Impact melt deposits have been identified in small, simple impact craters within the lunar highlands. Such deposits are rare, but have been observed in craters as small as 170 m diameter. The melt occurs as well-defined pools on the crater floor, as well as veneers on the inner crater wall and stringers of material extending over the rim and away from the crater. Model calculations indicate that the amount of melt formed in craters 100–2000 m diameter would amount to a few to ∼106 m3, representing <1% of the crater volume. Thus, significant, visible impact melt deposits would not be expected in such small craters as most of the melt material that was formed would be ejected. Variations in the properties of the projectile or the target cannot account for the amount of observed melt; the amount of melt produced is largely insensitive to such variations. Rather, we suggest that these small melt-containing craters represent near-vertical impacts in which the axes of melting and melt motion are essentially straight down, toward the base of the transient cavity. For a given event energy under vertical impact conditions, the volume of melt produced would be greater than in an oblique impact and the momentum of the material would be directed vertically downward with minimal lateral momentum such that most of the melt is retained within the crater interior. Since vertical impacts are relatively rare, such small craters with visible, interior melt deposits are rare. While we focus here on the highlands, such craters also occur on the maria.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 Impact melt is typically observed in large (multikilometer diameter) simple and larger complex impact craters on the terrestrial planets (Figure 1). The melt forms pools on the crater floor, fills low regions on faulted terrace blocks and in depressions on the exterior ejecta blanket; it also occurs as flows and veneers on the inner walls and outer rim. While it is typical of large simple and complex impact craters, significant volumes of impact melt have rarely been observed in small (<few km) simple lunar craters. Previous studies of lunar crater melt deposits have established that they are of shock-generated melt material, rather than volcanic in origin [e.g., Howard and Wilshire, 1975; Hawke and Head, 1976; Schultz, 1976]. Those studies also suggested that deposits of impact melt for craters having diameters less than a few kilometers would be relatively rare.
 Using Lunar Reconnaissance Orbiter Camera (LROC) images [Robinson et al., 2010] of the lunar highlands, a number of simple impact craters, as small as about 170 m in diameter, have been recognized with pools of impact melt on their floors (Figures 2 and 3). Such deposits are only observed unequivocally in fresh craters and are rare even then. While numerous small lunar craters have flat floors, the nature of that floor material beneath the surface regolith (debris versus impact melt) typically cannot be determined.
 This work demonstrates that significant quantities of impact melt can be produced, retained and recognized in small impact craters, at diameters which are smaller than would be expected based on theoretical estimates. Further, it provides insight into the cratering process in general, particularly the mechanisms of energy distribution within the transient cavity and the formation and distribution of impact melt. The observation that pools of impact melt can form even in such small craters indicates that there are considerably more source areas for melt rocks in the Apollo and Luna collections than previously recognized. If the proposed mechanism (vertical or near-vertical impacts) were correct, it would allow for a better estimate of the frequency of such events.
2. Previous Work
 In the early work on lunar impact crater morphology, deposits with fluid-like morphology associated with impact craters were considered to be volcanic in origin [e.g., Strom and Fielder, 1968, 1970]. Later, from observations made of a number of lunar craters [Shoemaker et al., 1968; Guest, 1973] and based on observations at terrestrial impact structures [Dence, 1971; Grieve et al., 1977], it was suggested that such material could be shock-generated melt. Howard and Wilshire  examined some 20 lunar craters and concluded that the melt deposits had the same distribution characteristics as other ejected material and, thus, the melt material was transported as part of the excavation process and therefore was not volcanic in origin. That such materials are impact-derived melt has subsequently become the canonical interpretation.
 A number of studies have examined different aspects of impact-melt deposits, although most of those have focused on large craters such as Copernicus and Tycho (tens of km in diameter); only a few studies have considered smaller craters. Howard and Wilshire  suggested that most craters >20–30 km have impact-melt material on their floors, and further noted that some craters in the size range of 1–5 km had ponded, apparently uncracked, material around their margins. For craters <30 km, they indicated that flows typically would be absent and that perched ponds were common at craters >40–50 km. In cases of oblique impact, the melt was concentrated in the downrange direction, an aspect more recently considered by a number of studies [e.g., Gault and Wedekind, 1978; Pierazzo and Melosh, 1999, 2000; Schultz and D'Hondt, 1996; Anderson et al., 2004a].
