All DLE craters in this study have, at least, some of the attributes of fresh craters, but none of the DLE craters investigated appears to be completely pristine. Even the fresh appearing crater Bacolor (Figure 4), which will be used here extensively to illustrate the morphologic features of DLE craters, has a few superposed small impact craters, and fine-scale morphologic features such as dunes that indicate some postimpact modification.
 In spite of the effects of modification by the Martian surface processes, the ejecta deposits of DLE craters share a number of morphologic features that distinguish them from other Martian craters [Barlow et al., 2000], including (1) two ejecta layers, (2) strong radial texture composed of grooves and ridges that extends continuously from their rims to the outer edge of their continuous ejecta blankets, and (3) the apparent lack of secondary impact craters. We find the inner layer is typically ∼1–2 R (here R is defined as crater radius measured from the crater rim) wide compared with the outer layer and extends out to ∼3–6 R. This is in reasonable agreement with the ejecta mobility (EM) of DLE craters measured by Barlow [2005b] for the Viking-based database of 1.49 R average for the inner layer and 3.24 R average for the outer layer. Bridges and Barlow  and Barlow [2005b] also find that the outer edge of the inner ejecta layer (Γ = ∼1.04) of DLE craters is similar to that of SLE craters (Γ = ∼1.10), but less lobate than that of their outer ejecta layer (Γ = ∼1.14) or MLE ejecta (Γ = ∼1.18). Neither the inner or outer ejecta layers terminate in a relatively narrow distal rampart in a manner like that of SLE or MLE craters, but instead thicken more gradually toward their outer perimeters. High-resolution images of DLE crater ejecta deposits reveal few features that appear to be blocks on their surface of the ejecta, unlike the examples found on fresh MLE and SLE craters [Mouginis-Mark et al., 2003; Barnouin-Jha et al., 2005; Baratoux et al., 2005]. This lack of blocks may partly account for the remarkable absences of secondary craters around DLE craters (at least out to at least 11 R). In the rest of this section, we will describe the morphology of the ejecta deposits, and the (lack of) secondary craters of DLE craters in more detail.
 From the rim crest outward to ∼0.05–0.1 R, the slopes are steep, commonly near the angle of repose (Figure 5). Beyond this is a sharp break in slope of the surface to more gentle slopes.
Figure 5. Profiles from raw MOLA data showing that the ejecta layers of DLE craters are quite diverse. Each profile starts at the rim crest of the crater. For all craters a depression or “moat” (denoted by “M” on each profile) surrounds the rim of the crater, with the relative depth of this moat generally related to the size of the crater. In addition, the inner ejecta layer typically terminates in a broad distal rampart. The size of this rampart is also generally related to the size of the crater. The outer ejecta layer is thinner than the inner layer and is generally of relatively uniform thickness. Like the inner ejecta layer the outer layer terminates in a broad low rampart. Profiles are taken from the following craters and MOLA profiles: (a) 13.4 km crater at 34.9°N, 102.6°E, MOLA orbit 19296; (b) 16.2 km crater at 45.4°S, 25.7°E, MOLA orbit 16620; (c) 8.3 km crater at 42.41°N, 344.98°E, MOLA orbit 20051; (d) Bacolor crater at 33.0°N, 118.6°E, MOLA orbit 20322 (see Figure 4 for location); (e) the crater Arandas, 24.8 km in diameter, located at 42.4°N, 345.0°E, MOLA orbit 12307 (see also Figures 9a and 9b); and (f) 17.4 km crater at 38.0°S, 16.4°E, MOLA orbit 13058.
Download figure to PowerPoint
 The rims of the DLE craters measured in this study are nearly the same as those predicted from empirically derived rim height to diameter relationship by Pike  for lunar craters, and recent work by Steward and Valiant  that take into account regional strength difference in Mars terrains. For example, the rim of the freshest appearing DLE crater, Bacolor, is ∼680 m high (with a measurement error of ∼10%), compared with ∼788 m predicted by Pike , and ∼710 m predicted by Steward and Valiant . Considering the predicted uplift of the bedrock at the rim of ∼430 m [Melosh, 1989; Roddy, 1978], the initial thickness of ejecta at the rim of Bacolor might have been as much as ∼250 m. In addition, it should also be noted that rim uplift may be accentuated in water-rich targets [Greeley et al., 1980; Steward et al., 2000] and as a result may indicate that little, if any, ejecta is missing from the rim, a possibility that is consistent with the morphologic freshness of Bacolor.
