Identification of mercurian volcanism: Resolution effects and implications for MESSENGER

Transcription

1 Meteoritics & Planetary Science 37, (2002) Available online at Identification of mercurian volcanism: Resolution effects and implications for MESSENGER S. M. MILKOVICH1*, J. W. HEAD1 AND L. WILSON2 1Department of Geological Sciences, Brown University, Providence, Rhode Island 02912, USA 2Department of Environmental Science, Lancaster University, Lancaster LA1 4YQ, U.K. *Correspondence author's address: (Received 2002 January 15; accepted in revised form 2002 May 17) (Presented at the Workshop on Mercury, The Field Museum, Chicago, Illinois, 2001 October 4 5) Abstract The possibility of volcanism on Mercury has been a topic of discussion since Mariner 10 returned images of half the planet's surface showing widespread plains material. These plains could be volcanic or lobate crater ejecta. An assessment of the mechanics of the ascent and eruption of magma shows that it is possible to have widespread volcanism, no volcanism on the surface whatsoever, or some range in between. It is difficult to distinguish between a lava flow and lobate crater ejecta based on morphology and morphometry. No definite volcanic features have been identified on Mercury. However, known lunar volcanic features cannot be identified in images with similar resolutions and viewing geometries as the Mariner 10 dataset. Examination of high-resolution, low Sun angle Mariner 10 images reveals several features which are interpreted to be flow fronts; it is unclear if these are volcanic flows or ejecta flows. This analysis implies that a clear assessment of volcanism on Mercury must wait for better data. MESSENGER (MErcury: Surface, Space ENvironment, GEochemistry, Ranging) will take images with viewing geometries and resolutions appropriate for the identification of such features. INTRODUCTION Ever since the arrival of Mariner 10 at Mercury in there has been debate over the existence of volcanic features on the surface of the planet. Mercury has two types of plains, smooth and intercrater; they are the most extensive terrain on the planet (Strom et al., 1975). The smooth plains are characterized by flat or gently rolling topography overlain by wrinkle ridges. The intercrater plains are level to gently rolling as well, and are found between and around the cratered terrain that covers much of the mercurian surface (Spudis and Guest, 1988). Through comparisons with lunar mare, many researchers (e.g., Murray et al., 1975; Strom et al., 1975) concluded that these plains were formed by volcanic processes; the smooth plains look very similar to the lunar mare and are quite widespread. Several plains fill up basins and craters while others sit in low-lying areas between craters. The widespread distribution of the intercrater plains indicates extensive resurfacing early in Mercury's history and many researchers conclude that the morphology of the plains is most consistent with volcanism as a resurfacing agent (Murray et al., 1975; Strom et al., 1975; Dzurisin, 1978; Kieffer and Murray, 1987). Others consider that they are more similar to some of the lunar light plains formed by crater ejecta (Wilhelms, 1976). Wilhelms pointed out that the pro-volcanism arguments are similar to arguments used to demonstrate that certain light plains (the Cayley Formation) on the Moon were volcanic; Apollo 16 samples proved these plains to be made up of impact breccia. The plains may have been formed by fluidized ejecta, impact melt, overlapping ejecta deposits from multiple craters, or secondary crater ejecta. This debate can only be resolved if landforms typical of extrusive volcanism (e.g., domes, shields, sinuous rilles, calderas, flow fronts) are clearly identified or clearly found not to exist on the surface of Mercury. Schultz (1977) pointed out that lunar volcanic features are small, localized, and nearly undetectable at resolutions comparable to those of Mariner 10. Work by Malin (1978) indicated that the viewing geometry of Mariner 10 may have been insufficient for the identification of such volcanic features. He addressed this issue by analyzing lunar images under Mariner 10 resolutions ( 1 2 km) and various lighting conditions (0 30 ) for volcanic features. Since the mercurian images are at about the same resolution as pre-1964 Earthbased telescopic images of the Moon, Malin used such images from the Consolidated Lunar Atlas (Kuiper et al., 1967) with approximately the same viewing conditions as the Mariner 10 data. Malin was only able to identify lunar volcanic domes and flow fronts under these conditions, and found that volcanic Prelude preprint MS# Meteoritical Society, Printed in USA.

