Lunar Floor-Fractured Craters: Classification, Distribution, Origin and Implications for Magmatism and Shallow Crustal Structure

Lunar floor-fractured craters: Classification, distribution, origin and implications for magmatism and shallow crustal structure

The MIT Faculty has made this article openly available. Please share how this access benefits you. Your story matters.

Citation Jozwiak, Lauren M., James W. Head, Maria T. Zuber, David E. Smith, and Gregory A. Neumann. “Lunar Floor-Fractured Craters: Classification, Distribution, Origin and Implications for Magmatism and Shallow Crustal Structure.” Journal of Geophysical Research 117, no. E11 (2012). Copyright © 2012 American Geophysical Union

As Published http://dx.doi.org/10.1029/2012je004134

Publisher American Geophysical Union (AGU)

Version Final published version

Citable link http://hdl.handle.net/1721.1/85651

Terms of Use Article is made available in accordance with the publisher's policy and may be subject to US copyright law. Please refer to the publisher's site for terms of use. JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 117, E11005, doi:10.1029/2012JE004134, 2012

Lunar floor-fractured craters: Classification, distribution, origin and implications for magmatism and shallow crustal structure Lauren M. Jozwiak,1 James W. Head,1 Maria T. Zuber,2 David E. Smith,2 and Gregory A. Neumann3 Received 17 May 2012; revised 29 September 2012; accepted 3 October 2012; published 15 November 2012.

[1] Floor-Fractured Craters (FFCs) are a class of lunar craters characterized by anomalously shallow floors cut by radial, concentric, and/or polygonal fractures; additional interior features are moats, ridges, and patches of mare material. Two formation mechanisms have been hypothesized—floor uplift in response to shallow magmatic intrusion and sill formation, and floor shallowing in response to thermally driven viscous relaxation. This study combines new Lunar Orbiter Laser Altimeter (LOLA) and Lunar Reconnaissance Orbiter Camera (LROC) data to characterize and categorize the population of FFCs and map their distribution on the Moon, and uses variations in floor-fractured crater morphology and regional distribution to investigate the proposed formation mechanisms. The population of FFCs was categorized according to the classes outlined by Schultz (1976). The distribution of these FFC categories shows an evolution of crater morphology from areas adjacent to lunar impact basins to areas in the lunar highlands. We propose that this trend is supportive of formation by shallow magmatic intrusion and sill formation—crustal thickness determines the magnitude of magmatic driving pressure, and thus either piston-like floor uplift for high magnitude, or a convex floor profile for low magnitude. Predictions from previous studies modeling viscous relaxation are inconsistent with the observed altimetric profiles of FFCs. Hence our analysis favors FFC formation by shallow magmatic intrusion, with the variety of FFC morphologies being intimately linked with location and crustal thickness, and the driving pressure of the intrusion. Data from the GRAIL (Gravity Recovery and Interior Laboratory) mission will help to test these conclusions. Citation: Jozwiak, L. M., J. W. Head, M. T. Zuber, D. E. Smith, and G. A. Neumann (2012), Lunar floor-fractured craters: Classification, distribution, origin and implications for magmatism and shallow crustal structure, J. Geophys. Res., 117, E11005, doi:10.1029/2012JE004134.

1. Introduction and Background classification system based primarily on observed morphological differences between floor fractured craters and fresh [2] Floor-Fractured Craters (hereafter referred to as FFCs) unaltered impact craters. Schultz [1976] defined six sub classes are characterized by their shallow, often plate-like floors for FFCs, summarized in Table 1. A map showing the dis- and contain radial, concentric, and/or polygonal fractures; tribution of all FFCs over the lunar surface was published additional interior features can include moats, ridges, pits of by Schultz [1976], although the map did not distinguish mare material, and dark-haloed pits. These craters were first between the different classes of FFC for individual points described in detail by Schultz [1976], who examined them and a list of craters with coordinates and classification was using Lunar Orbiter and Apollo images, and established a not provided. [3] As described by Schultz [1976], FFCs share several 1Department of Geological Sciences, Brown University, Providence, characteristics, most notably their fractures. The fractures Rhode Island, USA. can be radial, concentric, or polygonal, with the differences 2Department of Earth, Atmospheric, and Planetary Sciences, contributing to the definition of the different categories Massachusetts Institute of Technology, Cambridge, Massachusetts, USA. 3 of FFCs. FFCs are also characterized by a shallow floor, Solar System Exploration Division, NASA Goddard Space Flight although the shallowing varies widely between craters, any- Center, Greenbelt, Maryland, USA. where from 20 to 85% of the predicted crater floor depth Corresponding author: L. M. Jozwiak, Department of Geological [Schultz, 1976]. On the basis of new image and altimetry data Sciences, Brown University, Providence, RI 02912, USA. from the Lunar Reconnaissance Orbiter (LRO) we have ([email protected]) undertaken a detailed analysis of the global distribution and ©2012. American Geophysical Union. All Rights Reserved. characteristics of FFCs in order to 1) revisit and assess the 0148-0227/12/2012JE004134

E11005 1of23 E11005 JOZWIAK ET AL.: FLOOR-FRACTURED CRATER DISTRIBUTION AND ORIGIN E11005

Table 1. List of the Defining Characteristics for Each Class of FFCa Crater Class Class Characteristics 1 Deep floor, radial/concentric fractures, crescent shaped patches of mare material along walls, central peak complexes 2 Well-defined wall scarp, uplifted central region/convex up floor profile, concentric fractures 3 Wide moat between crater wall scarp and crater interior, radial/polygonal fractures, terracing on wall opposite the nearby mare 4a V-shaped moat, radial/concentric fractures, convex up floor profile 4b V-shaped moat with pronounced inner ridge on the interior side, subtle fractures, irregular convex up floor profile 4c V-shaped moat, hummocky interior 5 Degraded crater walls, radial/polygonal fractures 6 Mare-flooded interiors, concentric fracture pattern near wall aCraters in each class need not possess all of the listed characteristics, although they must possess a majority in order to be classified. classification scheme of Schultz [1976], 2) provide more into the moat region. Wall slumps and terraces may be pres- quantitative data on crater morphometry, 3) assess the global ent, and are generally located on the area of the crater farthest distribution of FFCs and members of their subclasses, and 4) away from the basin interior; correspondingly, the moat is revisit and reassess theories of origin (e.g., viscous relaxation often deepest and most visible on the side of the crater closest or magmatic intrusion and piston-like uplift). The proposed to the basin (Figure 3b). The interior of the crater has a central theories of origin predict different end-member morphologies, peak complex and is covered with hummocky material, such as the height of the crater rim crest, although it is difficult excepting parts of the moat which may contain mare units. to distinguish the processes by purely quantitative standards, There is a small subset of Class 3 craters that are located because the exact amount of deformation experienced may inside the mare units of a basin, and lack central peaks, floor vary widely by crater. We first describe the classification fractures, and hummocky material; however, they have dis- scheme of Schultz [1976], outline the major theories of origin, tinctly shallow, flat floors and wide, pronounced moats. describe the data sets used in our analyses, and then address These craters exhibit a steeper wall slope than Class 3 FFCs the four goals outlines above. not located inside mare units, and their uplifted central region Schultz is more pronounced. Figure 3a illustrates the Class 3 FFC 1.1. Classification Scheme of [1976] characteristics in crater Gassendi. Examples of Class 3 craters [4] Although FFCs share broad characteristics, they can be include Gassendi (Figure 10a), Taruntius (Figure 10c), readily divided into subcategories based on their morphol- Lavoisier; craters Haldane and Runge are examples of ogy; Schultz [1976] described six subcategories for FFCs. the wide-moat-mare subset. We found these categories to be accurate classifications, and [8] Class 4a (Figure 4): Class 4 is divided into 3 sub employ the descriptions when classifying FFCs. A summary categories—4a, 4b, and 4c—all class 4 craters possess a of subclass characteristics is presented in Table 1. shallow “v” profile moat separating the wall scarp from the [5] Class 1 (Figures 1 and 10i): Generally large craters— crater interior. Class 4a has a less pronounced “v” moat and diameters between 50 km and 300 km with an average of very low bounding inner ridge, and diameters between 4 km 140 km—with deep floors, central peak complexes, and and 38 km, with a majority of diameters between 10 km and extensive wall terraces. The floor has either radial or con- 20 km. These craters have strong radial and concentric centric fractures, or both. Among the defining characteristics fractures across a hummocky floor that is generally convex are crescent-shaped mare deposits that often occur along the up in profile. Class 4a is similar to Class 2, but distin- edge of the crater wall scarp. Figure 1 highlights the char- guished by the “v” moat, radial fractures, and more diffuse acteristics and distribution of Class 1 FFCs as demonstrated wall scarps. Examples of Class 4a are Bohnenberger and by the crater Humboldt. Examples of Class 1 craters include Airy; Figure 4a depicts the Class 4a characteristics of crater Humboldt (Figure 10i), Atlas, Schülter, and Cardanus. Bohnenberger. [6] Class 2 (Figures 2 and 10g): These craters are mid-size [9] Class 4b (Figures 5 and 10e): Craters in this category craters—diameters between 13 km and 75 km with an average are between 7 km and 45 km in diameter, with an average of 35 km—with a well-defined wall scarp and a convex-up diameter of 25 km. They are dominated by a pronounced “v” floor profile with an uplifted central floor. The most defining moat and a high inner ridge; this inner ridge is often as high characteristic is the series of strong concentric features as the crater rim crest, and in cases such as Gaudibert comprised of wall terraces and floor fractures. Figures 2a and (Figures 5a and 10e), the ridge is more prominent than the 10g show the Class 2 FFC characteristics displayed by the actual crater rim crest. The rim slope (between the rim and crater Vitello. Examples of Class 2 craters include Vitello, the inner ridge) is steeper than in Class 4a or 4c craters, a Encke, and Davy. result likely related to the greater height of the inner ridge in [7] Class 3 (Figures 3, 10a, and 10c): Class 3 craters vary Class 4b FFCs. The interior of Class 4b craters is hummocky greatly in size—diameters between 12 km and 170 km, and may contain subtle radial fractures; the combination of however a majority have diameters between 30 and 60 km— a high moat ridge and hummocky interior yield a rugged but all have a wide moat feature separating the wall terraces interior. Examples of Class 4b craters are Gaudibert from the central crater floor. The fractures in these craters (Figure 10e) and Stoney; Figure 5a illustrates the Class 4b are either polygonal or radial in nature, and do not extend characteristics of Gaudibert.

