During the First Earth-Moon Flyby

Transcription

1 JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 98, NO. E9, PAGES 17,207-17,231, SEPTEMBER 25, 1993 Galileo Observations of Post-Imbrium During the First Earth-Moon Flyby Lunar Craters ALFRED S. McEWEN AND LISA R. GAt)DIS U.S. Geological Survey, Flagstaff, Arizona GERHARD NEUKUM AND HARALD HOFFMAN German Aerospace Research Establishment, Berlin, Germany CARLE M. PIETER$ AND JAMES W. HEAD Brown University, Providence, Rhode Island Copernican-age craters are among the most conspicuous feature seen on the far side and western limb of the Moon in the Galileo multispectral images acquired in December Among the new morphologic observations of far-side craters are bright rays, continuous ejecta deposits, and dark rings associated with probable impact-melt veneers. These observations suggest that the mapped age assignments of severalarge far-side craters (Ohm, Robertson, and possibly Lowell and Lenz) need revision. New crater size-frequency measurements on Lunar Orbiter imagesuggest the following age reassignments: Hausen (170 km diameter), Pythagoras (120 km), and Bullialdus (61 km) from Eratosthenian Upper Imbrian, and Carpenter (60 km) and Harpalus (39 kin) from Copernican to Eratosthenian. Colors and albedos of craters (away from impact-melt veneers) are correlated with their geologic emplacement ages as determined from counts of superposed craters; these age-color relations are used to estimate the emplacement age (time since impact event) for other Copernican-age craters. These agecolor relations indicate a probable Copernican age for 27 far-side or western limb craters larger than 10 km diameter that were not previously mapped as Copernican. The apparent deficiency of Copernican craters on the far side compared with the near side in published geologic maps is not present in our data. Age-color trends differ between mare and highland regions and between the interiors and continuous ejecta of the craters. Similar trends are established for color and albedo versusoil-maturity indices for the returned lunar samples, with distinct trends for mare and highland soils. However, the mare versus highland offsets are reversed in the two comparisons. These relations can be explained by variations in regolith thicknesses and rates of mixing with relatively fresh, crystalline ejecta. Therefore, the soil-maturity trends represent longer geologic time periods in regions with thinner regoliths, such as the maria. 1. INTRODUCTION possible over only a small part of the Moon's surface. The Unlike the Earth's surface, where erosion and other next-best method (in the absence of radiometric dates) is use processes obscure the impact record, the surface of the Moon of the size-frequency counts of superposed craters [e.g., Hartmann, 1968; Neukum and KOnig, 1976; Guinness and contains an accurate and complete record of the impact flux over the Earth-Moon region during the past few billion years. Amidson, 1977; Chapman and McKinnon, 1986], but sufficiently high-resolution images are not yet available for much of the Deciphering that record (or just its most recent part) could Moon, especially on the far side. Thus, the presence or resolve current controversies over the magnitude and the prominence of rays and ejecta deposits, as well as crater periodicity or episodicity of the asteroid and comet flux in the neighborhood of Earth and their effect on mass extinctions and morphology [Soderblom, 1970; Trask, 1971], have been the key the evolution of life [e.g., Shoemaker and Wolfe, 1986; Grieve factors for distinguishing between the Copemican and et al., 1988]. In principle, this could be accomplished through Eratosthenian time-stratigraphic systems. However, ray remote sensing combined with radiometric dates for key units, visibility depends not only on age, but also on crater size (and such as impact melts of large Copemican craters with ejecta volume), compositional differences between ejecta and extensive ray systems (for example, Copemicus, Tycho, and substrate, and phase angle. For example, the crater Jackson). Copemicus, the type example of a rayed crater [Shoemaker, Understanding the stratigraphy of crater deposits is essential 1962], may be one of the oldest rayed craters on the Moon, for determining the geologic history of the Moon [Shoemaker but the rays are conspicuous because it is a large crater that and Hackmann, 1962; Mutch, 1972; Wilhelms, 1987]. deposited bright highland ejecta on a mare substrate [Pieters Preferably, geologic ages are assigned from direct stratigraphic et al., 1985]. Also, bright rays seen at low phase angles may observations, such as superposition of a unit on the rays of be undetectable (from albedo alone) at high phase angles Copemicus (or vice versa), but for post-imbrium units this is [Fedoretz, 1952; Baldwin, 1963]. Crater morphology can be difficult to interpret, especially where photography is poor, Copyright 1993 by the American Geophysical Union. and Neukum and KOnig [ 1976] have documented a tendency for photo-geologists assign younger ages than are indicated by Page number 93JE crater counts to the larger craters (and we provide additional /93/93JE examples in section 3 below). We believe that the age and 17,207

2 17,208 MCEWEN ET AL.: GALILEO OBSERVATIONS OF LUNAR CRATERS stratigraphic assignments of some post-imbrian craters (i.e., Copemican versus Eratosthenian, and subdivisions within the Copemican) can be significantly improved through the use of multispectral observations. We have five main objectives for this paper: (1) to describe 1968] and further elaborated from studies of returned Apollo samples [e.g., McKay et al., 1972, 1974]. As micrometeorites bombard the lunar surface, soil grains are comminuted to smaller sizes, and dark agglutinates are formed from the fine component by melting and agglomeration. Larger impacts the Galileo EM1 (first Earth-Moon flyby) imaging excavate fresh subsurface material that buries or is mixed with observations of post-imbrium craters, including bright rays, continuous ejecta, and dark crater rings; (2) to present new nearby soils. With continued surface exposure and small-scale impact "gardening," a steady state may be reached in which counts of the size-frequency distributions of crater superposed the mean grain size and agglutinate content do not change on large isolated post-imbrium craters, for determination of geologic emplacement ages; (3) to quantitatively relate multispectral measurements of crater materials to geologic until interrupted by a rare larger impact that buries the soil and resets the surface exposure age. An impact that excavates to depths much greater than the regolith thickness will emplacement ages, providing a basis for mapping Copemican emplace or expose crater floor materials and continuous ejecta units; (4) to compare and contrast these results with soilmaturity parameters and regolith-evolution models; and (5) to with exposure ages near zero. Buried soil may be reexcavated by a subsequent event and returned to the surface to undergo describe the production of multispectral mosaics from the further maturation. In addition to small-scale impact Galileo EM1 images (appendix)ø The organization of the paper follows these objectives. We use three data sets: (1) Galileo solid state imaging (SSI) multispectral observations of the Moon's westem near side and far side [Belton et al., 1992a]; (2) new measurements of the frequencies of craters superposed on the floors and/or continuous ejecta blankets of larger craters, extending the original results of Neukum and Ki nig [1976]' and (3) measurements of spectral reflectivity (J. Adams et al., Lunar Samples Spectral Reflectance Atlas, unpublished, 1974) and soil maturity [Morris et al., 1983] on the returned samples of lunar soils. The spectral reflectivities of lunar soils are controlled primarily by (1) the steady state maturity, and (2) the mineralogy of the underlying bedrock or megaregolith. Improved understanding of the spectral effects of variations in the steady state soil maturity will improve our ability to gardening, which may be thought of as "in situ" processing, and large nearby events that reset the soil exposure age, medium-sized or large distant impacts will contribute "outside" materials (i.e., outside the local regolith) that are mixed into the soil. Outside materials may originate either from vertical mixing from below the local regolith, or from lateral mixing from greater distances. In regions of thin regolith, this outside contribution will be dominated by fresh crystalline materials, whereas in regions of thick regolith there will be a greater proportion of soils with a prior exposure history. Another way of stating this is that an impact event of a given magnitude will excavate only regolith where the target regolith is thick and can be considered part of the "in situ" gardening, whereas the same event over a thin regolith will excavate and expose fresh crystalline materials. The absence of constraints on the rate of replenishment by interpret item (2), especially for the use of crater compositions fresh "outside" materials has prohibited evaluation of as probes of crustal stratigraphy [e.g., Pieters, 1986]. The steady state maturity (or simply "maturity" hereafter) results from both the soil's cumulative age of exposure to the space environment and steady state horizontal and vertical mixing of fresh crystalline materials [e.g., Ba a, 1990]. "Maturity" is sometimes misused to indicate "exposure age", but a soil's maturity is limited to a steady state value determined by the rate of mixing with fresh crystalline materials, which in tum is a function of regolith thickness, the crystallinity and quantitative regolith evolution models [McKay and Basu, 1983; Basu, 1990]. The analysis in this paper represents progress toward resolving this limitation, but additional radiometric ages of Copernican units (preferably impact melts) are needed. Nevertheless, multispectral data provide information on regolith thicknesses. Regolith thickness is determined by geologic emplacement age, the coherence of the geologic unit, and topography (i.e., the preferential downslope movement of ejecta [Soderblom, 1970]). coherence of the underlying strata, and topography. In this paper we show that steady state mixing with fresh crystalline 2. GALILEO SSI OBSERVATIONS materials is especially important for understanding the agecolor relations of Copemican craters. "Lunar soil" is used in this paper to refer to the finer grained fraction of the unconsolidated material (regolith) at the lunar surface [cf. Heiken et al., 1991]. In addition, we will follow the distinction made by Shoemaker and Morris [1970] between regolith and a fragmental geologic unit, such as an ejecta blanket, in which the regolith is a strictly surficial layer Galileo SSI obtained multispectral images of the western near side and far side of the Moon during the December 1990 Earth flyby [Belton et al., 1992a]. New data of scientific significance include multispectral observations of the western limb and far side. These observations are unique in several ways. They provided the first digital multispectral data for the Moon's far side, except for the Mariner 10 vidicon images of the Moon's eastern hemisphere [Danielson et al., 1975; Robinson et al., of debris whose thickness and other characteristics are a 1992]. Furthermore, SSI has a charge-coupledevice (CCD) function of surface processes and exposure time. Hence, a new ejecta deposit consisting of fragmented bedrock has no regolith, although the process of regolith development begins detector [Klaasen et al., 1975], which, compared with a comparable vidicon-tube imaging system, provides a 100-fold improvement in broad-band sensitivity and about twice the almost immediately. Fragmental geologic units do contribute spectral response (0.4 to 1.0 gm). The 1.0-gm spectral region to the "megaregolith"[hartmann, 1973]. is especially important for characterizing lunar soils and Most soil-evolution models are based on the scenario first mineralogy. Initial lunar science results were presented by described from Surveyor observations [e.g., Shoemaker et al., Belton et al. [1992a], and more detailed results are presented