Hawke and Head  examined melt on crater rims and observed rim ponds or flows of melt at craters having diameters as small as 4 km; they recognized 18 craters with diameters <20 km that had rim ponds or flows. In most cases, crater rims were coated with a veneer of melt that had locally coalesced into flows or ponds. Hawke and Head [1977a, 1977b, 1979] further examined interior melt deposits and concluded that they occurred in simple craters (<15 km) as ponds of low-albedo material on the floors, occasional dark streaks on the walls, and discontinuous veneers on the rims and crater walls (some of which had flowed onto the crater floors). The smallest crater they studied having interior melt deposits was 750 m in diameter, but they suggested that melt might be present in smaller craters. At the large end of the simple crater size range (around 10 km), exterior deposits of melt become common and flows of melt, as opposed to veneers, define the typical morphology. Schultz  reviewed the morphology of impact craters of various sizes. With respect to the morphology of their floors, it was noted that the flat floors of some might represent impact-melt deposits based on the their morphology (e.g., viscous flow, cooling cracks, low albedo, etc.).
 While impact-melt deposits were observed in a few small simple craters, the earlier work generally concluded that such deposits would be rare and the lower limit at which such deposits would form was undefined. The conclusion that such deposits would rarely be associated with small impact craters was also consistent with theoretical and experimental studies at the time, which suggested that most of the impact melt that was produced would be ejected from the crater interior rather than concentrated on the floor [e.g., Gault et al., 1968; Gault and Wedekind, 1978; Cintala and Grieve, 1998]. If only small volumes of melt were produced, the amount remaining in the interior would be small and unlikely to collect into a well-defined pool because of rapid cooling.
 These early studies were limited by the resolution, illumination geometry, and coverage of the preponderance of available imaging data (e.g., Lunar Orbiter). LROC and other recent lunar imaging system data sets provide high-resolution coverage over much larger fractions of the lunar surface at a variety of illumination geometries that allows better observation of small-diameter impact-crater morphology. This extensive, high-resolution coverage has allowed the recognition and evaluation of aspects of small impact-crater morphology that were not previously possible.
3. Data Collection
 LROC (Lunar Reconnaissance Orbiter Camera) images were examined to identify candidate craters. Images were processed and map-projected using the U. S. Geological Survey ISIS (Integrated Software for Imagers and Spectrometers) software [Gaddis et al., 1997; Anderson et al., 2004b]; the morphometric parameters were then compiled from the map-projected images. Because of the enormous number of LROC images, a complete review of all images was not practical. Therefore, images of highlands areas were randomly selected and examined. Appendix A lists the data set used in the study, including the location of each crater, its diameter and melt-pool dimensions. The image number is also included.
4. Recognition Criteria
 For craters having diameters larger than a few kilometers, the melt materials are easily recognized as widespread, floor-filling deposits and as ponds perched on the terraced rims; in some cases, well-defined flows occur and the walls may exhibit flows or a veneer of impact melt. Melt material typically has a complex morphology with flow texture, cracks, blocky areas, and festoons, and typically has a lower albedo than surrounding materials and is easily distinguished from debris material from the crater walls which covers the floor [e.g., Howard and Wilshire, 1975; Schultz, 1976]. Figure 1 shows Copernicus crater with its characteristic melt pool on the floor and ponds within the terraces on the rim.
 For smaller craters with more limited volumes of melt material, the morphology noted in the large craters does not usually occur. Therefore, a critical aspect of this study is to identify materials that are impact melt, as opposed to materials formed by other processes. Flat-floored craters are common, but few flat-floored craters have exposed melt materials.
 The criteria used here to define floor melt deposits include: smooth; flat surface; low albedo; embayment relations with surrounding materials; abrupt change in slope between the floor materials and the crater walls; and impact craters having diameters of a few to a few tens of meters with a morphology suggestive of impact into a viscous (incompletely solidified) material. In a few cases, tension cracks typical of larger melt volumes are observed. The flat surface of the floor may be continuous across its extent or it may have kipukas of underlying impact materials and/or rocks around the margin or the interior. In cases of hummocky areas on the floor, it is not clear if those hummocks are covered with impact melt. Figure 3 shows expanded views of the morphology of material on crater floors considered to be impact-melt material.
 A flat, level surface results from the freezing of the molten material along a surface of equal gravitational potential. That material may be thick enough to bury the underlying topography such that the surface is smooth (Figure 3a); in some cases it is not sufficiently thick and underlying materials protrude through the melt deposit as kipukas. When a floor is rough and covered with blocks and the melt deposit is thin enough, the melt will embay those blocks (Figure 3b).