2 1210 Milkovich et al. features cannot be identified in images with Sun elevation angles above 25. Only 15% of the surface was imaged with Sun elevation angles <20 (Malin, 1978). Smooth plains were identified in 40% of the images, or 18% of the surface (Spudis and Guest, 1988). Due to the limited coverage of mercurian plains at low Sun angles, Malin was not surprised that volcanic landforms have not been identified on Mercury. He concluded that the only way to identify volcanic landforms in the Mariner 10 data is to look near the terminator. In such images he identified two possible volcanic domes on Mercury; these are positive topographic features which are reminiscent of lunar domes but are near scarps and other structural features and are not unequivocally volcanic. One of these possible domes is located near Discovery Scarp, and can be seen in Fig. 1. Recent recalibration of Mariner 10 color data has allowed identification of several features spectrally similar to volcanic flows and pyroclastic fissure eruptions (Robinson and Lucey, 1997). The recalibrated data (Robinson and Taylor, 2001) and microwave and mid-infrared observations (Jeanloz et al., 1995) indicate a low abundance of FeO in the crust of Mercury; this may imply that basalt is absent on the mercurian surface (Sprague et al., 1994; Jeanloz et al., 1995) or that the mantle has a low FeO content (Robinson and Taylor, 2001). The question of the mode of formation of the mercurian plains remains unanswered. This study assesses the best way of identifying volcanic landforms on Mercury by addressing several key issues. These issues are what distinguishes an ejecta flow from a volcanic plain, what would mercurian volcanism look like, how would we find evidence for mercurian volcanism in images, and what do we actually see in the Mariner 10 images. Finally, we discuss the upcoming MESSENGER (MErcury, Surface, Space ENvironment, GEochemistry, Ranging) spacecraft mission to Mercury and examine its prospects for identifying possible volcanic landforms. LOBATE EJECTA FLOWS Outline of Process Early analyses of lunar ejecta deposits relied heavily on terrestrial analogs, particularly small impact craters such as Meteor Crater, Arizona, and craters resulting from highexplosive and nuclear events. The small sizes of these meant that the ejecta velocities were such that most near-crater ejecta was emplaced as an "overturned flap" and at relatively low velocities. Consideration of the full temporal and spatial range of planetary craters, however, soon showed that the energetics of impact ejecta emplacement were quite different. The pioneering work summarized in Oberbeck et al. (1977) and Oberbeck (1975) showed that in order for ejecta to be transported from the impact site to distances of tens to thousands of kilometers, the ejecta must have a very high initial ejection velocity. Furthermore, due to the lack of an atmosphere, FIG. 1. Feature near Discovery Scarp. This structure between the arrows has been suggested to be a dome, but identification is difficult. deceleration was minimal, and the ejecta would reimpact with the same velocity it left the initial cavity of excavation. The fundamental implication of this fact was that ejecta would not be emplaced as a passive "blanket", but that it would locally impact at very high velocity, excavating many multiples of its own mass, to produce an ejecta deposit which would consist of a small component of the primary ejecta and a large component of the locally excavated material. Thus, the final "ejecta deposit" would be emplaced in a very dynamic event, involving significant energetic erosion of the substrate, huge clouds and surges of laterally moving ejecta, and related landslides and debris flows, and ponding of ejecta. When the

3 Identification of mercurian volcanism 1211 event was over, the ejecta "deposit" would consist primarily of locally excavated material, rather than an "ejecta blanket" of predominantly excavated material. In a series of papers, Oberbeck and colleagues (see summary in Oberbeck, 1975) showed that smooth plains could reasonably be interpreted as accumulations of locally ponded ejecta from such basin-scale impact events. In principle, the composition of these deposits should be characterized by a mixture of primary and secondary ejecta material, and be dominated by secondary ejecta at increasingly greater distances from the transient cavity due to the increasingly higher ejecta velocities at greater radial distances. Thus, ejecta spectral signatures might be dominated by primary target material near the rim, but increasingly by local material excavated by the secondary impacts and incorporated into the ejecta deposits at greater radial ranges. Indeed, the detection of cryptomaria (Head and Wilson, 1992; mare volcanic deposits underlying higher albedo ejecta deposits) showed the importance of mixing of basin ejecta with local material (e.g., Mustard and Head, 1996). In most cases, there was evidence of ejecta emplacement by radial sculpting of highland terrain and smoothing of intervening low terrain by ponded ejecta. Some very fresh lunar craters, such as Tsiolkovsky, have lobate flow-like features on their rims (e.g., Wilhelms, 1987, plate 3.24). However, there was little evidence of discrete flow fronts or lobes in association with the emplacement of these smooth plains at basin scales (e.g., Oberbeck, 1975), making difficult the distinction between smooth plains formed by different mechanisms such as effusive volcanism and ejecta ponding. In some cases there is evidence for lobate flow-like features that appeared to represent late-stage movement and emplacement of these ejecta deposits. In the next section, we examine the two most well-developed examples of these known from lunar basins, and assess their origin, associated features, and the ability to recognize them under different conditions. Lunar Examples from the Orientale Basin One distinctive flow lobe has been identified 1000 km south-southeast of the center of the Orientale Basin, in the midst of radial ejecta facies near the crater Inghirami (Moore et al., 1974, their Fig. 7; Wilhelms, 1987, his Fig. 4.4D). This flow lobe is 30 km wide, about km long, and extends away from the vicinity of the end of a major secondary crater chain radial to the Orientale Basin (Fig. 2). The flow lobe is 400 m thick and is convex outward away from the basin rim. It appears to be composed of ejecta, rather than lava flows, and to have been formed in the terminal stages of the emplacement of the Orientale Basin ejecta deposit. The evidence for this includes (1) its crispness, (2) its superposition on ejecta plains and related deposits, (3) its spectral similarity to adjacent radially textured ejecta deposits, (4) its contrast to the spectral characteristics of adjacent mare deposits, (5) its apparent topographic control by newly formed ejecta topography, and (6) its scarp height and thickness, much greater than any known lava flows on the Moon (e.g., Head and Wilson, 1992). Clementine color composite images of the area show that the lobate flow is virtually invisible, being indistinguishable from the surrounding textured ejecta and smooth plains. Lunar Orbiter and Clementine images (Fig. 2c,d) show that the origin of the lobe blends with the more proximal ejecta facies and may be related to the flow of mobilized ejecta from the end of the large secondary crater chain. A second distinctive lobate flow is located just inside the crater Grimaldi (Fig. 3a d). This lobate feature is similar to the one near the crater Inghirami and has been described in Wilhelms (1987; his Fig. 4H). The structure itself is 35 km wide, up to 60 km long, and is generally concave outward from the Orientale Basin, with its local direction influenced by the presence of Grimaldi. Its spectral characteristics are indistinguishable from those of the surrounding ejecta deposits and contrast distinctly with those of the adjacent mare deposits on the floor of Grimaldi. This example too appears to have been derived from the latestage movement of ejecta of the Orientale ejecta deposit. Lessons for the Recognition of Ejecta-Related Plains on Mercury On the basis of these examples and our general knowledge of the emplacement of crater and basin ejecta (e.g., Oberbeck, 1975), we can reach the following conclusions: (1) most ejectaemplaced plains fill low-lying areas between radially textured uplands; (2) these plains can have multiple elevations due to the elevation of the topography prior to the ejecta emplacement event; (3) these plains distal to the basin will have mineralogic characteristics typical of a mixture of the projectile material dominated by the local substrate; (4) in most cases, lobate ejecta boundaries are not seen in the smooth plains; (5) in some cases, ejecta lobes are seen, but they tend to be very localized and to be related to late stage phases of the event; and (6) the spectral characteristics of these plains are virtually indistinguishable from the adjacent ejecta, but distinguishable from mare deposits. On the basis of these characteristics, we can further assess the nature of plains on Mercury and assess the possibility that they may be of impact ejecta origin. Of course, in order to apply these observations effectively, one must keep in mind the differences in how the cratering processes and ejecta emplacement is manifested on Mercury due to variations in impact velocities and gravity (e.g., Gault et al., 1977; Pike, 1988; Schultz, 1988). A comparison of Orientale Basin on the Moon with Caloris Basin on Mercury clearly demonstrates the morphological differences in ejecta emplacement on the two bodies. The flow fronts identified above are located in the portion of the Orientale ejecta blanket interpreted to be the Inner Facies of the Hevelius Formation. The Inner Facies contain elongate ridges and troughs generally radial to the basin. Many of the ridges display lobate edges similar to those of viscous flow fronts and are