 In addition, on the basis of the predicted uplift and topographic relief, the thickness of ejecta at the break in slope at the base of the rim may be <∼100 m around Bacolor, much less than predicted [McGetchin et al., 1973]. This apparent thickness of ejecta may indicate that (1) ejecta have been preferentially eroded from near the rim; (2) rim uplift was not as great as predicted, or (3) the ejecta emplacement process resulted either in less material being deposited on the rim than predicted by present models or collapse because of water in the target materials causing the rim to flow outward [Barnouin-Jha et al., 2005]. The present data do not provide information to resolve this question.
2.2. Inner Ejecta Layer
 The morphology of the inner ejecta layer of DLE craters is different than MLE and ballistic ejecta impact craters on Mars. The ejecta in the inner layer of DLE craters is generally relatively thin close to the rim, but thickens outward to form a broad, low rampart-like ridge on the outer edge (Figure 5). This is generally similar to SLE crater ejecta, but the rampart around SLE craters is narrower. This contrasts with ballistic ejecta craters whose ejecta thins rapidly and continuously outward. For MLE craters, the ejecta rapidly thins near the rim but decreases slightly outward in each layer to their terminal rampart [Garvin and Frawley, 1998; Garvin et al., 2000].
 In cross section, the inner ejecta layer of all DLE craters is typically thinnest from the rim out to ∼0.5 R (e.g., Figures 5a and 5d) forming a depression that resembles a “moat” around the crater. However, around some DLE craters the topography of the moat can be quite subdued (Figure 5c) in places. On the basis of MOLA measurement (assuming the preexisting surface under the ejecta was a flat plain), ejecta in this moat can be only a few tens of meters or less thick [Garvin and Frawley, 1998; Garvin et al., 2000].
 An estimate of ejecta thickness in this zone can be made based on the degree of partial burial of small preexisting impact craters by using their predicted initially morphology compared with the present shape. In the case of the small craters found in the inner ejecta layer of Bacolor shown in Figure 6 the ejecta thickness (at their locations of ∼0.3 R) are estimated to be ∼45 m thick based on the degree of their burial. In Figure 6 crater “a” is ∼1.2 km in diameter while crater “b” is ∼1.75 km diameter. A MOLA profile (orbit 11007) crosses the center of the crater “a” indicates its rim relief is ∼20 m. Assuming that it was fresh before burial, a rim height of ∼65 m (using the rim height to diameter relationship from Roddy  for Meteor Crater) and an ejecta thickness of ∼45 m are estimated. Estimating an ejecta thickness at crater “b” requires scaling its rim height from its diameter [Pike, 1977] (no MOLA trace cross it) and suggests an initial rim height of ∼96 m. Considering the steep appearance of the outer slopes of the rim (assuming them between ∼20°–25°) then the present relief of the rim is ∼42–55 m. As a result the ejecta deposit thickness is estimated to be ∼48 ± 6 m, consistent with that of the crater “a”. This value is also consistent with ejecta thicknesses reported by Garvin and Frawley  for Martian fluidized ejecta craters the size of Bacolor.
Figure 6. Inner ejecta layer of Bacolor crater (see Figure 4 for location). Crater “a” is ∼1.2 km in diameter, while crater “b” is ∼1.75 km diameter. Using the predicted shape of these craters compared with their present shape suggests that they have been partially buried by ∼45 m of ejecta. This is part of THEMIS image V09670012.
Download figure to PowerPoint
 Outward from the moat, starting at 0.3–0.5 R, the inner ejecta layer of DLE craters typically thickens into a broad annular ridge that resembles a broad rampart (see Figure 5). For Bacolor, ejecta near the outer edge of this annular ridge is ∼200 m thick. On large DLE craters (i.e., >∼15 km diameter) this ridge abruptly terminates in a steep scarp of jumbled and elongate blocks. However, on smaller DLE craters the scarp is lower relief and such blocks are uncommon.