 In cases where the crater floor is covered with melt material, there is an abrupt break in slope at the contact with the crater wall (steep crater wall, flat crater floor) marking the edge of the melt pool. Depending upon the geology, that break in slope may be covered locally by younger debris shed from the crater walls; in those cases, however, the morphology of the debris lobes is easily recognized by the presence of boulders and linear striation along the length of the deposit. Craters with flat floors that are not composed of melt (or are old enough such that a well-defined regolith has formed) do not have an abrupt topographic discontinuity. Typically, such surfaces have a fine texture characteristic of regolith surfaces and if any significant topography is present, the regolith develops the so-called “elephant hide” texture due to downslope regolith creep. It is possible that such craters have melt material at depth, but they are not interpreted to be present at the surface in a pristine state.
 Veneers of melt material have been identified coating and flowing down the walls of large craters. In some cases, the material is thick enough that it displays well-defined flow texture and tension cracks. There are numerous cases in these small craters in which the inner walls appear to be covered with a smooth material with lobate margins on the lower part of the wall.
Figure 4 illustrates such an example and demonstrates that caution must be exercised in the interpretation of impact melt coating crater walls. Figure 4 shows North Ray crater, a 1-km impact crater at the Apollo 16 landing site. Bright, smooth lobes of material are observed extending downslope on the inner crater walls in the vertical LROC view, and such material might be interpreted as impact melt flowing down the walls. These lobes do represent flow of material but, based on analysis of Apollo surface observations and images [Ulrich, 1973] (Figure 5), the material is instead clastic debris. Such morphology is often observed in small craters, but since impact-melt veneer cannot be easily distinguished from clastic debris at these resolutions, it is not used as a melt-identifying criterion.
 Given the diameters of their parent craters, the melt pools are relatively small. For craters up to 3000 m in diameter, the pools have diameters of up to about 800 m. Figure 6 plots the crater rim diameter against the melt-deposit diameter for craters that contain well-defined melt pools.
 A second type of feature interpreted to be impact melt are dark stringers of material that extend from the crater interior over the crater rim and across the ejecta blanket (Figures 7a and 7b). This material was deposited late in the cratering process as it overlies the rim and the continuous ejecta. Such stringers are thin and a few meters wide. Individual particles that compose the stringers are below the limit of resolution, but the relation with the crater indicates that its source is the crater and the low albedo suggests it is impact melt.
5. Melt Generation and Crater Scaling
 In their investigation of the effects of impact melting on the Moon, Cintala and Grieve  suggested impact-melt deposits would be extremely rare in simple lunar craters because the volume of melt generated would be very small relative to the volume of the cavity itself. The volume of melt retained within a given crater is a complex function involving an array of factors whose form has not yet been determined completely. In any lunar hypervelocity impact, crater formation is accompanied by the generation of a volume of impact melt determined by many of the same factors controlling the crater's dimensions. The physical and mathematical relationships governing the two, however, are considerably different. Further complicating prediction of the final results of that impact is the poorly understood process of melt ejection from the cavity, which ultimately controls the amount of melt remaining in the crater versus that ejected from the crater.
5.1. Crater Scaling
 Craters treated in this paper fall in a size range that has received relatively little attention in cratering mechanics studies. Much larger events — those leading to complex craters — have instead drawn the preponderance of investigative attention, a situation arguably begun because they are so enormous. For large, complex craters, the daunting physical and computational complexities due to target strength and structure can be ignored to the first order because the effects of gravity on crater formation are so much greater [Chabai, 1965; Gault, 1974; Holsapple, 1993; Holsapple and Schmidt, 1982; Schmidt and Housen, 1987]. At the other extreme of the size spectrum, the dimensions of small craters formed by impact into rock are governed by the strength of the target. The dimensions of the craters described here lie between these two limiting sizes have formed well within the gravity regime [e.g., Holsapple, 1993; Nolan et al., 1996]. Because the craters studied are located in the highlands within the regolith and megaregolith [Hartmann, 1973; Head, 1976; Short and Forman, 1972], the target can be considered to be highly pulverized material with little to no strength common to coherent rock. The megaregolith has been estimated to be of the order 2–3 km thick and intact bedrock not reached until depths of 20–30 km [Short and Forman, 1972; Toksöz et al., 1972; Head, 1976]. While rocks are certainly present in the regolith and megaregolith, they do not form a coherent, strong target but simple represent large clasts.