3 MCEWEN ET AL.' GALILEO OBSERVATIONS OF LUNAR CRATERS 17,209 here and in three companion papers in this issue [Greeley et al., this issue; Head et al., this issue; Pieters et al., this issue]. The basic SSI data set used in this paper and the companion papers in this special section is a coregistered set of multispectral mosaics consisting of the best SSI coverage of the illuminated hemisphere, corrected to normal albedo (Table 1' see also Figures 1-2 and Plate 1). Data processing for the mosaics included radiometric calibration, geometric control and reprojection, and photometric normalization, which are described in detail in the appendix. Images were acquired through seven bandpasses: VLT ( gm), GRN ( gm), RED ( gm), 727 ( gm), 756 ( gm), 889 ( gm), and 1MC ( gm); mosaics have been produced in all except the 727 bandpass (because the data are largely redundant with the RED and 756, but noisier due to the narrowness of the bandpass). Images at each wavelength are highly correlated, and all of the albedo images appear similar (e.g., Figure la), but color ratios (Figures lb-ld) reveal the subtle color distinctions of interest. Especially interesting are color-ratio composites [Johnson et al., 1977] that allow us to see the variation in two or three color ratios simultaneously (Plate 1, bottom). Derivation of relative and "absolute" reflectance spectra from the mosaics is described by Pieters et al. [this issue]. Among the most obvious features seen in the albedo maps and color ratios (Figure 1 and Plate 1) are the relatively recent craters, ejecta, and rays. Craters and other features discussed in this paper are labeled in Figure 3. The surface albedo of lunar soils decreases with time due to the formation of dark Well-known craters with bright rays on the western near side, which were imaged during EM1, include Tycho, Copernicus, Aristarchus, Kepler, Olbers A, and Byrgius A. Bright crater rays on the far side have not been well described, with the exception of publications in the former USSR based on Zond photographs acquired at low phase angles [Lipskil et al., 1975, p. 26] (Figure 4). Conspicuous farside rays seen in both the SSI and Zond images radiate from the craters Ohm, Lenz D, Lowell, Vavilov, and Jackson (cf. Figures 1, 3, 4, and Plates 1-2). The rays of Jackson are especially impressive; one ray extends at least 3600 km to the south, and another extends at least 1800 km to the southwest (Figure 1 and Plate 1). Jackson was imaged previously at a low phase angle at the extreme limb by one of the Zond spacecraft [Lipskii et al., 1975, po 26], but the full extent of these rays could not be seen. Also seen for the first time by Galileo is a bright, northeast-trending ray crossing the eastern portion of the South Pole-Aitken basin, which projects back to the 75-km-diameter Copernican crater Stevinus on the southeasternear side; this ray has a length of about 3000 km. These observations of bright rays indicate that two or more errors are present in the geologic map of this region [Scott et al., 1977]. Ohm was mapped as Eratosthenian, and the large nearby crater Robertson was mapped as Copernican. However, Ohm has obvious bright rays and ejecta (Plates 1-2 and Figure 4) and must be Copernican. Also, a ray of Ohm is superposed over Robertson, so Robertson probably belongs to the Eratosthenian or Upper Imbrian system. In addition, Scott et al. [1977] mapped the 66-km-diameter crater Lowell as Upper Imbrian, but prominent rays extending east and northwest from Lowell suggest that a Copernican age assignment may be correct. An alternative possibility is that a smaller Copernican crater responsible for the rays is superimposed on Lowell. The best available images, LO IV- 194-M at a large sun angle to see the topography and Zond-8 (Figure 4) at a small sun angle to see the rays, appear most consistent with the latter interpretation, but the resolutions are marginal. Lenz may also be incorrectly mapped (see discussion in section 4.2). Ejecta deposits within one to five crater diameters are distinctive around the rayed craters (Plates 1-2); some are agglutinitic glass by micrometeoroid impacts [Adams and McCord, 1970; Nash and Conel, 1973; Adams and Charette, 1975]. In addition, with increasing soil maturity, the continuum slope from-0.75 to 1.5 gm steepens and the 1-gm band decreases in depth [McCord and Johnson, 1969' Adams and McCord, 1971; McCord et al., 1976; Charette et al., 1976; Pieters, 1993]. Recent crater deposits (other than glassy impact melt) contain relatively abundant exposures of fresh crystalline materials due to (1) relatively little exposure to the space environment; (2) thin regoliths' and (3) steep slopes. The thin regolith and steep slopes both result in higher rates of soil replenishment with fresh crystalline materials. Hence, compared with surrounding mature soils, recent crater materials have relatively high albedos (Plate 2, top) and flat continuum slopes (e.g., high GRN/756 ratio, Figure lc), and radially asymmetric. The ejecta distributions around both Jackson and Ohm resemble that from an impact angle of 45 ø at White Sands Missile Range [Moore, 1976], and the ejecta around Olbers A ¾esemble that from an impact angle 5 ø above horizontal produced in the laboratory [Gault and Wedekind, 1978]. relatively deep 1-gm absorptions (e.g., high 756/1MC ratio, Figure ld). Combining the GRN/756 and 756/1MC ratios 2.2. Rings of Impact-Melt Veneer results in the GRN/1MC ratio (Plate 2, bottom), a useful Relatively dark rings surround the rims of Copemicus, maturity index [McCord et al., 1976; Johnson et al., 1977' Tycho, Aristarchus, Zucchius, Olbers-A, and other large Johnson et al., 1991]. The effects of the high 756/1MC and GRN/756 ratios are to make the more crystal-rich soils appear bright yellow in the color-ratio composite (Plate 1, bottom). TABLE 1. SS] Sequences Processed Furthermore, those craters with the most conspicuous ejecta Mapping Resolution, Subspacecraft, de 8 Phase Angle, Sequence km/pixel Latitude Longitude deg deposits and rays, which must be among the most recent craters [e.g., Shoemaker, 1962; Baldwin, 1963], also have the Lunmap highest albedos and 756/1MC and GRN/756 color ratios. Mapca Lunmap Bright-Rayed Craters Lunmapl Lunmapl Lunmap

4 ,210 MCEWEN ET AL.' GALILEOBSERVATIONS OF LUNAR CRATERS +8O " ".i: i - '",... " ' '.,.: --- ::.' ' :.... : : g..:.:':? :--.?,...-?.,. + o '"' -. """"'? ". "...,.: '...;...,..... S,{: t;--,,..: '... '.... ß... ' ;. -' '.; ,.:......,..,......,..., ,. '...., '...,.' s ' -.,; _.:... '..,' 4.- ':... -:.,.,"' ,...,..... :,-. e.:.- ;, '?. ::-. '-:.',> '... ]g s:.:<::,. m.":' 7'...,., ::.../?, e {W."'..' :-. : ::.:.:.' '...::.::' '. :' '.. '? ::?s'.'.t-"' '--. r, '...- ß '; "': ) '... ' ' :z... ' " ' " "-' :......:.:.;...." f '::.;:::a......,, ?:.:: :::..-::: ::. ;:, , c ,,... -.L, ,..--,,: -, " : :_..: l O 120 lo Fig. l a. GRN filter mosaic. Figures l a-ld are all in simple cylindrical projection. Curved white line outlines regions of most reliable data (see section A.4).,...: :".::.,-'-... -,..,.,...,I...: , ?.. :.. : , :..,, , ß.,,.... ß."' -' ' '- ß... '... f ;'..2- : '.-- - '. +80.";;"""' ':' -..'..'-".','.?-' -'. " '...-. ':: -...%.,..2 ;?.....';'. :.'u ' " ß....?...?. : , , r...,, --. -, , -., -,. -.'. ß '...?,.....,.., ,, ß.. _ g... ' '.;.,. -...; ?,,. ",;'.? "-.:.., ,...,...,.....,...,.,... -,.,.,,.. -- ' - ' - ' - :,-...'.. '"-? -? -.. : ::'.: f.;,. ' --: " -' ' ' -, ,...,..,..,.. "-' '" ' ' - '"..., ,', ' i.... ',....,-- '..' '.'..'...' :,..,...,...,,...,...,,.... ':.<;,.' "g;t....-".., '....,. f.', f' ;--..'?. W,,, , :. :...., :. ; t -.-, ' '...'-.:..-:... ' %-.? ;., :.., -. :--.., :: :.:....?. c'{ ",?f:(.-?' :: :-d'? d'.:4' f :' ' '. :'(.....:?...:.,..:.f- ; -. :: c ,;'. :.x. :--. :': '...?.....'.: :.' - "- :;.;: ß. l.i'. 'b:..' ß...:..?...:. :.....,:. -2O....., '...,..... :. i '..-:-.,. ß. -..,, ,... > ß -4O ", :...., " ß O l O 120 io O Fig. lb. VLT/GRN ratio.

5 MCEWEN ET AL.' GALILEO OBSERVATIONS OF LUNAR CRATERS 17, I - ' -5;:.::.. ß.:..?: :.o:..,.,..,... ::,: :....:,,.:-.:...-i?;,,-:-' 4,';.,, Fig. lc. GRN/756 ratio. +80 :--,... -?.'}.." :', O... '... '... ". ':...'. *'}..': , : ,'-".....a.,'..,....:.... -:;-... :....:. ß.'... '..'... --"*:' *' *:'J '"..,...,:' 1. - '.,.,:' '... **. ' **-,t, ',,- :,.:.... %.. ':.x,... :,. --,. ". +20 ß **:,..½x:****... :.* '*. '.. ' ' *..: ':' z..?-' ß :.:-. *..,x.../ ; :*., ,.-...,:-, ;:;. ß. ' '..-.".,. ;:..;.-.;.,:**".....,,-?::. <,...:?.. _.,.,,...,....,..,..,.. -2O, :... ':;..,.., *',' %':. ½;}.;' J.'.';, 2 '".. -.>a.., :..', ';,c**a..:--:,. : - -4O ' '.,,.S.... :. "-... '..... % ½... :..'-/'-:'..-.. '..".-/:.-* '-... :< "'""?:'- ':/;...,....: :...?...:..;;;;..% ;...;...:......_ Fig. ld. 756/1MC ratio.

6 17,212 MCEWEN ET AL.: GALILEO OBSERVATIONS OF LUNAR CRATERS Fig. 2. 1). Map of seams between each mapping sequence (Table Copemican-age craters, best seen on Earth-based images acquired at low phase angles. Most of these westernear side dark rings can also be seen in the SSI EM1 mosaics, but at lower contrast at the phase angle of this portion of the mosaic (57 ø ) than at full Moon. Howard and Wilshire [1975] proposed that the dark halo around Tycho was due to the presence of low-albedo ponds and thin veneers of fine-grained recrystallized impact melt. Hawke et al. [1979] suggested that a significant component of impact-generated glass was also required to account for the color properties of the dark rings. Smrekar and Pieters [1985] confirmed that the reflection spectra of several of these regions are consistent with the presence of Fe-bearin glass, in addition to pyroxene and Febearing feldspar. The dark rings extend up to one crater diameter from the rim, similar to the distribution of flows and ponds of likely impact-melt origin [Hawke and Head, 1977]. However, the dark rings are generally radially symmetric, whereas the flows and ponds have more asymmetric distributions. This suggests that the dark rings are primarily due to a finely dispersed and rapidly quenched (i.e., glassy) component of melt, rather than to the thicker melt flows and ponds which are likely to be holocrystalline. Dark rings of similar appearance are present around the farside craters Ohm, Vavilov, and Jackson (Plate 1, top, and Figure 4). These rings (including the near-side rings) also have color properties that are distinct from those of surrounding crater materials (Figure 1 and Plate 1, bottom). Specifically, the rings are redder ( um) and have relatively low values in the 756/1MC ratio (Figure ld), which are consistent with the presence of Fe-bearing glass. The spectral properties are also consistent with the effects of increasing soil maturity, but the presence of glassy impact melt is a more likely explanation because (1) these are young craters; (2) the concentration around crater rims is consistent with morphologic evidence for impact melts and the expectation that melting is concentrated in the deepest ejected materials; and (3) the rings are seen only around the largest craters, consistent with the theoretical expectation that only the most energetic impacts will produce large quantities of melt [O'Keefe and Ahrens, 1975; Grieve et al., 1977]. In addition to thin glassy veneers that are among the last ejecta deposits, another contributor to the dark rings may be impact-melt splashes (IMS's) such as those returned from the Apollo 16 landing site [See et al. 1986]. IMS's are significantly larger than soil agglutinates but smaller than fragments of holocrystalline melt sheets. These glassy materials occur as splashes that drape various host rocks, sometimes on several sides. See et al. [1986] proposed that the splashes were deposited on the host rocks inside growing crater cavities or during ballistic flight. Reddish materials similar to the dark Copernican rings are present around the rims and on the floors of the large craters Hausen (170-km diameter) and Pythagoras (120-km diameter; see Plate 1, bottom). These craters are much older than Copernican, probably Upper Imbrian (-3.6 b.y.) on the basis of new crater counts described below (section 3), so soilmaturity effects cannot be ruled out as readily as with the Copernican craters. However, impact melts have been identified around these craters from morphologic evidence [Hawke and Head, 1977], so we speculate that sufficient quantities of impact-melt splash and veneer were produced by these very energetic events to preserve a distinctive spectral signature in spite of more than 3 b.y. of regolith evolution. Furthermore, these units are redder ( gm) than surrounding mature highland soils. An alternative or additional hypothesis for the origin of the reddish materials around Hausen and Pythagoras is that these large craters excavated layers of relatively Fe-rich mafic materials, perhaps due to intrusions associated with nearby mare volcanism [e.g., Wilhelms, 1987; Head and Wilson, 1992], and the dark reddish deposits may be rich in both mafic breccias and Fe-bearing melt glass. To fuel further speculation, it is interesting to note that the spectral properties of the highlands materials within the South Pole/Aitken basin are also similar to that of the dark rings [Murchie et al., 1992; Pieters et al., this issue]. However, extensive glass has not been recognized in deposits from the Imbrium basin [Spudis et al., 1988], which is the largest wellstudied (and sampled) lunar basin, and suspected impact melt around Orientale basin lacks the relatively dark, reddish spectral properties [Head et al., this issue]. Thin glassy veneers on the surface would probably be largely buried or reworked by Imbrian resurfacing. 3. FREQUENCIES OF SUPERPOSED CRATERS We have examined the multispectral and superposed crater frequencies of large isolated craters, mostly of Eratosthenian and Copernican ages, to avoid complications due to secondaries (as they affect superposed crater counts) and spatially nonuniform regolith mixing from other nearby, large, and younger impacts (as it affects regolith-evolution models). The new crater size-frequency measurements (Table 2), made on Lunar Orbiter images, refine and extend the results of Neukum and KOnig [1976]. Our total data set for the region of the Moon imaged by Galileo includes 11 "mare" craters and 9 "highland" craters (Table 3). "Mare" and "highland" in this case refer to the regional terrain surrounding the crater; several of the "mare" craters excavated highland materials underlying the mare flows. Copernicus could have been grouped either