2of23 E11005 JOZWIAK ET AL.: FLOOR-FRACTURED CRATER DISTRIBUTION AND ORIGIN E11005

Figure 1. Class 1 floor-fractured craters. (a) Crater Humboldt shown with LOLA topography overlain with LROC-WAC image data. Distinguished as a Class 1 FFC by the dark patches of mare material along the wall, and the radial and concentric floor fractures. (b) Transect of Humboldt, indicated by the line from A-A′ in Figure 1a. Marked on the profile are the locations of the class 1 FFC features. (c) Size-frequency of Class 1 FFCs, binned by diameter; Class 1 FFCs are among the largest. (d) Areal distribution of Class 1 FFCs.

[10] Class 4c (Figure 6): Although alluded to by Schultz The diameters of class 4c craters range from 8 km to 39 km [1976], these types of craters were not previously a separate with an average diameter of 15 km. In a manner similar category of FFC. We find, however, that the class is different to other class 4 craters, 4c craters contain a “v” shaped moat, enough from 4a and 4b to merit its own sub-category. but beyond this similarity, their characteristics are different.

3of23 E11005 JOZWIAK ET AL.: FLOOR-FRACTURED CRATER DISTRIBUTION AND ORIGIN E11005

Figure 2. Class 2 Floor-fractured craters. (a) Crater Vitello shown with LOLA topography overlain with LROC-WAC image data. Class 2 distinguishing features include the strong concentric fractures and the uplifted central region. (b) Transect of Vitello, indicated by the line from A-A′ in Figure 2a. Marked on the profile are the locations of class 2 FFC features, note also the smooth transition between crater wall and crater floor, and the uplifted central crater region. (c) Size-frequency distribution of Class 2 FFCs binned by diameter, like a majority of the FFC population, they range between 10 and 50 km in diameter. (d) Areal distribution of Class 2 FFCs.

4of23 E11005 JOZWIAK ET AL.: FLOOR-FRACTURED CRATER DISTRIBUTION AND ORIGIN E11005

Figure 3. Class 3 floor-fractured craters. (a) Crater Gassendi shown with LOLA topography overlain with LROC-WAC image data. Class 3 is distinguished by a flat, plate-like floor, a wide moat, and radial/polygonal fractures. (b) Transect of Gassendi, indicated by the line from A-A′ in Figure 3a. Marked on the profile are the locations of class 3 FFC features, note in Gassendi the terracing on the crater wall opposite the basin, and the pronounced moat on the crater wall adjacent to the basin. (c) Size-frequency distribution of Class 3 FFCs, although a wide range of diameters occur, the majority of Class 3 craters are 30–40 km in diameter. (d) areal distribution of Class 3 FFCs; they are located almost exclusively just inside basins, or at basin margins.

5of23 E11005 JOZWIAK ET AL.: FLOOR-FRACTURED CRATER DISTRIBUTION AND ORIGIN E11005

Figure 4. Class 4a floor-fractured craters. (a) Crater Bohnenberger shown with LOLA topography overlain with LROC-WAC image data. Class 4a is distinguished by a “v”-profile moat, radial or concentric fractures, and a convex-up crater floor. (b) Transect of Bohnenberger, indicated by the line from A-A′ in Figure 4a. Marked on the profile are the locations of class 4a FFC features, note despite the large fracture cutting through the center of the crater floor, the floor has an overall convex up profile. (c) Size-frequency distribution of Class 4a FFCs. (d) Areal distribution of Class 4a FFCs; note that Class 4a FFCs are more uniformly located beyond basin margins, and into the lunar highlands.

6of23 E11005 JOZWIAK ET AL.: FLOOR-FRACTURED CRATER DISTRIBUTION AND ORIGIN E11005

Figure 5. Class 4b floor-fractured craters. (a) Crater Gaudibert shown with LOLA topography overlain with LROC-WAC image data. Class 4b is distinguished by a “v”-profile moat with a very pronounced inner ridge occasionally higher than the crater rim crest, a hummocky convex up central floor region, and subdued fractures. (b) Transect of Gaudibert, indicated by the line from A-A′ in Figure 5a. Marked on the profile are the locations of class 4b FFC features, note how the ridge nearly matched the elevation of the crater rim crest. (c) Size-frequency distribution of Class 4b craters. (d) Areal distribution of Class 4b FFCs, as with Class 4a, they are located farther from the basin margins than Class 3 FFCs.

7of23 E11005 JOZWIAK ET AL.: FLOOR-FRACTURED CRATER DISTRIBUTION AND ORIGIN E11005

Figure 6. Class 4c floor-fracture craters. (a) Unnamed crater located at 56.3 S, 143.9 W shown with LOLA topography overlain with LROC-WAC image data. Class 4c is distinguished by a “v”-profile moat, although the interiors vary between craters, and floors can be either convex up or concave down in profile. (b) Transect of the crater located at 56.3 S, 143.9 W, indicated by the line from A-A′ in Figure 6a. Marked on the profile are the locations of class 4c FFC features. (c) Size-frequency distribution of Class 4c craters, the vast majority having diameters 10–20 km. (d) Areal distribution of Class 4c FFCs.

8of23 E11005 JOZWIAK ET AL.: FLOOR-FRACTURED CRATER DISTRIBUTION AND ORIGIN E11005

The interiors are shallower than fresh craters, and appear topography, particularly the long-wavelength topography flat, although most have a slightly concave up floor profile. [Daneš, 1965; Baldwin, 1968; Cathles, 1975; Hall et al., The surface is hummocky and largely lacking fractures. 1981; Dombard and Gillis, 2001]. In the most basic sys- The texture of the floor varies between craters, but they tem, that which assumes constant Newtonian viscosity, vis- are united by evidence of post emplacement modification. cosity decreases with depth (a result of temperature Class 4c craters are often small and unnamed, and occasion- increasing with depth), and as a result, long wavelength fea- ally difficult to identify as craters. A named example of a tures, such as the crater depression, tend to relax; this results Class 4c crater is Colombo M, the Class 4c characteristics of in an apparently shallower crater depth and an upbowed floor the crater located at 56.3S, 143.9W are shown in Figure 6a. [Parmentier and Head, 1981; Hall et al., 1981; Solomon [11] Class 5 (Figure 7): Class 5 craters are old, degraded et al., 1982; Dombard and Gillis, 2001]. The lunar crust craters with strong radial, concentric, and/or polygonal frac- can also be modeled as an extended Maxwell solid whose tures cutting their floors; they have diameters ranging from rheology has three components: linear elasticity (as seen in 12 km to 177 km, with the vast majority between 45 km and constant Newtonian viscosity models), solid state ductile 100 km. The interiors of these craters are old highland creep, and plasticity [Albert et al., 2000]. Modeling the crater material, and often contain slumps from the degraded walls. system using this more complex elastoviscoplastic relaxation Like the rest of the crater, the rim crest is heavily degraded. process, results in the same relaxed topography and bowed Examples of Class 5 craters include Von Braun, Repsold, floor as seen previously, but with the addition of floor frac- and Alphonsus. Figure 7a shows the Class 5 features of crater tures. These floor fractures are the result of a cold shallow Von Braun. crust containing the crater overriding warmer, less viscous [12] Class 6 (Figure 8): Class 6 craters are easily identified deeper material. The system strain created by this situation by their distinctive interiors which are completely flooded leads to upward flexure of the bottom of the crater floor, by mare material; excluding the extreme diameters of 5 km and this flexure causes bending stresses on the crater floor and 700 km, Class 6 FFC diameters range between 50 km that may yield floor fractures [Dombard and Gillis, 2001]. and 200 km. Additionally, Class 6 craters have a distinctive Both the Newtonian viscosity model [Parmentier and Head, concentric fracture pattern close to their wall scarp. Some 1981; Hall et al., 1981; Solomon et al., 1982; Dombard and craters have central peaks and radial fractures, although these Gillis, 2001] and elastoviscoplastic model [Dombard and are not requirements. Examples of Class 6 craters include Gillis, 2001] have been used to model the effects of viscous Pitatus and Fracastorius, and the Class 6 features of crater relaxation on craters, but are limited by the accuracy of input Pitatus are shown in Figure 8. parameters describing the lunar rheology, and the assumption of axisymmetric deformation. 1.2. Theories of Origin of FFCs [15] A problem arises then in determining which of these [13] There are currently two proposed formation mechan- mechanisms formed floor-fractured craters, as both mechanisms for FFCs: magmatic intrusion leading to sill formation isms are predicted to produce the large-scale characteristic [Brennan, 1975; Schultz, 1976; Wichman and Schultz, 1995] features of the craters. Our study uses the new high-resolution and viscous relaxation [Masursky, 1964; Daneš, 1965; topographic and image data from the Lunar Orbiter Laser Cathles, 1975; Schultz, 1976; Hall et al., 1981; Dombard Altimeter (LOLA) and Lunar Reconnaissance Orbiter Camera- and Gillis, 2001]. The process of magmatic intrusion has Wide-angle Camera (LROC-WAC), respectively, to analyze been modeled by Johnson and Pollard [1973], a schematic the smaller-scale characteristics that define the different for this process is shown in Figure 9. Energy from the initial classes of FFC, and also to characterize and plot the distri- crater impact forms a highly fractured region and brecci- bution of these classes on the lunar surface. This distribution, ated lens beneath the crater [Baldwin, 1963; Roddy, 1968; and the high-resolution data allow us to investigate in a new Roddy, 1977]. A magma filled crack (dike) propagates upward way the formation of FFCs, in particular the ability of the from the magma source region in the mantle [Wilson and proposed theories to explain the distribution of FFC types, Head, 1981]. When the propagating dike reaches this level, and the more detailed differences in addition to the funda- it ceases vertical propagation and begins to spread laterally. mental differences from fresh craters illustrated by their The magma then spreads beneath the crater forming an axi- shallow floors and floor-fractures. symmetric sill broadly consistent with the dimensions of the crater floor. The sill then begins to grow vertically, uplifting 2. Data Set and Analysis of FFCs the overlying crater floor; eventually the pressure becomes sufficient to cause piston-like uplift of the crater floor. [16] The data used in this analysis were drawn from LOLA Depending on the extent of the intrusion, additional faults, and LROC-WAC data sets. The LOLA data set that served graben, and mare units may be formed on the crater walls as a global lunar base map was composed of tiles at 512 m/ and floor [Schultz,1976].Schultz [1976] examines this pro- pixel resolution. This basemap was overlaid with a hillshade posed process extensively, and shows models for increasing generated from the 512 m/pixel LOLA data, and LROC- amounts of crater modification in response to a growing WAC imagery. The craters Humboldt (1), Vitello (2), Gassendi intrusion, including the formation of graben and terraces and (3), Bohnenberger (4a), Gaudibert (4b), Von Braun (5), and the floor fractures observed in the craters. Pitatus (6) were identified as exemplary for their class (listed [14] On the other hand, the shallow crater floors and floor in parentheses), because they contained all or a majority of fractures could be the result of viscous relaxation, sketched the defining characteristics for their specific class. Profiles of in Figure 9b. Viscous relaxation occurs when high thermal these craters (Figures 1b 2b, 3b, 4b, 5b, 6b, 7b, and 8b) were gradients within a given region lead to a decrease in material compiled from the gridded data to investigate the variations strength and a corresponding relaxation in the regional in crater morphometry between FFCs and fresh craters

