Reexamination of Early Lunar Chronology With GRAIL Data:Terranes, Basins, and Impact FluxesAlexander J. Evans1,2 , Jeffrey C. Andrews-Hanna1,2 , James W. Head III3 ,Jason M. Soderblom4 , Sean C. Solomon5,6 , and Maria T. Zuber4

1Lunar and Planetary Laboratory, University of Arizona, Tucson, AZ, USA, 2Southwest Research Institute, Boulder, CO, USA,3Department of Earth, Environmental and Planetary Sciences, Brown University, Providence, RI, USA, 4Department of Earth,Atmospheric and Planetary Sciences, Massachusetts Institute of Technology, Cambridge, MA, USA, 5Lamont-Doherty EarthObservatory, Columbia University, Palisades, NY, USA, 6Department of Terrestrial Magnetism, Carnegie Institution ofWashington, Washington, DC, USA

Abstract Flooding of the lunar surface by ancient mare basalts has rendered uncertain the ages of lunargeochemical terranes and several impact basins. Here we combine craters having recognizable surfaceexpressions with craters identified only by their gravitational signatures in Gravity Recovery and InteriorLaboratory data to reassess the chronological sequence of lunar impact basins and the ages of major lunargeochemical terranes. Our results indicate that although volcanically flooded regions are deficient incraters with diameters greater than 20 km by more than 50% relative to unflooded regions, craters withdiameters greater than 90 km can be readily recognized either by topography or by gravity anomaly. Onthe basis of the areal density of craters with diameters greater than 90 km we conclude that (1) theSerenitatis basin could be as young as the Imbrium basin; (2) the areal density of craters within theProcellarum KREEP Terrane is significantly lower than that for the South Pole-Aitken basin and theFeldspathic Highlands Terrane; (3) if the youngest age of final crystallization of the lunar magma ocean isadopted as a lower bound on the age of the Procellarum KREEP Terrane, a minimum age of approximately4.3 Ga is inferred for ~40% of lunar impact basins, including South Pole-Aitken; and (4) the flux ofimpactors capable of forming craters with diameters of at least 90 km decreased substantially through theNectarian and Imbrian periods.

1. Introduction

Establishing the absolute and relative chronology of ancient lunar events is of fundamental importance toour understanding of early Solar System history and the evolution of rocky planetary bodies. In this endeavor,the Moon has a unique role, as it is the only planetary body from which both absolute and relative ages canbe calibrated, from the radiometric dating of returned samples and the areal density of impact craters,respectively (Hiesinger et al., 2011; Le Feuvre & Wieczorek, 2011; Marchi et al., 2009; Neukum et al., 1975,2001). In contrast to the relatively young and heavily modified surfaces of many other planetary bodies inthe Solar System, a majority of the lunar surface has been well preserved since antiquity and thus containsthe most comprehensive record of early impact cratering presently known to exist.

For those few lunar impact basins from which samples have been returned from impact-associated deposits,radiometric age dating has provided constraints on their absolute ages (Wilhelms, 1987). Yet with the excep-tion of the Imbrium basin and perhaps the younger Orientale basin, substantial uncertainty remains in theabsolute ages of lunar basins, as exemplified by the prominent South Pole-Aitken (SPA) and Nectaris basins(Fischer-Gödde & Becker, 2011; Korotev et al., 2002; Lawrence et al., 2003; Maurer et al., 1978; Papike et al.,1998; Warren, 2003; Wilhelms, 1987). Although SPA is generally accepted to be the oldest as well as the lar-gest recognizable basin on the Moon, there is only a lower bound on its absolute age of ~3.92 Ga (Wilhelms,1987). This lower bound is based on the canonical age inferred for the stratigraphically younger Nectarisimpact. Yet this frequently cited age of 3.92 Ga for the Nectaris impact is likely to be incorrect, as the samplesused to date Nectaris radiometrically are probably ejecta from a younger impact event (Fischer-Gödde &Becker, 2011; Lawrence et al., 2003; Warren, 2003). Other lunar samples identified as associated withNectaris have argon-argon ages as old as 4.26 Ga (Warren, 2003) and have been interpreted to establish amuch older age for the basin, although the correlation between these samples and the Nectaris basin is alsoin question, as a recent analysis of one of these samples indicates that the abundances of radiogenic

EVANS ET AL. 1596

Journal of Geophysical Research: Planets

RESEARCH ARTICLE10.1029/2017JE005421

Key Points:• The record of lunar craters withdiameters greater than 90 km is wellpreserved by a combination of gravityanomaly and topography

• The South Pole-Aitken basin andProcellarum KREEP Terrane formedmore than 4.3 billion years ago

• Approximately 40% of lunar basinsformed prior to 4.3 billion years ago

Supporting Information:• Supporting Information S1

Correspondence to:A. J. Evans,alex@lpl.arizona.edu

Citation:Evans, A. J., Andrews-Hanna, J. C., Head,J. W., III, Soderblom, J. M., Solomon, S. C.,& Zuber, M. T. (2018). Reexamination ofearly lunar chronology with GRAIL data:Terranes, basins, and impact fluxes.Journal of Geophysical Research: Planets,123, 1596–1617. https://doi.org/10.1029/2017JE005421

Received 15 AUG 2017Accepted 24 MAY 2018Accepted article online 31 MAY 2018Published online 5 JUL 2018

©2018. American Geophysical Union.All Rights Reserved.

elements are too high to be consistent with the nominal composition observed within that basin (Normanet al., 2016). As the Nectaris basin provides the principal chronologic anchor for the transition from thepre-Nectarian period (4.53–3.92 Ga) to the Nectarian period (3.92–3.85 Ga), the uncertainty in its age hasmajor implications for the durations of lunar periods and epochs. Furthermore, a substantial change in theabsolute age of Nectaris would have a major effect on interpretations of the early lunar impact rate and con-straints on an early impact cataclysm for the inner Solar System (e.g., Bottke & Norman, 2017).

In addition to the uncertain absolute ages of many lunar basins, there likely exists substantial error in relativesurface ages as determined from crater densities. Although one might generally presume that an older sur-face has a greater concentration of craters than a younger one, the establishment of relative ages from craterdensities is complicated by the variability introduced by secondary craters formed by ballistic ejecta fromnearby impacts (McEwen & Bierhaus, 2006; Xiao & Strom, 2012), regions with saturated crater densities(Chapman & Jones, 1977), personal biases that may arise in the classification of craters (Robbins et al.,2014), time-variable impact rates (Le Feuvre & Wieczorek, 2011; Neukum et al., 2001), and, most significantly,nonuniform resurfacing (Phillips et al., 1992; van der Bogert et al., 2017).

Despite rigorous analyses to reduce the effects of some of these potential sources of error on the Moon (e.g.,Fassett et al., 2012; McEwen & Bierhaus, 2006; Robbins, 2014), uncertainty remains as a result of impact ratesthat are uncalibrated to reliable absolute ages for much of lunar history (1–3 Ga and greater than 4 Ga) andpartial or complete resetting of crater densities by multiple volcanic resurfacing events that primarilyoccurred on the lunar nearside and within basin impact structures (Hiesinger et al., 2011; Le Feuvre &Wieczorek, 2011; Marchi et al., 2009; Neukum et al., 2001; van der Bogert et al., 2017). The uncalibrated lunarcratering rates prior to ~4 Ga introduce substantial ambiguity in the interpretation of crater age assessmentsfor the most ancient lunar surfaces (e.g., Le Feuvre & Wieczorek, 2011; Morbidelli et al., 2012; Neukum et al.,1975; Neukum & Ivanov, 1994). Many ancient lunar surfaces have similar crater densities, a result that mayindicate the surfaces formed contemporaneously. However, similar crater densities may also result from sur-faces that have reached a state of crater saturation (e.g., Gault, 1970; Richardson, 2009) or from a time-variable cratering rate, for example, by which different surfaces formed at different times within a sustainedperiod of relatively lower cratering rates prior to a later period of greater impact bombardment. Attempts toassign relative ages to the major impact basins through the application of traditional crater size-frequencyanalyses are often frustrated by the extensive deposits of basaltic plains, or maria, that preferentially floodedand presently obscure the primary surfaces of major impact basins (Budney & Lucey, 1998; Evans et al., 2016;Gong et al., 2016; Head & Wilson, 1992, 2017; Hiesinger et al., 2000, 2010, 2011; Hörz, 1978; Whitten & Head,2015a, 2015b; Zhong et al., 2000). To estimate the relative ages of these mare-flooded basins, previous work-ers either used a patchwork of unflooded surfaces of small area (e.g., Fassett et al., 2012) or made adjustmentsfor the size-frequency distributions (SFDs) of mare-covered regions (Hartmann & Wood, 1971), but bothof those methods inject uncertainty and potential sample bias, especially for heavily flooded basinssuch as Serenitatis (Fassett et al., 2012; Wilhelms, 1987). More comprehensive treatments have augmentedthe traditional crater size-frequency analyses with stratigraphic inferences to establish the relative agesand a chronologic sequence of lunar basins (e.g., Fassett et al., 2012; Wilhelms, 1987), but the uncertain craterdensity of the premare nearside surface nonetheless remains an obstacle to establishing a reliablerelative chronology.

Of particular interest is the age of Serenitatis basin, a topic that has long been debated in the scientific litera-ture (e.g., Fassett et al., 2012; Hartmann & Wood, 1971; Head, 1975a; Spudis et al., 2011; Stuart-Alexander &Howard, 1970; Wilhelms, 1987). The age of Serenitatis has important implications for the bombardment his-tory of the Moon, since a late-Nectarian or early-Imbrian age close to that of Imbrium could support thenotional idea of a major lunar cataclysm at ~3.9 Ga, whereas a pre-Nectarian age for Serenitatis may be con-sistent with a steadily declining bombardment (e.g., Spudis et al., 2011). Geologic evidence, including radio-metric dating of the Apollo 17 Taurus-Littrow samples posited to have been derived from units emplacedduring the Serenitatis impact event (Jessberger et al., 1978), indicates a late-Nectarian age (Wilhelms,1987). However, on the basis of Lunar Reconnaissance Orbiter Camera (LROC) images (see Chin et al.,2007), Spudis et al. (2011) suggested that these samples may not be from units associated with theSerenitatis impact event. The highly degraded appearance of Serenitatis has also led several investigatorsto suggest that the basin may be of pre-Nectarian age (Hartmann & Wood, 1971; Head, 1975a; Wilhelms,1987). Instead, the apparently degraded state of Serenitatis may be at least partly the result of its close

10.1029/2017JE005421Journal of Geophysical Research: Planets

EVANS ET AL. 1597

proximity to the younger and larger Imbrium basin, from a combination of ejecta (Fassett et al., 2012; Head,1975b) and seismic shaking (Kreslavsky & Head, 2012) from the Imbrium impact. Although crater density esti-mates for the Serenitatis basin have yet to yield a reliable and consistent determination of its relative age, inpart, because of the extensive mare flooding (Hartmann & Wood, 1971; Wilhelms, 1987), two recent analysessuggest that Serenitatis has a crater density consistent with the formation of the basin during the pre-Nectarian epoch (Fassett et al., 2012; Spudis et al., 2011).

In addition to constraining the absolute ages of major impact basins, lunar samples also constrain the timingof cooling and fractional crystallization of the early lunar magma ocean. Samples returned from the Moonindicate that much of its crust formed as a direct consequence of magma ocean crystallization, which yieldedat least one of the three major geochemical provinces now observed on the surface (Figures 1a and 1b): theFeldspathic Highlands Terrane (FHT; Jolliff et al., 2000). The second province, the SPA Terrane (Jolliff et al.,2000), is associated with the early formation of the SPA basin, and the third province, the ProcellarumKREEP Terrane (PKT; Jolliff et al., 2000), does not have a clearly constrained origin or age.

The FHT occupies more than 60% of the lunar surface area and is themost ancient terrane on theMoon (Jolliffet al., 2000), with an estimated age of ~4.4–4.5 Ga inferred from the radiometric ages of lunar anorthositesamples (Elkins-Tanton et al., 2011; see also Barboni et al., 2017). As shown in Figure 1, the FHT is character-ized by low FeO and Th abundances and elevated topography (Jolliff et al., 2000; Wilhelms, 1987). The FHT isgenerally held to be the surface manifestation of the primary feldspathic crust formed by the crystallizationand buoyant rise of plagioclase feldspar during cooling of the lunar magma ocean (Elkins-Tanton et al., 2011;Warren, 1985).