7 MCEWEN ET AL.' GALILEO OBSERVATIONS OF LUNAR CRATERS 17, Plate 1. Color presentations of SSI lunar mosaics. (Top) 1MC, 756, and GRN filter albedo images composited as red, green, and blue, respectively. (Bottom) Color ratio composite consisting of 756/1MC as red, GRN/756 as green, and VLT/GRN as blue. Curved white line outlines region of most reliable data (see section A.4). Simple cylindrical projections.

8 17,214 MCEWEN ET AL.' GALILEO OBSERVATIONS OF LUNAR CRATERS '., -.: g :.?..,..:. :: :,..,:.;,....,... >..:.,}.'....-,- :"....,-,.:?t..' -:.?'..:..:..1:.:-.%,...?....,...:......,..-,...-; ;....:...:_ _....::....., :..... :... ::.,.... : : A::.....a. -. : L}..:- :-.7..,:. :...:..: :... ß :..,, :....-:-. -.a...,.... :. :... ':....,,,.,..., " -...,.?,.... ;. -..: :..,,.,...,......:.... :,-' :.:....:: :., ' i,,, ' ' ','.... ',... ß,... ß.., " '.,' ß ß,... -*....:-. '.,- if-.:, ', ,.,,. : &.L,. -..,,..- - '.,... * '.".,..... v:...,-....:,...? ::... :.., ' 7,.., ;.... :'.-e --...,: '-:.. :.... :.. : ß ' - '..- : '.'.. ::.-:...::.:...:. '-- - ":'.-:-... :::::: " -. ' ' - ß,, ,.,, ,...,..., :.., ,-- -. ', c, ', :.. : ,:..? ,, 0 ::.,..,,.,,.,... '. _... :..'.. :.....,, ,...:,,...., :?-. :. : :..-.? :.. :. :......,...,...,...:..,..,... :,,..,.., -,.., -,....., ' -.: '.....-,....:.... :.:......',.:..:.: ::.: ,, - :.,......,, t:......,,. e..,., :., :.,, :,,.,.., ß..a :.'-..:,....:<...., ': , -...,.. :...-:...,.. -..,.... ß :,:. --,, 1. ' - "::.'.' :.-... :, :.,..,..,:,....:.,,.,.,...,.:.,. :... '..':',: [!,,-:-:., ,,,..,.:.? :...,.,...,.....,.... 7;:..,, :...,.,.,.-....,,....:.,:.,-...,-.:.- :... '-:'-.-:. -': : ' :' '-:-: -.'-', - t.,." %:-,..,.,.. -,., --., :.:....--:. :: -::. : -,...- -,. :. :. :.,... --,,,,--...,:-.., , :.,, -.. :: -'.,, S,..'. :.:.i..-:' -".-...,,:: :-:-.. :.., '.... :. ".

9 MCEWEN ET AL.' GALILEO OBSERVATIONS OF LUNAR CRATERS 17,215 MOON +8O +6O [ r' O O -4O -6O -8O Plate 2. Color-classified images. (Top) GRN-filter normal albedo, grouped into 12 categories. (Bottom) GRN/1MC ratio, grouped into five categories. Simple cylindrical projections.

10 17,216 MCEWEN ET AL.: GALILEO OBSERVATIONS OF LUNAR CRATERS Fig. 4. Zond-8 full-disk image of the Moon's western hemisphere, including the Orientale region and several bright-rayed craters. TABLE 2. New Measurements of Crater Size-Frequency Distributions Crater Cumulative N,* Fit Images$ (D=I km)x10-3 Range I' km Aristarchus LO IV-198-M, 201-M, 202-M Olbers A LO IV-174-H2 Kepler LO IV-137H1,H2 Philolaus LO IV-164-H2 Copernicus LO V-150-M Zucchius LO IV-154-H3 Carpenter LO IV-164-H2 Flaresteed LO IV-143-H3 Harpalus LO IV-158-H3 Hausen LO IV- 193-H2 Pythagoras LO IV-176-H2 *Cumulative crater frequency N (km-2) reduced to diameter of 1 km. In some cases these are the averages of two solutions, either from two measurementsuch as separate counts over crater floors and continuous ejecta, or fits over two size intervals. 1'Size interval used for fits to Neukum et al. [1975] calibration distribution. $Identifier of images used for crater counts. LO designates Lunar Orbiter [Hansen, 1970]. way; we chose to assign it to the "highland" regional terrain type. With five exceptions, these crater counts are consistent with the geologic maps shown by Wilhehns [ 1987], in which crater assignments in earlier mapping [e.g., Wilhehns and McCauley, 1971] were updated on the basis of the results of Neukum and KOnig [1976]. Three of the exceptions are the large craters Hausen (170-km diameter), Pythagoras (120-km diameter), and Bullialdus (61-km diameter) which, according to our crater counts, should be reclassified from Eratosthenian to Upper Imbrian. Other relatively large craters, Carpenter (60-km diameter) and Harpalus (39-km diameter), should perhaps be reclassified from Copernican to Eratosthenian. These are additional examples [cf. Neukum and K6nig, 1976] of how photogeologic crater classification tends to underestimate the ages of larger craters. One further potential reclassification applies to Diophantus (18-km diameter), mapped as Eratosthenian by Wilhehns and McCauley [1971], but which has a crater frequency slightly less than that of Copernicus. However, note that the crater frequencies for all six of the craters discussed above are near the boundaries of the relevant lunar geologic time periods; given the uncertainties in crater counts, perhaps formal reclassification should await more compelling evidence.

11 MCEWEN ET AL.: GALILEO OBSERVATIONS OF LUNAR CRATERS 17,217 TABLE 3. Lunar Crater Data Crater Abbre- Cumulative N,* Age, - GRN ALBEDO GRN/968 Diameter, viation (D=I km)x10-3 b.y. I$ CE$ I$ CE$ km Highlandsõ Hell Q HEL <0.08 < Tycho TYC Olbers A OLB Philolaus PHI Copernicus COP Zucchius ZUC Carpenter CAR Hausen HAU Pythagoras PYG Mariaõ Aristarchus ARI Kepler KEP Diophantus DIO Flamsteed FLA Delisle DEL Euler EUL Eratosthenes ERA Harpalus HAR Timochams TIM Lambert LAM Bullialdus BUL *Cumulative crater frequency N (km-2) reduced to diameter of 1 km, either from this work (Table 2) or from Neukum and KOnig [1976]. -Age model is from numerical solution of N (D=I) = 5.44 x 10-4(e '93' -1) x 10 '4 x t, where t is time in billion years [Neukum, 1983; see also Greeley et al., this issue]. $I denotes crater interior and CE denotes continuous ejecta. No I values are given for Flamsteed, as the interior was flooded by mare lava. No spectral values are given for Eratosthenes because of apparent contamination by ejecta and rays from Copernicus. Bullialdus may also be contaminated by a ray from Tycho. õ"highlands" "Maria" groups refer to the dominant composition of the surrounding terrains. Several of the "maria" craters excavated highlands materials. Copernicus could be classified either way, but is included with "highlands" here. 4. SPECTRAL REFLECTANCE VERSUS CRATER FREQUENCY parameter by previous workers [McCord et al., 1976; Johnson et al., 1977, 1991]. The broad bandwidth of the SSI 1MC The use of multispectral observations for age assignments filter, gm half-width at half-height of spectral response of Copernican craters was first proposed by Charettet al. (see Figure 3 of Pieters et al. [this issue], helps to minimize the [1976]. They demonstrated a correlation between 1-gm band effects of pyroxene mineralogy (i.e., the position of the band depth and the percentage of magnetic agglutinates returned minimum) on the GRN/1MC ratio values. Note the close lunar samples and between 1-gm ban depths from telescopic correspondence between relatively high albedo (relative to spectra of highlands craters and the D L measurements of mare or highlands location) and high GRN/1MC ratio of crater $oderblom and Boyce [1972]. DL is a model-dependent materials (Plate 2). parameter related to geologic emplacement age [$oderblom and To measure GRN/1MC and GRN albedos for crater Lebofsky, 1972], which correlates well with size-frequency materials, we first extracted small images containing each cou.nts on large Copernican craters [Wilhelms, 1987, p. 256]. However, the 1-gm band depths also vary significantly with the mineralogy, even when restricted to highlands [Charett et al., 1976] (Figure 5), which limits the usefulness of these relations for assigning relative ages to craters that lack radiometric dates or size-frequency counts. Because the SSI spectral capability includes only a broad 1-gm band and does not cover the-1.5-gm region (needed to map the crater from the mosaics and applied a subpixel color registration algorithm to avoid spurious color ratio values. All albedo and ratio values have been calibrated from telescopic measurements of standard areas, as described in detail by Pieters et al. [this issue; see also McCord et al., 1981], and the values are normalized to illumination and emission angles of 20 ø and 0 ø, respectively, from the surface normal. We avoided measurements over areas of suspected impact melt, gm continuum slope), we cannot make precise measurements described above in section 2.2. Where several measurements of band depth. Therefore, a different spectral parameterization were made of the interior or continuous ejecta, the of soil maturity is both desirable and necessary for the present measurements were averaged. Crater interiors can be effort. We have chosen two simple parameters: visible (GRN) normal albedo and the GRN/1MC ratio (e.g., Plate 2). The latter ratio includes both the effects of visible continuum slope (GRN/756 ratio; Figure ld), and 1-gm band depth (756/1MC complex, perhaps consisting of crystalline central peaks or terrace walls, slumping wall materials, ponds of impact melt, and breccia lenses. Ideally we would distinguish between these units in our spectral measurements, but we lack the ratio; Figure ld); a simila ratio has been used as a maturity needed spatial resolution in EM1 images.