 This distribution of ejecta in the inner layer is illustrated by the ejecta thickness function (ETF) [McGetchin et al., 1973; Garvin and Frawley, 1998] for DLE, compared with other types of craters. The ETF describes how ejecta thin away from the parent crater. Garvin and Frawley  found that the exponent of the ETF function (that describes the rate of change in ejecta thickness outward) was commonly larger for Martian craters (average ∼−4, but has considerable variability) than predicted by the equations of McGetchin et al.  for ballistic ejecta (i.e., −3 ± 0.5). Garvin and Frawley  suggested that this was the result of the more rapid thinning of ejecta away from fluidized ejecta craters than away from ballistic ejecta craters. The measurements of Garvin and Frawley  are consistent with the ETF exponent of ∼−4.5 we found for Bacolor, out to ∼0.5 R of the rim (i.e., the outer edge of the moat of DLE craters). However, outward from the outer edge of the moat of Bacolor to the outer edge of its inner ejecta layer (at ∼1R) the ETF exponent becomes positive as ejecta thickness increases. In addition, as much as 25% of the total ejecta may be in this outer ring of material in the inner ejecta layer of large DLE craters.
 One of the most prominent and diagnostic features of the inner ejecta layer is its radial texture. This texture is unique to DLE craters and is composed of closely spaced, straight, flat-floored, steep-sided grooves cut into the surface of the inner layer (Figure 7). These features commonly start on the ejecta flap deposits (sometimes even on the steep slopes below the crater rim crest) and extend across the entire inner ejecta layer and many times connect uninterrupted with morphologically different radial grooves and ridges on the outer ejecta layer (see Figure 1). This suggests that the same process produced the radial features on both layers and operated continuously from the rim (or very near it) out to the outer edge of the outer ejecta layer.
Figure 7. Striations (grooves and ridges) that have developed near the rim of Bacolor. Such features are common on the rims of most DLE craters and suggest that the process that formed them operated from the rim outward. In addition to the radial groove and ridges, sets of troughs develop roughly concentric to the center of the crater on DLE craters. Individual trough segments are generally several hundred meters across and 1–3 km long. In places the grooves cut across the troughs indicating that the troughs formed before the grooves formed. This is part of THEMIS image V12141005. See Figure 4 for location. Inset shows higher-resolution view of the same striations on the ejecta blanket. Mars Orbiter Camera (MOC) image E1003462 is 6.4 m/pixel resolution.
Download figure to PowerPoint
 Individual grooves can be as wide as 300 m, and on the basis of estimates from MOLA profile data, a few tens (generally < 20 m) of meters deep, or as narrow as a few tens of meters and a few meters deep (“Grooves” in Figure 7). The depth of these features appears to be directly related to their width. In addition, the slopes along the sides of the groove are typically relatively steep. Locally, the tops of the ridges in between the grooves are typically flat and appear to be at the same general elevation level on either side of a particular groove, no matter the size of the groove or the width of the ridge. These relationships suggest that the relief on the radial features is a result of erosion of the grooves into the ejecta deposit of the inner layer instead of deposition of materials to form the ridges.
 In addition, within the inner ejecta layer, the grooves and ridges cross small preexisting features such as small impact craters (see Figure 6) or hills without changing direction. For large obstacles, such as the hill ∼0.4 R northwest of the rim of the 16 km diameter crater shown in Figure 8a (from its morphology and distance from the crater rim this hill appears to be a mesa like other mesas found in that region and not a block of displaced rim), most of the radial grooves and ridges terminate partly up its craterward side and others continue again a short distance on the other side, but maintain the radial direction (Figure 8b.). The grooves and ridges close to the mesa, on the craterward side, are <50 m wide and <100 m apart. There is no evidence for ejecta deposits on top of this obstacle, although, smooth material is found on the lee side of this mesa. This smooth material is most likely a deposit formed by deceleration of ejecta in the shadow of the obstacle. Considering the momentum of ejecta flowing across the surface (modeled as the velocity required for momentum to roll a rounded meter diameter basalt boulder to the summit of the hill), in order for these obstacles to be topped by the ejecta, it must have exceeded 50 m/s, indicating that ejecta flow must have been at least that velocity. Such an interpretation is consistent with the observation of Baloga et al. .
Figure 8a. A 16 km diameter crater in W. Utopia Planitia (located at 35.3°N, 88.1°E) illustrating two examples of ejecta flow diverted by preexisting obstacles. The locations of Figures 8b and 11a, which provide higher-resolution views, are shown. This is a mosaic of THEMIS images I09721009 and I10657011.
Download figure to PowerPoint
Figure 8b. An ∼5 km long and ∼50–75 m high hill ∼0.4 R from the rim of the crater (see Figure 8a for location). Striations run up to this hill and become subdued, or die out, over it but resume on the downrange side (labeled “a”). Ejecta also appear to have been deposited as an apron in the lee of the hill (labeled “b”). This is THEMIS image V10657012.