4 1212 Milkovich et al. FIG. 2. Ejecta flow lobe associated with Orientale Basin and near the crater Inghirami. (a) Lunar Orbiter image of flow (subframe of image IV-180-H1). (b) Sketch map of area in (a). (c) Clementine image of flow and surrounding deposits. (d) Lunar Orbiter image of flow and surrounding deposits (image IV-180-H1). oriented facing away from the central basin (McCauley, 1987). The Inner Facies are made up of the outward ground surge of material characteristic of an ejecta flow (McCauley, 1987). Surrounding the Inner Facies are the Outer Facies, made up of rolling plains and weakly lineated terrain thought to be the distal ends of the ejecta blanket. The Outer Facies are recognizable up to one basin diameter out from Orientale (McCauley, 1977). The outermost ejecta facies recognized around Caloris is the Van Eyck Formation, a lineated terrain (Spudis and Guest, 1988) proposed to be equivalent to the Inner Facies of Orientale (McCauley, 1977). The lineated terrain around Caloris is recognizable out to a distance of one basin diameter (McCauley, 1977), where it is embayed by smooth plains. These smooth plains are found radially outside of Caloris (Spudis and Guest, 1988) and contain possible flow fronts (discussed below). If the Van Eyck Formation and the Inner Facies are indeed equivalent, then ejecta emplacement on the Moon and Mercury must be different. The Inner Facies do not extend proportionally as far from the associated basin as do the Van Eyck Formation. This may imply that lobate flows travel further on Mercury than on the Moon. Alternatively, ejecta flows may have a different morphology on the two bodies due to gravitational differences. There are several reasons why the smooth plains encircling Caloris Basin are unlikely to be crater ejecta deposits. The most probable location for lobate ejecta deposits, based on a

5 Identification of mercurian volcanism 1213 FIG. 3. Ejecta flow lobe associated with Orientale and located just inside the crater Grimaldi. (a) Lunar Orbiter image of flow (subframe of image IV-168-H3). (b) Sketch map of area in (a). (c) Clementine image of flow and surrounding deposits. (d) Lunar Orbiter image of flow and surrounding deposits (image IV-168-H3). comparison with Orientale Basin, is closer to the rim of Caloris (either in the Van Eyck Formation or other terrains closer to the basin rim). The candidate flow fronts (discussed below) observed in the smooth plains are oriented in the wrong direction their scarps face towards the basin rim rather than away, as would be expected from an outward flow. VOLCANISM Theoretical Considerations of Mercurian Volcanism We have discussed above how some have argued that the plains deposits could be basin ejecta, similar to those found at the lunar Apollo 16 landing site (Wilhelms, 1976; Oberbeck et al., 1977), raising the possibility that there are no identifiable volcanic units on Mercury (for a summary of the controversy, see Spudis and Guest, 1988). Fundamental questions remain about the presence and nature of basic magmatic processes on Mercury: Did Mercury, like the Moon (Head and Wilson, 1992; Head et al., 2000), form a primary crust which served as a barrier to, and filter for, magma ascent? Did secondary crustal formation (e.g., analogous to that of the lunar maria) occupy a