12 ,,, 17,218 MCEWEN ET AL.: GALILEOBSERVATIONS OF LUNAR C TERS O ' _ 0.82 E e- - highlands (a) Crater Interiors--,--mr materials are shown in Figure 5. The correlations are clearly...! -A--R-!... i... significant for craters younger than Copernicus (linear OFEL ß :... ß... L... :... : ',,, ß [...'..,.....c o... [ ;... e...;,f.,:a.....! :... :... i'... HARJ.:....BILL., i i PHf"*' ':' " --I 1' LAM ß i :: i, 'F, IM I PYG : : :OO. I "'q", ', !... [... i"a;...? "8'" log cumulative crater frequency (D=I) (b) Crater Ejecta -e--highlands --O-- maria o, ?...!ari ß ß ß :: FLA...,--... :--'--.,-- EP-... ' -... '[... TYC! '% ' ' HAR:: :....'...';... PHI. _ DEl.. '::, LAM HAU 1,'z co _ Ti '' O 1 o ' ' CAR log cumulative crater frequency (D=I) O.3O (D 0.25 o 0.20 o. 5 E '-010 O ' (c) Crater - G- - highlands Interiors--B--maria, 1 i "4 o.! ; ; ARI ' ', O : ; 4 ; L'-..j.¾...c..... :... _ &:.:: : :::... 'i... :: :: e,..,-.. j log cumulative crater frequency (D=I) Fig. 5. Plots of superposed crater frequencies (N) versus SSI data: (a) N versus GRN/1MC ratio for crater interiors; (b) N versus GRN/1MC for continuous ejecta blankets; (c) N versus GRN albedo for crater interiors. Crater names abbreviated to three letters; see Table 3. Dashed lines show linear fits to data points. Correlations between log N (cumulative crater frequency per km 2 reduced to diameter of 1 km) and both the 0.56/0.99 gm color ratios and 0.56-gm normal albedos of the crater correlation coefficients > 0.8), except for GRN versus log N for continuous crater ejecta. For craters older than Copemicus, the ratio and albedo values tend to converge on values typical of the surrounding regions. Note that the interiors of most Copernican craters have higher albedos and GRN/1MC values than do the continuous ejecta from the same crater (Figure 6; this relation is also clearly seen in Plates 1 and 2). There are several reasons for this relation, including (1) mixing of ejecta with older soils, both during initial emplacement and from subsequent impact gardening; (2) downslope movements on steep interior crater walls, continuously exposing fresh crystalline materials; and (3) the presence of bedrock or breccia units exposed on crater floors or central peaks. Spectral properties of the crater interior and ejecta converge with time, and they are nearly indistinguishable for craters older than Copemicus (Figure 6) Mapping of Copernican Units We have used the correlations described above to estimate and map the geologic emplacement ages of many near-side and far-side Copernican craters for which counts of superposed craters are not available (typically because the imaging from spacecraft such as Lunar Orbiter is too poor). We chose to classify on the basis of the GRN/1MC values only, rather than albedo, because there are large regions of intermediate albedo, such as the South Pole-Aitken basin, Schickard-Schiller cryptomare, and Fra Mauro Formation, for which we lack sufficient crater counts for independent correlations, whereas the GRN/1MC ratio is more globally uniform (cf. Plate 2, bottom). We first stenciled the GRN/1MC ratio into images of either highland or mare regions, as distinguished on the global geologic map of Wilhelms [1987, Plate 12]. Large craters on the geologic map were assigned to either the highland or mare stencil on the basis of surrounding units. Several of the large craters in mare regions excavated highlands materials, so our classification is imperfect. Next, the linear correlations shown in Figures 5 and 6 were used to classify each pixel of each stenciled ratio into one of five categories (see Table 4). For the Eratosthenian-Copernican boundary we used the GRN/1MC value corresponding to the superposed crater frequency of Copemicus, and for the boundary between younger and older Copernican units (Cc 2 versus Cc,), we used the age of Kepler, the oldest crater in our data set mapped as Cc 2 [Wilhelms and McCauley, 1971]. Classified mare and highland images were then merged, and the final unit map is shown in Plate 3. Our color classification is compared with those from various geologic mappers and from the superposed crater counts in Table 5. In general, the results agree with the nearside geologic map [Wilhelms, 1987; Wilhelms and McCauley, 1971], where low-phase albedo data and thermal measurements [Saari and Shorthill, 1967] were available to aid the mappers. Several exceptions are discussed below, but where crater counts are available (Table 3), these age assignments are more consistent with the color assignments. Albedo and thermal

13 MCEWEN ET AL.' GALILEO OBSERVATIONS OF LUNAR CRATERS 17,219 1.o -0' - ejecta (a) Highland Craters--e--interiors el-b ß : I. : I, i ,...,,,,... ',,..C)... r.: :11...,,.... :... I: wc; i..: :! 0.75 I"... o.o 1.o log cumulative crater frequency (D=I) T= i :.-'..-, Sharonov and Crookes (previously mapped as Copernican by fi: 0.85 " e- - ejecta (b) Mare Craters --O--interiors,.,! I i... r... -i '.....,,,j.....'!?......b.u..l..._i ' ß... ; i ' BJL ' YAM TIM r. BUL ---] log cumulative crater frequency (D=l) Fig. 6. Plots of superposed crater frequencies (N) versus SSI data: (a) N versus GRN/1MC ratio for highland craters; (b) N versus GRN/1MC for mare craters. Crater names abbreviated to three letters; see Table 3. TABLE 4. Geologic Assignments of GRN/1MC Color Units data for the far side or limb regions (as seen from Earth) available to the geologic mappers [Scott et al., 1977; Stuart- Alexander, 1978; Wilhelrns et al., 1979] were much poorer than the near-side data, so we consider the color classifications to be more believable over these regions. Among the far-side craters, three of the color assignments that differ from the published geologic maps (those for Ohm, Robertson, and possibly Lowell) can probably be reclassified to agree with the color age assignments on the basis of the presence of bright rays and ejecta around Ohm and Lowell (or a crater superimposed on Lowell), and the superposition of a ray from Ohm over Robertson, as discussed above in section 2.1. In addition to these new age assignments, the color classification provides more detailed far-side information than the geologic maps in two ways: (1) Copernican craters are subdivided into younger and older units (which was not attempted on the farside geologic maps); and (2) many relatively small craters on the far side are indicated as Copernican in Plate 3 (the far-side geology has not been mapped at a scale below 1:5 million). Ours is the first attempt to assign Copernican ages to 27 craters larger than 10 km diameter. Several additional craters, not listed in Table 5 because they are outside of the region of most reliable multispectral data, nevertheles stand out as conspicuous features in the color and albedo (Plate 2) and are probably Copernican. These additional Copernican craters include the far-side craters Stuart-Alexander [1978]), and several km diameter craters southeast of Bailly and west of Lacroix (cf. Figure 3 and Plate 3). Certain lunar regions or craters with unusual compositions may be misclassified by our color-ratio method. For example, craters in cryptomare regions will tend to be classified too young because color-ratio values for the highlands were applied, but they may have excavated mare-rich materials with higher GRN/1MC color ratios (see Table 4). Crater materials rich in olivine or augre will have deeper 1MC bands than typical soils of comparable age, and they may also be classified too young. Fe-rich impact melt is relatively dark and red, as discussed in section 2.2, and is classified too old. The possible effects of exposures of pure anorthosite are discussed in section 5 below. Careful geologic interpretations are needed to avoid these possible errors. GRN/1MC Ratio Geologic Classification* Color 4.2. Discussion of Individual Craters Mare Highland (Plate 3) Aristarchus is one of the most conspicuous GRN/1MC anomalies on the half of the Moon imaged during EM 1 (Plates 2, bottom, and 3), and it is one of the most recent craters, as >0.89 >0.82 Cc 2 crater interiors or red indicated by the counts of superposed craters and by superposition near-crater ejecta. of Aristarchus rays over nearby young craters such as Kepler Cq crater interiors or yellow There appear to be at least two smaller and younger Copernican Cc 2 ejecta. craters superimposed on the ejecta north and northwest of Aristarchus, which contribute to the complex shape of the Cq or Cc 2 ejecta. green GRN/1MC anomaly. However, this is also a geologically complex area [Moore, 1965] and is compositionally very Cq ejecta or older blue heterogeneous [Lucey et al., 1986]. than Copernican Another strong albedo and GRN/1MC bright spot occurs <0.79 <0.765 Older than Copernican none (gray) near the double crater Lenz. This bright crater was first seen in photographs acquired by the Soviet Zond missions [Lipskii, *Cq denotes older Copernican crater; Cc 2 denotes younger Copernican 1975] (see Figure 4). The spot appears to be centered about crater. 25 km east of Lenz in the SSI mosaics CLenz D" on Figure

14 .., 17,220 MCEWEN ET AL.' GALILEO OBSERVATIONS OF LUNAR CRATERS MnnN,... ß "'. -.'.,½ '., * ;,l&.2,. -½ " Z ' ".....';! r; '.,-. "' '" '" 80 ø0 0 Plate 3. Classified map of Copernican-age surface units, superimposed on USGS airbrush map. See Table 4 for explanation of color categories. Simple cylindrical projection. 3), but the accuracy of the geometric control may be worse and the flux rate has remained constant, then the age of Lenz than 25 km over this area of the Moon. Scott et al. [1977] A or D could be about 12 m.y.. mapped Lenz (the double crater) as Copernican, on the basis The near-side crater Diophantus has GRN/1MC values of the bright rayed crater photographed by Zond 8. Wieczorek consistent with a younger Copernican age (Plate 3 and Table and Mendell [ 1993] report a very strong thermal anomaly near 5), but it was mapped as Eratosthenian by Wilhelms and Lenz, which they identified with a small (12.5-km diameter) McCauley [1971] and Wilhelms [1987, p. 125, Table 7.2] unnamed crater about 5 km southeast of Lenz (they labelled it "Lenz A"). The thermally enhanced ejecta blanket has a diameter of about 140 km, about the same size as the bright indicated that Copernicus ray material is superposed over Diophantus. Moore [1965] showed superposition of ray material from Aristarchus over Diophantus ejecta; this could ejecta blanket seen from Zond 8 and the GRN/1MC anomaly account for a high GRN/1MC ratio for an Eratosthenian crater. measured by Galileo, so these three observations probably It is also possible that Diophantus itself excavated materials of correspond to the same feature. If the young feature is due to unusual composition. However, the superposed crater Lenz A or D rather than Lenz, then the Copernican frequency [Neukum and KOnig, 1976] indicates an age slightly assignment to Lenz may be incorrect. Wieczorek and Mendell classified Lenz A as the youngest of 17 Copernican and Eratosthenian craters in their dataset, younger than Olbers ^ or Aristarchus (class 2), or Kepler (class 3). Based on the strength of the GRN/1MC ratio values, we would classify these four craters into the same three relative age groups. The anomaly near Lenz has the highest GRN/1MC value (-1.0) of any highland crater in the dataset, and its extrapolated crater younger than Copernicus (CQ), and the inclusion of secondaries from Copernicus and Adstarchus might bias the crater count towards an older than actual age estimate. We conclude, for now, that the correct relative age assignment for Diophantus is uncertain, but it is probably either younger Eratosthenian or older Copernican. The large craters Carpenter and Philolaus have been mapped as Copernican [Lucchitta, 1978] on the basis of crater frequency (N; see Figure 6a) is 10-5, about 1/8 of Tycho's dating by Neukum and Ki nig [1976]. Although outside the frequency. If Tycho is about 100 m.y. old [Drozd et al., 1977] limit of reliable SSI data (appendix), these craters appear to

