Stofan 1990. Coronae on Venus: Morphology, classification and distribution. Icarus 87, 452–474. Schenk, P. M. 1991. Ganymede and Callisto: Complex crater formation and planetary crusts. J. Geophys. Res. 96, 15,635–15,664. Schenk, P. M. 1996. Origins of palimpsests and impact basins on Ganymede. Lunar Planet. Sci. XXVII, 1137–1138. Shoemaker, E. M. 1996. The age of Europa’s surface. Europa Ocean Conf., San Juan Capistrano, CA, Nov. 12–14, p. 65. [conference abstract] Smith, B. A., L. A. Soderblom, R. Beebe, J. Boyce, G. Briggs, M. Carr, S. A. Collins, A. F. Cook II, 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. Suomi, and J. Veverka 1979. The Galilean satellites and Jupiter: Voyager 2 imaging science results. Science 206, 927–950. Squyres, S. W. 1980. Topographic domes on Ganymede: Ice vulcanism or isostatic upwarping. Icarus 44, 472–480. Squyres, S. W., D. M. Janes, G. Baer, D. L. Bindschadler, G. Schubert, V. L. Sharpton, and E. R. Stofan 1992. The morphology and evolution of caronae on Venus. J. Geophys. Res. 97, 13,611–13,634. Stofan, E. R, V. L. Sharpton, G. Schubert, G. Baer, D. L. Bindschadler, D. M. Janes, and S. W. Squyres 1992. Global distribution and characteristics of coronae and related features on Venus: Implications for origin and relation to mantle processes. J. Geophys. Res. 97, 13,347–13,378. Thomas, C. 1997. Is Tyre an ice cauldron? Lunar Planet. Sci. XXVIII, 1427–1428. Thomas, P. J., and G. Schubert 1988. Power law rheology of ice and the relaxation style and retention of crater on Ganymede. J. Geophys. Res. 93, 13,755–13,762. Turtle, E. P., and E. Pierazzo 1998. Contraints on the size of the Vredefort impact crater from numerical modeling. Meteoritics, in press. Turtle, E. P., C. B. Phillips, A. S. McEwen, J. M. Moore, R. Greeley, and the Galileo SSI Team 1998. Numerical modeling of multi-ring impact craters on Europa: Implications for subsurface structure. Lunar Planet. Sci. XXIX, No. 1325. Lunar and Planetary Institute, Houston. [Abstract] [CD-ROM] Wood, C. A. 1981. Possible terrestrial analogs of Vahalla and other ripplering basins, Multi-ring Basins. Proc. Lunar Planet. Sci. 12A, 173–180. Zahnle, K., H. Levison, and L. Dones 1998. Cratering rates on the Galilean satellites. Lunar and Planet. Sci. XXIX, No. 1902. Lunar and Planetary Institute, Houston [Abstract] [CD-ROM]