6 1214 Milkovich et al. significant part of the resurfacing history of Mercury? Did basaltic volcanism contribute to the resurfacing history of Mercury or did the formation of the iron core so alter the mantle geochemistry that other rock types dominate any eruptives? Does the tectonic history of Mercury (which exhibits significant global compressional deformation) mean that extrusive volcanism was inhibited or precluded by the state of stress in the lithosphere in its early to intermediate history? We now address these questions through an assessment of the ascent and eruption of magma under mercurian conditions under a variety of settings. It is commonly assumed that the large iron/silicate ratio of Mercury is the result of its early modification by a giant impact (Benz et al., 1988; Tyler et al., 1988; Tonks and Melosh, 1992). The extent of the differentiation of Mercury prior to this event is moot as regards surface features and surface chemistry if most of the original crust and much of the original mantle were stripped away and lost into the Sun (Benz et al., 1988), and much of the remaining crust and mantle were left as a molten magma ocean (e.g., Benz et al., 1988; Tyler et al., 1988; Tonks and Melosh, 1992). However, the extent of any early differentiation and the details of the impact event are significant for the size of the iron core and for the bulk compositions of the residual unmelted mantle and the postulated overlying magma ocean. In particular, it is not automatically safe to assume that the mantle which is related to the present crust was very close to chondritic in composition. Whatever the composition of the mantle after the giant impact, there is uncertainty about the extent to which solidstate convection could have proceeded in that mantle (Benz et al., 1988; Jeanloz and Morris, 1986; Spohn, 1991). The vigor of any convection that was occurring would have varied with time as a function of the ever-decreasing amounts of longlived radioactive heat sources and the ever-increasing temperature difference across the mantle as a result of the cooling of the overlying magma ocean. If convection in the mantle did take place, then partial melting should have occurred and buoyancy forces should have been able to extract melts; these would have been of basaltic composition if the mantle were chondritic but clearly other compositions are possible (e.g., Jeanloz et al., 1995). Bodies of buoyant melt would have risen, probably diapirically, until they encountered one or other of two kinds of trap (Head and Wilson, 1992; Wilson and Head, 2001). The first option is a density trap, that is, a level at which melts became neutrally buoyant on encountering less dense overlying rocks: the base of the crust is an obvious possibility for this. The second option is a rheological trap, that is, a level at which the surrounding rocks are no longer able to deform in a plastic fashion fast enough to relax the stresses caused by the buoyancy forces. Magma present in a rheological trap will still in general be buoyant and may be able to rise further by opening a brittle crack in the overlying rocks (i.e., by propagating a dike). However, any dike which forms will have a vertical extent limited by the volume of magma available and the distribution of stresses in the lithosphere (Head and Wilson, 1992; Wilson and Head, 2001). Multiple such dikes may form (Head and Wilson, 1992; Wilson and Head, 2001), but they will stall at shallower depths if they encounter density traps. If the rheology of the rocks at all points around the periphery of a magma body stalled at a rheological trap can support stresses, an excess pressure may be created in the magma body by density changes consequent on chemical evolution. Such evolution would be the consequence of fractional crystallization which could conceivably lead to supersaturation of any relatively insoluble volatile phase and gas bubble formation, and this could aid dikes in penetrating much further into the overlying rocks, thus overcoming density traps to some extent (Head and Wilson, 1992). A magma body stalled at a density trap may also acquire an excess pressure due to chemical evolution, enabling it to migrate to shallower depths, if the rocks surrounding it are elastic. The optimum mechanism for allowing mantle melts denser than the crust to penetrate to shallow depths as intrusions or to erupt at the surface is to have a rheological trap a short distance below a density trap (Wilson and Head, 2001). It is then possible for dikes to grow both upward into the crust and downward into the mantle (which behaves elastically on the short timescales of dike propagation (Wilson and Head, 2001) and to use the positive buoyancy of the melt in the mantle to offset the negative buoyancy in the crust (Solomon, 1975). How do these ideas relate to Mercury? If partial melting began in the mantle while a magma ocean produced by a giant impact was still liquid, any melts rising buoyantly through the mantle that were also less dense than the magma ocean would, of course, have mingled with the ocean. Any such melts denser than the ocean would have ponded beneath it, with perhaps some small amount of mechanical mixing occurring at the boundary. Only after the ocean had solidified could mantle melts penetrate the resulting crust, and then only under the circumstances described above. The complete range of possible density structures for Mercury after the solidification of any magma ocean, and the consequent range of density contrasts between lithosphere and possible mantle melts, is not very different from those that apply to the Moon. As a result the possible outcomes range from no magmas reaching the surface, through magmas reaching the surface only in topographic lows but otherwise stalling as (possibly shallow) intrusions, to circumstances in which large volumes of magma reach the surface. The last option would be favored by the fortuitous proximity of rheological and density traps near the crust mantle boundary and could be related to the widespread plains-forming units if these are in fact volcanic. Magmas formed by the chemical evolution of mantle melts in rheological traps would be the best candidates for the pyroclastic deposits associated with the crater Homer. Ultimately these issues will only be clarified by better identification of the composition of surface units, currently rendered difficult both by the spatial and spectral resolution of available data (see following discussions)

7 Identification of mercurian volcanism 1215 and the complications due to space weathering of Mercury's surface (Hapke, 2001; Noble and Pieters, 2001). Identifying Volcanism in Images The Moon makes a useful analog for Mercury. Mercury's crust is spectrally similar to the lunar highlands (Blewett et al., 1997). Mare basalt deposits are the primary volcanic feature on the Moon and are often compared to the mercurian plains (e.g. Murray et al., 1975; Strom et al., 1975). Lunar Volcanic Features A wide variety of landforms are associated with the lunar maria (Head and Wilson, 1992). These include regional dark mantling deposits (Fig. 4a) which are interpreted to be caused by pyroclastic eruptions. Glass FIG. 4. Lunar volcanic landforms. (a) Dark mantling deposits on the floor of Alphonsus (LO V-116-M). (b) Sinuous rilles. Hadley Rille from the Clementine Lunar Basemap. (c) Lava flow units. Large flow in Mare Imbrium (LO V-160-H2,H1, V-161-H2,H1). (d) Shield volcanoes in Mare Insularum (LO IV-133-H1). (e) Volcanic complex in Marius Hills (LO V-211-M).