15 MCEWEN ET AL.: GALILEO OBSERVATIONS OF LUNAR CRATERS 17,221 TABLE 5a. Age Assignments for Near-Side Copernican Craters TABLE 5b. Age Assignments for Far-Side Copernican Craters Crater Name Diameter,* Geologic Age Assisnment I' km Plate 3 Geologic Crater Map Frequency Crater Name Diameter,* Geolol ic Aloe Assisnment ' km Plate 3 Geologic Crater Map Frequency Tycho 85 Cc2 Cc2 Cc: Olbers A 43 Cc2 Cc2 Cc: Aristarchus 40 Cc2 Cc2 Cc 2 Kepler 32 Cc2 Cc2 Cc: Lalande 24 Cc: Cc2 -- Horrebow 24 Cc2 Ec -- Bouger 23 Cc: Cc -- Cleostratus J 20 Cc 2 Ec -- Byrgius A 19 Cc 2 Cc 2 -- Diophantus 18 Cc 2 Ec Cc Sharp A 17 Cc 2 Ec -- Phocylides N 15 Cc2 Isc : -- Galvani B 15 Cc 2 Icl -- Krafft H 15 Cc2... Darney 15 Cc 2 Cc -- Einstein Z 4 13 Cc2... Plato M 8 Cc 2... Rumker E 7 Cc2 Cc -- Lichtenberg B 5 Cc 2 Cc -- Hell Q 4 Cc2 Cc2 -- La Condamine S 4 Cc 2... Zucchius 64 Cc Cc /Cc2õ Cc Robinson 24 Cc Ic -- Sharp B 21 Cc Cc -- Strove H 21 Cc Ec -- Mairan A 16 Cc Cc -- Harding A 14 Cc Cc -- Sirsalis F 13 Cc Cc -- Euclides 11 Cc Cc -- Copernicus 93 Cc or Ec Cc Cc Gassendi A 33 Cc or Ec Cc -- Briggs B 25 Cc or Ec Cc -- Lichtenberg 20 Cc or Ec Cc -- Harpalus 39 >C Cc Ec Timocharis 34 >C Cc Ec Euler 28 >C Cc Ec Dash indicates not mapped. *From Andersson and Whitaker [1982]. 'Ccl, older Copernican; Cc2, younger Copernican; Cc, Copernican (undivided); Ec, Eratosthenian; Icp Lower Imbrian; Ic2, Upper Imbrian; >C, most likely older than Copernican. $Mapped as Orientale secondary by Wilhelms et al. [1979]. õmapped as Cc by Wilhelms and McCauley [1971] and as Cc z by Offield [1971]. have GRN/1MC values consistent with an age older than Copernicus (Plate 3). New determinations of the frequencies of superposed craters (Table 2) indicate that Carpenter may indeed be older than Copemicus. The spectral properties of Philolaus may be highly affected by impact melts. Superposition relations indicate that relative ages decrease in the order Pythagoras-Carpenter-Philolaus [Lucchitta, 1978], consistent with the crater frequencies in Table 2. These cratering events probably occurred over a period of more than 2 b.y. (Table 3), yet all would be classified to about the same relative age on the basis of the seven morphologic stages for relative age-dating of large craters of Pohn and Offield [1970], widely used by lunar geologic mappers. Another discrepancy in the near-side classifications (Table 5) is apparent for the crater Zucchius, which was mapped as Jackson 71 Cc 2 Cc Lowell (?)õ 66 Cc 2 Ic 2 Ohm 64 Cc 2 Ec Galois C 22 Cc2 -- Galois B 20 Cc 2 -- Lewis P$ 20 Cc2 Cc Petzval Wn 20 Cc 2 -- Lowell W 18 Cc2 Ec Butleroy X$ 18 Cc 2 -- Grigg F ' 15 Cc2 -- Lenz D 12 Cc 2 -- Fizeau R 4 11 Cc2 -- near Wood S 10 Cc 2 -- Lippmann W$ 10 Cc2 -- Vavilov 99 Cc Cc Engelhardt 43 Ccl Ic 2 Das 39 Cc Cc Guthnic 36 Ccl Cc Golitsyn 36 Cc -- Gerasimovich D 26 Cc Ic2 Focas 22 Cc Ec Golitsyn J 20 Cc Ec Robertson 88 >C Cc Dash indicates not mapped. *From Andersson and Whitaker [1982]. 'Ccl, older Copernican; Ccz, younger Copernican; Cc, Copernican (undivided); Ec, Eratosthenian; I% Lower Imbrian; Ic 2, Upper Imbrian; C, most likely older than Copernican. $For these unnamed craters, we applied a letter designation using the far-side azimuthal scheme relative to the nearest named crater (see"notes on Base" in U.S. Geological Survey [1980]). õspectral data may apply to a crater superimposed on Lowell, as discussed in text. Cc 2 by Offield [1971] and as Ccl by Wilhelms and McCauley [1971]. Zucchius has a crater frequency identical with that of Copemicus (Table 3). The crater interior has a GRN/1MC ratio value consistent with the younger Copernican (Cc2) classification, but the data are near the limit of reliability (illumination angles > 70ø; see appendix). The GRN/1MC ratio of the continuous ejecta of Zucchius varies, but it is most consistent with a Cc classification. Therefore, results are inconsistent from both the geologic mapping and the color method, but all agree that Zucchius is Copernican. The spectral interpretations may be complicated both by impact melts [Hawke and Head, 1977] and by cryptomare associated with the Schickard-Schiller region [HawIce and Bell, 1981; Hawke et al., 1991 ] Near-side/Far-side Abundances of Copernican Craters The apparent deficiency (on published geologic maps) of Copernican craters on the far side compared with the near side [cf. Wilhelms, 1987] is not present in the portions of the near side and far side covered by our Copernican-unit map (Figure 7), confirming the suspicion that this apparent deficiency was due, at least in part, to a paucity of low-phase images of the far side. In fact, our spectral map shows more Copernicanage surface coverage on the far side (Table 6). However,

16 17,222 MCEWEN ET AL.: GALILEO OBSERVATIONS OF LUNAR CRATERS Fig. 7. Comparison of near-side and far-side abundances of Copernican-age units. Sinusoidal equal-area projection; vertical white line delineates longitude 90 ø. Four brightness levels correspond to categories in Table 4, with white for red, light gray for yellow, medium gray for green, and dark gray for blue. See Table 6 for percent coverage by each unit. TABLE 6. Near-Side/Far-Side Comparison of Copernican-Age Color Units (Figure 7) Unit* % Near Side % Far Side The linear trends between N and 0.56/0.99 micron ratio differ from maria to highlands (Figure 5) and between the interiors and continuous ejecta of Copernican craters (Figure 6). Similar trends are established for color and albedo versus soil maturity indices for the returned lunar samples, again with distinct trends for mare and highlands soils (Figure 8). However, the mare versus highland offsets are reversed in the two comparisons: any particular GRN/1MC ratio value corresponds to a smaller N (younger emplacement age) and a Blue 8 13 larger maturity index (older exposure age) for highland trends Green 5 9 than for mare trends. These trend offsets may be explained by Yellow variations in regolith thicknesses, which influence the rates of Red mixing with relatively fresh ejecta [McKay and Basu, 1983; Basu, 1990]. The maria have thinner regoliths than do most * Color as shown on Plate 3. highland areas, so mare soils undergo a higher rate of vertical and horizontal mixing with fresh ejecta from nearby impacts. given the calibration uncertainties (appendix), the difference A similar explanation may apply to the different trends seen in near-side/far-side resolutions, the different compositional in continuous ejecta blankets and in crater interiors: the units, different phase angles, etc., we do not consider this difference to be significant. The impacting flux is unlikely to significantly favor one hemisphere over another, but it is possible that the rougher far-side topography [Kaula et al., 1974] could result in some differences in soil maturity due to downslope movements. interiors undergo a higher rate of mixing with fresh ejecta due to (1) greater near-surface abundances of blocky or massive crystalline rock in the crater interior and in the blocky ejecta concentrated near the crater rim, and (2) relatively steep slopes on crater walls and central peaks, where regoliths are kept thin by downslope mass movements and preferential downslope movement of ejecta from impacts that are small relative to the 5. SPECTRAL REFLECTANCE VERSUS slope [Soderblom, 1970]. Therefore, soil-maturity parameters, SOIL MATURITY IN APOLLO SAMPLES: which are related to a soil's cumulative exposure age, correspond to a range of emplacement ages for the underlying EFFECTS OF REGOLITH DEPTH geologic unit, depending on the geology of the sun'ounding area. Regolith thickness should increase with net accumulated impact flux, so the younger maria are expected to have generally thinner regoliths than the older highlands. This relation is confirmed by a variety of measurements and observations (summarized by Moore et al. [1980, pp. B13-

17 - - _ -. MCEWEN ET AL.' GALILEOBSERVATIONS OF LUNAR CRATERS 17, ' (a) - ß- maria Lunar Soils -- < -- highlands Maturity ( IIFe O) - ß- maria (b) Lunar Soils--O--highlands o o -. o o --. o o '0-.. o o o o ø Orientale and Grimaldi have been reported [Spudis et al., 1984; Hawke et al., 1991]. At least one spot with a very high' GRN/1MC ratio occurs on a massif of the Inner Rook Ring of Oftentale, at a location where pure anorthosite has been observed [Hawke et al., 1991; see Figure 2 of Lucey, 1992]. From the reflectance spectrum of anorthite [Pieters, 1986, 1993], we would expect pure anorthosite to have a relatively high albedo and GRN/756 ratio and a relatively low 756/1MC ratio in the SSI data. However, the basin-ring units shown in Plate 3 all correspond to areas with relatively high 756/1MC ratios (Figure l d), as well as high albedos (Plate 2) and GRN/756 ratios (Figure lc), consistent with low soil maturity. In fact, none of the SSI EM1 data are obviously consistent with the expected signature of nearly pure anorthosite, although the resolutions are comparable with that of the published telescopic spectra (5-10 km). Perhaps the spectral range and resolution of SSI do not permit the unique identification of pure anorthosites. The hypothesis that less mature soils over rugged terrain could affect the spectral properties of basin rings is supported by the preferential occurrence of high albedo and GRN/1MC values where ring massifs are crossed by bright rays. Examples include the northern rings of Orientale, crossed by bright rays from Lowell and Olbers-A, and Hertzsprung, crossed by rays of Jackson and Vavilov or Vavilov D (see Plates 1-3). Small-scale landsliding may have occurred, perhaps analogous to the Apollo 17 bright landslide triggered by secondaries from Tycho [Lucchitta, 1977]. 6. SUMMARY AND CONCLUSIONS Copernican-age craters are among the most conspicuous features seen on the far side and western limb of the Moon in the Galileo multispectral images acquired in December o 4o 6o 8o lo0 Among the new observations of far-side craters are bright Maturity ( I/F e O) rays, continuous ejecta deposits, and dark rings associated with probable impact-melt veneers. Colors and albedos of Fig. 8. Plots of soil maturity index (ljfeo [see Morris et al., Copernican craters (away from impact-melt veneers) are 1983] versus spectral properties of retumed lunar samples, correlated with their geologic emplacement ages as determined convolved to SSI bandpasses. (a) ljfeo versus GRN/1MC ratio; from counts of superposed craters. These results are (b) l/feo versus GRN normal albedo. compared with relations among the colors, albedos, and soilmaturity parameters of returned lunar soils. Three data sets B 14]). The youngest (Copernican and Eratosthenian) mare are utilized: (1) Galileo SSI multispectral observations of the regions of Oceanus Procellarum have some of the thinnest Moon's western near side and far side; (2) new measurements mare regoliths, from 1 to 5 m thick [Quaide and Oberbeck, of the frequencies of craters superimposed on the floors and/or 1968; Moore et al., 1975]. The relatively young flows of continuous ejecta blankets of large post-imbrium craters in the Oceanus Procellarum also show a regionally elevated Moon's western hemisphere; and (3) measurements of spectral GRN/1MC ratio value (Plate 2, bottom). The thinnest regolith reflectivity and soil maturity on the returned samples of lunar measured, m, is on the rim flank of Tycho, determined soils. Production of the SSI multispectral mosaics is described from the depth of trenches excavated by the Surveyor 7 surface in the appendix. sampler [Scott and Robertson, 1968]. From these limited We have examined the multispectral and superposed crater measurements, we can conclude that regolith depths are frequencies of large isolated craters, mostly of Eratosthenian typically 1 to 8 m over mare plains, 8 to 15 m over highlands, and Copernican ages, to avoid complications due to and less than 1 m over young Copernican units. secondaries (as they affect superposed crater counts) and Some areas of high GRN/1MC ratio mapped as potential spatially nonuniform regolith mixing from other nearby, large, Copernican-age units (Plate 3) correspond to mountain rings and younger impacts (as it affects regolith-evolution models). surrounding Oftentale, Grimaldi, Hertzsprung, Apollo, and The new crater size-frequency measurements refine and extend South Pole-Aitken basins; these units may be interpreted as previous results; our total dataset for the region of the Moon exposures of nearly pure anorthosite, areas of immature soils imaged by Galileo includes 20 craters. Our crater counts due to downslope movements on steep slopes, or both.,suggest the following age reassignments: Hausen (170 km Exposures of nearly pure anorthosite on the inner rings of diameter), Pythagoras (120 km), and Bullialdus (61 km) from