Figure 1. Maps of lunar terranes and basins: (a) FeO (weight %; Lawrence et al., 2002), (b) surface thorium concentration(ppm; Lawrence et al., 2003), (c) lunar craters with D ≥ 90 km on a shaded-relief map of topography (Barker et al., 2016;Smith et al., 2010), and (d) QCMAs (outlined in magenta) and surface mare deposits (outlined in dark blue; Nelson et al.,2014) shown on a lunar morphological base map (Speyerer et al., 2011). The Feldspathic Highlands Terrane (FHT), SouthPole-Aitken (SPA) basin, and Procellarum KREEP Terrane (PKT) are outlined in (a–c) and labeled in (a) and (b). The basinsfrom Table 1 are outlined in (a) and (b) and labeled in (b). In (c), surface craters (black) cataloged by Head et al. (2010),QCMAs (magenta and black) identified by Neumann et al. (2015), and QCMAs (magenta) identified by Evans et al. (2016) areoutlined. In (d), QCMAs (magenta and white) identified by Neumann et al. (2015) are outlined. All maps are in Eckert-IVprojections. QCMA = quasi-circular mass anomaly.

10.1029/2017JE005421Journal of Geophysical Research: Planets

EVANS ET AL. 1598

Unlike the FHT and the SPA Terrane, for the PKT the origin and age are less clear. The PKT is a geochemicalprovince, often delineated by the 3.5-ppm contour of the surface concentration of Th, characterized by aregional enrichment of KREEP (material with an enhanced concentration of K, rare earth elements, P, andother incompatible elements), low-lying topography, high surface abundances of FeO and Th, and extensivedeposits of lunar maria (Hiesinger et al., 2003; Jolliff et al., 2000; Smith et al., 2010; Wieczorek & Phillips, 2000).Although KREEP is postulated to have crystallized beneath the feldspathic crust as the residuum of the magmaocean (i.e., urKREEP), the cause of its predominant surface expression within the PKT remains undetermined(Elkins-Tanton et al., 2011; Hess & Parmentier, 2001; Jolliff et al., 2000; Warren, 1985; Wilhelms, 1987).Returned lunar samples yield a preponderance of crystallization ages for urKREEP material between 4.48 and4.34 Ga (see review by Borg et al., 2015), with a recent analysis yielding an age of 4.389 ± 0.045 Ga (Borg et al.,2015). These ages likely represent either crystallization of the residuum from the lunar magma ocean or crystal-lization from a subsequent widespreadmelting event (Borg et al., 2015). To be conservative, in our analyses thatfollow we will assume a young age of 4.3 Ga for the crystallization of urKREEP.

An ancient impact basin in the region of Oceanus Procellarum has often been invoked to account for thelocalization of KREEP on the lunar nearside (Cadogan, 1974; Whitaker, 1981; Wilhelms, 1987) and the presencethere of low-Ca pyroxene (Nakamura et al., 2012). The gravitational and physiographic characteristics of theregion, however, are inconsistent with an ancient impact structure (Andrews-Hanna et al., 2014).Nonetheless, the absence of these characteristics does not preclude the possibility that a Procellarumbasin-forming impact could have occurred sufficiently early in lunar history that extreme heating and/orthe presence of a subsurface magma ocean markedly diminished the gravitational and physiographicsignature of the basin and its rim structure. Unfortunately, the extensive mare flooding of this region has,to date, prevented a reliable crater density estimate for the premare PKT, thereby hindering a comprehensiveevaluation of the circumstances surrounding its formation.

There is general consensus on the absolute age of the Imbrium basin and the relative ages of many otherbasins. Nonetheless, there remain several unanswered questions regarding lunar terrane and basin chronol-ogy, in particular the age of Serenitatis, its implications for the flux and provenance of early lunar impactors,and the age of the PKT. In this investigation, we use both craters with a recognizable surface expression (Headet al., 2010), here termed surface craters, and craters inferred to be fully buried by volcanic or impact depositsand identified from quasi-circular mass anomalies (QCMAs; Evans et al., 2016; Neumann et al., 2015) pre-served in the lunar gravitational field and revealed by analyses of the gravity field mapped by the GravityRecovery and Interior Laboratory (GRAIL) mission (Konopliv et al. 2013; Lemoine et al., 2013; Zuber, Smith,Lehman, et al., 2013; Zuber, Smith, Watkins, et al., 2013). On the basis of the distribution of both types of fea-tures, we examine the ages of formation of major lunar terranes, the chronological sequence of impact basinformation, and changes in the lunar impactor flux prior to ~3.8 Ga.

2. Areal Distribution of Craters and QCMAs

From the GRAIL-derived lunar gravity field, Evans et al. (2016) identified 104 QCMAs within and near the lunarnearside maria having diameters between 26 and 300 km (Figure 1d). Moreover, they proposed that mostQCMAs are impact craters that lack surface topographic expression because of burial by mare basalt or mate-rial ejected by younger impact events. Several lines of evidence support this inference. The most compellingargument is that the most widespread, quasi-circular features on the Moon are impact craters. Thus, by infer-ence, the QCMAs are also likely to be impact craters. Evans et al. (2016) also showed that the QCMAs withdiameters less than 80 km follow the same trend of Bouguer anomaly versus diameter as that for unfilled sur-face craters. For diameters greater than 80 km, the QCMAs generally form two groups, one with large positiveBouguer anomalies consistent with infilling of impact craters by high-density mare material, and one withsmall negative Bouguer anomalies consistent with impacts into early mare deposits that have a greateramount of lower-density feldspathic material interior to the feature than their surroundings. The latter grouphas been suggested to be the result of partial infilling of the crater interior with feldspathic material ejectedby nearby impacts or preferential uplift of feldspathic material within the crater interior during the crater-forming process. Both large positive and small negative Bouguer anomalies are observed for partially filledcraters on the lunar nearside hemisphere (Zuber, Smith, Watkins, et al., 2013). This observation providesfurther evidence that QCMAs are buried craters. Furthermore, large craters of volcanic origin (i.e., calderas)

10.1029/2017JE005421Journal of Geophysical Research: Planets

EVANS ET AL. 1599

are not favored to form under lunar ascent and eruption conditions (Head & Wilson, 1991), and none in thisdiameter range have been observed (Head & Wilson, 2017). We assume hereafter that all QCMAs representburied craters. We recognize that the QCMA detection techniques employed by Evans et al. (2016) withthe lunar gravitational field are susceptible to false identifications by up to ~10%, but such a percentage offalsely identified buried craters would serve only to reinforce the conclusions of this study (see section 4).

In addition to the QCMAs of Evans et al. (2016), we incorporate 10 QCMAs identified by Neumann et al. (2015)that are not present in either the QCMA data set of Evans et al. (2016) or the crater catalog of Head et al.(2010). Neumann et al. (2015) analyzed the lunar gravitational data and identified large, positive quasi-circular features in the Bouguer and free-air anomaly fields for signatures consistent with the central positiveBouguer anomalies of basin-size impact structures. Scaling relations developed between the diameter of thecentral positive Bouguer anomaly and the diameter of the main rim for well-preserved basins were used toestimate the rim diameter of these gravitationally identified basin structures (Neumann et al., 2015).Because of the wide areal distribution and the small number of QCMAs included from the study ofNeumann et al. (2015), our conclusions are not fundamentally altered by their incorporation. However, wechose to include these QCMAs so that we may analyze the relative ages of the proposed Fecunditatis andAustrale North basins.

The contribution of QCMAs to the crater record may be assessed with a cumulative SFD N(D), where N is thenumber of craters of diameter D (in kilometers) or greater per unit area (106 km2). In order to assess the relia-bility of the QCMAs in determining relative ages, we tested if the combined population of surface craters andQCMAs follows the expected SFD. To do so, we compared the normalized cumulative SFD of surface cratersplus QCMAs for mare units to the normalized cumulative SFD of surface craters on mare and nonmare units(Figure 2a). For this comparison, we used the mare and nonmare unit boundaries constructed by Nelson et al.(2014) from LROC data (see Figure 1d). The normalization of cumulative SFDs was accomplished by settingthe value to unity at D = 20 km. The cumulative SFD of surface craters on nonmare units was taken as repre-sentative of a primary unmodified crater population. As illustrated in Figure 2a, the QCMAs have a cumulativeSFD that is skewed toward larger craters; nearly half of the QCMAs are greater than 90 km in diameter,whereas craters with diameters greater than 90 km comprise less than ~20% of the population of surface cra-ters in mare and nonmare units. A noteworthy observation from Figure 2a is that the normalized cumulativeSFD of surface craters plus QCMAs for mare units provides a poorer fit to the normalized cumulative SFD ofsurface craters on nonmare units than the normalized cumulative SFD of only surface craters on mare units.As discussed below, this effect is likely the result of incomplete recovery of a population of buried craters onmare units that is skewed toward smaller diameters.

To compare directly the mare and nonmare regions across a range of diameters, we determined ratios of theincremental SFD of impact craters in nonmare regions to that for mare regions with and without the QCMAsincluded. The incremental SFD is the number of craters within a given bin of diameters per standard area(106 km2). For our analyses, we took bins at 20-km-diameter intervals centered on a given diameter D, andeach bin includes craters of diameter D ± 20 km (Figure 2b). For each bin, the crater density was estimatedfrom areal maps constructed with a 500-km-radius moving window average (see example in Figures 2cand 2d), and the errors were calculated from the weighted standard error on the mean (Cochran, 1977;Gatz & Smith, 1994). To minimize the influence of large regions resurfaced by impact basins, we excludedthose areas interior to basins having diameters greater than 650 km. However, our analyses are insensitiveto both the radius selected for the moving window and the basin diameter cutoff (see Figure S1 of thesupporting information). The ratios shown in Figure 2b are minimally affected by the exclusion of impactstructures with diameters as small as 250 km.

If the incremental SFD of the combined set of surface craters and QCMAs for mare regions represents a popu-lation of impact structures derived from the same population of projectiles as those that formed the surfacecraters preserved in the nonmare regions, then the data in Figure 2b would show a horizontal line. Moreover,if the ages of the surface in nonmare regions and the premare surface in mare regions were similar, a value ofunity would be expected for the ratio of incremental SFDs.

The ratio of incremental SFD in nonmare regions to that in mare regions with QCMAs included (in Figure 2b)shows a negatively sloping line with increasing diameter and values greater than unity at small diameters(D < 90 km), suggestive of a population of smaller buried craters (D < 90 km) that were not recovered as

10.1029/2017JE005421Journal of Geophysical Research: Planets

EVANS ET AL. 1600

resolved QCMAs. Given the demonstrated ability to recover QCMAs smaller than 30 km in diameter and withamplitudes of less than 2 mGal (Evans et al., 2016), the deficit of smaller buried craters (D < 90 km) in thecrater population identified by Evans et al. (2016) on the lunar nearside is unlikely to be purely an effect ofthe methodology applied to identify buried craters.

At larger diameters, D ≥ 90 km (the smallest craters included in the 110-km-diameter bin), the ratio of incre-mental SFDs follows a horizontal line with a value near unity, indicating that for such features, there is noapparent deficit of craters in mare regions compared with nonmare regions when QCMAs are included.Thus, the ratio of incremental SFDs with QCMAs included indicates that the record of craters having dia-meters greater than 90 km on the Moon is well preserved by the combination of gravity and topographyinformation. In contrast, more than 60% of the expected total of premare craters between 20 and 90 km indiameter in mare regions are not recovered from gravity or topography information.

The ratio of the incremental SFD in nonmare regions to that in mare regions determined without QCMAs var-ies between values of 1 and 3. Variations in the observed ratio are at least partially due to the variable ages ofthe nearside maria (e.g., Hiesinger et al., 2011) that have been binned together in this analysis as well as

Figure 2. (a) Normalized cumulative size-frequency distributions (SFDs; D ≥ 20 km) for lunar crater populations, defined toequal unity at D = 20 km. Crater populations shown include nonmare regions (red), QCMAs or buried craters (gray), mareregions without buried craters included (black), and mare regions with buried craters included (blue). (b) Ratio of incre-mental SFD for craters in nonmare regions to that in mare regions, shown with (blue) and without (black) QCMAs.Incremental SFDs are determined at intervals of 20 km in diameter and for 40-km-diameter bin sizes. Errors follow from theweighted standard error of the mean. The red line denotes a 1:1 ratio. (c) N(90) and (d) N(20), each averaged over acircular window of 500-km radius, determined from QCMAs (Evans et al., 2016; Neumann et al., 2015) as well as cratersidentified from Lunar Orbiter Laser Altimeter (LOLA) topographic data (Head et al., 2010), are shown in Eckert-IV projec-tions. For (c) and (d), panels on the left include only craters identified from LOLA topographic data; panels on the rightinclude QCMAs as well as craters identified from LOLA topographic data. The basins listed in Table 1 are outlined andlabeled. QCMA = quasi-circular mass anomaly.