Download figure to PowerPoint
 In addition to these radial structures, there are features that form in roughly concentric patterns with the crater center. Some of these features are graben-like, flat-floored, steep-sided troughs ∼300–400 m wide and ∼20–30 m deep that develop most commonly in the moat area. They are ∼1–3 km long (“Troughs” in Figure 7), but typically interconnect to form chevron or scallop patterns. In places, they grade into outward facing scarps a few to tens of meters high. The radial grooves and ridges cut across the floors and walls of these troughs (“Grooves” in Figure 7) indicating that the troughs formed before the radial striations, but how much before cannot be determined. The generally concentric nature and graben-like morphology of the troughs suggests that they may be extensional deceleration features produced as the ejecta halted. Alternatively, these features could be deflation features produced as the ejecta out gassed, although this requires outgassing to be extraordinarily rapid in order to occur before groove formation. These features are not observed in the ejecta of other types of fluidized ejecta craters and may indicate differences in rheological behavior between the ejecta of the inner layer of DLE craters and other types of craters.
 In addition to these troughs, on some craters like Arandas (43°N, 344°E), the outer edge of the inner ejecta layer has developed sets of ridges that follow the outline of the edge of the ejecta layer and wrap around preexisting obstacles [Carr et al., 1977] (Figures 9a and 9b). The geometry of these ridges suggests that they are produced as the ejecta flowed around and piled up against the obstacles shortly before it stopped. Consequently, these features appear to be compression features similar to pressure ridges.
Figure 9a. Ridges on some DLE craters such as Arandas (24.8 km, located at 42.4°N, 345.0°E). The box shows the location of Figure 9b. This is a mosaic of THEMIS images V03346003, V09812010, V11035005, V11347008, V12882007, V13194006, and V13793015.
Download figure to PowerPoint
Figure 9b. Detailed view of the ridges (arrowed) upslope of preexisting craters. These ridges are concentric with the edge of the inner ejecta layer and wrap around preexisting features such as the crater at bottom left. These are most likely to be pressure ridges that developed as the leading edge of the ejecta began to slow and the materials accumulated upslope. This is THEMIS image V09812010.
Download figure to PowerPoint
 The inner ejecta layer of larger DLE craters (i.e., >∼15 km diameter) commonly terminates in a scarp of hummocks, jumbled and elongate blocks of ejecta (“a” in Figure 10), while for smaller DLE craters (i.e., <∼15 km diameter) this zone may have not developed (see Figure 1). Typically, the width of this zone of jumbled blocks is proportional to crater size. On large DLE craters such as Bacolor, this zone can be as much as 6 km wide, with >100 m of relief above the lower outer ejecta layer. Most of the grooves on the inner ejecta layer terminate at this zone indicating that this zone formed after the grooves, although how long afterward cannot be determined form the data. On small DLE craters where this zone is absent or not well developed, the radial grooves frequently cross it uninterrupted to the outer layer (Figure 1).
Figure 10. Transition zone at Bacolor between the inner ejecta layer and the outer ejecta layer abruptly terminating outward in scarps over 100 m high. This zone typically includes elongate blocks of ejecta concentric with the edge of the zone, suggesting that they may be slump blocks that indicate failure of the slopes along the scarp (labeled “a”). In a few places, small patches of ejecta with radial grooves have been preserved in this zone in spite of the slumping that has occurred along the scarp (labeled “b”). While most radial grooves terminate in this zone, suggesting slumping occurred after formation of the grooves, a few of the largest grooves have been preserved and can be traced across the zone and out into the outer layer. This is part of THEMIS image V12141005. See Figure 4 for location.
Download figure to PowerPoint
 Some of the largest of the blocks along this scarp form ridges several hundred meters wide and several kilometers long that parallel the outline of the scarp. The nature of the large blocks suggests slope failure along the scarp, causing sliding and slumping. This implies that the mechanical properties of the ejecta were first sufficient to bring to rest a thick layer of ejecta, but later at some unspecified time the strength of this layer was exceeded by the shear stress exerted by its own weight and it slumped. Such slumps and slides have been identified from Viking images for the MLE crater Bamburg [Mouginis-Mark, 1979b], but THEMIS data reveal that this slumping is much more common at many other craters.