9of23 E11005 JOZWIAK ET AL.: FLOOR-FRACTURED CRATER DISTRIBUTION AND ORIGIN E11005

Figure 7. Class 5 floor-fractured craters. (a) Crater Von Braun shown with LOLA topography overlain with LROC-WAC image data. Class 5 is distinguished by a flat plate-like floor, radial or polygonal fractures, and an old, degraded appearance. (b) Transect of Von Braun, indicated by the line from A-A′ in Figure 7a. Marked on the profile are the locations of Class 5 FFC features, note the extremely flat floor and prevalent wall slumps. (c) Size-frequency distribution of Class 5 craters, a large span of diameters are present, and are mostly larger than a majority of FFCs. (d) Areal distribution of Class 5 FFCs, note locations close to basin margins, in particular to the west of Oceanus Procellarum.

10 of 23 E11005 JOZWIAK ET AL.: FLOOR-FRACTURED CRATER DISTRIBUTION AND ORIGIN E11005

Figure 8. Class 6 floor-fractured craters. (a) Crater Pitatus shown with LOLA topography overlain with LROC-WAC image data. Class 6 is distinguished by a mare-flooded interior and a concentric fracture pattern near the crater wall. (b) Transect of Pitatus, indicated by the line from A-A′ in Figure 8a. Marked on the profile are the locations of Class 6 FFC features. (c) Size-frequency distribution of Class 6 FFCs, which are larger than average FFCs. (d) Areal distribution of Class 6 FFCs, note proximity to basins.

(Figure 10), as well as between FFC classes. The results of compiled using the global distribution map of Schultz [1976]; these analyses are described in the following section. A global this list was then used as a guide for corroboration of the search for FFCs was undertaken using the LROC and LOLA currently identified FFCs and identification of any previously databases. An initial list of currently recognized FFCs was unidentified FFCs found in our global search. We identified

11 of 23 E11005 JOZWIAK ET AL.: FLOOR-FRACTURED CRATER DISTRIBUTION AND ORIGIN E11005 approximately 10–20 previously unnoted FFCs, and removed class (Figure 10); these craters were selected because they approximately 10 from the existing lists because the LOLA contained all of the identifying characteristics of their class, and LROC data showed that the craters did not share the thus providing an end-member example for each class, and characteristics of FFCs. Using the LOLA and LROC-WAC assisting comparisons between classes. These representative global data, craters were located, and those that fit the char- FFCs were then compared to Copernican-aged craters acteristics of FFCs were recorded in a newly generated list of all lunar FFCs, provided as auxiliary material to this paper.1 The morphology of individual craters was studied in detail Figure 9. Diagrammatic cross-sectional representations of using both LOLA altimetry and LROC Narrow-Angle Cam- a fresh complex crater and the predictions of two candidate era (NAC) image data. These analyses were all carried out crater modification processes proposed to explain the charac- using the ArcGIS program platform. teristics of floor-fractures craters. (a) Fresh complex craters are characterized by a depth-diameter (d/D) relationship of 2.1. Floor-Fractured Crater Diameter Distribution approximately d = 1.044D0.301[Pike, 1980], and an exterior and Morphometry rim crest height (h) that represents the height of the crater [17] The diameter frequency distribution of FFCs rim crest above the level of the surrounding uncratered sub- (Figures 1c, 2c, 3c, 4c, 5c, 6c, 7c, and 8c) highlights the strate. Exterior crater rim crest height is a combination of predominance of FFCs with diameters between 10 km and structural uplift and ejecta thickness and is represented in this 40 km. This could have implications for the preferential diagram by a horizontal dashed line labeled ORCH (original scale of activity for the FFC formation mechanism. Current rim crest height) for reference in b and c. The cater interior is gravity data resolution cannot define gravity anomalies in characterized, from the rim crest inward, by wall terraces areas as small as a majority of these craters. However, formed by rim slumping in the terminal stages of the cratering GRAIL (Gravity Recovery and Interior Laboratory) mission event, impact melt overlying a breccia lens emplaced in the data will acquire data at this scale, and could reveal the terminal stages of the impact event as impact melt settled to formation mechanism, as will be discussed in section 3.3. the crater floor above the breccias lens, and a central peak, [18] For detailed morphologic and morphometric charac- uplifted and emplaced in the latter part of the excavation and terization, a representative crater was selected from each FFC modification phase. (b) Complex craters undergoing viscous relaxation [e.g., Hall et al., 1981; Dombard and Gillis,2001] experience a wavelength-dependent change in topography. 1Auxiliary materials are available in the HTML. doi:10.1029/ Long-wavelength topography (in this case, at the scale of the 2012JE004134. crater itself) decreases in amplitude, while short-wavelength topography, such as the central peaks of the shape of the rim crest itself, is much less affected. This results in an overall decrease in the depth of the crater (d) and a decrease in the original rim crest height (ORCH), as the original crater topography smooths out at long wavelengths. Local (short-wavelength) topographic features, such as the central peaks and the crater rim crest tend to remain prominent, however. Evidence for a fresh complex crater undergoing viscous relaxation (Figure 9b) would be a decrease in both the crater depth (d) and the crater exterior rim crest elevation (h), and a distinctive crater interior topographic profile [e.g., Dombard and Gillis, 2001]. (c) Floor- fractured craters have also been interpreted as complex craters intruded by dikes and sills [e.g., Schultz, 1976; Wichman and Schultz, 1996]. In this scenario, a dike propagating vertically from the mantle toward the surface encounters a low-density breccias zone at the base of the crater floor, becomes neutrally buoyant [see Head and Wilson, 1992], and spreads laterally to form a sill. Continued pressurization of the sill causes vertical thickening and uplift of the crater floor and central peaks, with the most extreme deformation focused at the lateral margins of the sill. The predicted result of sill emplacement is the fractur- ing and faulting at the margins of the crater floor, uplift of the floor and central peaks (with some associated floor deformation), formation of a marginal trough or moat as the crater floor is elevated above the base of the wall terraces, and a decrease in the value of (d), the depth to the crater floor. In the sill intrusion model, the exterior rim height, h, should remain unaltered. These predictions of the behavior of complex impact craters (Figure 9a) that have undergone either viscous relaxation (Figure 9b), or sill intrusion and floor uplift (Figure 9c), can be used to distinguish the origin of floor-fractured craters using new image and altimetry data.