10.1029/2017JE005421Journal of Geophysical Research: Planets

EVANS ET AL. 1601

regional variability in the number of craters that are embayed by mare units. For the largest craters, the ratioof incremental SFDs without QCMAs approaches unity, indicating that the identification of the largest impactstructures has been minimally influenced by mare volcanism.

Thus, on the basis of ratios of incremental SFDs, we can make the following inferences about the comparisonof relative ages between nonmare areas and the surfaces underlying mare material in volcanically floodedbasins and terranes: (1) as has been previously established, the substantial burial of craters by mare volcanismin mare areas precludes reliable comparisons of ages of the underlying terranes from crater SFDs using sur-face craters alone, and (2) sets of craters having diameters greater than 90 km and constructed with the addi-tion of information on buried craters provide the best available basis for making reliable age comparisons forsuch areas.

Crater areal density maps constructed with a 500-km-radius moving window average (shown in Figures 2cand 2d) also illustrate the affect that the inclusion of QCMAs has on nearside mare regions. With QCMAsincluded, most nearside lunar mare regions are indistinct relative to their surroundings in the map of N(90)(Figure 2c), although the lower areal density values within several major impact basins (e.g., Orientale,Crisium, Imbrium) can still be observed. In contrast, without QCMAs, the lunar nearside mare region is clearlydistinct from the surrounding nonmare regions in the map of N(20) (Figure 2d).

Given that mare volcanism had little apparent effect on the incremental and cumulative SFDs of the com-bined set of surface craters and QCMAs for D ≥ 90 km in mare regions, we chose this crater diameter cutoffto assess the relative ages of lunar geochemical terranes and major impact basins.

3. Age Determination With Buried Craters Included

The age of an exposed surface determined via crater size-frequency analyses is generally referred to as thecrater retention age (Hartmann, 1966). For size-frequency analyses that include QCMAs (i.e., interpreted asburied craters), applying the term crater retention age to the derived age estimates would be inappropriate.Instead, we use the term crustal crater retention age to reflect those ages determined from crater size-frequency analyses that include QCMAs. Such a designation more accurately reflects the length of time thatrecognizable gravitational and topographic signatures of impact features are retained within the crust. Forregions where the original surface is obscured, specifically the PKT and mare-flooded basins, we employthe crustal crater retention age to characterize the relative ages of these regions.

The rationale for using densities of recognized impact craters to determine absolute and relative (craterretention) ages of exposed surfaces is well established (e.g., Neukum et al., 1975) and requires that the follow-ing criteria be met:

1. The craters included are predominantly a result of primary impacts rather than ballistic impact of ejecta(secondary impact craters);

2. The diameter ranges of the craters used in the size-frequency analyses are minimally affected by cratersaturation;

3. The crater population was reset at the time of formation for the exposed surfaces, and later craters thatformed on the exposed surfaces were preserved; and

4. The impact rate, at a given time, did not substantially vary across the planetary surface.

Below, we evaluate the applicability of the above criteria to the crustal crater retention age and highlight thevalidity of each of the above criteria with respect to using QCMAs as well as surface craters to determinerelative ages.

3.1. Secondary Craters

Secondary craters, or secondaries, are the class of impact craters formed by material ejected from impacts ofextra-planetary projectiles and have been unambiguously identified on the lunar surface for D ≤ 30 km(Wilhelms, 1976). On the Moon, secondaries are estimated to account for a substantial fraction of the impactcraters at D < 20 km, but they are found to be significantly less abundant for diameters between 20 and30 km (Head et al., 2010; Wilhelms et al., 1978). The consistency of previous N(20) crater age assessments withstratigraphic relationships and the absence of statistically significant changes in N(20) for annular zonesaround large basins reinforce the inference regarding the negligible influence of secondaries at D ≥ 20 km

10.1029/2017JE005421Journal of Geophysical Research: Planets

EVANS ET AL. 1602

(Fassett et al., 2012; Head et al., 2010). As the smallest craters and QCMAs that have been used for crater ageassessments in this study are 90 km in diameter, well above the size threshold at which the influence of sec-ondaries is significant (Guo et al., 2017), we can confidently neglect the influence of secondaries for our craterage assessments.

3.2. Crater Saturation

Crater saturation equilibrium occurs when the crater density of a surface remains unchanged in time, becausethe formation of each new crater, on average, leads to the removal of a preexisting crater of similar diameter(Gault, 1970). For the ancient lunar highlands and many pre-Nectarian basins, N(20) values are within therange of saturation equilibrium values (Fassett et al., 2012; Head et al., 2010; van der Bogert et al., 2017).The claim of saturation equilibrium at D ≥ 20 km on the Moon has been previously invoked to account forthe anomalously low crater densities for the SPA basin, which are not consistent with the generally acceptedstratigraphic relationship indicating that SPA is the most ancient recognizable basin on the Moon (Wilhelms,1987; see also Fassett et al., 2012). If saturation equilibrium has been reached for pre-Nectarian surfaces onthe Moon, it may be impossible to distinguish the chronologic sequence of the earliest impact events onthe basis of crater density alone. In sections 4 and 5, we consider further the implications of saturation equili-brium for the interpretation of relative ages from crater densities.

3.3. Areal Distribution of Cratering

The ability to establish relative ages between surfaces on the basis of crater densities also requires a suffi-ciently random areal distribution of impacts across the surfaces considered so that variations in crater densi-ties between surfaces can be primarily attributed to their age differences. It has also been suggested thatvariations in crater densities have been influenced by spatial variations in impact rates and nonuniform crus-tal and upper mantle temperatures across the Moon (Le Feuvre & Wieczorek, 2011; Miljković et al., 2013). Thesimilar N(90) values on the nearside and farside (Figure 2c), however, suggest a lack of significant differencesin the impact crater population between these two hemispheres for the diameter range that we treat here.For this reason, we assume that spatial variations in impact rates do not significantly affect our results. Forbasins with D ≥ 200 km, hydrocode modeling has demonstrated that crust and upper mantle temperatures,within expected ranges, can exert an influence on the final basin diameter for impacts of a given energy(Miljković et al., 2013). Although temperature-induced variations in basin size may affect the precise craterSFD for D ≥ 200 km, relative ages determined from N(90) values are dominated by craters that are well belowthe diameter threshold (200 km) at which this effect has been suggested and therefore are not affected by achange in final basin size at larger diameters. Thus, following previous workers and in the absence of satura-tion effects, we assume that variations in cumulative crater densities between regions primarily reflect differ-ences in age.

3.4. State of Preservation of the Surface and Crust

Accurately constraining the age of an exposed surface from crater size-frequency analyses relies on theassumption that the crater record was effectively reset at the time of the last resurfacing event. This assump-tion becomes more complex in the type of analyses that we are presenting, as we consider both surface cra-ters and impact features buried beneath the exposed surface. Of specific concern is whether the gravitationalsignature, or more generally the crustal signature, of an impact crater is erased by a superposed basin-forming impact.

The absence of obvious gravitational and crustal signatures of preexisting simple and complex craters withinand along the Orientale basin rim (defined by the Cordillera ring; Wilhelms, 1987) suggests that the basin-forming process effectively erases such signatures. As Orientale is the best-preserved and youngest multiringimpact basin to have formed on the Moon, its preimpact site was likely to have been more densely crateredthan the preimpact sites of other, older impact basins. Accordingly, the Orientale impact would have had thegreatest probability among all basins of preserving preexisting simple and complex craters. We can, there-fore, infer with high likelihood that the other major impact basins did not preserve gravitational signaturesof preexisting simple or complex craters to a significant extent.

As the diameters of QCMAs from Evans et al. (2016) used in this study (less than 200 km) overlap the sizerange of the central positive Bouguer anomalies for larger basins (Neumann et al., 2015; see also Bakeret al., 2017), we must also consider the possibility that a fraction of the positive QCMAs could be

10.1029/2017JE005421Journal of Geophysical Research: Planets

EVANS ET AL. 1603

central mantle uplifts associated with buried peak-ring basins thatpredate the basin of interest, as has been suggested by Sood et al.(2017) for one of the QCMAs identified by Evans et al. (2016).Although central positive Bouguer anomalies of preexisting basinshave not been conclusively identified interior to basin excavationcavities, central positive Bouguer anomalies associated with preexist-ing basins are observed interior to the main rim of superposedbasins, as exemplified by the Asperitatis, Orientale SW, andSerenitatis North basins, which are contained at least partially withinthe outer rings of the Nectaris, Orientale, and Serenitatis basins,respectively (Neumann et al., 2015).

Because the signatures of central positive Bouguer anomalies in olderbasins are not necessarily erased outside the excavation cavities ofyounger superposed basins, we cannot reject the possibility a priorithat some QCMAs interior to basins represent preexisting structures.The diameters of the central positive Bouguer anomalies associatedwith impact basins extend down to 84 km and are approximately halfthe diameters of the outer rims of their corresponding basins(Neumann et al., 2015). Thus, it follows that there is an ambiguitybetween a complex crater in the diameter range 100–200 km thatformed after the basin of interest and was subsequently buried by

maria to produce a QCMA, and a peak-ring basin or protobasin in the diameter range 200–400 km that pre-dated the basin of interest and was subsequently buried by the ejecta from that basin, resulting in a QCMAfrom the central mantle uplift. Although this ambiguity cannot be resolved for each QCMA with a large posi-tive Bouguer anomaly in this study, we can confidently state that such preexisting structures must comprise aminority of QCMAs, for several reasons.

Notably, on the basis of the number of QCMAs with large positive Bouguer anomalies (36) relative to the totalnumber of central positive Bouguer anomalies associated with recognized basins (66; Neumann et al., 2015),the majority of QCMAs cannot represent mantle uplifts of ancient basins unless the mare-flooded regionswere bombarded by a disproportionately larger number of basin-forming impactors per area than the restof the Moon. Moreover, the cumulative SFD for lunar craters in the FHT and SPA (Figure 3a) dictates that com-plex craters of a diameter greater than D should be more abundant by a factor of ~4.5 than basins with a dia-meter greater than 2D (i.e., those with central positive Bouguer anomalies of diameter D). Thus, at a givendiameter, the gravitational signatures of the central mantle uplifts of buried basins in mare-flooded areasshould be much less common than the signatures of volcanically buried craters. Although some individualQCMAsmay be the expressions of mantle uplift beneath preexisting basins, the population as a whole cannotsatisfy this interpretation.

To test for a statistically significant presence of preimpact structures, we calculated N(90) values for the annu-lus interior to the main rim diameter and exterior to the central positive Bouguer anomaly for each basin inTable 1. By applying a randomness detection method (runs test; Bradley, 1968) to the set of N(90) valuesordered by stratigraphic (age) group, we can reject the null hypothesis of randomness at the 90% significancelevel. This result remains unchanged for all possible arrangements of basins within a given group. Therefore,the observed decrease in N(90) values with age is statistically significant and validates our assumption thatthe density of craters and QCMAs within the annulus bounded by the main rim and the central positiveBouguer anomaly is reset at a significant level by basin-forming impacts.

Accordingly, we make the assumption that most QCMAs within and along basin rims are superposed ontoand thus postdate their respective basin structures and were later buried beneath ejecta or mare material.If this assumption were to be invalidated, an N(90) value inclusive of QCMAs would overestimate the relativeage of a given region. However, as demonstrated in the next section, this effect would serve only to reinforceour conclusions in sections 4 and 5.

The absence of a well-preserved basin rim for Asperitatis, Serenitatis North, and Orientale SW suggests thatthese basins likely predate the overlapping Nectaris, Serenitatis, and Orientale basins, respectively. We

Figure 3. Cumulative size-frequency distributions (SFDs) of preserved impactcraters for the major lunar terranes. The cumulative SFDs without QCMAs areshown for the FHT proxy region (green) and SPA (black). The PKT region is shownwith and without QCMAs in blue and magenta, respectively (the portion of thefigure with crater diameter D < 90 km is shaded gray). FHT = FeldspathicHighlands Terrane; PKT = Procellarum KREEP Terrane; SPA = South Pole-Aitken;QCMA = quasi-circular mass anomaly.