 Occasionally, preexisting topographic features such as small mesas or craters are found in this disrupted zone. The way the ejecta flowed around or over such features provide information about the flow properties of the ejecta at that distance from the crater [Baloga et al., 2005]. For example, the double-layered mesa located ∼11.8 km (1.36 R) from the rim crest of a 17.3 km diameter crater in western Utopia Planitia (at 36.1°N, 87.9°E) (Figure 11) is in the outer edge of the disrupted zone between ejecta layers. MOLA data (orbit 15435) show that this mesa has a maximum height of ∼373 m above the surrounding terrain, with the lower bench of the mesa at ∼150 above the surrounding terrain. The mesa has created a barrier to the radial flow of the ejecta with the inner ejecta layer butting up against it on the side facing the crater up to the level of the lower bench. Thin deposits of ejecta can be traced across the bench and diverted around the upper part of the mesa, but little, if any, is found on top or on the lee side of either level of the mesa (Figure 10 MOC image). This suggests that the outward flow of the inner ejecta layer halted at the mesa, while ejecta of the outer ejecta layer flowed around, but not over or fell from above on to the summit of the mesa. This implies that the velocity of the ejecta at the boundary between ejecta layers was too low for momentum to carry it over the summit of the upper mesa, or a velocity of the <∼75 and 100 m/s (modeled as the velocity required for momentum to roll a rounded meter diameter basalt boulder to the summit of the hill). This is consistent with that found by Baloga et al.  for SLE and MLE craters and is a factor of 2–3 times slower than if they transported along ballistic arc to this location. It should also be kept in mind that while ejecta deposits are not apparent on the summit of the mesa, it is also possible that ejecta may have been deposited there but drained off or are too thin (or scours too small) to be observed with MOC images of ∼7 m/pixel.
Figure 11. (a) A 17.3 km diameter crater within W. Utopia Planitia (at 36.1°N, 87.9°E) clearly demonstrating the ability of preexisting obstacles to divert the ejecta flow from DLE craters in the zone between ejecta layers. This is THEMIS image V10345016. (b) High-resolution MOC image showing that while the outer layer of ejecta can be traced over the top of the bench of this obstacle and diverted around the upper part of the mesa, no ejecta or scours are found on the summit of the obstacle. This is part of MOC image E1200669 with a resolution of 6.4 m/pixel.
Download figure to PowerPoint
2.3. Outer Ejecta Layer
 From Viking images, the outer ejecta layer of DLE craters was interpreted to be morphologically similar to the MLE craters ejecta [e.g., Carr et al., 1977], but this no longer appears to be the case, based on our analysis of the THEMIS, MOLA and MOC data. The new data have revealed that, like the inner ejecta layer, the outer ejecta layer has unique attributes that distinguish it from ejecta layers of other fluidized ejecta craters as well as from the inner ejecta layer.
 Like the inner ejecta layer, one of the most distinctive characteristics of the outer ejecta layer is its radial texture. This texture is composed of numerous radial grooves, troughs, and ridges (Figure 12). Frequently, the troughs are bounded with levee-like ridges that give them a leveed channel appearance that may have formed as a result of the low-velocity flow of ejecta. These landforms can be as much as 150 m across and a few meters to tens of meters deep, nearly that of the thickness of the ejecta deposit.
Figure 12. Flow paths (labeled “a”) within the outer ejecta layer of Bacolor showing that, in some instances, the flow is not exactly radial to the parent crater. Local obstacles or topographic depressions divert the flow. In the right center part of the image a small tongue of ejecta (labeled “b”) extends past a major lobate ejecta scarp, indicating a complicated flow history during emplacement of the ejecta. This is part of THEMIS image V05451015. See Figure 4 for location.
Download figure to PowerPoint
 While the direction of the grooves on the inner ejecta layer is relatively unaffected by the presence of preexisting obstacles, the radial texture of the outer ejecta layer appears to curve around preexisting topographic features (“a” in Figure 12), although frequently the radial texture also can be traced over low-relief obstacles. In order for the ejecta to develop flow lines with these characteristics, it must be traveling as a near surface flow at relatively low velocity. This is consistent with the channel-like morphology and the relief of these features relative to the layer thickness, and suggests that the radial features are produced by deposition and not erosion like the grooves on the inner ejecta layer.