12 of 23 E11005 JOZWIAK ET AL.: FLOOR-FRACTURED CRATER DISTRIBUTION AND ORIGIN E11005

Figure 10. LROC-WAC images of Floor-Fractured craters and the Copernican-aged, un-modified impact craters they were compared against. (a) Crater Gassendi compared against (b) Crater Tycho; (c) Crater Taruntius compared against (d) Crater Aristillus; (e) Crater Gaudibert compared against (f) Crater Kepler; (g) Crater Vitello compared against (h) Crater Glushko; (i) Crater Humboldt compared against (j) Crater Hausen (Eratosthenian-aged).

(Figure 10), with diameters comparable to the diameters of Copernican craters were used because they represent the the selected FFCs. The motivation for this comparison was youngest lunar craters [Wilhelms, 1987], and as such have twofold: 1) to quantify the morphologic differences between undergone the least deformation and modification, and for FFCs and normal fresh craters to assess the differences and the purposes of comparison can be used to represent fresh their geographic locations at the craters, and 2) to use these impact craters. data to help determine the formation mechanism of FFCs.

13 of 23 E11005 JOZWIAK ET AL.: FLOOR-FRACTURED CRATER DISTRIBUTION AND ORIGIN E11005

average floor depth for the crater. Copernican craters such as Tycho have a narrow frequency peak at the deepest depths corresponding to their crater floor (Figure 11b), and in many cases the deepest floor depth was also the most frequent. In FFCs, however, there is a larger spread of depths for the crater floor (Figure 11a). This spread arises from floor fractures, which expose small areas of greater depth, as well as occasional wall slumps and convex floor profiles. To miti- gate these contributions, the most frequent floor depth was used to represent the average floor depth. [21] Using ArcGIS, a line was traced along the entirety of the rim crest, a corresponding profile generated, and from this an elevation for the average rim crest height was obtained. It should be noted that when tracing the rim crest, areas with post-impact modification to the rim crest, such as subsequent cratering, were omitted. From the terrain surrounding the crater, profiles were taken to represent the average background elevation. This average elevation was then subtracted from the background elevation to yield the exterior rim crest height. [22] Crater classes were assigned using a combination of qualitative and quantitative methods. Crater morphology was the largest factor in crater classification. Craters displayed a majority of class-specific characteristics (Table 1) that allowed them to be categorized. At times there is ambiguity between the morphologic characteristics of a crater, at which point LOLA topography was used to generate detailed crater profiles. These profiles resolved the crater classification, as they revealed finer scale crater features that helped to distinguish between different classes, for example the presence of a moat to distinguish between Class 2 (Figure 2) and Class 4 (Figure 4). Figure 11. Plot of frequency of floor depth for craters (a) Gassendi, a FFC, and (b) Tycho, an un-modified impact 2.2. FFC and Fresh Crater Comparison crater. In Figure 11a, Gassendi is a Class 3 FFC and has a [23] Floor Fractured Craters exhibit distinct morphologic much broader spectrum of frequent floor depths, indicating differences from fresh impact craters. Many of the differ- a floor that is wider and flatter than the floor of crater Tycho ences are localized to the crater floor. All classes of FFC (Figure 11b). Tycho displays narrow bands of frequent floor contain fractures on their floors. The fractures vary in width depths corresponding to the classical bowl-like shape of depending on the crater class; the fractures also vary in ori- impact craters. entation, and can be radial, concentric, or polygonal. Many of the fractures, especially the larger ones are interpreted to be graben. [19] Crater diameter was calculated using a 3-point circle [24] Certain classes of FFC have moat features (Figures 3a, fit. On each crater, three points were selected on the rim crest 3b, 4a, 4b, 5a, 5b, 6a, and 6b) with Figure 12 depicting this and then used to form a circle. The selected points were specific type of morphologic feature. Wide moats (Figures 3a, spaced in equidistant locations on the rim crest around the 3b, and 12a) are separated from the central crater floor by crater; the points were also selected to avoid areas where the concentric fractures or strong topographic differences. crater wall or rim crest had been modified by post-impact Topographically, the moats are either flat, or slope down events, such as additional crater impacts on the rim. Most toward the wall scarp forming a broad “U” shape. Wide of the craters were not perfectly circular, so to account moats comprise, on average, 40% of the crater floor; although for this, each crater was fit three times, and the average of other crater morphologies, such as terraces, can decrease these fits was taken as the crater diameter. the percent of floor coverage. Other craters have a narrow [20] Given the precision of LOLA data, 30 m spatial moat feature that has a characteristic “v” shaped profile resolution and 1 m vertical resolution [Smith et al., 2010; (Figures 4a, 4b, 5a, 5b, 6a, 6b, 12b, and 12c). This moat type Zuber et al., 2010], detailed crater floor depth measurements is identified in topographic profile by the “v” shape formed were made. In each crater, profiles were taken from rim crest between the interior crater wall and the domed central crater to rim crest trending west- east, northwest-southeast, and floor region. In Class 4b craters, this moat is bounded on the southwest-northeast; these data were compiled and a depth inside by a ridge (Figures 5b and 12c). The ridge varies in frequency plot was constructed with depths binned in 25 m height from almost indistinguishable from the elevation of intervals. Although the plots showed significant variation in the interior crater floor (Figures 4b, 6b, and 12b) to being shape between Copernican craters and FFCs, in both cases near the height of the crater rim crest (Figure 5b). We note the most frequent floor depth was used to represent the

14 of 23 E11005 JOZWIAK ET AL.: FLOOR-FRACTURED CRATER DISTRIBUTION AND ORIGIN E11005

Figure 12. LOLA topographic profiles depicting the range of FFC moat morphology. (a) Wide moat from crater Haldane. The interior of crater Haldane is has a visually distinct flat central region surrounded by a clear wide moat, indicated in the LOLA topographic profile. (b) “V” profile moat without an inner ridge from crater Bohnenberger. Note how the meeting of the wall scarp and the domed central crater floor create a “v” shape, for which the “v” profile moat is named. (c) “V” profile moat with an inner ridge from crater Gaudibert. Crater Gaudibert hosts the most topographically distinct inner ridge, with the elevation of the inner ridge exceeding the crater rim crest in places. The “v” moat is created by the meeting of the crater wall with the ridge material. that not all Class 4a and 4c craters have a discernible ridge (Figures 3b and 7b). This is distinctly different from fresh feature; however, they do all have the “v” profile. craters which display the classic parabolic profile (Figure 11b) [25] FFCs have distinctly broad, shallow floors, although and do not have a distinct central floor region except central the floor can tend to be concave up (Figure 6b), concave peaks at larger diameters. down (Figure 2b), or flat to irregular in topographic profile

15 of 23 E11005 JOZWIAK ET AL.: FLOOR-FRACTURED CRATER DISTRIBUTION AND ORIGIN E11005

Figure 13. Depth versus diameter relationship for FFCs plotted with the same relationship for fresh impact craters. The craters used are shown in Figure 9. Note the distinctly shallower trend for FFC depths.

[26] This fundamental difference in crater floor interior the modification to FFCS occurred primarily to their floor, between fresh craters and FFCs is readily observed in floor not their rim (compare Figures 11a and 11b and Figures 13 depth versus frequency distributions as shown in a comparison and 14). This relationship holds true even for FFCs with an of the depth frequency distributions for FFC Gassendi, and inner ridge, such as Class 4b, demonstrating further that the the Copernican-aged crater Tycho (Figure 11). Fresh crater modification occurs primarily in the central floor region of depth distributions (Figure 11b) show a frequency spike the crater. corresponding to the crater rim crest, a small one at the central 2.3. FFC Areal Distribution peak (if present), and then a large spike at the deepest floor depth. On rare occasion there are a few depths deeper than the [29] A plot of the entire population of FFCs on the lunar most frequent depth, however, they are negligible by com- surface (Figure 15) shows that the majority of FFCs are parison. FFC crater depth distributions (Figure 11a), on the located in close vicinity to the lunar maria and the mare-filled other hand, display the spikes at the rim crest height and impact basins, with a significant population also located in central peak, but have a spread of very frequent depths, with the South Pole Aitken basin, and in the highlands, especially the most frequent values corresponding to the broad uplifted the highlands west of Oceanus Procellarum. This distribution central crater floor depth. There is a significant frequency of was also noted by Schultz [1976] but the distribution of the depths deeper than the most frequent one, and these depths subclasses was not given there. We will examine the distri- come from the interior of the fractures and also the moats bution of individual FFC classes and use this as a basis to present in certain classes. assess further the proposed formation mechanisms. [27] Comparison of the fresh (Tycho) and floor fractured [30] Class 1 FFCs (large craters with radial and con- (Gassendi) craters (Figure 11) also readily shows that the centric fractures, and wall- adjacent patches of mare material) range of depths is very different, with Tycho having almost (Figure 1d) do not constitute a large population (8 craters), twice the topographic range as Gassendi. Indeed, the new and with one exception, are not located directly adjacent data confirm that FFCs typically differ from fresh impact to basins. This location away from direct basin edge contact craters in their depth to diameter ratio (Figure 13). Pike is intriguing given that the hallmark of Class 1 FFCs are [1980] demonstrated a relationship between crater diameter the patches of mare material along the crater wall margins and crater depth for lunar craters (red squares in Figure 13). (Figure 1a). This supports the hypothesis that the Class 1 FFCs are shallower than fresh craters, and thus follow a FFCs were formed by magmatic intrusion as opposed to shallower regression when depth is plotted against diameter viscous relaxation, if the mare patches are areas where the (blue diamonds in Figure 13). This was noted by Schultz intruding magma reached the surface. [1976], and is corroborated by the LOLA data measured [31] Class 2 FFCs (medium size craters with strong con- (Figure 13). We also investigated the depth to diameter ratio centric fractures and convex up floor profiles) (Figure 2d) for individual FFC classes, plotting approximately 70% are located along the edges of maria and basins, with a few of the craters for each class. All classes had a shallower also located inside the South Pole Aitken (SPA) basin. There regression than would be predicted by Pike [1980]; however, is no apparent correlation between the age of the FFC and there were no observable trends beyond this shallower the age of the basins. The distinguishing feature of Class 2 relationship. We make these data available in the auxiliary craters, concentric fractures around an uplifted central region material. (Figure 2b), can be explained as uplift caused by either mag- [28] Pike [1980] also documented a relationship between matic intrusion or viscous relaxation. There are five Class 2 crater diameter and exterior crater rim crest height in fresh FFCs located in highland regions. Several of these resemble craters (red squares in Figure 14). FFCs plot along this same Class 4 FFCs, however, lack the “v” shaped moat (compare regression (blue diamonds in Figure 14), demonstrating that Figure 2b with Figure 5b). It is unclear if this is a reflection