10.1029/2017JE005421Journal of Geophysical Research: Planets

EVANS ET AL. 1604

therefore excluded these basins from the population of superposed impact structures used to calculate thecrater densities for the latter group of younger basins.

4. Results: Relative Ages of Lunar Terranes and Major Impact Basins

The reliability of surface ages determined by previous workers (e.g., Fassett et al., 2012; Hartmann & Wood,1971; Wilhelms, 1987) depends on the accuracy of the employed crater density estimation methods.Commonly, crater SFDs used to assess relative and absolute ages include crater diameters of 20 km or smallerto improve precision (e.g., Hartmann & Wood, 1971; Neukum, 1983), but both crater saturation and the pre-sence of the lunar maria can introduce biases that hinder the reliable interpretation of such crater-based ageassessments, and some such analyses yielded results at variance with stratigraphic inferences, most notablyfor the SPA and Serenitatis basins.

The use of N(90) for crustal crater retention age assessments provides ages that are generally unbiased bylater volcanism. The elimination of this bias is particularly important for the major impact basins on the near-side that were substantially flooded by maria. To ensure further that reliable estimates of N(90) can be deter-mined, we chose to investigate regions with sufficient areal extents that a crater with a diameter of 90 km,interior to the region of interest, will occupy less than 2% of the region’s total area. On the basis of this criter-ion, we conducted crater size-frequency analyses for areal extents of at least 3.2 × 105 km2, equivalent to acircular area with a diameter of approximately 650 km. We determined crater-size frequency distributionsby applying the buffered crater density approach (Fassett et al., 2012) to all superposed crater structureswithin a given region of interest. All N(D) values were normalized to an area of 106 km2. We determinedthe error for N(D) values at the 68% confidence level (one standard deviation) as prescribed by the methodol-ogy of the Crater Analysis Techniques Working Group (1979). For testing whether two N(D) values are statis-tically distinguishable, we applied the E-Test, a hypothesis test for the comparison of Poisson means(Krishnamoorthy & Thomson, 2004).

For each basin, we used N(90) values, inclusive of QCMAs, determined from the full region interior to themainrim diameter as the primary metric for comparison. We regard the cumulative crater SFD as a valid estimate ofthe true areal density of craters over this size range. As noted by Hartmann and Wood (1971), the differences

Table 1Cumulative Crater Density N(90), N(64), and N(20) for Lunar Basins Listed in Order of Decreasing Inferred Stratigraphic Age

Basin name

Agea N(90)b

N(64)c N(20)cPeriod Group Incl. QCMAs Excl. QCMAs

South Pole-Aitken (SPA)1 PN 1 16 ± 2 16 ± 2 33 ± 3 156 ± 7Procellarum PN 1 11 ± 1 5 ± 1 — —Fecunditatis (F)1 PN 3 10 ± 4 3 ± 2 — —Australe North (A)1 PN 3 15 ± 4 15 ± 4 — —Nubium (Nu)1 PN 3 10 ± 4 8 ± 3 33 ± 7 195 ± 18Smythii (Sm)1 PN 5 13 ± 4 13 ± 4 28 ± 6 225 ± 19Coulomb-Sarton (CS)1 PN 5 11 ± 4 11 ± 4 32 ± 14 271 ± 54Nectaris (N) N 10 9 ± 3 5 ± 3 17 ± 5 135 ± 14Mendel-Rydberg (MR) N 10 6 ± 4 6 ± 3 12 ± 5 125 ± 17Serenitatis (S)1 N 11 5 ± 2 1 ± 1 28 ± 20 298 ± 60Humorum (H) N 11 4 ± 3 3 ± 2 12 ± 6 108 ± 21Crisium (C) N 11 3 ± 1 3 ± 1 8 ± 3 113 ± 11Imbrium (I) I 12 6 ± 2 1 ± 1 4 ± 2 30 ± 5Orientale (O) I 12 0 0 1 ± 1 21 ± 4

Note. Units for N(D) are number of craters per 106 km2. Overlapping basins are excluded from theN(90) values, unless thebasin is denoted by “1.” QCMA = quasi-circular mass anomaly.aThe inferred stratigraphic ages are binned by lunar period and group as defined byWilhelms (1987). The notations “PN,”“N,” and “I” denote the pre-Nectarian, Nectarian, and Imbrian periods, respectively. The basin group number denotesrelative age inferred from stratigraphic relationships, from oldest to youngest. bThe N(90) value was determined bothwith (“incl. QCMAs”) and without (“excl. QCMAs”) QCMAs for the region interior to the main basin rim. cN(64) and N(20)values are from Fassett et al. (2012). A dash symbol indicates that no value was listed for that region by Fassett et al.(2012).

10.1029/2017JE005421Journal of Geophysical Research: Planets

EVANS ET AL. 1605

in the derived crater densities will not be directly proportional to differences in absolute age because of acratering rate that was nonuniform in time, but instead such values establish relative (crustal crater retention)ages. Thus, we use the terms crater density and relative age interchangeably.

4.1. Major Lunar Geochemical Terranes and Pre-Nectarian Basins

The Pre-Nectarian period is generally recognized to have persisted from ~4.53 Ga until the time of theNectaris impact. Although the age of the Nectaris impact is often accepted as ~3.92 Ga, radiometric agesof material inferred to have been emplaced during the Nectaris impact event indicate that the Nectarisimpact may have occurred as early as 4.26 Ga (Warren, 2003).

During the Pre-Nectarian period, at least two of the three major geochemical provinces on the lunar surfacewere formed: (1) the FHT and (2) the SPA Terrane. The third major geochemical province, the PKT (Jolliff et al.,2000), although ancient, has not been conclusively confined to a specific lunar epoch. The formation of thesegeochemical provinces constrains the timing of the earliest events in lunar history, including crystallization ofthe lunar magma ocean and the SPA impact. For the purposes of this work, we assume that the SPA basininterior (Garrick-Bethell & Zuber, 2009) is representative of the SPA Terrane as defined by Jolliff et al. (2000).

Below, we use the crater densities inclusive of QCMAs to reexamine relative age constraints on lunar geo-chemical terranes and pre-Nectarian basins. For the geochemical terranes, we use the regions outlined inFigure 1 to assess relative age. We compare our results to the generally accepted chronologic sequence ofWilhelms (1987).4.1.1. Feldspathic Highlands TerraneThe formation of the primary crust, and hence the FHT, is arguably the most ancient event preserved atthe lunar surface. For the full FHT shown in Figure 1a, we determined an N(90) value of 15 ± 1. However,as the crater density of the full FHT may have been affected by the subsequent formation of youngerbasins and the other geochemical terranes, we also consider a single, continuous region within theFHT, 1490 km in diameter, centered at 70°N, 143°E, that is devoid of basins with diameters greater than650 km and that does not border other geochemical terranes (see Figure 4a). For this proxy region, wedetermined the N(90) value to be 14 ± 3, similar to the N(90) value of 15 ± 1 determined from thediscontinuous but full FHT region (see Table 2). Hence, we may conclude that the influence of youngerbasins and other geochemical terranes on the N(90) value of the full FHT is negligible.4.1.2. South Pole-Aitken and Procellarum KREEP TerraneFor the ancient SPA basin, we determined an N(90) value of 16 ± 2. The significance of this result is twofold: (1)unlike analyses with N(20) values (see Table 1), N(90) values for SPA, within error, are consistent with thechronologic sequence established by stratigraphic analyses; and (2) the relative age of the SPA basin is indis-tinguishable from that of the FHT region. The latter could be the result of crater saturation within the FHT andSPA basin, nearly contemporaneous ages for the formation of the FHT and SPA basin, or a significant reset ofthe crater population within the FHT by ejecta from the SPA basin.

For the full PKT region (see Figures 1a and 4a), we find that the N(90) value is 11 ± 1. This N(90) value of the fullPKT is statistically indistinguishable from those N(90) values obtained for the western portion of the PKTdevoid of major impact basins (see supporting information) and the interior of the proposed Procellarumimpact structure (see Figure 5), 11 ± 2 and 11 ± 1, respectively. Moreover, the N(90) value of the PKT is sig-nificantly (99% confidence level) lower than the N(90) value for the SPA basin of 16 ± 2. In fact, as shown inFigure 3, even with QCMAs included, the N(D) values of the SPA basin remain greater than those of the PKTfor all D ≥ 20 km, further supporting the robustness of this result. If a small fraction of the QCMAs within thePKT are not buried craters, the true N(90) value for the PKT would be lower, thereby increasing the signifi-cance of this result.

Although an ancient Procellarum impact basin has been invoked to account for the localization of low-Ca pyr-oxene (Nakamura et al., 2012) and KREEP on the lunar nearside (e.g., Cadogan, 1974), the gravitational andphysiographic characteristics of the region are inconsistent with those of an ancient impact structure(Andrews-Hanna et al., 2014). Nonetheless, the absence of these characteristics does not necessarily precludethe possibility of a Procellarum impact event for which little or no topographic or gravitational evidenceremains. Regardless of whether a Procellarum basin-forming impact occurred, the lower crater density ofthe PKT (and the suggested Procellarum basin) relative to the SPA basin requires that the crater populationwithin the PKT was significantly reset at, or crater retention inhibited until, a time after the SPA impact event.

10.1029/2017JE005421Journal of Geophysical Research: Planets

EVANS ET AL. 1606

Figure 4. Maps of superposed craters for geochemical terranes and selected basins. (a) The Feldspathic Highlands Terrane(FHT) proxy region, Procellarum KREEP Terrane (PKT), and South Pole-Aitken (SPA) Terrane are shown in orthographicprojections of the lunar morphological base map (Speyerer et al., 2011) in the upper, middle, and bottom panels, respec-tively. Maps are centered on the regions of interest, which are outlined in black and white and labeled. The SPA interior, theregion 500-km interior to the main rim of SPA basin and the nominal location of the SPA Terrane, is enclosed by theinnermost outline of black and white in the bottom panel. (b) The Nectaris, Imbrium, and Orientale basins are outlined inblack and white in the upper, middle, and bottom panels, respectively. The positive Bouguer anomalies (Neumann et al.,2015) central to each of the basins are also outlined in black and white. Craters identified from topography byHead et al. (2010; yellow) and buried craters identified from gravity anomalies by Neumann et al. (2015; green) and Evanset al. (2016; cyan) are shown for each area. Craters with diameters greater than 90 km are shown with thicker outlines. Thesouth and north poles are marked with the circle and diamond symbols, respectively.

10.1029/2017JE005421Journal of Geophysical Research: Planets

EVANS ET AL. 1607

Three scenarios could account for the younger crustal crater retention age of the PKT relative to the SPAbasin: (1) the deposition of impact ejecta from the Imbrium impact obliterated some of the large craters(D ≥ 90 km) within the PKT such that they are not recoverable as QCMAs; (2) the primary crust within thePKT formed after the primary crust within the SPA basin, as well as after the FHT; or (3) the precursor of lunarKREEP-rich material, commonly referred to as urKREEP, produced sufficient subsurface heat across the PKT tomodify crater formation or delay the onset of crater retention for large craters (D ≥ 90 km) until after the timeof the SPA impact event. For the first scenario, the higher crater density of the PKT than for the Imbrium basin(see Tables 1 and 2) precludes the complete obliteration of the population of craters with D ≥ 90 km withinthe PKT during or after the Imbrium impact event. Although the obliteration of a partial population of craterswithin the PKT by the Imbrium impact is possible, the similar N(90) values for the full PKT and the westernportion of the PKT (i.e., the portion farthest from Imbrium) along with the absence of any substantial devia-tion in the ratio of the incremental SFD of nonmare craters to that of PKT craters (Figure S7) suggests thateither such a scenario did not occur or was not important to the record of preserved craters.