8 1216 Milkovich et al. beads from these eruptions are spread tens to hundreds of kilometers and are found in a number of lunar samples. The eruptions must continue for a relatively long time to build up the observed deposits. Thus the dark mantling deposits appear to mark locations of rapid and sustained eruptions (Head and Wilson, 1991). Sinuous rilles (Fig. 4b) are often associated with these deposits as well as the maria. Sinuous rilles are meandering channels generally an order of magnitude larger and often much more sinuous than terrestrial lava channels and are found primarily in the lunar maria. They are thought to be sites of eruptions and thermal erosion. Sinuous rilles may be the sites of the rapid and sustained eruptions which produced dark mantling deposits (Head and Wilson, 1991). Lava flow units (Fig. 4c) are mapped on the basis of topographic, albedo, and color boundaries but individual flows are usually hard to identify due to impact degradation of individual flow unit boundaries (Head and Wilson, 1991). Flow fronts are more easily identified in near-terminator images (Head and Lloyd, 1973). About 50 small shield volcanoes (Fig. 4d) are identified. They are characterized by a circular to irregular outline and slopes <5 ; nearly all of these features are found in association with mare. The shields range from 3 to 20 km in basal diameter but no shields larger than 20 km are seen (Head and Gifford, 1980). Such volcanoes with diameters over 50 km are common on Earth, Mars, and Venus; this difference is attributed to the lack of lunar shallow neutral buoyancy zones affecting eruption processes (Head and Wilson, 1991, 1992). Also conspicuous in their absence are calderas; these represent shallow magma reservoirs at shallow neutral buoyancy zones. Finally, several volcanic complexes (Fig. 4e) with unusual concentrations of the above features are seen; these are likely sites of multiple high-volume eruptions and the source regions of much of the surrounding lava. The Marius Hills area ( km 2 ) contains 20 sinuous rilles and over 100 domes and cones while the Aristarchus Plateau/Rima Prinz region ( km 2 ) contains 36 sinuous rilles (Whitford-Stark and Head, 1977). We assess the detectability of such features under MESSENGER viewing geometry and lighting conditions. Observations Important volcanic sites were selected for this study representing the range of volcanic features described above. The volcanic complexes of Marius Hills ( 12 N, 310 E) and Aristarchus Plateau/Rima Prinz ( 25 N, 315 E) contain an assortment of features including sinuous rilles and volcanic domes. The Apollo 15 landing site near Hadley Rille ( 25 N, 3 E) is a classic example of a sinuous rille. Mare Imbrium contains many lava flow fronts; the example used here is at 33 N, 335 E. Additional volcanic shields are found in Mare Insularum ( 7 N, 330 E). Images were taken from the Clementine lunar basemaps at 100 and 500 m/pixel, and 2.5 km/pixel resolutions. For comparison, the average resolution of Mariner 10 images was 1.5 km/pixel while the highest resolution images were 90 m/pixel (Spudis and Guest, 1988). The anticipated average resolution for MESSENGER is 250 m/pixel and the highest resolution images will be 25 m/pixel (Solomon et al., 2001). Image Resolution Figure 5 shows the three images for Hadley Rille, a 120 km long, 1.5 km wide sinuous rille located at the edge of Mare Imbrium. Figure 5a shows the rille at 100 m/pixel; here the feature is clearly identifiable. At 500 m/pixel (Fig. 5b) this rille is still easily identified, although its outline is pixelated. However, at 2.5 km/pixel resolution (Fig. 5c) the rille is not recognizable. Resolutions of ~500 m/pixel (or slightly lower) are required to identify this feature. Rima Prinz is a volcanic complex containing an extensive series of sinuous rilles (Fig. 6). Four of these rilles are marked with arrows in Fig. 6a; the rilles are clearly recognizable in the 100 m/pixel image. In the 500 m/pixel image (Fig. 6b), two of these rilles are still identifiable; however the smaller rilles (upper left and furthest right) are much harder to locate. Finally, at 2.5 km/pixel none of the rilles can be recognized. Resolutions between 100 and 500 m/pixel are needed for rille identification. Figure 7 contains the same volcanic domes in Mare Insularum as Fig. 4d; they are much more difficult to identify in the Clementine lunar basemap image due to their different Sun elevation angles (see following section on viewing geometry). Arrows in Fig. 7a point to the locations of these FIG. 5. Hadley Rille ( 25 N, 3 E) at multiple resolutions. (a) 100 m/pixels. (b) 500 m/pixels. (c) 2.5 km/pixels. All images from the Clementine lunar basemap.

9 Identification of mercurian volcanism 1217 FIG. 6. Rima Prinz at multiple resolutions. (a) 100 m/pixels. (b) 500 m/pixels. (c) 2.5 km/pixels. All images from the Clementine lunar basemap. FIG. 7. Domes located in Mare Insularum at multiple resolutions. (a) 100 m/pixels. (b) 500 m/pixels. (c) 2.5 km/pixels. All images from the Clementine lunar basemap. FIG. 8. Flow front in Mare Imbrium at multiple resolutions. (a) 100 m/pixels. (b) 500 m/pixels. (c) 2.5 km/pixels. All images from the Clementine lunar basemap. domes; the leftmost dome is easiest to identify based on albedo differences with the surrounding material. This dome is recognizable in the 100 m/pixel image (Fig. 7a). In the 500 m/ pixel image (Fig. 7b) it is difficult to distinguish the dome from other albedo variations in the area, and in the 2.5 km/pixel image none of the features in this region are identifiable. To identify small domes, resolutions between 100 and 500 m/pixel are necessary. Flow fronts are also very difficult to identify in the Clementine lunar basemap. Figure 8 contains the same Imbrium flow front found in Fig. 4d. The flow front is outlined in Fig. 8a based on close comparison with low Sun Lunar Orbiter images. In several locations the edge of the flow corresponds to albedo boundaries. While the flow front itself is not visible in any resolution figure, these boundaries are recognizable in both the 100 and 500 m/pixel images (Fig. 8b

