Main

Ganymede's surface can be divided into older, heavily cratered, dark terrain (∼34% by area, based on a global 2-km-resolution mosaic produced by P.M.S.) and younger, resurfaced, bright terrain (∼66%). The ages13 and nature of the resurfacing, if determined, would place important constraints on Ganymede's evolution (for example, on the role of tidal heating, which has been proposed as a trigger for internal melting and resurfacing on Ganymede11,12). Voyager-based analyses concluded that bright terrain, including both smooth and grooved terrain units, probably formed by resurfacing of shallow structural graben, troughs or rifts in Ganymede's dark, ancient crust1,2,4. Although volcanism was inferred, no direct evidence was observed to indicate how the resurfacing occurred. At least 18 caldera-like features were identified globally7,14, but their role in resurfacing was unclear.

Except for a high-standing ridged unit in one of the calderas10,15, sought-after evidence for individual lava flows or eruption centres has not been resolved in high-resolution (<300 m per pixel) Galileo images to date. This suggests that flow fronts have been destroyed by fracturing, impact erosion or mass wasting16, or are too low in relief to be resolved by Galileo, or that bright terrain simply did not form by extrusion of lavas. The lack of clear morphological evidence for volcanism, together with the discovery of pervasive faulting in grooved terrains in Uruk Sulcus17, has led to (or revived7) an alternative model: bright terrain might be mostly formed by pervasive tectonic disruption and brightening of older terrains, a process referred to as “tectonic resurfacing”8,9,18.

Regional differences in elevation between resurfaced terrains of different ages can help to distinguish emplacement mechanisms. Although crosscutting relationships have long suggested that bright terrain may be lower than dark terrain1,2,3,4,5—in that bright terrain is nearly always tectonically confined—offsets were never quantitatively determined. As of this writing, targeted Galileo stereo observations and derived topographic maps (digital elevation models, or DEMs) of bright terrain show only local-scale features, and do not include terrain boundaries19. To map the relative topography of resurfaced units, we have combined Voyager and Galileo images to produce eight regional-scale stereo pairs. DEMs were generated using stereogrammetric autocorrelation software20. The areal coverage and associated high-resolution CCD (charge-coupled device) images in these stereo mosaics offer a notable improvement over low-resolution Voyager-only stereo pairs21.

We have identified two sites that are best suited for investigation of the roles of volcanism and tectonism. The Sippar Sulcus site (Fig. 1) features numerous crosscutting bright terrain units and calderas. Geologic mapping by us and Wilhelms22 indicates that rugged, reticulate terrains (which are intensely deformed by multiple groove sets) are the oldest units in this region. Crosscutting these are numerous younger (less cratered) grooved terrain units, characterized by varying degrees of relief interpreted as variations in intensity of tectonic deformation. Although relative ages of grooved units are difficult to correlate regionally, the oldest in Sippar Sulcus tend to be the most intensely tectonically deformed (that is, they have the greatest areal density of—or highest degree of relief across—grooves). The last non-impact geologic events in this region were the formation of a number of cuspate, open-ended calderas7,14 (seven can be seen in Fig. 1a) and a distinctive 15–25-km wide swath of smooth terrain; all crosscut grooved terrain units.

Figure 1: Bright terrain units and topography within Sippar Sulcus, Ganymede.
figure 1

The three dominant structural styles—grooved terrain, smooth terrain and reticulate terrain—are represented. a, Galileo image mosaic obtained with a resolution of 180 m per pixel, superposed on regional Voyager images of resolution 1 km per pixel. The swath of smooth terrain crosses the scene diagonally from upper right to centre left. Irregularly shaped enclosures are interpreted as calderas14,15. The numerous bright patches are due to secondary impacts from the large crater Osiris, ∼600 km to the ESE of the centre of the mosaic (30° S, 184° W). Images used are FDS 20637.41, 20637.53 and 20637.56 (Voyager) and 394532413–39453278 (Galileo). b, DEM of topography of the same scene. Relative elevation values have been colour-coded and merged with the Galileo mosaic. The DEM has a horizontal resolution and sample spacing of 7 km and 1 km, respectively; vertical resolution is 0.2 km. The inset shows a geologic map highlighting areas of grooved terrain (g, black), reticulate terrain (r, grey), smooth terrain (s, white), calderas (hatched), and locations for Fig. 3a and b (upper and lower box, respectively).