The remaining two scenarios—late formation of the PKT crust and delayed retention of large craters withinthe PKT—are not readily distinguishable from crater size-frequency analyses, but either process would havelikely concluded by the end of magma ocean crystallization, as discussed below. Furthermore, both scenariosare consistent with the predicted crystallization sequence for the cooling lunar magma ocean (e.g., Elkins-Tanton et al., 2011), which calls for urKREEP to crystallize later than and at a lower temperature than the feld-spar that formed the FHT. At the time of crystallization of the subsurface urKREEP layer, the primary crust ofthe PKT would have been sufficiently thick to preserve the crustal signature of impact structures with dia-meters greater than ~150 km (Miljković et al., 2016), and the crustal temperature would have been too low

Table 2Cumulative Crater Density N(90) for Major Lunar Geochemical Terranes

Terrane

Proxy region Full region

N(90) QCMAs N(90) QCMAs

South Pole-Aitken Terrane 16 ± 2 <1 — —Feldspathic Highlands Terrane (FHT) 14 ± 3 <1 15 ± 1 <1Procellarum KREEP Terrane (PKT) 11 ± 2 7 11 ± 1 7

Note. Units for N(90) are number of craters per 106 km2. The “Full Region” denotes the area as defined by Jolliff et al.(2000), and the “Proxy Region” refers to the SPA interior, the FHT proxy region, and the western portion of the PKT(see the supporting information) for the SPA Terrane, FHT, and PKT, respectively. A dash symbol indicates values thatwere not computed. “QCMA” indicatesN(90) values for the QCMA population only. QCMA = quasi-circular mass anomaly.

Figure 5. Map and cumulative size-frequency distribution (SFD) for craters within the Procellarum basin suggested byWhitaker (1981). The proposed Procellarum basin is outlined in black and white in the left panel on the lunar morpholo-gical base map using the same scheme as in Figure 4. The right panel compares the crater cumulative SFD for Procellarum(black-dashed) with those of SPA (black), Nectaris (green), Imbrium (blue), and Orientale (magenta). SPA = South Pole-Aitken.

10.1029/2017JE005421Journal of Geophysical Research: Planets

EVANS ET AL. 1608

(less than 1300 K, the crystallization temperature for urKREEP; Rutherford et al., 1996; Wieczorek & Phillips,2000) to enable substantial viscous relaxation of impact structures (e.g., Kamata et al., 2015). Taken togetherwith the previously noted absence of a significant diameter-dependent trend in the ratio of incremental SFDof nonmare craters to that of PKT craters at large diameters (D ≥ 90 km), we may then conclude that after thecrystallization of urKREEP, neither the crater formation process nor viscous relaxation had a prominent effecton the retention of craters with D ≥ 90 km. Consequently, we may use the youngest age inferred for the finalcrystallization of urKREEP, ~4.3 Ga (Borg et al., 2015), as an approximate lower bound on the absolute age ofthe crust underlying the PKT. This lower bound is also consistent with the ages of the oldest mare basalt unitswithin the PKT region, radiometrically dated at ~4.3–4.2 Ga (Dasch et al., 1987; Taylor et al., 1983). Becausethese basalts likely postdated emplacement of the PKT basement, the ages of these samples provide an inde-pendent lower bound on the absolute age of the PKT crust. Given the greater N(90) value of SPA relative tothe PKT, it then follows that the SPA impact must have occurred prior to ~4.3 Ga.4.1.3. Other Pre-Nectarian BasinsThe N(90) values for the pre-Nectarian basins considered in this study vary between 10 and 16. The largeerrors associated with the N(90) values of the pre-Nectarian basins prevent the determination of statisticallysignificant differences in crater density among them. The proposed pre-Nectarian Fecunditatis and Nubiumbasins also yield N(90) values, as well as SFDs for D ≥ 90 km (see the supporting information), concordant withthose for basins and surfaces of pre-Nectarian age. This age constraint neither supports nor refutes their inter-pretation as impact basins. By combining the crater densities within comparatively young, unambiguous pre-Nectarian impact structures—specifically Ingenii and Poincaré—we determined an N(90) value of 19 ± 8 forbasins of stratigraphic group 4 (of 12 groups, with 1 being the oldest) in the chronology of Wilhelms (1987).This N(90) value is greater than the value of 11 ± 1 determined for the PKT at the 90% confidence level. Thus,on the basis of the chronologic sequence of the pre-Nectarian impact basins as ordered by Wilhelms (1987)and the minimum absolute age inferred for the PKT, the aggregated N(90) value determined from thesebasins indicates that all pre-Nectarian basins and impact structures identified within stratigraphic age groups1–4 have an absolute age of at least ~4.3 Ga. Accordingly, the basins classified within stratigraphic age groups1–4, which account for ~40% of the total number of basins and 32% of the “definite” basins chronologicallyordered by Wilhelms (1987), would have likely formed prior to ~4.3 Ga. It should be noted that the implica-tions of this constraint for the crater chronology of the Moon, and the existence and timing of a late heavybombardment in particular (e.g., Kamata et al., 2015; Morbidelli et al., 2012, 2018), depend heavily on thepoorly constrained absolute age for Nectaris as well as the absolute ages for the younger pre-Nectarianbasins (stratigraphic age groups 5–9) that are not considered herein due to their small sizes (D < 650 km).

4.2. Nectarian and Imbrian Periods

We examine next the basin-forming impacts that occurred after the pre-Nectarian period, during theNectarian and Imbrian periods. The Nectarian period spans the interval between the ages of the Nectarisand Imbrium impacts, often accepted as 3.92 and 3.85 Ga, respectively (e.g., Wilhelms, 1987). However, asnoted above, the Nectaris impact has been suggested to have occurred as early as 4.26 Ga (e.g., Warren,2003; Wilhelms, 1987). For the subsequent Imbrian period, we examine those remaining basins that wereformed between the Imbrium impact and the last major impact event, Orientale, which may have occurredas late as ~3.8–3.7 Ga (e.g., Whitten et al., 2011; Wilhelms, 1987).4.2.1. NectarisFor the Nectaris basin (Figure 4b), we determined an N(90) value of 9 ± 3, similar to the N(90) value of 11 ± 1obtained for the PKT region. Nectaris is the first basin of certain impact origin in the chronologic sequence ofWilhelms (1987) that exhibits an N(90) value significantly different (90% confidence level) from that of SPA.The similar N(90) values for the Nectaris basin and PKT are consistent with nearly similar absolute ages forthe two regions. An age of 4.12–4.26 Ga for the Nectaris basin, as indicated by a subset of KREEP-poor sam-ples posited to be Nectaris impact-melt breccias (Warren, 2003), is not very different from the minimum ageof ~4.3 Ga inferred for the PKT. This age range does not, however, exclude the possibility of the younger agefor Nectaris.4.2.2. Imbrium and SerenitatisThe N(90) value for the Imbrium basin, determined for the region interior to the main topographic rimdefined, in part, by the Apennine mountains, is 6 ± 2, significantly lower (99% confidence level) than valuesfor the pre-Nectarian basins and the FHT. Although the N(90) value for Imbrium is considerably higher with

10.1029/2017JE005421Journal of Geophysical Research: Planets

EVANS ET AL. 1609

the buried crater population included than if estimated from surface craters alone (see Figure 4b and Table 1),its relative age is consistent with the inferred chronology of the Moon and with stratigraphic evidence thatImbrium is one of the youngest lunar basins.

The relative age of Serenitatis, as noted above, has long been debated. Estimates of relative ages for theSerenitatis basin have been complicated because of the extensive regional mare flooding (Fassett et al.,2012; Hartmann & Wood, 1971; Wilhelms, 1987). The recovery of the QCMAs (Figure 6a) allows for animproved estimate of its crater density and chronologic position within the sequence of lunar basins. Asshown in Figure 6a, Serenitatis and Imbrium have indistinguishable N(90) values, 5 ± 2 and 6 ± 2, respectively.Additionally, the N(90) value for Serenitatis is significantly lower at the 95% confidence level than the aggre-gate N(90) value of 13 ± 3 determined for the pre-Nectarian impact basins in stratigraphic age group 5,Coulomb-Sarton and Smythii. Therefore, we can conclude that the Serenitatis basin is younger than thebasins within stratigraphic age group 5, although not necessarily younger than all pre-Nectarian basins.

The similar N(90) values for Imbrium and Serenitatis derived in this work are in agreement with the relativelyyoung age for Serenitatis proposed by Wilhelms (1987). Wilhelms (1987) cited several lines of evidence insupport of this suggestion, including the fresh appearance of the massifs of Serenitatis, as noted duringthe Apollo 17 mission (Evans & El-Baz, 1973).

However, on the basis of more recent image and topographic data, Spudis et al. (2011) and Fassett et al.(2012) suggested that the Serenitatis basin-forming impact may have occurred during the pre-Nectarianepoch, as previously proposed in some of the earliest relative age classifications for Serenitatis (Head,1974; Stuart-Alexander & Howard, 1970). Fassett et al. (2012) based their suggestion of a pre-NectarianSerenitatis, in part, on the identification of possible Nectaris sculptured ejecta at Serenitatis and a high

Figure 6. Maps and cumulative size-frequency distributions for craters within the (a) Serenitatis and (b) Crisium basins.Figure scheme is the same as for Figure 5. SPA = South Pole-Aitken.

10.1029/2017JE005421Journal of Geophysical Research: Planets

EVANS ET AL. 1610

N(20) value of 298 ± 60 in a region of small area (~8×104 km2) east of Mare Serenitatis. Yet three noteworthyconcerns are raised by the pre-Nectarian age classification as suggested by Fassett et al. (2012): (1) Thederived N(20) value of 298 ± 60 for Serenitatis in that study is higher than the N(20) value of any other basin,including the N(20) value of 156 ± 7 for the oldest recognizable basin on the Moon, SPA; (2) the derived cra-ter density values for Serenitatis were classified by Fassett et al. (2012) as “uncertain” because of observa-tional challenges; and (3) there is conflicting stratigraphic evidence for the relative age of Serenitatis, alsoreflected by the question mark associated with the stratigraphic classification of Serenitatis in Table 1 ofFassett et al. (2012).

A possible explanation for the high N(20) values determined for Serenitatis is that the count areas from whichthese values were determined have been contaminated by secondary craters from the Imbrium impact, asnoted by Wilhelms (1987) for the area east of Mare Serenitatis. Fassett et al. (2012) also acknowledged thatthe influence of secondary craters from the Imbrium impact could account for the high density of craters withD ≥ 20 km at Serenitatis, but ultimately, they discounted this explanation by the argument that Crisiummate-rials immediately to the east should have been similarly contaminated by secondaries. Yet contamination ofthe Serenitatis region by Imbrium secondaries would yield the high N(20) value for Serenitatis despite its rela-tively low N(90) value. Alternatively, the high N(20) values for Serenitatis could be the result of including sev-eral craters in the count area for Serenitatis that predate that impact (e.g., Wilhelms, 1987).

The possibility that the true N(20) and N(90) values for Serenitatis are higher than those of Imbrium andNectaris cannot be completely excluded on the basis of this work. If the interpretation of a pre-Nectarianage for the Serenitatis impact event is correct, then our results further restrict the Serenitatis impact to haveoccurred during the late pre-Nectarian, in the time interval spanning stratigraphic age groups 6 and 9 ofWilhelms (1987).

4.2.3. Mendel-Rydberg and HumorumMendel-Rydberg and Humorum basins are classified as Nectarian in age (Wilhelms, 1987). According to thechronology of Wilhelms (1987), Mendel-Rydberg is younger than the Nectaris basin but older thanSerenitatis and Humorum. No buried craters are found within the main rim of the Mendel-Rydberg basin,and its N(90) value of 6 ± 4 is consistent with the chronology of Wilhelms (1987), although the relative agederived here is not significantly younger than Nectaris or older than Serenitatis and Imbrium.

Humorum has an N(90) value of 4 ± 3, lower than that of the Nectaris and Imbrium basins. Humorum isconsidered to be late-Nectarian in age and within the same stratigraphic age group as Serenitatis. Theerror associated with Humorum does not allow for a statistically significant relationship to be establishedbetween Humorum and Imbrium. Accordingly, we deem this result, within error, to be consistent with thepreviously established chronologic position of the Humorum impact event as prior to that of Imbrium butafter Nectaris.4.2.4. CrisiumThe center of the Crisium basin is characterized by unusually thin crust (Wieczorek et al., 2013; see alsoMiljković et al., 2015), substantial mare basalt infill (Head et al., 1978), deeply penetrating thrust faultsbeneath surface contractional features (Byrne et al., 2015), and no evident extensional features (Solomon &Head, 1980). Possible Serenitatis ejecta superposed on the Crisium rim and the lack of obvious Crisium ejectanear Serenitatis suggest a pre-Serenitatis and likely Nectarian age for Crisium (Wilhelms, 1987). Alternatively,an oblique Crisium impact that occurred post-Serenitatis could account for the lack of definitive Serenitatisejecta at Crisium and the absence of Crisium ejecta near Serenitatis (Stickle & Schultz, 2011). AlthoughLuna 20 samples of postulated Crisium ejecta have been radiometrically dated at 3.84 ± 0.04 Ga (Podoseket al., 1973), within the original suspected age range for the Imbrium-forming impact (3.87–3.82 Ga), theyounger-than-expected age measurement led some authors to classify the samples as secondary ejecta fromImbrium (Adams et al., 1978; Wilhelms, 1987).