 MOLA-derived topographic data indicate that the outer ejecta layer is commonly only a few meters to tens of meters thick and varies little in thickness (see Figure 4) much like the fluidized ejecta deposits around SLE and MLE craters [Garvin and Frawley 1998; Garvin et al., 2003; Barnouin-Jha et al., 2005; Baloga et al., 2005]. In addition, the outer ejecta layer commonly thickens to a low rampart-like ridge near the outer edge where it terminates in low, lobate scarps. Some of these lobes overlap indicating a complex emplacement history within this layer. In contrast to rampart craters (i.e., SLE and MLE craters) that have pronounced distal ramparts that may be 80–170 m high around large craters [Garvin and Frawley, 1998; Mouginis-Mark and Baloga, 2006], MOLA profiles (Figures 13a and 13b) reveal that, even on large DLE craters such as Bacolor, the height of the DLE rampart is considerably lower, <35 m, and may be as low as 20 m.
Figure 13a. Distal southern margin of the Bacolor ejecta blanket showing the location of the four MOLA profiles (labeled “a”–“d”) illustrated in Figure 13b. These MOLA data confirm that DLE craters, while they still have distal ramparts, have much less pronounced relief than either SLE or MLE craters. This is a mosaic of THEMIS images V11829005 and V12141005.
Download figure to PowerPoint
Figure 13b. Profiles of the distal edge of the ejecta blanket of Bacolor shown in Figure 13a. Maximum height of the scarp at the edge of the ejecta is ∼35 m, and craterward (to the left) from the rampart the outer ejecta layer is of the order of 10 m thick. Topographic data derived from the following MOLA orbits (a) 15195, (b) 19972, (c) 13560, and (d) 12818.
Download figure to PowerPoint
 In a few locations, thin subtle, ridge-like features extend off and outward radial from the lobate edge of the outer layer (see “b” in Figure 12). These features are typically a few tens of meters wide with a few meters of relief (judging from their shadows), and extend for a few kilometers or less from the edge of the outer ejecta layer. On Bacolor, these ridge-like features appear to become more subdued and grade into faint radial texture that may extend outward for nearly a crater radius from the outer edge of the outer ejecta layer. However, because these features are so small, their exact size and extent away from the outer edge of the outer ejecta layer cannot be determined. This outer ejecta layer may be caused by flow separation, as predicted by the model of Schultz  or observed at Mount St. Helens [Kieffer, 1981] where the winds generated by the advancing flow carried fine grain materials up and over the near surface atmosphere as the ground hugging debris halted.
 These subtle ridge-like features may be the erosional remnants of a thin, easily eroded, outer ejecta deposit such as that observed on some of the freshest DLE craters found in high-latitude regions of Mars (Figure 14). In THEMIS images taken at 36 m/pixel resolution, IR images taken at ∼100 m/pixel resolution and MOLA profiles, such fresh craters have the attributes common to all other DLE craters (e.g., inner ejecta layer with a pronounced inner moat and broad rampart whose surface is cut by straight grooves cut into the inner layer, lack of secondary craters outward form the continuous ejecta), but with the addition of a thin (∼10–25 m thick based on MOLA profile data), relatively smooth ejecta deposit that extends outward from the outer edge of the second ejecta layer (Figure 14). The outer edge of this highly irregularly shaped ejecta layer extends from the outer edge of the second ejecta layer to an average of ∼6–8 R from the crater rim. In addition, there are narrow strands of ejecta that extend to ∼10–12 R from the crater rim. This ejecta layer terminates in a low, steep scarp, but does not appear to have distal rampart ridges.
Figure 14. Fresh, high-latitude impact crater having the same general morphologic characteristics as all DLE craters but including a thin (uniformly thick at <∼10–25 m), highly irregularly shaped, outer ejecta layer located beyond the second ejecta layer. The crater is 13.0 km in diameter and 1201 m deep and is located at 72°N, 38°E. Boxes show the locations of enlargements in Figure 15. This is a mosaic of THEMIS VIS images V10709002, V11358008, V09798018, and V10085009 and IR image I03669002. The resolution of the VIS images is ∼36 m/pixel, and the IR image is ∼100 m/pixel.