18 17,224 MCEWEN ET AL.: GALILEO OBSERVATIONS OF LUNAR CRATERS Eratosthenian to Upper Imbrian, Carpenter (60 km) and Harpalus (39 km) from Copemican to Eratosthenian, and Diophantus (18 km) from Eratosthenian to Copernican. Correlations between log N (cumulative crater frequency The Clementine mission will carry four imaging systems and a laser altimeter to the Moon in early If this sensor test mission is a complete success, it will return global multispectral imaging in 11 bandpasses from 0.41 to 3.0 txm per km 2 reduced to diameter of 1 km) and both the 0.56/0.99 at scales of m/pixel, selected high-resolution (10-30 txm color ratios and 0.56 gm normal albedos of the crater materials are clearly significant for craters younger than Copernicus. These results are used to estimate the geologic emplacement ages of many other near-side and far-side Copernican craters. The apparent deficiency of Copernican craters on the far side compared with the near side in published geologic maps is not present in our data, confirming the suspicion that this apparent deficiency was due (at least in part) to a paucity of low-phase images of the far side. The mapped age assignments of several large post-imbrium farside craters (Ohm, Robertson, and possibly Lowell and Lenz) m/pixel) and thermal IR images, and altimetry profiles under each of more than 300 polar orbits. Such a dataset would enable much more detailed studies of craters and many other aspects of the lunar surface. Global remote sensing will provide a wealth of information for relative age dating, but does not provide the absolute age information needed to calibrate the impact flux over the Earth- Moon system during the past billion years. Additional sample-return missions are needed. At the present time, no absolute age dates are available for any Copernican crater older than 50 million years (e.g., North Ray crater [Behrmann need revision. et al., 1973; Marti et al., 1973]). Inferred or indirect The linear trends of N versus 0.56/0.99 gm ratio differ between mare and highland regions and between the interiors radiomettic ages for Tycho (100 m.y.) and Copemicus (850 m.y.) are highly controversial (see discussions in Neukurn and and continuous ejecta of the craters. Similar trends are Ki nig [1976] and Wilhelms [1987]). A series of simple established for color and albedo versus soil-maturity indices for the returned lunar samples, again with distinct trends for mare and highland soils. However, the mare versus highland offsets are reversed in the two comparisons: any particular unmanned probes for the return of samples of regolith over impact melt ponds associated with about nine key craters with extensive bright rays, such as Copernicus, Tycho, Stevinus, Aristarchus, Anaxagoras, Hayn, Giordano Bruno, King, and 0.56/0.99 txm ratio value corresponds to a smaller N (younger Jackson, could provide global absolute age calibration. emplacement age) and a larger maturity index (older exposure age) for highland terrains relative to mare terrains. These APPENDIX: PRODUCTION OF trends and the different trends for soils formed over crater MULTISPECTRAL MOSAICS interior and continuous ejecta deposits can both be explained by variations in regolith thicknesses and rates of mixing with Systematic processing of the imaging data is described relatively fresh, crystallin ejecta. Therefore, the soil-maturity below, including radiometric calibration, geometric control, trends and parameters represent longer geologic time periods for samples of thinner regoliths such as that over the maria. photometric normalization, and mosaicking [cf. Batson, 1987]. Preliminary plans exist at USGS in Flagstaff to archive the 7. FUTURE PROSPECTS EM 1 mosaics on compact disks (CDs), along with other lunar data sets, for distribution to the planetary science community. The Galileo Earth-Moon 2 (EM2) flyby in December 1992 CDs containing the Raw Experimental Data Record (REDR), provided observations of the north polar region and entire near or unprocessed images, are available from the Planetary Data side at resolutions as high as 1.1 krn/pixel. Compared to EM1 Facility, Jet Propulsion Laboratory. data, the EM2 data have improved calibration and signal:noise A. 1. Calibration because (1) the transparent dust cover has been removed; (2) the exposure times were long enough for about two times Several calibration issues related to "flat field" corrections improved signal:noise; (3) in-flight calibration images were acquired; and (4) higher spatial resolution and Target Motion Compensation (TMC) made the diffuse scattered light more uniform over each frame and color set (and thus, easier to normalize). We expect further improvements in the data from use of improved photometric normalizations and new algorithms for subpixel coregistration of the color data. and frame-to-frame consistency had to be addressed before mosaicking. Corrections to telescopic and laboratory spectra [Pieters et al., this issue] were applied after mosaicking. A transparent dust cover over SSI had two effects on the images: (1) it reduced the transmission, especially at the wavelength extremes (near 0.4 and 1.0 [xm)[breneman and Klaasen, 1988]; and (2) its reflections created "ghost" images [Belton et al., Geometric control will be tied to the unified control net of 1992b]. To complicate these problems, the in-flight calibration Davies et al. [ 1987]. We will also apply the new photometric images were not properly acquired, and only preflight ß models, processing algorithms, and geometri control to the EM1 data for a combined EM1-EM2 multispectral mosaic radiometric files, which were not acquired through the dust cover, were available. As a result, ting-shaped shadows from covering about 75% of the lunar surface. EM2 multispectral dust particles in the optical system ("dust donuts") are not observations of post-imbrium craters on the eastern near side will roughly double the number of craters available for correlation of spectral properties with frequencies of superposed craters. The higher spatial resolutions should allow detailed mapping of compositional units associated with completely removed by the pre-flight calibration files [see Belton et al., 1992b, Figure 5]. A dust donut correction file was created by filtering and stenciling an image of Venus' limb (picture number V645) that is very bland (except for dust donuts); this correction has been incorporated into the large, well-preserved craters such as Copernicus, Eudoxus, calibration files on the EM1 REDR CDs. Carpenter, Philolaus, Anaxagoras, and Hayn. The ghost images appear to be simple single reflections,

19 MCEWEN ET AL.' GALILEO OBSERVATIONS OF LUNAR CRATERS 17,225 displaced 23 pixels to the left and 2 pixels up, relative to each frame. The ghosts were removed by multiplying each frame by the ratio of brightnesses of ghost:primary (Table 7), repositioning the image 23 pixels to the left and 2 pixels up, and subtracting it from the original. An additional problem, not related to the cover, is a lowlevel, low-frequency brightness component over the images caused by scattered light, best seen next to a bright limb. From observations of Jupiter, Gaspra, Venus, and the Moon, we know that the larger the angular diameter of the bright object, the greater the magnitude and spatial scale of the observed off-limb scattered light. The scattered component appear symmetric about the target (planetary body) in or near the field of view in all filters except 1MC, but this could change when in-flight calibration is available. An attempt was made during the EM 1 encounter to remove the scattered light with a series of filters, and although successfully normalizing space to within one-half data number (DN) of zero, our efforts worsened the frame-to-frame consistency of the data on the illuminated disk, so the effort was abandoned. However, we found that the halos were similar within each matched set of seven images through each filter, i.e., with the camera pointing toward the same portion of the target, resulting in consistent color ratios. Use of the overlap region between different matched sets resulted in obviously inconsistent color ratios. Therefore, we proceeded to minimize this problem in two ways: (1) stenciling each seven-filter set of images to prevent overlap with adjacent sets in the mosaics; and (2) empirical frame-to-frame histogram matching (after photometric normalization) via linear leastsquared fits to normalize small residual offsets. These linear fits are very precise, with correlation coefficients typically around 0.99, so the slope and offset terms seem to describe the frame-to-frame offsets very exactly. Shadow values from Lunmap 8 images were used to fix the zero level. The multiplicative terms correlate with phase angle and wavelength, consistent with phase-function residuals (described below). A.2. Geometric Control and Reprojection Images from six of the lunar mapping sequences were chosen for optimal longitudinal coverage at the best resolutions and lowest phase angles (Table 1). The USGS shaded relief maps of the Moon [U.S. Geological Survey, 1980, 1981, 1992] were scanned into digital format and used as base maps for geometric control. Although the control of the airbrush maps is far from perfect, we judged them good enough relative to EM1 resolutions, and this procedure ensured good registration to the airbrush map for comparison of spectral units with the surface morphology. Camera angles were corrected by choosing tie points (to the airbrush map) and match points between SSI frames [Edwards, 1987]. Each frame was then reprojected into the sinusoidal equal-area projection for mosaicking. After mosaicking, the data sets were reprojected into simple cylindrical or other map formats. The mosaics were constructed from the Lunmaps 8, 12, 14, 23, and Mapcal01 sequences (cf. Table 1) to provide the best overall mosaic. Lunmap 16 images were excluded because they would provide only a sliver of data at better resolution, but would introduce an extra seam. A seam-removal technique was applied to the mosaic (after histogram matching) to average out residual calibration and photometricfunction errors. A.3. Photometric Normalization We modeled normal albedo with the Hapke photometric function and parameters from Helfenstein and Veverka [1987] for the disk-integrated Moon (w = 0.21, h = 0.07, S = 0.71, g = -0.1, and 0 = 20ø). Identical parameters were applied to all bandpasses. This model appears to provide an adequate limbdarkening correction over most areas, but overcorrects the limb darkening at high photometric latitudes (which correspond to planetary latitudes for these images), especially at higher phase angles (Figure 9). As a result, the highland terrains near the limb of Lunmap8 images, in particular, are too bright. The phase darkening of each frame (independent of illumination or emission angle) was clearly undercorrected by the Hapke function, i.e., higher-phase images remained systematically darker than lower-phase images. This systematic residual also varies with wavelength. Using the overlap between frames, we found linear correction factors for each filter (as part of the histogram matching described above). The correction factors (Table 8) increase systematically with phase angle and decrease with increasing wavelength. Therefore, application of these correction factors normalizes the phase reddening of the lunar regolith. These residual phase corrections are surprisingly large. One might have expected the disk-integrated Hapke fit to do a good job of normalizing the overall phase darkening of the Moon, at least in visible wavelengths. However, the viewing angle limitation from Earth-based observations insures biased phase curves. A different sample of the lunar near side is viewed at each phase angle, but the albedo variations from place to place are considerable. Because the darkest regions are near the center of the disk, the "well-known" lunar phase curve is not Filter VLT GRN RED MC TABLE 7. "Ghost Image" Fractional Values (Ghost Brightness/Direct Image Brightness) Filter TABLE Ghost/Direct Ratio VIO GRN RED < < Phase Function Corrections (Normalization to Lunmap12, 20 ø Phase Angle) Phase An le, deg