For Crisium, we determined an N(90) value, inclusive of QCMAs, of 3 ± 1. The low density of large craters withD ≥ 90 km observed interior to Crisium basin is consistent with the results of previous analyses (Fassett et al.,2012; Hartmann & Wood, 1971; Head et al., 2010) and with the formation of Crisium after that of the Nectarisbasin as proposed by Wilhelms (1987). Additionally, we note that our results favor, although do not require,that the Crisium basin was formed after the Imbrium basin. On the basis of N(90) values, Crisium may be of asimilar or younger absolute age than Imbrium.

10.1029/2017JE005421Journal of Geophysical Research: Planets

EVANS ET AL. 1611

4.2.5. OrientaleThe Orientale basin is unambiguously the youngest major impact basin preserved on the surface of the Moon(Wilhelms, 1987; see also Whitten et al., 2011). Unlike many major impact basins across the Moon, Orientale iswell preserved and not substantially obscured by thick mare basalt deposits. No superposed craters withD ≥ 90 km are found within the Orientale basin. The absence of superposed craters with D ≥ 90 km withinthe Orientale basin structure reinforces the conclusion that Orientale is the youngest impact basin on theMoon.

5. Possible Variations in the Early Lunar Impact Flux

As shown in Figure 7, with the possible exception of Crisium and Serenitatis, our results are in agreement withthe relative basin chronology detailed by Wilhelms (1987). With this relative chronologic sequence, we usethe crater densities of the geochemical terranes and major impact basins to examine the temporal evolutionof the impactor population incident on the Moon. This discussion is complicated by the fact that there areinconsistent constraints on the absolute ages of the major impact basins, a situation dependent on the geo-logical interpretation of the provenance of returned samples.

Among the pre-Nectarian basins we find no statistically significant difference in the density of superposed cra-ters and basins greater than 90 km in diameter. There are several possible scenarios that could account for thesimilar crater densities among the pre-Nectarian basins: (1) the basins formed over an extended period oftime, but there was a relatively low rate of impact craters formed with D ≥ 90 km during this period; (2) thebasins were formed over a short time interval; or (3) the surfaces of these basins have reached a state of cratersaturation equilibrium. We regard the first scenario as implausible, since the formation of as many as six majorimpact basins (D ≥ 650 km) in a period of less than ~200 Myr, inferred from the ~4.3 Ga age of the pre-Nectarian basin group (see Table 1), is suggestive of a high impact rate. Although we cannot exclude the sec-ond scenario, the greater impact rate required to have formed the basins within a short time interval wouldhave likely resulted in a trend of increasing crater density with age among the basins, unless the third scenario,crater saturation, applied. Therefore, in agreement with the conclusions of previous workers (e.g., Fassett et al.,2012; Head et al., 2010), we deem a saturated crater density as the most plausible explanation for the similarrelative ages among pre-Nectarian basins. Alternatively, although we do not resolve a statistically significantdifference in the density of superposed craters of diameterD ≥ 90 km among pre-Nectarian basins, our resultsdo not rule out the possibility of such a difference given the errors in the N(90) values for those basins.

Our results also indicate that there is a statistically significant decrease in the density of craters withD ≥ 90 kmfor basins formed after the youngest pre-Nectarian impacts that we consider here. As shown in Figure 7, this

Figure 7. N(90) values that include quasi-circular mass anomalies clustered by basin stratigraphic group according toWilhelms (1987). Basins are expanded on the horizontal axis for stratigraphic groups with multiple basins. Basins arelabeled by name and stratigraphic group in parentheses. The oldest definitive basin within each age period is shown with ared square, basins with disputed relative ages are indicated by green circles, and other basins are denoted by blue circles.SPA = South Pole-Aitken.

10.1029/2017JE005421Journal of Geophysical Research: Planets

EVANS ET AL. 1612

decline in the density of craters with D ≥ 90 km continued through the Nectarian period and into the earlyImbrian, terminating with the last major lunar impact basin, Orientale, which has no superposed crater struc-tures with D ≥ 90 km. The uncertainty in the absolute ages and derived N(90) values for the lunar geochemicalterranes and basins prevents any conclusive determinations regarding the timing of a lunar cataclysm or lateheavy bombardment. Given the combination of likely saturated crater densities in the pre-Nectarian and thestatistically significant decrease in crater density by the time of the Nectaris impact, however, a large changein impactor flux could have occurred prior to or accompanying the Nectaris basin-forming impact. If Nectarisis taken as the beginning of the cataclysm, then the similar crater densities of the PKT and Nectaris combinedwith the inferred minimum age for the PKT of 4.3 Ga is consistent with the timing for a cataclysm suggestedby some recent analyses (e.g., Morbidelli et al., 2012; van der Bogert et al., 2017).

N(20) values determined for Serenitatis and Imbrium, 298 ± 60 and 30 ± 5, respectively, indicate a substantialdifference in their crater densities despite potentially being of similar absolute age (Fassett et al., 2012; Spudiset al., 2011). The marked difference in the N(20) values determined by previous workers imposed a severeconstraint on the flux of lunar basin-forming impactors for scenarios under which Serenitatis and Imbriumbasins are of similar absolute age, requiring that up to half of the population of lunar basins formed withina very narrow time interval of ~50 Myr (Spudis et al., 2011). For this reason and others, several workers haveconcluded that a young absolute age (~3.8 Ga) assigned to Serenitatis might be incorrect and that samplesascribed to Serenitatis might instead be from Imbrium (Fassett et al., 2012; Spudis et al., 2011). Yet, given thestatistically indistinguishable N(90) values determined for Serenitatis and Imbrium in this study (see Table 1),we can eliminate the requirement for forming a large number of lunar basins within a ~50-Myr interval at~3.8 Ga in the event that Serenitatis has a young absolute age. We cannot, however, dismiss the possibilitythat such a geologically short interval of high lunar impact bombardment nonetheless occurred.

6. Summary

Although the record of premare impact craters has been modified by mare volcanism for crater diameters aslarge as 250 km, for diameters greater than 90 km that record is well preserved on the Moon by a combina-tion of topography and gravity. Our results demonstrate that themare-flooded regions are deficient in craterswithD ≥ 20 km bymore than 50% relative to unflooded regions, but themare-flooded and unflooded regionshave comparable areal abundances of surface and buried craters with D ≥ 90 km. Furthermore, we find thatthe total population of craters with diameters greater than 90 km provides a more complete crater recordthan do surface craters alone and can be used to reassess the relative ages of terranes and basins and thechronology for major geological events on the Moon. From the population of craters with diameters greaterthan 90 km identified in the gravity and topography data sets, we make the following conclusions:

1. The areal density of impact craters, including buried craters, yields relative ages for major basins that aregenerally consistent with the stratigraphic order and general lunar chronology of Wilhelms (1987).

2. The Procellarum KREEP Terrane (PKT) is similar in crustal crater retention age to the Nectaris basin, makingit younger than the South Pole-Aitken (SPA) basin. We infer a lower bound on the absolute age of the PKTof ~4.3 Ga, taken from the youngest crystallization age for urKREEP (Borg et al., 2015) and the oldest (~4.3–4.2 Ga) radiometric ages of mare basalts from the region (Dasch et al., 1987; Taylor et al., 1983).

3. The absence of gravitational and physiographic signatures indicative of a Procellarum basin rim structure(Andrews-Hanna et al., 2014) and the young relative age of the PKT are inconsistent with an ancientimpact event as the principal cause for the distinctive properties of this terrane. If the PKT was formedas a consequence of an ancient Procellarum impact, a later, unidentified secondary process that was com-pleted after the SPA impact event is required to remove most evidence for the basin in topography andgravity.

4. That the relative ages of several pre-Nectarian basins are significantly greater than the age of the PKTallows us to establish a minimum absolute age of ~4.3 Ga for the SPA impact basin and other pre-Nectarian basins within stratigraphic age groups 1–4 of Wilhelms (1987), indicating that ~40% of theknown lunar impact basins are older than ~4.3 Ga and likely predate any late heavy bombardment.

5. With and without the inclusion of the buried craters, the Serenitatis basin has a significantly lower abun-dance of craters with D ≥ 90 km than the pre-Nectarian basins Coulomb-Sarton and Smythii. Serenitatislacks more than half the population of craters with D ≥ 90 km per area seen for these pre-Nectarian basins.The similar N(90) values for Serenitatis and Imbrium suggest that the Serenitatis basin-forming impact is

10.1029/2017JE005421Journal of Geophysical Research: Planets

EVANS ET AL. 1613

most consistent with a late-Nectarian event, but the impact may have occurred as early as the impactsthat formed the pre-Nectarian basins in stratigraphic age group 6 of Wilhelms (1987).

6. The Crisium basin may have an absolute age similar to or younger than that of Imbrium.7. For pre-Nectarian basins, no statistically significant differences in their size-frequency distribution for

D ≥ 90 km can be resolved, a result likely indicating that a state of crater saturation equilibrium wasreached for basins in this age group. In contrast, the density of craters with D ≥ 90 km preserved by thecombination of gravity and topography declined through the Nectarian and Imbrian periods.

Inclusion of impact craters fully buried by volcanic or impact deposits yields a more complete record of thelunar crater population that has enabled a fresh assessment of the chronology of mare-flooded terranes andbasins. By incorporating buried craters identified from the gravity field (Evans et al., 2016; Neumann et al.,2015) along with craters recognized from their surface topography (Head et al., 2010), we have improved con-straints on early lunar chronology, particularly the relative impact sequence of the lunar basins, the absoluteages of lunar geochemical terranes and basins, and the impactor flux prior to ~3.8 Ga.

ReferencesAdams, J. B., Head, J. W., McCord, T. B., Pieters, C., & Zisk, S. (1978). Mare Crisium: Regional stratigraphy and geologic history. Geophysical

Research Letters, 5(4), 313–316. https://doi.org/10.1029/GL005i004p00313Andrews-Hanna, J. C., Besserer, J., Head, J. W. III, Howett, C. J. A., Kiefer, W. S., Lucey, P. J., et al. (2014). Structure and evolution of the lunar

Procellarum region as revealed by GRAIL gravity data. Nature, 514(7520), 68–71. https://doi.org/10.1038/nature13697Baker, D. M. H., Head, J. W., Phillips, R. J., Neumann, G. A., Bierson, C. J., Smith, D. E., & Zuber, M. T. (2017). GRAIL gravity observations of the

transition from complex crater to peak-ring basin on the Moon: Implications for crustal structure and impact basin formation. Icarus, 292,54–73. https://doi.org/10.1016/j.icarus.2017.03.024

Barboni, M., Boehnke, P., Keller, B., Kohl, I. E., Schoene, B., Young, E. D., & McKeegan, K. D. (2017). Early formation of the Moon 4.51 billion yearsago. Science Advances, 3(1), e1602365. https://doi.org/10.1126/sciadv.1602365

Barker, M. K., Mazarico, E., Neumann, G. A., Zuber, M. T., Haruyama, J., & Smith, D. E. (2016). A new lunar digital elevation model from the LunarOrbiter Laser Altimeter and SELENE Terrain Camera. Icarus, 273, 346–355. https://doi.org/10.1016/j.icarus.2015.07.039

Borg, L. E., Gaffney, A. M., & Shearer, C. K. (2015). A review of lunar chronology revealing a preponderance of 4.34–4.37 Ga ages. Meteoriticsand Planetary Science, 50(4), 715–732. https://doi.org/10.1111/maps.12373

Bottke, W. F., & Norman, M. D. (2017). The late heavy bombardment. Annual Review of Earth and Planetary Sciences, 45(1), 619–647. https://doi.org/10.1146/annurev-earth-063016-020131

Bradley, J. V. (1968). Distribution-free statistical tests. Englewood Cliffs, NJ: Prentice Hall.Budney, C. J., & Lucey, P. G. (1998). Basalt thickness in Mare Humorum: The crater excavation method. Journal of Geophysical Research,

103(E7), 16,855–16,870. https://doi.org/10.1029/98JE01602Byrne, P. K., Klimczak, C., McGovern, P. J., Mazarico, E., James, P. B., Neumann, G. A., et al. (2015). Deep-seated thrust faults bound the Mare