Schmidt and Housen  have shown that there could be two end-member cases for scaling impacts in the gravity regime, depending on the nature of the target. Dry, fragmental, porous targets, such as the lunar regolith and megaregolith, have little cohesive strength and no shear strength. Thus, except for perhaps very small impacts into the regolith, the sizes of craters formed in those targets should follow a relationship similar to that for dry sand derived empirically by Schmidt and Housen:
where Dd is the rim-crest diameter of the final crater formed in the dry regime, dp is the diameter of the assumed-spherical impactor, vi is the impact speed, and g is the gravitational acceleration.
 Because it is often assumed that the projectile and target have similar densities (ρp and ρt, respectively), the term involving the density ratio often is not included in scaling analyses, particularly as it appears only weakly as a cube root. Nevertheless, the targets considered below are porous, so the density ratio could be a significant factor.
 Competent rock, on the other hand, has little porosity, but it is also probably very rare at large spatial scales on the Moon. The highly fragmented nature of the highlands megaregolith has been discussed in some detail [Hartmann, 1973; Head, 1976; Short and Forman, 1972], and there is little doubt about its porous nature, at least at shallow levels. It would therefore seem that the dry-sand scaling relationship would be applicable for large craters, too.
 At the scales of large complex craters, however, when excavation exceeds the thickness of the megaregolith, it is likely that the highlands again take on a rock-like character, at least as far as the impact process is concerned. Not only can some of the porosity be lost due to the overburden at depth (see Head  for a review of lunar target properties as a function of depth), but impacts almost certainly had less influence on the physical state of the deep crust [e.g., Simmons et al., 1975; Todd et al., 1973]. Thus, in terms of the ability of the highlands to propagate stress waves at depths on the scales of large craters [e.g., Toksöz et al., 1972, 1974], they would appear to a large impactor to be little different from solid rock.
 A scaling relationship for wet, saturated sand was developed by Schmidt and Housen  to estimate the dimensions of craters formed in rock in the gravity regime. Such a target possesses no porosity, and thus transmits stress waves effectively, while having no tensile strength, permitting gravity to exert the dominant control over the cratering process. Because experiments for saturated sand completely in the gravity regime have not yet been conducted, the scaling relation proposed by Schmidt and Housen is an asymptotic solution to the problem, extrapolating available data. Nevertheless, until a more complete data set becomes available, this must be considered to be the best existing version. Expressed in a form similar to that of equation (1):
where Dw is the diameter in the wet regime and the other parameters are as defined above.
5.2. Melt Generation
 The creation of molten rock is a natural consequence of an impact that generates sufficiently high shock stresses. Such an impact raises the internal energy of both the target and projectile, and impact melt is produced when the residual internal energy after decompression in either is greater than that required for fusion. The impact and planetary literature are replete with contributions treating the formation of impact melt at varying levels of detail. Ahrens and O'Keefe , Melosh [1989, pp. 42–44], and Stöffler  (and references therein), for example, provide thermodynamic descriptions of the shock-melting process. Melt-related calculations presented here use the method described by Grieve and Cintala  and Cintala and Grieve . The physical and thermodynamic properties of the diabase, iron, and anorthosite can be found in the work of Cintala .
 Given a projectile whose size is fixed to facilitate this simplified overview, the volume of melt generated in an impact event is dependent on an array of factors. The densities, equations of state, and other physical properties (e.g., porosity, shape, etc.) of the impactor and target influence the peak shock stress, the rate of decay of the shock front, and ease with which each melts (or vaporizes). The impact speed is the most significant parameter, as its square appears in the equation determining the shock stress, and the impact angle modulates the overall melt volume [Pierazzo and Melosh, 2000]. Although any of those factors could be important in explaining the observations described here, at least one of them can be ruled out or minimized; because all of the craters treated here formed in the highly pulverized highlands, it can be assumed to the first order that the composition, porosity, and other physical properties of the target are constant. Thus, we concentrate on the projectile and its velocity components as potential reasons for observing melt in some of these small craters.
 Using the equations noted above (equations (1) and (2)), diameters estimated by the two scaling relationships are shown in Figure 8; the two expressions hold for different conditions, so those leading to the two curves are chosen accordingly. A density for anorthosite (2734 kg m−3) was used for the saturated-sand curve, as it applies to more or less solid highland rock. A regolith density of 1800 kg m−3, however, was assumed for the dry-sand scaling, since an equation of state for a lunar regolith exists [Ahrens and Cole, 1974] and should hold for porous materials for the purposes of this study.