16 of 23 E11005 JOZWIAK ET AL.: FLOOR-FRACTURED CRATER DISTRIBUTION AND ORIGIN E11005

Figure 14. External rim crest height versus diameter relationship for FFCs plotted with the same relationship for fresh impact craters. The crater populations used are shown in Figure 9. Note both FFC and fresh crater populations follow the same trend. of different formation mechanisms, or simply a difference in impact and geotherm uplift in the short term basin modifi- how the mechanism manifests itself in different lunar regions cation phase could create environments conducive to viscous or substrates. relaxation of subsequently formed craters [e.g., Hall et al., [32] Class 3 FFCs (wide moat with a shallow floor and 1981]. The distribution of these FFCs, however (Figure 3a), radial and polygonal fractures) (Figure 3d) are localized indicates that they form adjacent to basins with a wide range to the edges and close interiors of basins and mare units. of formation ages (e.g., Humorum, Nubium, Nectaris, Crisium, The distinguishing morphology of this class is a wide moat Imbrium, Serenitatis, etc.) [Fassett et al., 2012], and thus a (Figures 3a and 3b); in craters located at the edges of basin-related thermal anomaly is unlikely to have persisted basins, the moat is more pronounced on the basin interior side long enough to retain regional geothermal fluxes sufficient to (Figures 3a and 3b), and occasionally contains mare units. influence viscous relaxation of later craters [e.g., Bratt et al., Proximity of these craters to large impact basins could 1985a, 1985b]. A more direct correlation is with the time of implicate a role for thermal processes linked to basin forma- emplacement of the actual mare deposits themselves [e.g., tion. Localized thermal anomalies associated with heat of Hiesinger et al., 2010].

Figure 15. Areal distribution of all FFCs. FFCs tend to be located near mare-filled basins and impact basin margins, although there is also a significant population located in the lunar highlands. There is also a relationship between FFC class and location, as seen Figures 1d, 2d, 3d, 4d, 5d, 6d, 7d, and 8d.

17 of 23 E11005 JOZWIAK ET AL.: FLOOR-FRACTURED CRATER DISTRIBUTION AND ORIGIN E11005

classified or mentioned in the original Schultz [1976] classification. Craters of Class 4c are located away from the mare, though not necessarily deep in the lunar highlands in a manner similar to some of the Class 4a craters (Figure 4d). As mentioned previously, the craters have varied morphology, although all share a nominal “v” profile moat, shallow floor, and a fractured and/or hilly textured interior (Figures 6a and 6b). There is nothing specific in their spatial distribution or morphology to favor formation by either viscous relaxation or magmatic intrusion. [36] Class 5 craters (degraded craters with radial and polygonal fractures) (Figure 7d) are numerous (32 craters), and a majority are located in the highlands west of Oceanus Procellarum, although in positions closer to the mare than the Class 4 craters in this region (Figure 4d). They are located throughout highlands north of Mare Crisium, and a scattering of other places. The defining features of members of this class are their old, degraded appearance with strong radial Figure 16. Percent relaxation of a crater undergoing viscous and occasionally concentric fractures (Figures 7a and 7b). relaxation, as a function of diameter and time, modeled under The degraded nature of the crater suggests an older age for reasonable lunar thermal conditions. Little to no relaxation the host crater than other FFCs. The prevalence of these occurs for craters under 80 km diameter, a majority of craters near, but not on, the edge of Oceanus Procellarum FFCs, and at the longest timescales craters modeled with could be indicative of a relationship between the modification the largest diameter experience less than 10% relaxation. of these now Class 5 FFCs and the formation of Oceanus (From Dombard and Gillis [2001].) Procellarum, or the emplacement of mare units in Oceanus Procellarum [e.g., Hiesinger et al., 2010]. Although the highly subdued features of the crater (Figure 7a) may be the [33] Class 4a FFCs (small craters with “V” profile moat result of viscous relaxation, the generally flat floor profiles and radial or concentric fractures) (Figure 4d), are the most of Class 5 craters (Figure 7b) are not consistent with the numerous of Class 4 craters. They are located generally near flexure associated with viscous relaxation processes [Hall mare basin margins, although as in cases such as west of et al., 1981]. Oceanus Procellarum, they are not necessarily at the edges [37] There is a small population of Class 6 FFCs (mare- of an impact basin. There is also a significant population flooded interiors with concentric fractures along the outer of 4a FFCs scattered throughout the highlands. These craters edge of the crater floor) (Figure 8d) and they are all located contain both a “v” shaped profile occasionally with a subtle at the edge of a mare filled basin. This is not surprising inner ridge, and distinct radial and/or concentric floor fractures because Class 6 craters are defined by their smooth, mare (Figures 4a and 4b). Uplift alone does not uniquely indicate flooded interiors (Figure 8a). The other identifying feature a specific formation mechanism. However, the addition of of these craters is their distinct concentric fracture pattern the “v” profile (Figure 4b), suggests an area that may be around the edge of the crater interior (Figures 8a and 8b). defined by a subsurface magmatic intrusion that uplifted the As with the other craters, this fracture pattern is not suffi- central region of the crater, producing the “v” shaped moat in cient in itself to distinguish between a magmatic intrusion the uplift process. Long wavelength viscous relaxation might or viscous relaxation origin; their proximity to maria could favor more subtle topographic variations. suggest a formation mechanism involving magmatic processes [34] Similar to Class 4a, Class 4b (small craters with “V” or viscous relaxation due to the heat of basin formation profile moat, a pronounced inner ridge, and subdued fractures) or magmatic filling. (Figure 5d) craters are located primarily away from impact basins; indeed Class 4b craters are found to the outside of 3. Implications and Models for FFC Formation Class 4a craters relative to mare-filled basins, suggesting a relationship of Class 4a craters and Class 4b craters as a func- [38] In this section we present the major characteristics tion of increasing distance from an impact basin. The distin- of FFCs that might help to distinguish between an origin guishing morphologic features of 4b craters (Figure 5a) are by viscous relaxation and intrusion (Figures 16–19), and the pronounced “v” profile moat (Figure 5b), a very high inner make further assessments to assist in this distinction. bounding ridge (Figure 5b), and a hilly interior with subtle radial fractures (compare Figure 5b with other class profiles). 3.1. Formation Mechanism Implications of Crater If we postulate that Class 4b craters (Figure 5) are simply Morphology and Morphometry a spatial evolution of Class 4a craters (Figure 4), the different 3.1.1. Floor Depth morphology could be a result of thicker crust, warping and [39] The key morphologic and morphometric feature of muting the effects of the magmatic intrusion. At the same time, FFCs is their shallow floor (Figures 11a, 11b, and 13), which the generally muted and sinuous topography of 4b craters could be produced by either viscous relaxation or magmatic could be a result of deformation by viscous relaxation. intrusion. We compile in Table 3 the difference in the average [35] Class 4c (small craters with a “V” profile moat) crater depth and the theoretical depth of an unmodified crater (Figure 6d), is the only class in this study not distinctly

18 of 23 E11005 JOZWIAK ET AL.: FLOOR-FRACTURED CRATER DISTRIBUTION AND ORIGIN E11005

Figure 17. Half-space viscous relaxation model of Crater Lalande. As the model progressively shallows the floor of the crater, although it does not completely remove the crater Figure 19. Diagrammatic representation of crater modifi- rim crest, it does affect the elevation of the feature. Figure 12 cation as a result of laccolith formation. Df is the crater floor shows that FFCs and fresh craters have the same relationship diameter, used to represent the intrusion extent, and marked between rim crest height and diameter. Additionally, the by either an uplifted region or the cohesive smooth crater model does not produce a moat feature, a distinguishing floor unit. Wm is the thickness of the intrusion, taken to be characteristic of FFC classes 3, 4a, 4b, and 4c. (From Hall the difference of the observed FFC depth from the theoreti- et al. [1981].) cal depth of an unmodified crater of the same size [Pike, 1980]. Te is the effective flexural thickness of the floor as a result of the uplift, and the calculated values are listed in of that diameter [Pike, 1980]. These values could either Table 3. Hc is the length of the crustal magma column and be interpreted as the amount of viscous relaxation or the ds is the length of the column past the Moho, although neither postulated intrusion thickness. Investigations into the amount parameter was used in the calculation of magma overpressure. of topographic relaxation necessary to reproduce crater rc and rm represent the crustal density and magma density, respectively. (From Wichman and Schultz [1995].)