20 .. 17,226 MCEWEN ET AL.' GALILE OBSERVATIONS OF LUNAR CRATERS ::,,.. '..,.,..,.. :..,..½' ::..' ;.., ;. :%- :...:.., i-. - ', '" :-.:,..;:. 4..'... ;.-;:...,....,:.:..:-,;:?&]..--'... : '-::.:...{..: ;..::, ,½ :-.m.. Q-,. '-:. :' :.--: --.-.:..::.; ',,-;:-..?'... " Y:.:.:.";....:: :. _'... ': 2'.:.".%.. ':-:,: ':... "* d?:.':.;" ' :.' -'}::',,. '...,;"...,.½.....;...½ :.-...!; :- :.:--...::%...,...'-. :.-":;;4:: u.----:-::"' ':::..;... :. -;'".' :':.? ½"2, '.: ':"*'""; -.. :'.:.:....,... ; ;'-".::;, :-:" --' ½: ".:'.' --:--.;:'- 2";'- -...:: x.'--... ' '-...? :-' :. ' -....,.½ ' -.?.::;': "' './:- u:': -.;-:...:...,:.:-,.:.-'*.:.:...."... - ;.:.::...%-...: ½:,...,. ½;... -'-.. :.... :: ;.... m-'. Fig. 9. GRN filter albedo for each of six mapping sequences: Lunmap23 (upper left), Lunmap16 (upper fight), Lunmap14 (middle left), Lunmap12 (middle fight), Mapca101 (lower left), and Lunmap08 (lower fight). Simple cylindrical projections. Curved white lines outline regions of illumination angle less than 70 ø and emission angle less than 80 ø. Data have been grouped into five albedo categories to facilitate comparisons.

21 MCEWEN ET AL.: GALILEO OBSERVATIONS OF LUNAR CRATERS 17, :.. :.: :'.;.:. :. :.. :-,:....,.:. -.,.;...,..:....,... %:::..,:....: :.,:*.>:..:;.:.'..... '. ::""":;:.':' :,:.... ;' :. " ::... '..; ::.,::'. ß.e ::,. :.:.... :.; :..:, :::....::.:.e'?½.,. '% *.:..: : :...:. ½"....;;::::': ::...,,.,?½: -½½..., --...:".'""' "-"' ' m' '"' ':: :-..:-...-:., :::,:,,., :.:.:......,.. 2.,,.,.:.,, Fig. 10. GRN/1MC ratio for each of six mapping sequences: Lunmap23 (upper left), Lunmap16 (upper fight), Lunmap14 (middle left), Lunmap12 (middle right), Mapca101 (lower left), and Lunmap08 (lower right). Simple cylindrical projections. Curved white lines outline regions of illumination angle less than 70 ø and emission angle less than 80 ø. Data have been grouped into five categories to facilitate comparisons.

22 17,228 MCEWEN ET AL.: GALILEO OBSERVATIONS OF LUNAR CRATERS a very good estimate of the true phase darkening of the average lunar surface. To examine residual errors or variance in the data, a series of ratios of coregistered images from different mapping we trimmed off all data with a sun or emission angle exceeding 85 ø, which were clearly not useful data for spectral, studies. However, we chose not to trim data from 70o-85 ø illumination or 80o-85 ø emission, because they may be sequences were produced. These ratios show (1) high- qualitatively useful. For example, a local color/albedo frequency bright and dark parallel bands due to misregistration anomaly (compared with the surroundings) is of interest to (mostly subpixel magnitudes); (2) increasing noisiness toward terminator regions; and (3) relatively bright or dark terrain units due to scattering-function variations [Helfenstein and Veverka, 1987]. The scattering-function terrain variations are most pronounced at large sun angles or when ratioing images with significantly different phase angles, consistent with the note, in spite of the quantitative problems with the data. Only a small percentage of the data, as seen in the original point-perspective views, are affected by these problems, and we mosaicked the frames in a way that favors the "better" data. But there remain three problems: (1) the south polar region is covered at large sun angles in all frames; (2) the effects expected due to variations in macroscopic roughness' westernmost longitudes are covered only by the low-resolution [Hapke, 1984], but differences in single-particle phase function frames; and (3) there is a tradeoff between resolution and may also have some effect. These variations largely disappear photometric quality. Because of problem 3, the in the ratios of color ratios, as expected for macroscopic Lunmap seam includes data of poor photometric quality roughness variations. We expect to be able to use such information to map roughness variations over portions of the Moon imaged multiple times with a range of illumination and phase angles. on the Lunmap 14 side of the seam, to preserve the better spatial resolution. Problems 1 and 2 have no immediate solution; the major casualty is the south polar region and longitudes beyond about 160 ø W. Unfortunately, given the geometry and nature of this problem, all except the A.4. Analysis of Errors northeastern region of the South Pole/Aitken compositional anomaly are outside the region of data reliability. Perhaps The intrinsic limitations of our EM1 multispectral dataset trustworthy color data over this important region could be must be understood to avoid possible spurious interpretations. extracted in the future if we can (1) apply a rigorouscattering Comparison of reprojected and Hapke-normalized images correction model; (2) improve the geometricontrol; (3) apply reveals small systematic discrepancies with respect to an improved photometric function; and (4) average over large photometric coordinates (Figures 9-10). If the calibration, areas to improve the S:N due to the underexposure. signal-to-noise ratios (S:N), and Hapke normalizations were all perfect, then the albedo and color data from individual Acknowledgments. Many thanks to T. Becker, D. Cook, K. reprojections (e.g., Lunmaps 23, 16, 14, 12, 8, and Mapca101) Edwards, and E. Eliason for assistance with the image processing would be identical where they overlap (except for resolution and programming, to R. Wagner for crater count compilations, and topographic effects near the terminator). This is not the and to H. Moore, P. Spudis, B.R. Hawke, and an anonymous case. The albedo, as discussed previously, is too high at high reviewer for critical comments. This research was supported by latitudes (Figure 9). NASA's Galileo project and contract W-13,709 to USGS. The color data also become systematically poorer at large sun angles (Figure 10). The reasons for this are (1) poor overall S:N ratios because the images were underexposed, REFERENCES resulting in maximum DN values ranging from 45 to 130 out Adams, J.B., and M.P. Charette, Effects of maturation on the of a possible maximum of 255; (2) especially poor S:N at high illumination angles; (3) the SSI scattered-light problem, which reflectance of lunar regolith: Apollo 16--A case study, Moon, 13, , becomes significant over areas with low signal; (4) the fact Adams, J.B., and T.B. McCord, Remote sensing of lunar surface that the Hapke normalizations overcorrect the albedo at high mineralogy: Implications from visible and near-infrared latitudes, which in turn biases the frame matching; and (5) the reflectivity of Apollo 11 samples, Proc. Lunar Sci. Conf., 1st, geometric control is imperfect, resulting in poorer photometric-function corrections near the limbs and terminators. There also appears to be a small overall calibration problem for the low-resolution full-disk sequences (Lunmaps 16 and 23), perhaps due to the diffuse scattered light problem. If the diameter of the bright target approaches that of the camera, then the scattering can be reasonably modelled as a spatially uniform additive correction. For smaller targets, although the scattered light has a smaller magnitude, it is no longer uniform over the frame. The discrepancie seem to be significant for data acquired at sun angles exceeding -70 ø and/or viewing angles greater than -80 ø, and for images acquired at a spacecraft range of greater than about 106 km. These limits define the "line of data reliability" superposed on Figure 1 and Plates 1-3. The illumination and emission angle boundaries for each individual reprojection are overlain on Figures 9 and 10. In the mosaics, , Adams, J.B., and T.B McCord, Alteration of lunar optical properties: Age and composition effects, Science, 171, , Baldwin, R.B., The Measure of the Moon, 88 pp., University of Chicago Press, Chicago, II1., Basu, A., Recycled grains in lunar soils as an additional, necessary, regolith evolution parameter, Proc. Lunar Planet. Sci. Conf., 20th, , Batson, R.M., Digital cartography of the planets: New methods, its status, and its future, Photogramm. Eng. Remote Sens., 53, , Behrmann, C.J., G. Crozaz, R.J. Drozd, C.M. Hohenberg, C.E. Ralston, R.M. Walker, and D.E. Yuhas, Cosmic-ray exposure history of North Ray and South Ray material, Proc. Lunar Sci. Conf., 4th, , Belton, M.J.S., et al., Lunar impact basins and crustal