10 1218 Milkovich et al. FIG. 9. Two views of Marius Hills (~12 N, ~310 E) from ground-based telescopes. From The Consolidated Lunar Atlas (Kuiper et al., 1967). (a) Image C790. Sun angle is 26. (b) Image C4268. Sun angle is 6. and 8c, respectively). However, the 2.5 km/pixel image is so pixelated that the boundaries are unreliable. Viewing Geometry As mentioned in the above discussion of the Imbrium flow front and Insularum domes, viewing geometry plays a role in identification of volcanic features. This is also easily seen at Marius Hills, a volcanic complex with numerous, closely spaced domes as well as sinuous rilles, rimless depressions thought to be of volcanic origin, and mare ridges (Guest, 1971). Figure 9 shows two ground-based telescopic images of Marius Hills taken at different colongitudes. The image in Fig. 9a has a higher Sun elevation angle (26 ; Malin, 1978) and shows faint topographic features; however, it is difficult to interpret these features. In comparison, Fig. 9b was taken at low Sun elevation (6 ; Malin, 1978) and clearly contains domes and ridges. The only difference between these images is Sun elevation. Thus, lighting conditions are an important factor in identifying volcanic features in images. To summarize, in order to identify a volcanic feature on a planetary body, it is necessary to have 2 3 pixels containing that feature; thus the necessary resolution for identification is a function of landform size. For typical lunar volcanic landforms, resolutions of m/pixel are observed to be necessary for identification. Larger features such as sinuous rilles can be observed at 500 m/pixel, while smaller features such as domes require higher resolutions closer to 100 m/pixel. Additionally, many features such as domes and flow fronts require low Sun elevation angles in order to be identified. In order to unequivocally conclude that volcanism has been present in the history of a planet several clearly volcanic features ought to be identified. CANDIDATE FLOW FRONTS ON MERCURY Although the majority of Mariner 10 images are at resolutions and Sun elevation angles inappropriate for identifying volcanism, a small number of images were taken at <500 m/pixel and near the terminator (low Sun angle). An examination of these images reveal possible flow fronts which can be interpreted to have formed a number of ways. This demonstrates the difficulty of definitively identifying volcanic features. Lobate scarps are relatively steep escarpments tens to hundreds of kilometers in length found across the surface of Mercury. They are characterized by rounded crests and lobate outlines on the scale of a few to tens of kilometers. 17% of observed scarps appear on the smooth plains on crater floors and form boundaries between different terrains (Strom et al., 1975); Dzurisin (1978) classified these scarps as irregular intracrater scarps. These scarps have been interpreted as flow fronts (Strom et al., 1976) and as tectonic features (Dzurisin, 1978). We examine three examples of candidate lobate flow fronts in Mariner 10 data and discuss various options for interpretation. These three examples come from the Shakespeare Quadrangle of Mercury (H-3) as mapped geologically by J. E. Guest and R. Greeley (1983) and are arrayed to the east of the Caloris Basin rim. The first example (Fig. 10) is a 180 m/pixel image and the most distal from the Caloris Basin rim ( 600 km). It shows the contact of smooth plains (right) embaying the Van Eyck Crater rim (left), composed of radially textured Odin Formation, a facies of the Caloris Basin ejecta material. The contact (location denoted by arrows) is sharp, and the smooth plains appear to embay the inner base of the crater rim. The western margin of the smooth plains locally consists of a westwardfacing scarp that is lobate and flow-like and appears to conform to preexisting topography. This flow could be interpreted as either volcanic or as an ejecta flow. However, this may not be a flow at all; in some places it is also similar to wrinkle ridges in morphology. If the entire feature were made up of wrinkle ridges then this would be a tectonic feature formed by compressive stresses in the region. This last possibility seems unlikely due to the embayment relations between the flow and the preexisting topography; a wrinkle ridge is more likely to cut through preexisting topography.

11 Identification of mercurian volcanism 1219 FIG. 10. Smooth plains (center of image) embaying the inner rim of crater Van Eyck. Arrows indicate flow-like scarps. Mariner 10 image , 180 m/pixel. The second example (Fig. 11) is a 280 m/pixel image and is at an intermediate distance from the Caloris Basin rim ( 450 km). It shows the contact of smooth plains (lower middle of image) with another subunit of the Odin Formation (hummocky plains), a facies of the Caloris Basin ejecta material (left and upper part of image). The western margin of the smooth plains (denoted by white arrows) appears to embay the Odin Formation and to flood and cover the hummocks that characterize it in this region. The contact consists of a westward-facing scarp that is sinuous in nature and appears to embay local topography. This is a particularly good example to illustrate both: (1) the lobate, flow-like nature of the contact, and (2) its morphological similarity to wrinkle ridges that are pervasive in the smooth plains (lower center and right in image). Could this contact be a tectonic wrinkle ridge? Again, if this feature is a flow front it is impossible to tell if it is a lava flow front or a crater ejecta flow front. The third example (Fig. 12) is a 300 m/pixel image and is closest to the Caloris Basin rim ( 230 km). This image shows two westward-facing lobate scarps (white arrows) in the Nervo Formation (hummocky, rolling plains), a facies of the Caloris Basin ejecta interpreted to be fallback mixed with impact melt. The surrounding hills consist of the Odin Formation, a gradational facies of Caloris Basin ejecta. Patches of smooth plains are seen in the upper right and lower middle. The morphology of the lobate scarps is strongly suggestive of flow and embayment, but also bears similarities to wrinkle ridge morphology (compare to Fig. 10). It is also difficult to distinguish the morphology from lava flow fronts in the Imbrium Basin on the Moon (Fig. 2c) and Orientale Basin ejecta flow lobes also on the Moon. Through this examination of candidate flow front features on Mercury, we see that in all cases it is impossible to tell if