shallowing have been undertaken by both Dombard and Gillis [2001] and Wichman and Schultz [1995]. Model results for percent crater relaxation as a function of crater diameter under reasonable estimates of lunar conditions (Figure 15) show that the model predicts little to no shallowing over the range of FFC diameters. Dombard and Gillis [2001] conclude that “lunar crustal rocks are simply too rigid to have accommodated substantial relaxation over the age of the Moon.” 3.1.2. Exterior Rim Crest Height [40] The similarity of fresh crater and FFC exterior rim crest heights (relative to external topography) is an additional morphologic feature that makes FFC formation due to viscous relaxation improbable. The process of viscous relaxation affects primarily long wavelength topography, and thus the crater floor; it also affects the elevation of the crater rim crest, although this is not itself a long wavelength topographic feature. Figures 13 and 14 show the depth versus diameter and external rim crest height versus diameter, respectively, for both fresh craters and FFCs. The shal- Figure 18. Sketch of post-impact crater structure for both lower depth versus diameter relationship combined with the simple and complex craters. The exact shape and nature of unchanged rim crest height versus diameter relationship, the fractured bedrock and impact breccias zone remains demonstrate that the FFC modification process worked to unknown. (From Koeberl and Sharpton [2007].) shallow the floor without degrading any of the other crater

19 of 23 E11005 JOZWIAK ET AL.: FLOOR-FRACTURED CRATER DISTRIBUTION AND ORIGIN E11005

Table 2. Constants and Related Parameters Used in the Calculation axisymmetric, although observed profiles of fresh craters of Intrusion Flexural Thickness, and Driving Pressure, Listed in were, to a larger extent, axisymmetric. Table 3a 3.1.4. Crater Floor [42] Several of the FFC floor profile types appear less Symbol Definition Assumed Value consistent with viscous relaxation, or more consistent with ’ 10 Em Young s Modulus (basalt) 7.03 *10 Pa magmatic intrusion. FFC classes 1 (Figure 1b), 3 (Figure 3b), n Poisson’s Ratio 0.25 10 5 (Figure 7b), and 6 (Figure 8b) all have flat, plate-like floor Bm Elastic Modulus of Country Rock, 7.47 *10 Pa E/(1n2) profiles. Models for viscous relaxation show a convex floor k Magma Yield Strength 104 N/m2 responding to flexural uplift, or a sinuous floor that has been g Lunar gravitational acceleration 1.62 m/s2 warped by the process of viscous relaxation (Figure 17) [Hall 3 rm Magma Density 2900 kg/m 2 2 et al., 1981; Dombard and Gillis, 2001]. Class 4 craters, in gm Unit Magma weight (rmg) 4700 kg/m s particular class 4b craters, display uplifted or warped floor aFrom Wichman and Schultz [1996]. profiles (Figure 5b), however, their diameters do not exceed approximately 40 km, well below the range of even minimal modification from viscous relaxation (Figure 16) [Dombard and Gillis, 2001]. structure features. This morphology is more indicative of 3.1.5. Crater Moat the piston-like uplift predicted by magmatic intrusion than [43] One of the most important discrepancies between the feature-muting predicted by viscous relaxation (compare half-space models of viscous relaxation profiles seen in both Figures 17 and 19). Furthermore, models of viscous relaxa- Hall et al. [1981] and Dombard and Gillis [2001], is the tion show that the level of relaxation needed to replicate the inability to produce moats of any shape in the crater interiors shallow floor of FFCs change the elevation of the rim crest, (Figure 16). Steep “v” profile moats (Figures 4b and 5b) and making it higher than unmodified craters in the case of wide moats (Figure 3b) are key defining features of Class 4 moderate shallowing, and greatly lowering the rim crest and Class 3 FFCs respectively, and thus models of viscous in the case of extreme crater shallowing (Figure 17) [Hall relaxation must be further refined, or perhaps combined with et al., 1981]. other mechanisms, in order to produce these key features of 3.1.3. Crater Symmetry FFCs. [41] Models of viscous relaxation also assume axisym- [44] Alternatively, the moat could be interpreted as a metric deformation of craters [Hall et al., 1981], a result proxy for the dimensions of the underlying intrusion. In this rarely observed in FFCs. Most FFCs have generally uneven scenario, the intruding dike encounters highly brecciated rim crests, although this does not preclude formation by rock beneath the crater (Figure 18), where it then stalls and viscous relaxation, as later deformation could have caused forms a sill (Figure 19). The observed uplifted region of the the asymmetry of the rim. However, as described above, the crater interior (Df in Figure 19) would then show the lateral regions of intact crater rim crest are widely consistent with extent of the intrusion. In all observed FFCs the floor uplift rim crest heights of unmodified craters (Figure 14). Hence is confined to the interior of the crater, and in some cases (as the uneven rim crests are likely the result of other crater seen in Figure 2), the uplifted section represents only a degradation processes and not due to the FFC formation portion of the crater interior. The exact nature of this density mechanism. The interiors of FFCs also show heterogeneity. barrier beneath the crater is unknown, but its nature and Class 3 craters in particular, often have extensive wall slump- composition has important implications for the intrusion ing and terracing on one side of the crater interior with a formation hypothesis. It is hypothesized that the cratering corresponding wide moat on the other side separated by a process results in a highly fractured, brecciated lens beneath rather flat central floor region (Figures 3a and 3b). These the crater (Figure 18). This brecciated region is less dense craters are often located on the edges of mare, and thus the than the magma in the propagating dike, thus halting the asymmetric nature could be reflective of the initial local vertical propagation of the dike due to neutral buoyancy topography, dipping in toward the basin interior. Class 2 having been reached, and causing horizontal growth and sill craters have a convex up interior (Figure 2b), and examples formation. Although generally conceptualized from terres- such as Vitello have a distinctly elevated central floor, trial craters, experiments and modeling (Figure 18), the exact however, this elevated floor is not axisymmetric. Due to the morphology and extent of this brecciated lens remains high resolution of the LOLA data, we were able to determine unknown. The floor of FFCs often contains an uplifted that none of the crater profiles observed in this study were

Table 3. For Each Listed Crater, the Theoretical Driving Pressure Needed to Form the Intrusion, as Well as the Flexural Thickness of the Intrusion [Johnson and Pollard, 1973]a

Crater Crater Class Crater Diameter (D) Intrusion Radius (a) Intrusion Thickness (wm) Magma Driving Pressure (P0) Flexural Thickness (Te) Gassendi 3 109.67 km 47.58 km 1.97 km 11.34 MPa 4.20 km Taruntius 3 58.66 km 14.00 km 0.881 km 7.10 MPa 0.92 km Gaudibert 4b 33.99 km 5.76 km 0.14 km 5.95 MPa 0.49 km Vitello 2 41.71 km 11.69 km 1.03 km 4.28 MPa 0.58 km Humboldt 1 203.65 km 48.03 km 1.60 km 10.20 MPa 4.40 km

a The diameter was calculated using 3 points to fit a circle around the crater rim crest. The intrusion thickness (wm) was calculated to be the difference in the crater floor depth from the depth of an unmodified crater of the same diameter, calculated using Pike [1980].