Download figure to PowerPoint
 Remarkably, in places these outer ejecta deposits are dissected by chains of shallow elongate pits oriented approximately radial to the crater (arrows Figure 15). No such pits are found beyond this ejecta layer on the surrounding terrain. The depths of the pits appear to be no deeper than the thickness of the ejecta layer and consequently would be completely erased if the deposits were removed by erosion. While these pits have the appearance of secondary craters, their development on the outer ejecta deposit is inconsistent with current models for the timing of fluidized ejecta emplacement and secondary crater formation [e.g., Melosh, 1989; Mouginis-Mark et al., 2003; Mouginis-Mark and Baloga, 2006] and would be expected to have formed first and be overridden by the continuous ejecta deposits. However, if these pits are secondary craters the ejecta blocks that formed them appear to have been capable of excavating this thin ejecta layer but not the material beneath and/or beyond. This suggests that these ejecta blocks were composed of weak, easily disaggregated materials. Moreover, we suggest that the apparent ease at which this outer ejecta layer erodes compared with the other two ejecta layers (that remain around older craters after its disappearance) is also most consistent with this outer ejecta layer being composed of fine-grain, loosely consolidated, easily eroded materials. Alternatively the pits could be sublimation pits (i.e., alases) or some other type of erosional feature or these craters could be a new type of multilayered crater found only in high-latitude regions. Higher-resolution images (i.e., 18 m/pixel THEMIS, HiRISE and MOC images) of these craters and their ejecta deposits are required to confirm their exact nature.
Figure 15. Chains of pits (arrows) found in the outer ejecta deposit but not beyond of the crater. These pits lead away from the center of the crater and may be secondary craters from this crater. THEMIS VIS images (a) V11358008, (b) V10709002, and (c) V10709002. The resolution of these VIS images is ∼36 m/pixel. See Figure 14 for the location of these images.
Download figure to PowerPoint
2.4. Secondary Craters
 One of the most remarkable features of DLE craters is that they do not appear to have secondary craters, at least out to ∼11 R from the rim. Even though it had been suggested from Viking-based observations, that at least one DLE crater (i.e., Arandas crater) may have distant secondary craters [Schultz and Singer, 1980], new higher-resolution THEMIS VIS data do not support this contention. In contrast, most fresh SLE and MLE craters (see Figures 3 and 12) show secondary craters similar to those around primary lunar craters [Mouginis-Mark et al., 2003]. Furthermore, we can find no circular depressions within the ejecta layers that suggest that secondary craters were initially formed but were then buried during the outward radial flow of the ejecta.
 It is possible that DLE craters formed secondary craters, but they occur beyond 11 crater radii away from the primary crater. While such a relationship between the primary and its secondary is not observed on other planets where high-resolution images of fresh craters are available, it would not be a unique situation on Mars for secondary craters to be absent close to the primary. For example, in the case of the crater Zunil, which has abundant secondary craters, few are found within less than a ∼10 R [McEwen et al., 2005]. This may also be the case for DLE craters, but until additional THEMIS data are collected around fresh DLE craters we will be unable to determine with certainty if any DLE craters produce secondary craters.
2.5. Morphometry of DLE Crater Ejecta
 The unique geometry of DLE crater ejecta deposits, as illustrated by the profiles in Figure 3, shows important differences and similarities compared to the ejecta of other types of craters. These provide insight into ejecta emplacement mechanisms. For example, the general shape and volume of the inner ejecta layer of DLE craters are similar to ejecta around SLE craters, while the general shape of the outer ejecta layer of DLE craters and MLE craters ejecta and total ejecta volumes are also similar. Such similarities in distribution of mass may point to similarities in the mechanism of transport of mass around these craters.
 In particular, similarities in the overall shape of the inner ejecta layers of DLE craters and SLE craters ejecta (i.e., rapidly thinning ejecta outward to a terminal rampart) suggest similarities in their emplacement. In addition, the relatively short runout distance (i.e., 1–2 R) of the inner ejecta layers of DLE craters and the continuous ejecta of SLE craters (compared with ballistic ejecta from craters of the same size) are counter to the expectation that fluidized ejecta should be relatively more fluid and run out farther than ballistic ejecta. This is most likely a result of different emplacement regimes of these different types of ejecta. Ballistic ejecta are generally emplaced through ballistic sedimentation processes [Oberbeck, 1975], while the fluidized ejecta of SLE and DLE craters may be emplaced through a process involving flow [e.g., Wohletz and Sheridan, 1983; Schultz, 1992; Barnouin-Jha et al., 2005]. Several mechanisms have been proposed for producing relatively short runout distance for fluidized ejecta. These mechanisms generally involve unique target material or atmosphere characteristic.
