Stratigraphy, evolution, and volume of basalts in Mare Serenitatis




Abstract– Fourteen major basaltic units in Mare Serenitatis have been identified and mapped from differences in TiO2 wt%. The ages of these units have been inferred from their crater densities and reference to isotopically dated Apollo samples. It has been found that FeO and TiO2 wt% of the units do not show any apparent trend with time. However, the oldest units have much greater variation in FeO and TiO2 wt% than younger ones. No lateral trend in the age of the basaltic units is apparent within the basin. A vertical profile of Mare Serenitatis has been produced based on the depth of basalt within impact craters. The minimum depth of basalt has been estimated where craters have not exposed underlying highland material. The profile has been used to estimate the minimum volume of basalt within the basin to be ≈500,000 km3.


Basaltic lava flows on the Moon are indicative of volcanic activity and knowledge of their age should help to constrain the volcanic history of the Moon. Information about the volume of basalt erupted might be used in later work to estimate the amount of heating that was required to produce them. This in turn may improve our understanding of the magmatic, stratigraphic, and thermal evolution of the Moon. Clementine images at visible and ultraviolet wavelengths (UVVIS) can be used in conjunction with the algorithms of Lucey et al. (2000) to map discrete basaltic lava units of the lunar surface. A number of lunar maria have been mapped already from Clementine data using techniques similar to those used in this work, e.g., Mare Humorum and southeast Oceanus Procellarum (Hackwill et al. 2006); maria Nubium and Cognitum (Bugiolacchi et al. 2006); maria Tranquillitatis and Fecunditatis (Rajmon and Spudis 2004); the lunar nearside maria (Hiesinger et al. 2000); the Mare Serenitatis/Mare Tranquillitatis border area (Bell and Hawke 1995). This article describes how this information was used to map individual basaltic lava flows in Mare Serenitatis.

Mare Serenitatis (28.0°N/17.5°E) is one of the most prominent basalt-filled basins on the lunar surface. It is of Nectarian age and has a diameter of 740 km (Wilhelms 1987). Discrete basaltic units can be identified on the basis of statistically significant differences in TiO2 wt% between potential units. Once identified, the relative ages of these units can be deduced from their crater densities. Estimated ages of these units can be inferred by reference to the data in Schultz and Spudis (1983) and from Basaltic volcanism on the terrestrial planets (BVTP).


Four Clementine images were used for this project: FeO wt%, TiO2 wt%, “true” color, and false color at a resolution of 200 m/pixel. They are shown in Figs. 1a–d, respectively. Their boundaries are: 5°E–35°E, 15°N–40°N and they were prepared using Integrated Software for Spectrometers (ISIS) by the Lunar and Planetary Institute using the algorithms of Lucey et al. (2000). The FeO and TiO2 wt% images were created with ISIS cubes (USGS 2007), which provided the average FeO and TiO2 wt% for each 200 m2 pixel. The “true” color image (Fig. 1c) was prepared as a composite of images taken through three Clementine camera filters: 415 nm, 750 nm, and 950 nm, which were allocated to blue, green, and red, respectively. This gave a reasonable color representation of the lunar surface. The false color image was created from ratios of filtered images: 415 nm/750 nm controls the blue, 750 nm/415 nm controls the red, and 750 nm/950 nm controls the green. The rationale for this image is explained in Pieters et al. (1994). This image emphasizes differences in composition and surface maturity. In Fig. 1d, red represents low TiO2 mature basalt; blue represents high TiO2 mature basalt; green and yellow represent freshly exposed basalt; cyan may indicate either fresh highland soil or fresh mare basalt.

Figure 1.

 Clementine images. a) FeO. b) TiO2. c) “True” color. d) False color.

Lunar Orbiter IV high-resolution images were used for counting craters. These were preferred to the Clementine images because (i) their resolution extends down to ∼60 m, compared to 200 m/pixel in the Clementine images, and (ii) the incidence angle of the Sun is lower than in Clementine images, making craters more easily discernable.