20 of 23 E11005 JOZWIAK ET AL.: FLOOR-FRACTURED CRATER DISTRIBUTION AND ORIGIN E11005 central region, distinct from the central peak region highland regions [Hall et al., 1981; Wichman and Schultz, (Figure 2b). Combining this observation with the magmatic 1995; Dombard and Gillis, 2001]; however, the distribu- intrusion hypothesis, this uplifted floor region could yield tion of FFCs (Figures 1d, 2d, 3d, 4d, 5d, 6d, 7d, 8d, and 15), information about the morphology, and in particular the shows that FFCs occur not only along the edges of the lunar lateral extent of the brecciated lens. For the five FFCs whose basins, but also occur as significant populations in the lunar morphometry was studied in detail, the uplifted central floor highlands. Moreover, a majority of these highland craters are region covered approximately 50% of the entire crater floor. Class 4 FFCs, which (excluding crater Stoney, D = 45 km), The uplifted region of crater Gassendi was approximately have diameters less than 40 km; this small size would require 86% of the crater floor, however, the terraces on the northern an extremely high thermal gradient to cause modification wall of Gassendi (Figure 3a) obscure the edge of the moat, by viscous relaxation than gradients known from other data possibly affecting this result. Similarly, in crater Gaudibert, [e.g., Dombard and Gillis, 2001]. FFC distribution also it is unclear whether or not the inner ridge constitutes part of places severe constraints on the extent of potential thermal the uplifted central region. If only the central most region of gradients. In areas such as the crater located at (51.29N, Gaudibert is considered (Figure 5b), the percent floor uplift 121.67W) (Figure 15), a FFC is observed within the vicinity is 33%, however, if the ridge is included in the uplifted of an older, unmodified crater, implying an extremely region, the percent uplift becomes 58%. The discrepancy localized thermal gradient that would modify the younger with both of these craters arises from the nature of their crater, yet leave the older crater undisturbed. moat features, and the unknown origin of these features. 3.2.2. Kopff and Orientale It remains to be determined if moat formation is related to [48] One possible example of a FFC that could have the extent of the brecciated lens, or is formed as the result formed via viscous relaxation is Kopff, located in Mare of another facet of dike intrusion and sill formation. High- Orientale. Whitten et al. [2011] proposes that Kopff formed resolution gravity data from the GRAIL mission may help 10 m.y. after the Orientale basin, excavating the upper layer to resolve this question. of the solidified Orientale melt sheet. In this case Kopff 3.1.6. Floor Slope would have formed at a time when the local thermal gradient [45] From crater profiles, regardless of the overall floor was still relatively high, allowing for topographic relaxation profile shape (excepting features such as graben), the abso- before the crater was subsequently flooded by volcanic lute floor of the crater is approximately level, varying by only material. The rim crest height of Kopff is approximately tens of meters (e.g., Figures 3b, 4b, and 7b). The background 10% shallower than would be predicted given its diameter, elevation, however, slopes by as much as hundreds of meters however, this is still within the trend range for FFC rim crest over a region approximately 100 m in length, especially near heights. The depth to diameter ratio for crater Kopff also fol- basin edges. This can be observed in Figure 3b, where the lows the trend for FFCs, and is plotted as the black square in basinward rim crest of Gassendi is at a lower elevation pri- the Class 3 Depth v. Diameter plot found in the auxiliary marily due to the lower elevation of the region, but despite this material. However, as a broader FFC formation mechanism, the overall topographic profile of Gassendi is flat. Although this process would only be applicable to craters formed inside it is unlikely that viscous relaxation is dependent on regional basin regions and very soon after basin formation or mare slope, the regional slope would still be reflected in the final emplacement. These characteristics fit only a very small sub- relaxed crater. Uplift from magmatic intrusion, on the other set of all FFCs, those being the set of Class 3 FFCs which are hand, would not only be independent of regional slope, but contained within lunar basins, and often covered in mare also would erase the record of such a regional slope in the material. However, it should be noted that these craters (such area the intrusion is active. as Runge) have prominent wide moat features and plate-like 3.1.7. Summary central floor regions, features unsupportive of formation by [46] The key morphologic properties of FFCs include viscous relaxation. the floor depth, rim crest height, interior crater symmetry, presence or absence of a moat, and the texture and slope 3.3. Magmatic Intrusion Models of the crater floor. These properties would all be affected [49] Using the dimensions of the crater, and material by the initial FFC formation mechanism, and the LOLA and constants, a theoretical driving pressure for the magma can LROC-WAC data allow these properties to be investigated be calculated to provide a more quantitative assessment of in regards to the proposed formation hypotheses. All of the the magmatic intrusion model, represented schematically by above listed properties favor FFC formation by magmatic Figure 19. The intrusion thickness, wm, is taken to be the intrusion and sill formation, while remaining ambiguous, depth difference between the crater’s theoretical unmodified or unfavorable in relation to formation by viscous relaxation. depth (calculated using Pike [1980]), and the current depth of the crater, which was interpreted to be the most frequent 3.2. Implications of Crater Areal Distribution crater floor depth. The axisymmetric radius of the intrusion, for Formation Mechanisms a, was measured from the uplifted region of the crater inte- 3.2.1. Crater Size rior, a region that was easily identifiable from the crater [47] The distribution of FFCs appears to favor formation profile. The dimensions of the modeled intrusion are listed in by intrusion rather than viscous relaxation for the following Table 3, and a schematic of the crater parameters is given in reasons. In order for the amount of viscous relaxation seen in Figure 19. The physical assumption of the model from FFCs to occur, locally high thermal gradients are required; which the equations are derived, is that the uplifted crater though feasible for areas near the mare, such high thermal floor is supported entirely by the underlying intrusion; for a gradients would be more difficult to explain in the lunar full derivation we direct the reader to Johnson and Pollard

21 of 23 E11005 JOZWIAK ET AL.: FLOOR-FRACTURED CRATER DISTRIBUTION AND ORIGIN E11005

[1973]. Using equation (1), and material constants—k, the Future work will focus on analyzing the enhanced gravity magma yield strength, gm, the unit magma weight, and Bm, data in conjunction with the FFC classes and distributions the elastic modulus of the country rock—listed in Table 2 to elucidate a more detailed formation model for FFCs. [Johnson and Pollard, 1973], a magmatic driving pressure Pd was calculated 4. Conclusions ¼ ðÞðÞþ= : ð Þ Pd wm 2k a gm 1 [53] The availability of Lunar Reconnaissance Orbiter LOLA topographic data and LROC images has made possi- From this driving pressure, equation (2) [Johnson and ble a detailed investigation of FFCs, and a more quantitative Pollard, 1973] was used to calculate the flexural thickness approach to their classification. The depth versus diameter of the overlying crust (or roof), Te, relationship of the FFCs is distinctly shallower than the same ÀÁ regression for fresh craters (Figure 13); however, both FFCs ¼ : 3 = 4; ð Þ Pd 5 33wmBm Te a 2 and fresh craters share the same relationship between external rim crest height and diameter (Figure 14). Detailed crater Results of these calculations are displayed in Table 3. profiles allow for the observation of crater structures that [50] Both the magmatic driving pressure and flexural justify the further classification of FFCs into 6 subclasses. thickness scale with crater diameter, as would be expected. The distribution of these classes across the lunar surface has These calculations are similar to those of Wichman and implications for both their formation and also the geothermal Schultz [1995], wherein the driving pressure and flexural environment in which they evolved. This distribution of FFC thickness for Taruntius was modeled. The values presented classes supports formation by magmatic intrusion as the here are lower than those predicted by Wichman and Schultz crater morphologies evolve from regions adjacent to impact [1995]; however, this is expected given the refined topog- basins into the lunar highlands. Further work should be raphy data available for this study. done to investigate quantitatively this relationship. Of the [51] The calculated magma driving pressures shown in two proposed formation mechanisms for FFCs, viscous relax- Table 3 are feasible for an intrusion stalled several kilometers ation and magmatic intrusion, we favor magmatic intrusion on below the surface. An intrusion beneath crater Gassendi, for the basis of the data outlined above. Other works [Wichman example would require a driving pressure of 11.34 MPa. and Schultz, 1995; Dombard and Gillis, 2001] have shown Head and Wilson [1992] calculate that a magma overpressure the rheological improbability of formation by viscous relax- greater than 15.1 MPa is necessary to penetrate the nearside ation, and this work shows that the morphologic and mor- crust. Thus the model is consistent with the dike not pene- phometric features observed are inconsistent with viscous trating the floor of Gassendi. The two largest craters modeled relaxation, and instead support magmatic intrusion. Detailed (Gassendi and Humboldt), also have the largest flexural models of the evolution and emplacement of these intrusions thickness (Table 3), this could be attributed to a larger brec- can be carried out with the availability of this highly detailed ciated lens formed during crater formation and modification data set, and data from the GRAIL (Gravity Recovery and stage. Interior Laboratory) mission will help to test and refine these [52] In the magmatic intrusion model (Figure 19), a dike conclusions. propagates from the mantle, and is halted by a density [54] In this work we focus solely on Lunar FFCs, we note boundary at the base of the brecciated lens. The dike has a that floor-fractured craters have been identified on both Mars specific driving pressure necessary to fracture the anorthositic [Schultz, 1978; Korteniemi et al., 2006] and on Mercury crust [Joliff et al., 2006] and raise the negatively buoyant [Schultz, 1977; Head et al., 2009]. Martian floor-fractured magma. However, upon reaching the lower density zone of craters exhibit morphological characteristics such as shallow, the brecciated lens, the driving pressure is no longer great fractured floors and moats; however, the floors of many of enough to support vertical propagation, and the dike stalls. these craters have been highly dissected, and the overall Then the dike propagates laterally forming a sill, and eventu- morphologies of these craters have been heavily influenced by ally a laccolith. Remaining questions include the exact nature interactions with ice in the crustal material [Korteniemi et al., of this density boundary and how it relates to the stratigraphy 2006; Schultz and Glicken,1979;Costard and Dollfus,1987; underneath the overlaying crater (Figure 18). Detailed gravity Sato et al., 2010]. Schultz [1977] identified several candidate data for these craters, such as that which will be provided FFCs on Mercury in images from the Mariner 10 mission, by GRAIL, could help determine definitively whether FFCs using albedo differences within the crater floor region as an are formed by intrusion and uplift or by viscous relaxation. identifying feature. In images from the first flyby of the Young non-FFCs have been observed to have large negative MESSENGER mission, Head et al. [2009] identify a single Bouguer gravity anomalies [Sjogren et al., 1972; Scott, 1974; candidate floor-fractured crater, similar in morphology to lunar Janle, 1977; Dvorak and Phillips, 1977], whereas the older Class 4 FFCs, with a floor region consisting of two upbowed FFCs Humboldt and Petavius have, for their size, smaller lobes, and an association with nearby plains of volcanic origin. negative Bouguer anomalies relative to young non-FFCs This crater remains the only analogous floor-fractured crater [Dvorak and Phillips, 1978]. The mass accommodated by observed in the global coverage of Mercury with the MES- the floor uplift is not enough to account for the decreased SENGER MDIS imagery. It is postulated that the lack of floor- Bouguer anomaly according to models in Dvorak and fractured craters on Mercury is indicative of a planetary struc- Phillips [1978]. One interpretation for this disparity is that ture that does not favor small scale intrusive volcanism; rather intruding magma below the crater floor also contributes to the the thin mantle of Mercury seems to favor large-scale flood diminishment of the Bouguer anomaly underneath FFCs. basalt style eruptions [Head et al., 2012].