 Lower porosities would yield less melt, but would not change the conclusions of this analysis [e.g., Collins et al., 2011]. Otherwise, spherical, diabase impactors (3010 kg m−3) from 5 to 100 m in diameter were used to approximate rocky projectiles, impacting vertically at 16.1 km s−1, which is the root mean square (RMS) asteroidal value calculated for the Moon by Shoemaker and Wolfe .
 While the differences in crater diameter are apparent, it is perhaps more important to note that the corresponding differences in crater volume will scale as the cube of the diameter ratios, leading to volumes for the dry-sand scaling that are almost an order of magnitude smaller (Figure 9). It should be noted here that Cintala and Grieve  assumed that the saturated sand relationship held across the entire range of diameters they considered, with no distinction made between fragmental and coherent targets at the smaller end of the range.
 The craters under consideration here occur the highlands and they are small enough to have formed in the upper portion of the megaregolith. Thus, they are more justifiably described by the dry-sand scaling of Schmidt and Housen . Any effects on the cratering process due to variations in the megaregolith's properties at these scales are not apparent in the final craters' morphologies. While various studies have examined the question of the changes in morphology as a function of size [e.g., Howard, 1974; Cintala et al., 1977]; none has documented variations of crater morphology for a given size within the highlands. Hawke and Head  did note that highland craters contain equal or smaller amounts of impact melt relative to similar-sized mare craters.
 At first glance, then, it would appear that the craters exhibiting melt deposits could simply be due to their formation by iron impactors. As shown by Grieve and Cintala  and Cintala and Grieve , however, the difference in melt production between different impactor types is offset by the corresponding differences in crater dimensions. Figure 11 illustrates this effect at three different speeds for the two projectile types. While the envelope for both projectiles and all speeds extends over about a factor of two or so in the ratio of melt generated relative to the cavity volume, the important point is the proximity of the two impactors' curves at any of the three speeds. In a photogeological context, it would be extremely difficult to separate a crater formed by an iron impactor at, for example, 25 km s−1 from one formed by a diabase impactor at the same speed. While not shown here, the same could be said for ice projectiles [Cintala and Grieve, 1998].
 Currently it is extremely difficult to make a quantitative distinction between any of these craters in terms of impact speed. Given that, it also appears that there is not yet a way to identify the impactor type based on crater morphology. Thus, the presence or absence of impact melt in any of these craters cannot be used to point to a specific class of impactor, nor can it be used to imply an impact speed, other than the fact that it was high enough to generate an observable quantity of impact melt.
 As mentioned above, Cintala and Grieve  used the scaling relationship for saturated sand to determine the dimensions of the craters they treated in their study of lunar impact melting. It is instructive to compare the results of a diabase impactor into rock (saturated-sand scaling) and into regolith (dry-sand scaling). Figure 12 shows such a comparison for the diabase impactor and target types at 16.1 km s−1. The equation of state for the regolith was used for both targets, however, to emphasize the effect that would result simply from the difference in scaling relationships.
 As expected, the craters from the saturated-sand (“rock”) relationship are much larger, thus “diluting” the amount of impact melt relative to that for the dry-sand (regolith) scaling. For example, a 500-m crater formed under dry-sand scaling would have the same melt-to-crater volume ratio as that of a 5 km crater formed under saturated-sand scaling. While it is not the rule to find impact-melt ponds in 5-km lunar craters, such deposits are hardly a rarity. It must be noted here that these melt volumes represent the total amount of impact melt created, and while it is possible that a larger fraction of melt is ejected from smaller cavities as they grow, the implication is clear: use of a scaling relationship more appropriate for a highlands target in this size range results in a smaller crater for a given impact, leading in turn to greater ratios of melt volume to crater volume.
 It is clear from the observations that melt pools can form on the floors of small-diameter simple lunar craters. While not common, they occur in craters as small as 170 m, much smaller than previously recognized. Two questions are of interest in the interpretation of these observations: what, if anything, is different about the craters that exhibit melt, and why are melt deposits observed in such small craters? Well-defined melt pools are only observed in fresh craters. Such pools almost certainly have continued to form over time but are no longer obvious in older craters where a significant regolith has developed on the melt material and regolith has crept down the craters walls. Any veneer of melt on the crater wall will also be broken up over time and slump onto the crater floor.