 There is evidence that the concentration of volatiles in the target, and the atmospheric pressure during crater formation may play a key role in controlling the processes of emplacement of fluidized ejecta, while target material characteristics (lithology i.e., initial grain size and coherence) may also be of importance. Wohletz and Sheridan  examined data from dry and wet vapor explosion experiments and suggested that stubby ejecta flows are to be expected for craters where targets contain relatively little water. This is because less water in the ejecta generally meant more friction between particles during flow and as a result shorter transport distances. While this mechanism may predict shorter runout distance for fluidized ejecta, it does not explain how those distances can be shorter than dry ejecta emplaced ballistically. However, McSaveney and Davies  have suggested another possible mechanism that could account for the short runout distance for some fluidized ejecta even though the ejecta may have relatively low viscosity. They proposed that the runout of ejecta over a thick eroded substrate, such as the layered deposits found in the midlatitudes of Mars, would quickly remove energy from the flow because of the erosive nature of flowing particulate materials. While areas like the northern lowland plains [Head et al., 1998, 1999; Boyce et al., 2005] and southern highlands [Soderblom et al., 1973; Malin and Edgett, 2001] may contain easily erodible, layered deposits, DLE craters are also found in areas that show no evidence of such deposits, such as on Tharsis (e.g., Alba Patera). In addition, Schultz  has suggested that such parameters as grain size (lithology), variations in near-surface volatile content of the target and atmospheric pressure can have important effects on the runout distance of ejecta on Mars. While varying such parameters may result in relatively shorter runout distance of the inner ejecta layer, some require surface conditions on Mars to be different from craters in the same area at different times or a Martian crust with considerable lithologic heterogeneous.
 While runout distances are similar between SLE ejecta and the inner ejecta layer of DLE crater, the moats around the inner ejecta layer of DLE craters are narrower and the rampart broader and higher than around SLE craters. Schultz [1992, 2005] suggests that moats could be the result of intense erosion of fine-grain ejecta deposits by tornadic-strength vortices of atmospheric gas generated by an advancing ejecta curtain, although such erosion could also occur through the high-velocity outflow of particles and gas associated with base surge. Base surges, sometimes called column surges, are dense, ring-shaped cloud of ejected particles and gas generated by the collapse of a vertical explosion column of ejected fine-grain particles and gas (on Earth most of the gas is derived from the atmosphere) produced as a result of impact or explosion crater formation, and during some highly energetic explosive volcanic eruptions [Young, 1965; Gladstone and Dolan, 1977; Pohl et al., 1977; Kenkman and Schönian, 2005; Wohletz, 1998]. These turbulent, high-velocity clouds of ejecta sweep outward from the crater as a near-surface density flow and are also capable of substantial erosion. Both mechanisms are strongest and scour the surface more near the rim (i.e., in the area of the moats), probably entraining more materials as the flow outward. Outward, as the velocity in the flows drops and deposition is permitted. However, even taking into account materials eroded from the moat, for the outer ejecta layer to be mainly deposited from material carried aloft in a base surge would require nearly 30–40% of the total ejecta volume to be lofted in the explosion column. In any case, these mechanisms require that the ejecta deposits include large amounts of fine grain materials in order for the high-velocity winds to remove significant amounts of materials to produce the moats but only modestly eroding the small preexisting craters found in the moat zone (see Figure 5).
 Alternatively, the differences in the geometry of the moats may be a direct consequence of the nature of the materials and their mechanism of emplacement in the inner ejecta layer. For example, differences between DLE craters and SLE craters may only be that the ejecta were emplaced as a granular flow (probably after initial ballistic ejection from the crater) and that the ejecta of DLE craters contained materials with a different particle size distribution, gas concentration or grain size distributions compared with SLE crater ejecta [Wohletz and Sheridan, 1983].
 The outer ejecta layer of DLE craters and that of MLE craters also show similarities in overall planimetric shape. A comparison of their shapes with those of the inner ejecta layer of DLE craters and SLE craters suggests more fluid behavior because of their long runout distance. However, smaller-scale flow features, such as the radial texture, on MLE crater ejecta and ejecta of the outer ejecta layer of DLE craters are significantly different and may suggest different emplacement mechanisms. Additionally, these small-scale features (e.g., flow lines) suggest that the two ejecta layers of DLE craters were each emplaced by different mechanisms, while all ejecta layers of MLE craters appear to have been emplaced by the same process.