Full size image

Our DEM of Sippar Sulcus reveals that reticulate, grooved and smooth terrains lie at characteristically different relative elevations (Fig. 1b). The oldest and most strongly deformed units (reticulate terrain) are the highest standing. The youngest and least deformed unit (the smooth terrain swath) is the lowest, lying ∼100–500 m below older, crosscut lanes of grooved terrain, and ∼800–1,100 m below older, reticulate terrains (Fig. 1b). The elevations of grooved terrain units within Sippar Sulcus are variable, ranging between those of reticulate terrain and the youngest smooth terrain. The trapezoidal patch of smooth (or very lightly grooved) terrain at the southern margin of the DEM lies at an intermediate elevation. Although elevations of adjacent units are different, the elevation along the swath of smooth terrain is, to the limit of DEM vertical resolution, constant (or at most, gently sloping, with no undulations or breaks in slope) across our map area, a distance of over 500 km.

These relationships are further illuminated at the intersection of Erech Sulcus and northern Sippar Sulcus (Fig. 2a). The intensely grooved terrain within Erech Sulcus mostly lies at elevations comparable to adjoining dark terrain. These units are truncated by the northern margin of Sippar Sulcus, which is defined by a lane of smooth terrain ∼8 km wide and lying ∼750 m below adjacent dark terrain (Fig. 2b). Interior portions of Sippar Sulcus here consist of truncated packets of grooved material crosscut by topographically lower units of younger smooth material. Elevations of these grooved terrain units are intermediate between those of smooth terrain and the nearby dark terrains. Thus we observe the same general correlation between elevation and degree of tectonic deformation (that is, groove relief and density) at two distinct sites.

Figure 2: Terrain units and topography at the intersection of Erech Sulcus and northern Sippar Sulcus, Ganymede.
figure 2

a, Galileo image mosaic obtained with a resolution of 140 m per pixel. Erech Sulcus is the 75-km-wide band of grooved terrain extending between the two blocks of older, dark terrain. Portions of Sippar Sulcus lie to the south, trending at high angles to Erech Sulcus. Note the caldera at right, truncated by the narrow lane of smooth terrain. b, DEM of topography of the same scene. Relative elevation values have been colour-coded and merged with the mosaic. Horizontal resolution and sample spacing are 7 and 1 km, respectively; vertical resolution is 0.24 km. Scene is centred at 16° S, 177° W, roughly 500 km north of the centre of Fig. 1. Images used are FDS 20637.41 (Voyager), and 394533065 and 394533078 (Galileo).

Full size image

Uniformly low-standing topography of smooth terrains, especially in comparison to the variable topography of adjacent older grooved and reticulate terrain units, strongly suggests downdropping and flooding to an equipotential surface. In addition, the presence of only occasional bright lineations within the swath of smooth terrain suggests that tectonism (or at least deformation capable of destroying ridges hundreds of metres high) has not been an important process within this terrain. The embayment of individual older ridges and grooves along the margin of this swath22, confirmed in Galileo images10 (Fig. 3a), is also characteristic of flooding of low-lying topography, and is inconsistent with emplacement by tectonic disruption. Similar embayment of ridges is observed at several sites along a swath of relatively young smooth terrain near the south pole5,14 (see, for example, figure 1 in ref. 5) and elsewhere22, indicating that smooth terrains globally may have similar origins.

Figure 3: High-resolution views of Sippar Sulcus.
figure 3

a, Embayment of ridges and troughs (arrows) in older reticulate terrain10,22 by smooth terrain, indicative of flooding by low-viscosity materials. b, Irregularly shaped caldera and high-standing interior; the latter is interpreted to be a viscous flow10,15. Our DEM (Fig. 1b) shows that the elevation of the westernmost caldera floor material (arrow) is comparable to adjacent grooved material but decreases towards the east (right), where it is similar to nearby, lower-lying smooth terrains. Note the general lack of lineations within smooth terrain, which extends across the upper half of the image and crosscuts a similar but grooved unit at lower right. Both views are portions of Galileo image 394532478; resolution is 180 m per pixel. North is up. See Fig. 1b for locations.

Full size image

Tectonic resurfacing8,9,18, in contrast, should lead to intense groove formation within ‘new’ terrains and some preservation of broad-scale relative elevation differences of older crosscut and faulted terrains. Topographic downdropping is logically associated with tectonic resurfacing9,18, but the lowest units here are the least deformed and do not preserve vestiges of pre-existing relative elevation differences. We also find no tectonic evidence for shear (previously hypothesized7) within or adjacent to the smooth swath (Figs 1a and 3); nor can shear obviously explain the topography or topology of a depressed, dog-legged, smooth terrain swath. We also find no evidence that the two sides of the smooth terrain swath were originally together, as might be expected if the terrain formed by crustal spreading. We therefore conclude that volcanic resurfacing of shallow structural troughs (see, for example, ref. 4) is indeed primarily responsible for the formation of smooth terrain (at least in Sippar Sulcus and near the south pole14).