 Even among fresh highland craters, the presence of a melt pool is rare. This suggests that there is something unusual about the cratering events in which a melt pool is formed. It may be that an usually large amount of melt was produced or that the melt that was produced was retained in the crater, which could in turn be a function of the projectile properties or the target properties.
 The size of the melt pool compared with the crater diameter (Figure 6) suggests that the data occur in two fields, one for craters <1 km diameter, and a second for larger diameters. This change in the amount of melt in the floor pool is probably a function of both the relatively larger volume of melt produced in larger diameter craters and simply better retention of melt in deeper craters. As the volume of melt increases, the pool becomes better defined. A few of the melt deposits for the smallest craters are not continuous, but consist of several isolated pools. As the volume and depth of the pool increase, partly buried rocks are less frequent and are found only along the margin of the pool; presumably those toward the center are completely buried.
 As noted above, there is little significant difference in the volume of melt produced as function of projectile properties or impact speed. And, since these craters all occur in the highlands, the target properties do not offer an explanation. The remaining variable is the angle of impact.
 Modeling by Pierazzo and Melosh  shows that the peak shock pressures and temperatures are greater in vertical impacts when compared to non-vertical impacts. Those models also show that the volume of material which experiences a given shock pressure and temperature is greater for a vertical than oblique impact [Pierazzo and Melosh, 2000, Table 1]. The asymmetries in the distribution of heating within the target, as a function of impact angle, have also been noted by Sugita et al.  and Ernst et al. .
 Under conditions of a vertical impact, higher levels of shock and heating will occur at the highest levels will be relatively deep and at the crater center. This geometry may reduce the amount of melt that is ejected from the crater, relative to lower angle impacts allowing more to remain in the crater and form a pool on the crater floor. Figure 13 shows the model results form Pierazzo and Melosh  for a range of impact angles. With a vertical impact, the higher stresses would produce more melt relative to an oblique impact.
 In addition, in a vertical impact, the momentum of the melt would be directed downward into the crater floor rather than at some angle to the vertical. This orientation would allow more of the melt to be retained than in an oblique impact. The distribution of ejecta and melt as function of impact angle has been studied experimentally, theoretically and observationally. Gault and Wedekind  noted that the lower the impact angle, the greater the velocity of the downrange ejecta (and melt). At impact angles approaching 45° from vertical, melt focusing from the initial jetting and subsequent high-velocity portions of ejecta is greatest, with the focusing decreasing as the crater grows. As the impact becomes more and more oblique, more and more of the melt produced is ejected downrange.
 Impact-melt forming pools on the floors of simple craters as small as 170 m in the lunar highlands have been recognized in LROC images. These melt-containing craters appear to occur randomly throughout the highlands, although such deposits are rare and are recognized only in fresh craters. They almost certainly occur in more, older craters, but presumably the formation of an impact regolith and creep of the regolith down the crater wall has obscured those deposits. Modeling of the impact conditions suggests that differences in the projectile or target properties or impact speed are not responsible for the generation of melt in such small craters. These aspects do influence the volume of melt but are not sufficient to explain the occurrence of melt in small diameter (i.e., relatively low energy) craters.
 The preferred interpretation made here is that these small craters exhibiting obvious melt on their floors are the results of near-vertical impacts. Vertical impacts are relatively rare, consistent with the infrequent occurrence of such craters on the Moon. Vertical impact also produces the highest shock pressures and temperatures relative to oblique impacts and thus would produce more melt. The vertical momentum of the impact would tend to keep the melt that is produced within the crater compared with an oblique event, in which the melt has a higher lateral velocity component.
 This appendix provides a summary of the locations of the impact craters used in this study exhibiting melt material on their floor. Table A1 includes the LROC Narrow Angle image, the latitude and longitude of the crater center, the diameter of the crater and the diameter of melt pool on the floor.
Table A1. Listing of Impact Craters With Floor Melt Deposits
Crater Diameter (m)
Pool Diameter (m)
 We appreciate the careful reviews and suggestions made by Gordon Osinski and Peter Schultz that have served to improve the manuscript as well as the editorial handling by the editor, Mark Wieczorek, and an anonymous Associate Editor. This work was part of activities of the Lunar Science Institute Node at the Johns Hopkins University/Applied Physics Laboratory and the Lunar Reconnaissance Orbiter Camera Team.