22 of 23 E11005 JOZWIAK ET AL.: FLOOR-FRACTURED CRATER DISTRIBUTION AND ORIGIN E11005

[55] Acknowledgments. Thanks are extended to the Lunar Reconnais- Joliff, B. L., et al. (2006), Crystal chemistry of lunar merrillite and comparison sance Orbiter (LRO) mission personnel and to the planning and operations to other meteroritic and planetary suites of whitlockite and merrillite, Am. teams for both LOLA and LROC. This work was supported by the LRO Mineral., 91,1583–1595. LOLA team through grant NNX09AM54G to JWH. Koeberl, C., and V. L. Sharpton (2007), Terrestrial Impact Craters, 2nd ed., Lunar and Planet. Inst., Houston, Tex. Korteniemi, J., et al. (2006), Floor-fractures craters on the terrestrial References planets—The Martian perspective, in Proceedings of the First Interna- tional Conference on Impact Cratering in the Solar System, Eur. Space Albert, R. A., R. J. Phillips, A. J. Dombard, and C. D. Brown (2000), A test – of the validity of yield strength envelopes with an elastoviscoplastic finite Agency Spec. Publ., SP-612, 193 198. element model, Geophys. J. Int., 140, 399–409, doi:10.1046/j.1365- Masursky, H. (1964), A preliminary report on the role of isostatic rebound in the geologic development of the lunar crater Ptolemaeus, in Astrogeologic 246x.2000.00006.x. – Baldwin, R. B. (1963), The Measure of the Moon, Univ. of Chicago Press, Studies, Annu. Prog. Rep., A, pp. 102 134, U.S. Geol. Surv, Washington, Chicago, Ill. D. C. Baldwin, R. B. (1968), Rille pattern in the lunar crater Humboldt, J. Geophys. Parmentier, E. M., and J. W. Head (1981), Viscous relaxation of impact – craters on icy planetary surfaces: Determination of viscosity variation Res., 73,3227 3229, doi:10.1029/JB073i010p03227. – Bratt, S. R., S. C. Solomon, and J. W. Head (1985a), The evolution of with depth, Icarus, 47, 100 111, doi:10.1016/0019-1035(81)90095-6. Pike, R. J. (1980), Geometric interpretation of lunar craters, U.S. Geol. impact basins: Cooling, subsidence and thermal stress, J. Geophys. – Res., 90, 12,415–12,433, doi:10.1029/JB090iB14p12415. Surv. Prof. Pap., 1046- C,C1 C77. Bratt, S. R., S. C. Solomon, J. W. Head, and C. H. Thurber (1985b), The deep Roddy, D. J. (1968), Shock Metamorphism of Natural Materials, Mono, structure of lunar basins: Implications for basin formation and modifica- Baltimore, Md. tion, J. Geophys. Res., 90, 3049–3064, doi:10.1029/JB090iB04p03049. Roddy, D. J. (1977), Impact and Explosion Cratering, Pergamon, New York. Sato, H., K. Kurita, and D. Baratoux (2010), The formation of floor- Brennan, W. J. (1975), Modification of premare impact craters by volcanism – and tectonism, Moon, 12, 449–461, doi:10.1007/BF00577934. fractured craters in Xanthe Terra, Icarus, 207, 248 264, doi:10.1016/ ’ j.icarus.2009.10.023. Cathles, L. M. (1975), The Viscosity of the Earth s Mantle, 386 pp., – Princeton Univ. Press, Princeton, N. J. Schultz, P. H. (1976), Floor-fractured lunar craters, Moon, 15, 241 273, Costard, F. M., and A. Dollfus (1987), Thermokarstic evolution of impact doi:10.1007/BF00562240. Schultz, P. H. (1977), Endogenic modification of impact craters on craters on Mars, Lunar Planet. Sci. Conf., XVIII, Abstract 199. – – Daneš, Z. F. (1965), Rebound processes in large craters, in Astrogeologic Mercury, Phys. Earth Planet. Inter., 15(2 3), 202 219, doi:10.1016/ Studies, Annu. Prog. Rep., A, pp. 81–100, U.S. Geol. Surv., Washington, 0031-9201(77)90032-2. Schultz, P. H. (1978), Martian intrusions: Possible sites and implications, D. C. – Dombard, A. J., and J. Gillis (2001), Testing the viability of topographic Geophys. Res. Lett., 5, 457 460, doi:10.1029/GL005i006p00457. Schultz, P. H., and H. Glicken (1979), Impact crater and basin control of relaxation as a mechanism of the formation of lunar floor-fractured craters, – J. Geophys. Res., 106(E11), 27,901–27,909, doi:10.1029/2000JE001388. igneous processes on Mars, J. Geophys. Res., 84(B14), 8033 8047, Dvorak, J., and R. J. Phillips (1977), The nature of the gravity anomalies asso- doi:10.1029/JB084iB14p08033. – Scott, D. H. (1974), The geologic significance of some lunar gravity anoma- ciated with large young lunar craters, Geophys. Res. Lett., 4, 380 382, – doi:10.1029/GL004i009p00380. lies, Proc. Lunar Sci. Conf., 5th, 3025 3036. Sjogren, W. L., P. Gottlieb, P. M. Muller, and W. R. Wollenhaupt (1972), Lunar Dvorak, J., and R. J. Phillips (1978), Lunar Bouguer gravity anomalies: – Imbrian age craters, Proc. Lunar Planet. Sci. Conf., 9th, 3651–3668. gravity via Apollo 14 Doppler radio tracking, Science, 175,165168, Fassett, C. I., J. W. Head, S. J. Kadish, E. Mazarico, G. A. Neumann, D. E. doi:10.1126/science.175.4018.165. Smith, and M. T. Zuber (2012), Lunar impact basins: Stratigraphy, Smith, D. E., et al. (2010), Initial observations from the Lunar Orbiter Laser sequence and ages from superposed impact crater populations measured Altimeter (LOLA), Geophys. Res. Lett., 37, L18204, doi:10.1029/ from Lunar Orbiter Laser Altimeter (LOLA) data, J. Geophys. Res., 117, 2010GL043751. E00H06, doi:10.1029/2011JE003951. Solomon, S. C., R. P. Comer, and J. W. Head (1982), The evolution of impact basins: Viscous relaxation of topographic relief, J. Geophys. Hall, J. L., S. C. Solomon, and J. W. Head (1981), Lunar floor-fractured – craters: Evidence of viscous relaxation of crater topography, J. Geophys. Res., 87, 3975 3992, doi:10.1029/JB087iB05p03975. Res., 86, 9537–9552, doi:10.1029/JB086iB10p09537. Whitten, J., J. W. Head, M. Staid, C. M. Pieters, J. Mustard, R. Clark, Head, J. W., and L. Wilson (1992), Lunar mare volcanism: Stratigraphy, J. Nettles, R. L. Klima, and L. Taylor (2011), Lunar mare deposits eruption conditions, and the evolution of secondary crusts, Geochim. associated with the Orientale impact basin: New insights into mineralogy, his- – tory, mode of emplacement, and relation to Orientale Basin evolution from Cosmochim. Acta, 56, 2155 2175, doi:10.1016/0016-7037(92)90183-J. 3 Head, J. W., et al. (2009), Evidence for intrusive activity on Mercury from Moon Mineralogy Mapper (M ) data from Chandrayaan-1, J. Geophys. the first MESSENGER flyby, Earth Planet. Sci. Lett., 285, 251–262, Res., 116, E00G09, doi:10.1029/2010JE003736. doi:10.1016/j.epsl.2009.03.008. Wichman, R. W., and P. H. Schultz (1995), Floor-fractured craters in Mare Head, J. W., et al. (2012), Volcanic eruption process on Mercury, Lunar Smythii and west of Oceanus Procellarum Implications of crater modification by viscous relaxation and igneous intrusion models, J. Geophys. Planet. Sci. Conf., XLIII, Abstract 1316. – Hiesinger, H., J. W. Head, U. Wolf, R. Jaumann, and G. Neukum (2010), Res., 100, 21,209 21,218. Ages and stratigraphy of lunar mare basalts in Mare Frigoris and other Wichman, R. W., and P. H. Schultz (1996), Crater-centered laccoliths on the Moon: Modeling intrusion depth and magmatic pressure at the crater nearside maria based on crater size-frequency distribution measurements, – J. Geophys. Res., 115, E03003, doi:10.1029/2009JE003380. Taruntius, Icarus, 122, 193 199, doi:10.1006/icar.1996.0118. Janle, P. (1977), Structure of lunar impact craters from gravity models, Wilhelms, D. E. (1987), The geologic history of the Moon, U.S. Geol. Surv. J. Geophys., 42, 407–417. Prof. Pap., 1348, 302 pp. Wilson, L., and J. W. Head (1981), Ascent and eruption of basaltic magma Johnson, A. M., and D. P. Pollard (1973), Mechanics of growth of some – laccolithic intrusions in the Henry Mountains, Utah. I. Field observations, on the Earth and Moon, J. Geophys. Res., 86, 2971 3001. Gilbert’s model, physical properties and flow of magma, Tectonophysics, Zuber, M. T., et al. (2010), The Lunar Orbiter Laser Altimeter Investigation 18, 261–309, doi:10.1016/0040-1951(73)90050-4. on the Lunar Reconnaissance Orbiter Mission, Space Sci. Rev., 150, 209–241, doi:10.1007/s11214-009-9512-y.

23 of 23