12 1220 Milkovich et al. FIG. 11. Smooth plains (lower middle of image) in the Odin Formation. Arrows indicate margin of plains which appears to embay the Odin Formation. Mariner 10 image , 280 m/pixel. these features are due to volcanic or crater ejecta flows. Compositional data is needed to distinguish between types of flows; if the flow front also marks the boundary of two distinct compositional regions, then the flow is volcanic. Identification cannot be based on morphology alone. IMPLICATIONS FOR MESSENGER One of the major scientific goals of the MESSENGER mission is to determine the formation history of the surface of Mercury (Solomon et al., 2001). An important aspect of this is the role of volcanism in Mercury's history; MESSENGER needs to be able to search for volcanic landforms under suitable resolutions and viewing geometries to correctly address this issue. This requires adequate coverage of the planet at the resolutions and viewing geometries outlined above. The mission timeline and the anticipated image resolutions can be found in Solomon et al. (2001). The spacecraft camera system will build up a full global mosaic in the first 6 months after achieving orbit and cover the planet with stereo monochrome images at 250 m/pixel in the second 6 months. Two flybys prior to orbit insertion will allow 85% of the planet to be imaged in monochrome at 500 m/pixel and in color at an average resolution of 2.4 km/pixel. Since the spacecraft will be in an elliptical orbit, the narrow angle imager will be used during high altitude to achieve global monochrome image mosaics with an average 250 m/pixel resolution and a highest resolution of 25 m/pixel. Color images will average 1.1 km/pixel with targeted high-resolution color images at 300 m/pixel. The orbit itself starts in a dawn dusk configuration and moves through noon midnight back to dawn dusk many times during the mission (Solomon et al., 2001). Will this allow volcanic landforms to be identified? Most of the imaged 45% of the planet is covered with plains that may be volcanic in origin (Spudis and Guest, 1988). In the first 6 months the spacecraft will cover 50 % of the surface from near dawn dusk orbits. Even if the only plains of possible volcanic origin are the ones already observed, this orbit geometry will allow imaging of a portion of such plains at 250 m/pixel and with low Sun elevation angles. These are conditions favorable to the identification of volcanic landforms. Additionally, repeat coverage of the planet will allow for targeting of possible volcanic features with the high-resolution

13 Identification of mercurian volcanism 1221 FIG. 12. Lobate scarps in the Nervo Formation. The arrows indicate 2 lobate scarps. Mariner 10 image , 300 m/pixel. camera. Furthermore, MESSENGER multispectral imaging and spectrometer data will permit further assessment and characterization of candidate volcanic plains (Solomon et al., 2001; Robinson and Taylor, 2001) as well as distinguish between possible volcanic flow fronts and lobate ejecta flows. If volcanism exists on Mercury's surface, MESSENGER is well-equipped to find it. CONCLUSIONS The formation mechanism of the plains of Mercury has been debated since Mariner 10 revealed their existence. Both crater ejecta flows and volcanism have been suggested as mechanisms. Definitive identification of volcanism on Mercury requires concrete evidence such as an image of an unequivocal volcanic landform. No such image has been found; however, this does not mean that volcanism has been ruled out. Both resolution and viewing geometry play an important role in what sorts of features are visible in an image. To successfully identify volcanic features such as those on the Moon both low Sun angle ( 10 ) and 500 m/pixel resolution or better are needed. While Mariner 10 did not cover a significant portion of Mercury within these requirements, MESSENGER will be able to do so. The question of the existence of volcanism on Mercury must wait until then to be addressed more confidently. Acknowledgements Thanks are extended to Barbara Cohen and Laszlo Keszthelyi for helpful reviews which improved the manuscript. We gratefully acknowledge support from the NASA Planetary Geology and Geophysics program and the MESSENGER mission to J. W. H. Editorial handling: W. K. Hartmann REFERENCES BENZ W., SLATTERY W. L. AND CAMERON A. G. W. (1988) Collisional stripping of Mercury's mantle. Icarus 74, BLEWETT D. T., LUCEY P. G., HAWKE B. R., LING G. G. AND ROBINSON M. S. (1997) A comparison of mercurian reflectance and spectral quantities with those of the Moon. Icarus 129, DZURISIN D. (1978) The tectonic and volcanic history of Mercury as inferred from studies of scarps, ridges, troughs, and other lineaments. J. Geophys. Res. 83,