23 MCEWEN ET AL.: GALILEO OBSERVATIONS OF LUNAR CRATERS 17,229 heterogeneity: New western limb and farside data from Galileo, Science, 255, , 1992a. Belton, M.J.S., et al., Imaging of Venus from Galileo: Early results and camera performance, Adv. Space Res. 12, no. 9, 1992b. Breneman, H., and K. Klaasen, Solid-State Imaging Subsystem calibration report: Part 1, JPL Publ., D-5880, 214 pp., Chapman, C.R., and W.B. McKinnon, Cratering of planetary satellites, in Satellites, edited by J.A. Burns and M.S. Matthews, pp , University of Arizona Press, Tucson, Charette, M.P., L.A. Soderblom, J.B. Adams, M.J. Gaffey, and T.B. McCord, Age-color relationships in the lunar highlands, Proc. Lunar Sci. Conf., 7th, , Danielson, G.E., K.P. Klaasen, and J.L. Anderson, Acquisition and description of Mariner 10 television science data at Mercury, J. Geophys. Res., 80, , Davies, M.E., T.R. Colvin, and D.L. Meyer, A unified lunar control network: The nearside, J. Geophys. Res., 92, 14,177-14,184, Drozd, R.J., C.M. Hohenberg, C.J. Morgan, F.A. Podosek, and M.L. Wroge, Cosmic-ray exposure history at Taurus-Littrow, Proc. Lunar Sci. Conf., 8th, , Edwards, K.E., Geometric processing of digital images of the planets, Photogramm. Eng. Remote Sens., 53, , Fedoretz, V.A., Photographic photomerry of the lunar surface, Publ. Astron. Observ. Khar'kov State Univer., 2, , Gault, D.E., and J.A. Wedekind, Experimental studies of oblique impact, Proc. Lunar Planet. Sci. Conf., 9th, , Greeley, R., et al., Galileo imaging observations of lunar mafia and related deposits, J. Geophys. Res., this issue. Grieve, R.A.F., M.R. Dence, and P.B. Robertson, Cratering processes: As interpreted from the occurrence of impact melts, in Impact and Explosion Cratering, edited by D.J. Roddy, R.O. Pepin, and R.B. Merrill, pp , Pergamon, New York, Grieve, R.A.F., V.L. Sharpton, J.D. Rupert, and A.K. Goodacre, Detecting a periodic signal in the terrestrial crateting record, Proc. Lunar Planet. Sci. Conf, 18th, , Guinness, E.A., and R.E. Arvidson, On the constancy of the lunar cratering flux over the past 3.3 x 109 yr, Proc. Lunar Sci. Conf. 8th, suppl. 8, Geochim. Cosmochim. Acta, , Hansen, T.P., Guide to lunar orbiter photographs, NASA Spec. Publ. SP-242, 125 pp., Hapke, B.W., Bidirectional reflectance spectroscopy, III, Correction for macroscopic roughness, Icarus, 59, 41-59, Harmann, W.K., Lunar crater counts, VI, The young craters Tycho, Aristarchus, and Copernicus, Lunar Planet. Lab. Commun., 7, , Hartmann, W.K., Ancient lunar mega-regolith and subsurface structure, Icarus, 18, , Hawke, B.R., and J.F. Bell, Remote sensing studies of lunar dark- halo impact craters: Preliminary results and implications for early volcanism, Proc. Lunar Planet. Sci. Conf., 12th, , Hawke, B.R., and J.W. Head, Impact melt on lunar crater rims, in Impact and Explosion Craterins, edited by D.J. Roddy, R.O. Pepin, and R.B. Merrill, pp , Pergamon, New York, Hawke, B.R., D. MacLaskey, and T.B. McCord, Multispectral imaging of lunar crater deposits, LPI Contrib. 394, pp , Lunar and Planet. Inst., Houston, Tex., Hawke, B.R., et al., Remote sensing studies of the Orientale region of the Moon: A pre-galileo view, Geophys. Res. Lett., 18, , Head, J.W., and L. Wilson, Lunar mare volcanism: Stratigraphy, eruption conditions, and the evolution of secondary crusts, Geochim. Cosmochim. Acta, 56, , Head, J.W., et al., Lunar impact basins: New data for the western limb and far side (Orientale and South Pole-Aitken Basins) from the first Galileo flyby, J. Geophys. Res., this issue. Heiken, G., D. Vaniman, and B.M. French (Eds.), Lunar Sourcebook, 736 pp., Cambridge University Press, New York, Helfenstein, P., and J. Veverka, Photometric properties of lunar terrains derived from Hapke's equation, Icarus, 72, , Howard, K.A., and H.G. Wilshire, Flows of impact melt at lunar craters, J. Res. U.S. Geol. Surv. 3, , Johnson, J.R., S.M. Larson, and R.B. Singer, Remote sensing of potential lunar resources, 1, Nearside compositional properties, J. Geophys. Res., 96, 18,861-18,882, Johnson, T.V., D.L. Matson, R.J. Phillips, and R.S. Saunders, Vidicon spectral imaging: Color enhancement and digital maps, Proc. Lunar Sci. Conf, 6th, , Klaasen, K.P., M.C. Clary, and J.R. Janesick, Charge-coupled device television camera for NASA's Galileo mission to Jupiter, Opt. Eng., 23, , Kaula, W.M., G. Schubert, R.E. Lingenfelter, W.L. Sjogren, and W.R. Wollenhaupt, Apollo laser altimetry and inferences as to lunar structure, Proc. Lunar Sci. Conf. 5th, , Lipskii, Y.N. (Ed.), Atlas of the Farside of the Moon, Part 3, 239 pp., Moscow, Lucchitta, B.K., Crater clusters and light mantle at the Apollo 17 site: A result of secondary impact from Tycho, Icarus, 30, 80-96, Lucchitta, B.K., Geologic map of the north side of the Moon, U.S. Geol. Surv. Map , Lucey, P.G., Remote sensing of the Orientale basin of the Moon: Setting the stage for Galileo's first encounter with the Moon, Planetary Geosciences , NASA Spec. Publ. SP-508, 35-38, Lucey, P.G., B.R. Hawke, C.M. Pieters, J.W. Head, and T.B. McCord, A compositional study of the Aristarchus region of the Moon using near-infrared reflectance spectroscopy, Proc. Lunar Planet. Sci. Conf. 16th, Part 2, J. Geophys. Res., 91, suppl., D344-D354, Marti, K., B.D. Lightner, and T.W. Osborn, Krypton and xenon in some lunar samples and the age of North Ray Crater, Proc. Lunar Sci. Conf., 4th, , McCord, T.B., and T.V. Johnson, Relative reflectivity of selected areas of the lunar surface, J. Geophys. Res., 74, , McCord, T.B., C.M. Pieters, and M.A. Feierberg, Multi-spectral mapping of the lunar surface using ground-based telescopes, Icarus, 29, 1-34, McCord, T.B., et al., Moon: Near-infrared spectral reflectance, a first good look, J. Geophys. Res., 86, 10,883-10,892, McKay, D.S., and A. Basu, The production curve for agglutinates in planetary regoliths, Proc. Lunar Planet. Sci. Conf. 14th, Part 1, J. Geophys. Res. 88, suppl., B193-B 199, McKay, D.S., G.H. Heiken, R.M. Taylor, U.S. Clanton, D.A. Morrison, and G.H. Ladle, Apollo 14 soils: Size distribution and particle types, Proc. Lunar Sci. Conf., 3rd, , 1972.

24 17,230 MCEWEN ET AL.: GALILEO OBSERVATIONS OF LUNAR CRATERS McKay, D.S., R.M. Fruland, and G.H. Heiken, Grain size and evolution of lunar soils, Proc. Lunar Sci. Conf., 5th, , Moore, H.J., Geologic map of the Aristarchus region of the Moon, U.S. Geol. Surv. Map, 1-465, Moore, H.J., Missile impact craters (White Sands Missile Range, New Mexico) and applications to lunar research, U.S. Geol. Surv. Prof. Pap., 812-B, B1-B47, Moore, H.J., et al., Correlation of photogeology and remote sensing data along the Apollo 14 bistatic-radar ground track, I, A working compendium, Interagency Rep. Astrogeol. 75, U.S. Geol. Surv. Open File Rep., , 88 pp., Moore, H.J., J.M. Boyce, G.G. Schaber, and D.H. Scott, Lunar remote sensing and measurements, U.S. Geol. Surv. Prof. Pap B, 78 pp., Morris, R.V., R. Score, C. Dardano, and G. Heiken, Handbook of lunar soils, NASA Planet. Materials Branch, Publ. 67, Houston, 914 pp., Murchie, S.L., et al., Spectral properties of Orientale Basin materials from Galileo images (abstract), Lunar Planet. Sci.., XXIII, , Mutch, T.A., Geology of the Moon--A Stratigraphic View, 324 pp., Princeton University Press, Princeton, N.J., Nash, D.B., and J.E. Conel, Vitrification darkening of rock powders: Implications for optical properties of the lunar surface, Moon, 8, , Neukum, G., Meteoritenbombardement und datierung planerarer oberflachen, 186 pp., Habilitationsschrift, Ludwig Maximilianis Universitat, Munchen, Neukum, G., and B. K/Snig, Dating of individual lunar craters, Proc. Lunar Sci. Conf., 7th, , Neukum, G., B. K/ nig, and J. Arkani-Hamed, A study of lunar impact crater size-distributions, Moon, 12, , Offield, T.W., Geologic map of the Schiller quadrangle of the Moon, U.S. Geol. Survey Map, 1-691, O'Keefe, J.D., and T.J. Ahrens, Shock effects from a large impact on the moon, Proc. Lunar Sci. Conf., 6th, , Pieters, C.M., Composition of the lunar highland crust from nearinfrared spectroscopy, Rev. Geophys., 24, , Pieters, C.M., Compositional diversity and stratigraphy of the lunar crust derived from reflectance spectroscopy, in Remote Geochemical Analysis: Elemental and Mineralogical Composition, edited by C.M. Pieters and P. Englert, Cambridge University Press, New York, in press, Pieters, C.M., J.B. Adams, P. Mouginis-Mark, S.H. Zisk, J.W. Head, T.B. McCord, and M. Smith, The nature of crater ray.s: The Copernicus example, J. Geophys. Res., 90, 12,393-12,413, Pieters, C.M., et al., Crustal diversity of the Moon: Compositional analysis of Galileo solid state imaging data, J. Geophys. Res., this issue. Pohn, H.A., and T.W. Offield, Lunar crater morphology and relative age determination of lunar geologic units, I, Classification, U.S. Geol. Surv. Prof. Pap. 700-C, C153-C162, Quaide, W.L., and V.R. Oberbeck, Thickness determinations of the lunar surface layer from lunar impact craters, J. Geophys. Res., 73, , Robinson, M.S., B.R. Hawke, P.G. Lucey, and G.A. Smith, Mariner 10 multispectral images of the eastern limb and farside of the Moon, J. Geophys. Res., 97, 18,265-18,274, Saari, J.M., and R.W. Shorthill, Isothermal and isophotic atlas of the Moon, NASA Contr. Rep. CR-855, Scott, D.H., J.F. McCauley, and M.N. West, Geologic map of the west side of the Moon, U.S. Geol. Surv. Map, , Scott, R.F., and F.I. Robertson, Soil mechanics surface sampler, in Surveyor 7: A preliminary report, NASA Spec. Publ., SP- 173, , See, T.H., F. H6rz, and R.V. Morris, Apollo 16 impact-melt splashes: Petrography and major-element composition, Proc. Lunar Planet. Sci. Conf. 16th, Part 1, J. Geophys. Res., 91, suppl., E3-E20, Shoemaker, E.M., Interpretation of lunar craters, in Physics and Astronomy of the Moon, edited by Z. Kopal, pp , Academic San Diego, Calif., Shoemaker, E.M., and R.J. Hackmann, Stratigraphic basis for a lunar time scale, in The Moon, edited by Z. Kopal and Z.K. Mikhailov, pp , Academic, San Diego, Calif., , Shoemaker, E.M., and E.C. Morris, Physical characteristics of the lunar regolith determined from Surveyor television observations, Radio Sci., 5, , Shoemaker, E.M., and R. Wolfe, Mass extinctions, crater ages, and comet showers, in The Galaxy and the Solar System, edited by R. Smoluchowski et al., pp , University of Arizona Press, Tucson, Shoemaker, E.M., R.M. Batson, H.E. Holt, E.C. Morris, J.J. Rennilson, and E.A. Whitaker, Television observations from Surveyor 3, J. Geophys. Res., 73, , Smrekar, S., and C.M. Pieters, Near-infrared spectroscopy of probable impact melt from three large lunar highland craters, Icarus, 63, , Soderblom, L.A., A model for small-impact erosion applied to the lunar surface, J. Geophys. Res., 75, , Soderblom, L.A., and J.M. Boyce, Relative ages of some nearside and far-side terra plains based on Apollo 16 metric photography, part A, Photogeology, chap. 29, Apollo 16 Preliminary Science Report, NASA Spec. Publ., SP-315, 29-2 to 29-6, Soderblom, L.A., and L.A. Lebofsky, Technique for rapid determination of relative ages of lunar areas from orbital photography, J. Geophys. Res., 77, , Spudis, P.D., B.R. Hawke, and P. Lucey, Composition of Orientale basin deposits and implications for the lunar basinforming process, Proc. Lunar Planet. Sci. Conf. 15th, Part I, J. Geophys. Res., 89, suppl., C197-C210, Spudis, P.D., B.R. Hawke, and P. Lucey, Materials and formation of the Imbrium Basin, Proc. Lunar Planet. Sci. Conf., 18th, , Stuart-Alexander, D.E., Geologic map of the central farside of the Moon, U.S. Geol. Surv. Map, , Trask, N.J., Geologic comparison of mare materials in the lunar equatorial belt, including Apollo 11 and Apollo 12 landing sites, U.S. Geol. Surv., Prof. Pap. 750-D, D138-D144, U.S. Geological Survey, Map showing relief and surface markings on the lunar far side, Map A, scale 1:5,000,000, Flagstaff, Ariz., U.S. Geological Survey, Map showing relief and surface markings on the lunar polar regions, Map B, scale 1:5,000,000, Flagstaff, Ariz., U.S. Geological Survey, Map showing relief and surface markings on the lunar near side, Map B, scale 1:5,000,000, Flagstaff, Ariz., 1992.