Crisium lunar mascon. Earth and Planetary Science Letters, 427, 183–190. https://doi.org/10.1016/j.epsl.2015.06.022Cadogan, P. H. (1974). Oldest and largest lunar basin? Nature, 250(5464), 315–316. https://doi.org/10.1038/250315a0Chapman, C. R., & Jones, K. L. (1977). Cratering and obliteration history of Mars. Annual Review of Earth and Planetary Sciences, 5(1), 515–538.

https://doi.org/10.1146/annurev.ea. 05.050177.002503Chin, G., Brylow, S., Foote, M., Garvin, J., Kasper, J., Keller, J., et al. (2007). Lunar Reconnaissance Orbiter overview: The instrument suite and

mission. Space Science Reviews, 129(4), 391–419. https://doi.org/10.1007/s11214-007-9153-yCochran, W. G. (1977). Sampling techniques. New York: John Wiley & Sons.Crater Analysis Techniques Working Group (1979). Standard techniques for presentation and analysis of crater size-frequency data. Icarus,

37, 467–474. https://doi.org/10.1016/0019-1035(79)90009-5Dasch, E. J., Shih, C. Y., Bansal, B. M., Wiesmann, H., & Nyquist, L. E. (1987). Isotopic analysis of basaltic fragments from lunar breccia 14321:

Chronology and petrogenesis of pre-Imbrium mare volcanism. Geochimica et Cosmochimica Acta, 51(12), 3241–3254. https://doi.org/10.1016/0016-7037(87)90132-3

Elkins-Tanton, L. T., Burgess, S., & Yin, Q.-Z. (2011). The lunar magma ocean: Reconciling the solidification process with lunar petrology andgeochronology. Earth and Planetary Science Letters, 304(3-4), 326–336. https://doi.org/10.1016/j.epsl.2011.02.004

Evans, A. J., Soderblom, J. M., Andrews- Hanna, J. C., Solomon, S. C., & Zuber, M. T. (2016). Identification of buried lunar impact cratersfrom GRAIL data and implications for the nearside maria. Geophysical Research Letters, 43, 2445–2455. https://doi.org/10.1002/2015GL067394

Evans, R. E., & El-Baz, F. (1973). Geological observations from lunar orbit. In Apollo 17 preliminary science report, special publication SP-330(pp. 28–1–28–32). Washington, DC: NASA.

Fassett, C. I., Head, J. W., Kadish, S. J., Mazarico, E., Neumann, G. A., Smith, D. E., & Zuber, M. T. (2012). Lunar impact basins: Stratigraphy,sequence and ages from superposed impact crater populations measured from Lunar Orbiter Laser Altimeter (LOLA) data. Journal ofGeophysical Research, 117, E00H06. https://doi.org/10.1029/2011JE003951

Fischer-Gödde, M., & Becker, H. (2011). What is the age of the Nectaris basin? New Re-Os constraints for a pre-4.0 Ga bombardment history ofthe Moon. Lunar and Planetary Science, 42. Abstract 1414

Garrick-Bethell, I., & Zuber, M. T. (2009). Elliptical structure of the lunar South Pole-Aitken basin. Icarus, 204(2), 399–408. https://doi.org/10.1016/j.icarus.2009.05.032

Gatz, D. F., & Smith, L. (1994). The standard error of a weighted mean concentration—I. Bootstrapping vs other methods. AtmosphericEnvironment, 29, 1185–1193. https://doi.org/10.1016/1352-2310(94)00210-C

Gault, D. E. (1970). Saturation and equilibrium conditions for impact cratering on the lunar surface: Criteria and implications. Radio Science,5(2), 273–291. https://doi.org/10.1029/RS005i002p00273

Gong, S., Wieczorek, M. A., Nimmo, F., Kiefer, W. S., Head, J. W., Huang, C., et al. (2016). Thicknesses of mare basalts on the Moon from gravityand topography. Journal of Geophysical Research: Planets, 121, 854–870. https://doi.org/10.1002/2016JE005008

10.1029/2017JE005421Journal of Geophysical Research: Planets

EVANS ET AL. 1614

AcknowledgmentsThis work was conducted as part of theGRAIL mission, which was performedunder contract to the MassachusettsInstitute of Technology and the JetPropulsion Laboratory, CaliforniaInstitute of Technology. A. J. E. wassupported by a NASA Solar SystemExploration Research Virtual Institutegrant at the Southwest ResearchInstitute and by a NASA Lunar DataAnalysis Program grant to J. C. A. H.J. C. A. H. and J. W. H. were supported bygrants from the NASA GRAIL GuestScientist Program. J. C. A. H. and J. M. S.were supported by grants from theNASA Lunar Data Analysis Program. Wethank William Bottke, Prajkta Mane,Katarina Miljković, and the GRAILScience Team for assistance and com-ments. Additionally, we thank CalebFassett and two anonymous reviewersfor their constructive comments andsuggestions on an earlier draft. The dataused in this work are listed in the refer-ences, the tables, and the supportinginformation.

Guo, D., Liu, J., & Head, J. W. (2017). Spatial distribution and geometrics of Orientale secondary crater. Lunar and Planetary Science, 48.Abstract 2560

Hartmann, W. K. (1966). Martian cratering. Icarus, 5(1-6), 565–576. https://doi.org/10.1016/0019-1035(66)90071-6Hartmann, W. K., & Wood, C. A. (1971). Moon: Origin and evolution of multi-ring basins. The Moon, 3(1), 3–78. https://doi.org/10.1007/

BF00620390Head, J. W. III, Fassett, C. I., Kadish, S. J., Smith, D. E., Zuber, M. T., Neumann, G. A., & Mazarico, E. (2010). Global distribution of large lunar

craters: Implications for resurfacing and impactor populations. Science, 329(5998), 1504–1507. https://doi.org/10.1126/science.1195050Head, J. W. III, & Wilson, L. (1991). Absence of large shield volcanoes and calderas on the Moon: Consequence of magma transport

phenomena? Geophysical Research Letters, 18(11), 2121–2124. https://doi.org/10.1029/91GL02536Head, J. W. III, & Wilson, L. (1992). Lunar mare volcanism: Stratigraphy, eruption conditions, and the evolution of secondary crusts. Geochimica

et Cosmochimica Acta, 56(6), 2155–2175. https://doi.org/10.1016/0016-7037(92)90183-JHead, J. W. III, & Wilson, L. (2017). Generation, ascent and eruption of magma on the Moon: New insights into source depths, magma supply,

intrusions and effusive/explosive eruptions (Part 2: Predicted emplacement processes and observations). Icarus, 283, 176–223. https://doi.org/10.1016/j.icarus.2016.05.031

Head, J. W. (1974). Morphology and structure of the Taurus-Littrow highlands (Apollo 17): Evidence for their origin and evolution. The Moon,9(3-4), 355–395. https://doi.org/10.1007/BF00562579

Head, J. W. (1975a). Lunar mare deposits: Areas, volumes, sequence, and implication for melting in source areas. In Origins of mare basalts(pp. 66–69). Houston, TX: Lunar and Planetary Institute.

Head, J. W. (1975b). Processes of lunar crater degradation: Changes in style with geologic time. The Moon, 12(3), 299–329. https://doi.org/10.1007/BF02629699

Head, J. W., Adams, J. B., McCord, T. B., Pieters, C. M., & Zisk, S. H. (1978). Regional stratigraphy and geologic history of Mare Crisium. In R. B.Merrill & J. J. Papike (Eds.), Mare Crisium: The view from Luna 24 (pp. 43–74). New York: Pergamon Press.

Hess, P. C., & Parmentier, E. M. (2001). Thermal evolution of a thicker KREEP liquid layer. Journal of Geophysical Research, 106(E11),28,023–28,032. https://doi.org/10.1029/2000JE001416

Hiesinger, H., Head, J. W., Wolf, U., Jaumann, R., & Neukum, G. (2003). Ages and stratigraphy of mare basalts in Oceanus Procellarum, MareNubium, Mare Cognitum, and Mare Insularum. Journal of Geophysical Research, 108(E7), 5065. https://doi.org/10.1029/2002JE001985

Hiesinger, H., Head, J. W., Wolf, U., Jaumann, R., & Neukum, G. (2010). Ages and stratigraphy of lunar mare basalts in Mare Frigoris and othernearside maria based on crater size-frequency distribution measurements. Journal of Geophysical Research, 115, E03003. https://doi.org/10.1029/2009JE003380

Hiesinger, H., Head, J. W., Wolf, U., Jaumann, R., & Neukum, G. (2011). Ages and stratigraphy of lunar mare basalts: A synthesis. In W. A.Ambrose & D. A. Williams (Eds.), Recent advances and current research issues in lunar stratigraphy, Special. Paper 477 (pp. 1–51). Boulder, CO:Geological Society of America. https://doi.org/10.1130/2011.2477(01)

Hiesinger, H., Jaumann, R., Neukum, G., & Head, J. W. III (2000). Ages of mare basalts on the lunar nearside. Journal of Geophysical Research,105(E12), 29,239–29,275. https://doi.org/10.1029/2000JE001244

Hörz, F. (1978). How thick are lunar mare basalts? Proceedings Lunar Planetary Science Conference, 9, 3311–3331.Jessberger, E. K., Staudacher, T., Dominik, B., & Kirsten, T. (1978). Argon-argon ages of aphanite samples from consortium breccia 73255.

Proceedings Lunar Planetary Science Conference, 9, 841–854.Jolliff, B. L., Gillis, J. J., Haskin, L. A., Korotev, R. L., & Wieczorek, M. A. (2000). Major lunar crustal terranes: Surface expressions and crust-mantle

origins. Journal of Geophysical Research, 105(E2), 4197–4216. https://doi.org/10.1029/1999JE001103Kamata, S., Sugita, S., Abe, Y., Ishihara, Y., Harada, Y., Morota, T., et al. (2015). The relative timing of Lunar Magma Ocean solidification and the

Late Heavy Bombardment inferred from highly degraded impact basin structures. Icarus, 250, 492–503. https://doi.org/10.1016/j.icarus.2014.12.025

Konopliv, A. S., Park, R. S., Yuan, D.-N., Asmar, S. W., Watkins, M. M., Williams, J. G., et al. (2013). The JPL lunar gravity field to sphericalharmonic degree 660 from the GRAIL primary mission. Journal of Geophysical Research: Planets, 118, 1415–1434. https://doi.org/10.1002/jgre.20097

Korotev, R. L., Gillis, J. J., Haskin, L. A., & Jolliff, B. L. (2002). On the age of the Nectaris basin. In The Moon beyond 2002: Next steps in lunar scienceand exploration (p. 31). Houston, TX: Lunar and Planetary Institute. Abstract 3029

Kreslavsky, M. A., & Head, J. W. (2012). New observational evidence of global seismic effects of basin-forming impacts on the Moon fromLunar Reconnaissance Orbiter Lunar Orbiter Laser Altimeter data. Journal of Geophysical Research, 117, E00H24. https://doi.org/10.1029/2011JE003975

Krishnamoorthy, K., & Thomson, J. (2004). A more powerful test for comparing two Poisson means. Journal of Statistical Planning Inference,119(1), 23–35. https://doi.org/10.1016/S0378-3758(02)00408-1

Lawrence, D. J., Elphic, R. C., Feldman, W. C., Prettyman, T. H., Gasnault, O., & Maurice, S. (2003). Small-area thorium features on the lunarsurface. Journal of Geophysical Research, 108(E9), 5102. https://doi.org/10.1029/2003JE002050

Lawrence, D. J., Feldman, W. C., Elphic, R. C., Little, R. C., Prettyman, T. H., Maurice, S., et al. (2002). Iron abundances on the lunar surface asmeasured by the Lunar Prospector gamma-ray and neutron spectrometers. Journal of Geophysical Research, 107(E12), 5130. https://doi.org/10.1029/2001JE001530

Le Feuvre, M., & Wieczorek, M. A. (2011). Nonuniform cratering of the Moon and a revised crater chronology of the inner solar system. Icarus,214(1), 1–20. https://doi.org/10.1016/j.icarus.2011.03.010

Lemoine, F. G., Goossens, S., Sabaka, T. J., Nicholas, J. B., Mazarico, E., Rowlands, D. D., et al. (2013). High–degree gravity models from GRAILprimary mission data. Journal of Geophysical Research: Planets, 118, 1676–1698. https://doi.org/10.1002/jgre.20118

Marchi, S., Mottola, S., Cremonese, G., Massironi, M., & Martellato, E. (2009). A new chronology for the Moon and Mercury. Astronomy Journal,137(6), 4936–4948. https://doi.org/10.1088/0004-6256/137/6/4936