14 1222 Milkovich et al. GAULT D. E. AND WEDEKIND J. A. (1977) Experimental hypervelocity impact into quartz sand. II. Effects of graviational acceleration. In Impact and Explosion Cratering (eds. D. J. Roddy, R. O. Pepin and R. B. Merrill), pp Pergamon Press, New York, New York, USA. GUEST J. E. (1971) Centres of igneous activity in the maria. In Geology and Physics of the Moon: A Study of Some Fundamental Problems (ed. G. Fielder), pp Elsevier Publ. Co., Amsterdam, The Netherlands. GUEST J. E. AND GREELEY R. (1983) Geologic Map of the Shakespeare Quadrangle of Mercury. U.S.G.S. Misc. Inv. Map I 1408, 1:5,000,000. U.S.G.S., Denver, Colorado, USA. HAPKE B. (2001) Space weathering from Mercury to the asteroid belt. J. Geophys. Res. 106, HEAD J. W. AND GIFFORD A. (1980) Lunar mare domes: Classification and modes of origin. Moon Planets 22, HEAD J. W. AND LLOYD D. D. (1973) Near-Terminator and Earthshine Photography. NASA Spec. Pub 330, Sect. 4, NASA Washington, D.C., USA pp. HEAD J. W. AND WILSON L. (1991) Absence of large shield volcanoes and calderas on the Moon Consequence of magma transport phenomena. Geophys. Res. Lett. 18, HEAD J. W. AND WILSON L. (1992) Lunar mare volcanism: Stratigraphy, eruption conditions, and the evolution of secondary crusts. Geochim. Cosmochim. Acta 56, HEAD J. W., WILSON L., ROBINSON M., HIESINGER H., WEITZ C. AND YINGST A. (2000) Moon and Mercury: Volcanism in early planetary history. In Environmental Effects on Volcanic Eruptions: From Deep Oceans to Deep Space (eds. J. Zimbelman and T. Gregg), pp Kluwer Academic Publishers, New York, New York, USA. JEANLOZ R. AND MORRIS S. (1986) Temperature distribution in the crust and mantle. Ann. Rev. Earth Planet. Sci. 14, JEANLOZ R., MITCHELL D. L., SPRAGUE A. L. AND DE PATER I. (1995) Evidence for a basalt-free surface on Mercury and implications for internal heat. Science 268, KIEFFER W. S. AND MURRAY B. C. (1987) The formation of Mercury's smooth plains. Icarus 72, KUIPER G. P. ET AL. (1967) Consolidated Lunar Atlas. Univ. Arizona Press, Tucson, Arizona, USA. 24 pp. MALIN M. C. (1978) Surfaces of Mercury and the Moon: Effects of resolution and lighting conditions on the discrimination of volcanic features. Proc. Lunar Planet. Sci. Conf. 9th, MCCAULEY J. F. (1977) Orientale and Caloris. Phys. Earth Planet. Int. 15, MCCAULEY J. F. (1987) Orientale exterior. In The Geologic History of the Moon (ed. D. E. Wilhelms), pp U.S.G.S. Prof. Paper 1348, U.S. Geol. Survey, Denver, Colorado, USA. MOORE H. J., HODGES C. A. AND SCOTT D. H. (1974) Multiringed basins: Illustrated by Orientale and associated features. Proc. Lunar Sci. Conf. 5th, MURRAY B. C., STROM R. G., TRASK N. J. AND GAULT D. E. (1975) Surface history of Mercury: Implications for terrestrial planets. J. Geophys. Res. 80, MUSTARD J. F. AND HEAD J. W. (1996) Buried stratigraphic relationships along the southwestern shores of Oceanus Procellarum; implications for early lunar volcanism. J. Geophys. Res. 101, NOBLE S. K., PIETERS C. M., TAYLOR L. A., MORRIS R. V., ALLEN C. C., MCKAY D. S. AND KELLER L. P. (2001) The optical properties of the finest fraction of lunar soil: Implications for space weathering. Meteorit. Planet. Sci. 36, OBERBECK V. R. (1975) The role of ballistic erosion and sedimentation in lunar stratigraphy. Rev. Geophys. Space Phys. 12, OBERBECK V. R., QUAIDE W. L., ARVIDSON R. E. AND AGGARWAL H. R. (1977) Comparative studies of lunar, martian, and Mercurian craters and plains. J. Geophys. Res. 82, PIKE R. J. (1988) Geomorphology of impact craters on Mercury. In Mercury (eds. F. Vilas, C. R. Chapman and M. S. Matthews), pp Univ. Arizona Press, Tuscon, Arizona, USA. ROBINSON M. S. AND LUCEY P. G. (1997) Recalibrated Mariner 10 color mosaics: Implications for mercurian volcanism. Science 275, ROBINSON M. S. AND TAYLOR G. J. (2001) Ferrous oxide in Mercury's crust and mantle. Meteorit. Planet. Sci. 36, SCHULTZ P. H. (1977) Endogenic modification of impact craters on Mercury. Phys. Earth Planet. Sci. 15, SCHULTZ P. H. (1988) Cratering on Mercury: A relook. In Mercury (eds. F. Vilas, C. R. Chapman and M. S. Matthews), pp Univ. Arizona Press, Tuscon, Arizona, USA. SOLOMON S. C. (1975) Mare volcanism and lunar crustal structure. Proc. Lunar Sci. Conf. 6th, SOLOMON S. C. ET AL. (2001) The MESSENGER mission to Mercury: Scientific objectives and implementation. Planet. Space Sci. 49, SPOHN T. (1991) Mantle differentiation and thermal evolution of Mars, Mercury, and Venus. Icarus 90, SPRAGUE A. L. ET AL. (1994) Mercury Evidence for anorthosite and basalt from midinfrared ( µm) spectroscopy. Icarus 109, SPUDIS P. D. AND GUEST J. E. (1988) Stratigraphy and geologic history of Mercury. In Mercury (eds. F. Vilas, C. R. Chapman and M. S. Matthews), pp Univ. Arizona Press, Tucson, Arizona, USA. STROM R. G., TRASK N. J. AND GUEST J. E. (1975) Tectonism and volcanism on Mercury. J. Geophys. Res. 80, TONKS W. B. AND MELOSH H. J. (1992) Core formation by giant impacts. Icarus 100, TYLER A. L., KOZLOWSKI R. H. W. AND LEBOFKY L. A. (1988) Determination of rock type on Mercury and the Moon through remote sensing in the thermal infrared. Geophys. Res. Lett. 15, WHITFORD-STARK J. L. AND HEAD J. W. (1977) The Procellarum volcanic complexes; contrasting styles of volcanism. Proc. Lunar Sci. Conf. 8th, WILHELMS D. E. (1976) Mercurian volcanism questioned. Icarus 28, WILHELMS D. E. (1987) The Geologic History of the Moon. U.S.G.S. Prof. Paper 1348, U.S. Geol. Survey, Denver, Colorado, USA. 302 pp. WILSON L. AND HEAD J. W. (2001) Ascent and eruption of mare basalts on the Moon (abstract). Lunar Planet. Sci. 32, #1297, Lunar and Planetary Institute, Houston, Texas, USA (CD-ROM).