Tectonism remains crucial to Ganymede's geology, however. Extensional tectonism is responsible for downdropping blocks of dark terrain in the first place, and for extending both dark terrain and bright, resurfaced terrain to create lanes of grooved and reticulate terrain3,4. Nor do our observations contradict those at Uruk Sulcus, where grooved terrain formed entirely by tectonic extension (and some shear)9 from what Voyager identified as smooth terrain. This ‘smooth’ terrain, when seen at high resolution (75 m per pixel), is bright and flat-topped, with a structural imprint of subparallel fine crevasses and simple graben (see, for example, figure 5 in ref. 9). The important point is that this precursor bright terrain is flat overall, and only modestly tectonized; logically, it was originally emplaced as a volcanic unit. Likewise, the smooth terrains in Figs 1–3 are at most lightly tectonized (grooved).

The relative continuity of elevation and the lack of prominent volcanic textures or relief within smooth terrains studied here imply that volcanic resurfacing materials in smooth terrains have a relatively low viscosity. Based on terrestrial volcanic experience, resurfacing by solid-state flow of ice would be expected to produce flows of limited extent and substantial relief14. Such relief might go undetected in our DEMs, but high-resolution images (for example, Fig. 3) suggest that the relief of flow units within smooth terrain is very subtle and may be tens of metres or less. In contrast, the high-standing ridged deposit within the largest Sippar Sulcus caldera (Fig. 3b) has been inferred to be a relatively viscous flow10,15. Similarly high-standing (and sometimes ridged) materials exist within the other calderas10,14,15, but no similar morphological evidence for viscous flow outside of caldera margins has been identified to date. Our DEMs (and photoclinometric profiles14) indicate that caldera floor elevations are also variable, ranging between those of surrounding grooved terrains and adjacent smooth terrain (Figs 1b, 2b and 3b).

Resurfacing of smooth terrain on Ganymede may be in some respects analogous to that of the lunar mare, which are characterized by shallow flooding of topographic lows with low-viscosity lavas and a general lack of volcanic edifices. The lunar volcanic style is plausibly related to, among other things, the negative buoyancy of basaltic magma with respect to anorthosite-rich crust, and infrequent, high-volume eruptions23. On Ganymede, our data confirm that smooth terrain deposits are similarly confined to low-lying areas, indicating that smooth terrain magmas are negatively buoyant with respect to Ganymede crust as well (as first argued by Shoemaker et al.3). The long-standing problem of how a relatively dense, aqueous magma evolves out of Ganymede's icy mantle1,2,5,6 remains, but it is possible that tidal heating within the convecting portion of the upper mantle led to ascent of buoyant masses of partially molten ice, providing both the extensional stress and liquid for resurfacing12.

Our observations suggest a relatively simple model, in which each successive unit of bright terrain forms initially as a structural trough and is subsequently flooded by low-viscosity aqueous lava, which then freezes. Caldera formation may precede this structural downdropping, and certainly precedes the flooding that forms smooth terrain (ridged caldera floor material represents a higher-viscosity volcanic stage). Extension and groove formation can then follow smooth terrain flooding. Extension may further lower topographic levels9, but may be counteracted by isostatic adjustments owing to the high heat flows associated with groove formation24. The general correlation of elevation and deformation is consistent with decreasing intensity of groove formation over time. As most bright terrain on Ganymede is grooved, only the most recently formed smooth units have survived (relatively) undisrupted.

We find volcanic resurfacing to be an integral part of the process of bright terrain formation on Ganymede in at least two geographically distinct regions. High-resolution stereo observations of terrain boundaries taken during orbit G28 in May 2000, and transmitted to Earth over the following months, should indicate whether the topographic relationships we have described characterize bright terrains generally. It remains for future studies to unravel the global distribution, timing and relative importance of volcanism relative to tectonism on Ganymede.

Methods

To measure parallax, the stereogrammetry technique20,21 relies on matching of albedo patterns within a sample area centred on each pixel (this sample area is equivalent to the ‘footprint’ of the elevation measurement, 3–13 times the image resolution in these maps, depending on scene content). Some features (for example, bright crater ejecta) appear different in the stereo images due to photometric effects, and can contribute noise to the final DEM. Because no global topographic datum exists for Ganymede, our DEMs measure local relief only; they are not controlled with respect to the centre of Ganymede. Limb images indicate that there is little pronounced regional-scale topographic relief, consistent with low relief generally3. Assuming that the mean slope across each site is negligible, our stereo pairs were semi-controlled by selecting registration (or tie) points within geologic units assumed to lie at similar elevations. For the Sippar Sulcus site, tie points were located within older reticulate terrain. In the Erech Sulcus site, tie points were selected in older dark terrain.