Maurer, P., Eberhardt, P., Geiss, J., Grogler, N., Stettler, A., Brown, G. M., et al. (1978). Pre-Imbrian craters and basins: Ages, compositionsand excavation depths of Apollo 16 breccias. Geochimica et Cosmochimica Acta, 42(11), 1687–1720. https://doi.org/10.1016/0016-7037(78)90257-0

McEwen, A. S., & Bierhaus, E. B. (2006). The importance of secondary cratering to age constraints on planetary surfaces. Annual Review ofEarth and Planetary Sciences, 34(1), 535–567. https://doi.org/10.1146/annurev.earth.34.031405.125018

Miljković, K., Collins, G. S., Wieczorek, M. A., Johnson, B. C., Soderblom, J. M., Neumann, G. A., & Zuber, M. T. (2016). Subsurface morphologyand scaling of lunar impact basins. Journal of Geophysical Research: Planets, 121, 1695–1712. https://doi.org/10.1002/2016JE005038

Miljković, K., Wieczorek, M. A., Collins, G. S., Laneuville, M., Neumann, G. A., Melosh, H. J., et al. (2013). Asymmetric distribution of lunar impactbasins caused by variations in target properties. Science, 342(6159), 724–726. https://doi.org/10.1126/science.1243224

10.1029/2017JE005421Journal of Geophysical Research: Planets

EVANS ET AL. 1615

Miljković, K., Wieczorek, M. A., Collins, G. S., Solomon, S. C., Smith, D. E., & Zuber, M. T. (2015). Excavation of the lunar mantle by basin-formingimpact events on the Moon. Earth and Planetary Science Letters, 409, 243–251. https://doi.org/10.1016/j.epsl.2014.10.041

Morbidelli, A., Marchi, S., Bottke, W. F., & Kring, D. A. (2012). A sawtooth-like timeline for the first billion years of lunar bombardment. Earthand Planetary Science Letters, 355, 144–151. https://doi.org/10.1016/j.epsl.2012.07.037

Morbidelli, A., Nesvorny, D., Laurenz, V., Marchi, S., Rubie, D. C., Elkins-Tanton, L. T., et al. (2018). The timeline of the lunar bombardment—Revisited. Icarus, 305, 262–276. https://doi.org/10.1016/j.icarus.2017.12.046

Nakamura, R., Yamamoto, S., Matsunaga, T., Ishihara, Y., Morota, T., Hiroi, T., et al. (2012). Compositional evidence for an impact origin of theMoon’s Procellarum basin. Nature Geoscience, 5(11), 775–778. https://doi.org/10.1038/ngeo1614

Nelson, D. M., Koeber, S. D., Daud, K., Robinson, M. S., Watters, T. R., Banks, M. E., & Williams, N. R. (2014). Mapping lunar maria extents andlobate scarps using LROC image products. Lunar and Planetary Science, 45. Abstract 2861

Neukum, G. (1983). Meteoritenbombardement und Datierung Planetarer Oberflaechen, Habilitation Dissertation for Faculty Membership.Munich, Germany: University of Munich.

Neukum, G., & Ivanov, B. A. (1994). Crater size distribution and impact probabilities on Earth from lunar, terrestrial-planet, and asteroidcratering data. In T. Gehrels (Ed.), Hazards due to comets and asteroids (pp. 359–416). Tucson, AZ: University of Arizona Press.

Neukum, G., Ivanov, B. A., & Hartmann, W. K. (2001). Cratering records in the inner solar system in relation to the lunar reference system.Space Science Reviews, 96, 55–86.

Neukum, G., König, B., & Arkani-Hamed, J. (1975). A study of lunar impact crater size-distributions. The Moon, 12(2), 201–229. https://doi.org/10.1007/BF00577878

Neumann, G. A., Zuber, M. T., Wieczorek, M. A., Head, J. W., Baker, D. M. H., Solomon, S. C., et al. (2015). Lunar impact basins revealed by GravityRecovery and Interior Laboratory measurements. Science Advances, 1(9), e1500852. https://doi.org/10.1126/sciadv.1500852

Norman, M. D., Taylor, L. A., Shih, C.-Y., & Nyquist, L. E. (2016). Crystal accumulation in a 4.2 Ga lunar impact melt. Geochimica et CosmochimicaActa, 172, 410–429. https://doi.org/10.1016/j.gca.2015.09.021

Papike, J. J., Ryder, G., & Shearer, C. K. (1998). Lunar samples. Reviews in Mineralogy and Geochemistry, 36, 5.1–5.234.Phillips, R. J., Raubertas, R. F., Arvidson, R. E., Sarkar, I. C., Herrick, R. R., Izenberg, N., & Grimm, R. E. (1992). Impact craters and Venus resurfacing

history. Journal of Geophysical Research, 97(E10), 15,923–15,948. https://doi.org/10.1029/92JE01696Podosek, F. A., Huneke, J. C., Gancarz, A. J., & Wasserburg, G. J. (1973). The age and petrography of two Luna 20 fragments and

inferences for widespread lunar metamorphism. Geochimica et Cosmochimica Acta, 37(4), 887–904. https://doi.org/10.1016/0016-7037(73)90186-5

Richardson, J. E. (2009). Cratering saturation and equilibrium: A new model looks at an old problem. Icarus, 204(2), 697–715. https://doi.org/10.1016/j.icarus.2009.07.029

Robbins, S. J. (2014). New crater calibrations for the lunar crater-age chronology. Earth and Planetary Science Letters, 403, 188–198. https://doi.org/10.1016/j.epsl.2014.06.038

Robbins, S. J., Antonenko, I., Kirchoff, M. R., Chapman, C. R., Fassett, C. I., Herrick, R. R., et al. (2014). The variability of crater identificationamong expert and community crater analysts. Icarus, 234, 109–131. https://doi.org/10.1016/j.icarus.2014.02.022

Rutherford, M. J., Tonks, B., & Holmberg, B. (1996). Experimental Study of KREEP Basalt Evolution: The Origin of QMD and Granite at the Baseof the Lunar Crust. Lunar and Planetary Science, 27, 1113–1114.

Smith, D. E., Zuber, M. T., Neumann, G. A., Lemoine, F. G., Mazarico, E., Torrence, M. H., et al. (2010). Initial observations from the Lunar OrbiterLaser Altimeter (LOLA). Geophysical Research Letters, 37, L18204. https://doi.org/10.1029/2010GL043751

Solomon, S. C., & Head, J. W. (1980). Lunar mascon basins: Lava filling, tectonics, and evolution of the lithosphere. Reviews of Geophysics, 18(1),107–141. https://doi.org/10.1029/RG018i001p00107

Sood, R., Chappaz, L., Melosh, H. J., Howell, K. C., Milbury, C., Blair, D. M., & Zuber, M. T. (2017). Detection and characterization of buried lunarcraters with GRAIL data. Icarus, 289, 157–172. https://doi.org/10.1016/j.icarus.2017.02.013

Speyerer, E. J., Robinson, M. S., Denevi, B. W., & the LROC Science Team (2011). Lunar Reconnaissance Orbiter Camera global morphologicalmap of the Moon. Lunar and Planetary Science, 42. Abstract 2387

Spudis, P. D., Wilhelms, D. E., & Robinson, M. S. (2011). The Sculptured Hills of the Taurus Highlands: Implications for the relative age ofSerenitatis, basin chronologies and the cratering history of the Moon. Journal of Geophysical Research, 116, E00H03. https://doi.org/10.1029/2011JE003903

Stickle, A. M., & Schultz, P. H. (2011). Exploring the role of shear in oblique impacts: A comparison of experimental and numerical results forplanar targets. International Journal of Impact Engineering, 38(6), 527–534. https://doi.org/10.1016/j.ijimpeng.2010.10.025

Stuart-Alexander, D. E., & Howard, K. A. (1970). Lunar maria and circular basins—A review. Icarus, 12(3), 440–456. https://doi.org/10.1016/0019-1035(70)90013-8

Taylor, L. A., Shervais, J. W., Hunter, R. H., Shih, C. Y., Bansal, B. M., Wooden, J., et al. (1983). Pre-4.2 AE mare-basalt volcanism in the lunarhighlands. Earth and Planetary Science Letters, 66, 33–47. https://doi.org/10.1016/0012-821X (83)90124-3

van der Bogert, C. H., Hiesinger, H., Povilaitis, R. Z., Robinson, M. S., Meyer, H., & Ostrach, L. R. (2017). Regional lunar stratigraphy derived fromCSFDs extracted from the >5 km global crater catalog. Lunar and Planetary Science, 48. Abstract 1437

Warren, P. H. (1985). The magma ocean concept and lunar evolution. Annual Review of Earth and Planetary Sciences, 13(1), 201–240. https://doi.org/10.1146/annurev.ea.13.050185.001221

Warren, P. H. (2003). The Moon. In A. M. Davis (Ed.), Treatise on Geochemistry, Meteorites, comets, and planets (Vol. 1, pp. 559–599). Oxford:Elsevier.

Whitaker, E. A. (1981). The lunar Procellarum basin. In P. H. Schultz & R. B. Merrill (Eds.), Multi-ring basins: Formation and evolution (pp. 105–111).New York: Pergamon Press.

Whitten, J., Head, J. W., Staid, M., Pieters, C. M., Mustard, J., Clark, R., et al. (2011). Lunar mare deposits associated with the Orientale impactbasin: New insights into mineralogy, history, mode of emplacement, and relation to Orientale Basin evolution from Moon MineralogyMapper (M3) data from Chandrayaan-1. Journal of Geophysical Research, 116, E00G09. https://doi.org/10.1029/2010JE003736

Whitten, J. L., & Head, J. W. III (2015a). Lunar cryptomaria: Mineralogy and composition of ancient volcanic deposits. Planetary and SpaceScience, 106, 67–81. https://doi.org/10.1016/j.pss. 2014.11.027

Whitten, J. L., & Head, J. W. III (2015b). Lunar cryptomaria: Physical characteristics, distribution, and implications for ancient volcanism. Icarus,247, 150–171. https://doi.org/10.1016/j.icarus. 2014.09.031

Wieczorek, M. A., Neumann, G. A., Nimmo, F., Kiefer, W. S., Taylor, G. J., Melosh, H. J., et al. (2013). The crust of the Moon as seen by GRAIL.Science, 339(6120), 671–675. https://doi.org/10.1126/science. 1231530

Wieczorek, M. A., & Phillips, R. J. (2000). The “Procellarum KREEP Terrane”: Implications for mare volcanism and lunar evolution. Journal ofGeophysical Research, 105(E8), 20,417–20,430. https://doi.org/10.1029/1999JE001092

10.1029/2017JE005421Journal of Geophysical Research: Planets

EVANS ET AL. 1616

Wilhelms, D. E. (1976). Secondary impact craters of lunar basins. Proceedings of the Lunar Science Conference, 7, 2883–2901.Wilhelms, D. E. (1987). The geological history of the Moon, Prof. Paper 1348 (302 pp.). Denver, CO: U. S. Geological Survey.Wilhelms, D. E., Oberbeck, V. R., & Aggarwal, H. R. (1978). Size-frequency distributions of primary and secondary lunar impact craters.

Proceedings of the Lunar and Planetary Science Conference, 9, 3735–3762.Xiao, Z., & Strom, R. G. (2012). Problems determining relative and absolute ages using the small crater population. Icarus, 220(1), 254–267.

https://doi.org/10.1016/j.icarus.2012.05.012Zhong, S., Parmentier, E. M., & Zuber, M. T. (2000). A dynamic origin for the global asymmetry of lunar mare basalts. Earth and Planetary

Science Letters, 177(3-4), 131–140. https://doi.org/10.1016/S0012-821X(00)00041-8Zuber, M. T., Smith, D. E., Lehman, D. H., Hoffman, T. L., Asmar, S. W., & Watkins, M. M. (2013). Gravity Recovery and Interior Laboratory (GRAIL):

Mapping the lunar interior from crust to core. Space Science Reviews, 178(1), 3–24. https://doi.org/10.1007/s11214-012-9952-7Zuber, M. T., Smith, D. E., Watkins, M. M., Asmar, S. W., Konopliv, A. S., Lemoine, F. G., et al. (2013). Gravity field of the Moon from the Gravity

Recovery and Interior Laboratory (GRAIL) mission. Science, 339(6120), 668–671. https://doi.org/10.1126/science.1231507

10.1029/2017JE005421Journal of Geophysical Research: Planets

EVANS ET AL. 1617