The project comprised six stages:

  • • Identify and delineate the basaltic units within Mare Serenitatis.
  • • Determine the relative ages of these units from their crater densities.
  • • Use the crater densities from Schultz and Spudis (1983) and BVTP data to infer the ages of the identified units.
  • • Identify any lateral trends in eruptions across the lunar surface with time.
  • • Identify any trends in FeO and TiO2 wt% values with time.
  • • Estimate the minimum volume of basalt within Mare Serenitatis.

Identification and Delineation of Basaltic Units within Mare Serenitatis

It has been known since Apollo days that wide variations in FeO and TiO2 wt% exist between different locations on the lunar surface (Wilhelms 1987; Basaltic Volcanism Study Project (BVTP) (1981); Haskin and Warren 1991). These variations can be used to map geological units (Lucey et al. 2000). For this work, potential unit boundaries were identified from the four Clementine 200 m/pixel images (Fig. 1). The boundaries of many units are apparent on the FeO, TiO2 wt%, and false color images. Several are less clear but were accepted initially because they could be rejoined with adjacent ones later if found not to be statistically different.

Twelve FeO and 12 TiO2 wt% measurements were taken at random points within each potential unit. After averaging each set, it was noted that TiO2 wt% showed much greater numerical variation between units than FeO wt% and so these samples were used to compare potential units. Units were considered to be discrete when a significant difference (one standard deviation) existed between them. Potential, adjacent units were regarded as being the same unit if a significant difference did not exist between them. It is recognized that the Lucey et al. (2000) titanium algorithm may have shortcomings (e.g., Gillis et al. 2003), but it was used here as it is satisfactory for making comparisons. Units were identified only by this numerical method. Differences in color on the false color image and shades of gray in the FeO and TiO2 images are subjective and were not considered once the numerical method had been used to determine the units.

Determining the Relative Ages of the Identified Units from Their Crater Densities

The relative ages of basaltic units can be inferred from their crater size-frequency distribution; in general, the greater the density, the older the units (Neukum et al. 1975; Neukum 1977). Adobe Photoshop was used for crater counting and image analysis in this work. High-resolution Lunar Orbiter IV images were used for crater counting. The scales of these images were obtained from Anderson and Miller (1971). The images were scanned at very high resolution to ensure that no significant detail of the ∼60 m/pixel maximum resolution of the Lunar Orbiter IV images was lost. Craters with diameters >500 m were used to assess crater densities. Samples of ∼500 m diameter craters on the scanned images were compared with those on the original Lunar Orbiter IV frames, and no loss of important detail was noted. Obvious secondary craters were ignored. Craters on the border of a sample area were only included if more than half of their area occurred in the sample area. Sun angle can affect the apparent number of smaller craters (100–600 m diameter) on the lunar surface (Young 1975). Data from Bowker and Hughes (1971) show the Sun angle varies by <3° in the Lunar Orbiter IV images used in this work and so there was no need for any compensation.

500 m was chosen as the minimum crater diameter because (i) a large number would exist in each unit giving greater numerical variation between units, (ii) the Schultz and Spudis (1983) graph was derived from 500 m diameter craters and this can be used to infer unit ages directly for units having densities <0.05 crater/km2, and (iii) it allows the Schultz and Spudis (1983) graph to be easily extended using data from BVTP for densities >0.05 crater/km2.

The smallest diameter that can be used is probably 500 m because it is approaching the minimum resolution of some Lunar Orbiter IV images used for this work. However, this diameter avoids the problems in counting very small craters (<300 m). These problems were noted by Schultz et al. (1976) as (i) many craters are difficult to measure because they are asymmetric, (ii) small primary craters easily become subdued with erosion making them difficult to distinguish from secondary ones, and (iii) smaller craters erode and disappear more quickly than large ones and this can distort the number of craters being counted (Young 1976).

Previous studies involving crater densities (e.g., Hackwill et al. 2005; Hiesinger et al. 2000; Bugiolacchi et al. 2006) have typically sampled 10–40% of the area of each basaltic unit being studied. Neukum et al. (1975) show that flooding, blanketing, secondary cratering, super-positioning, infilling, abrasion, mass wasting, and volcanic craters can cause local inhomogeneities in an apparently single unit. For this study, >80% of the area of the major units was sampled. This large percentage was used in a somewhat crude attempt to minimize the effects noted by Neukum et al. (1975). Unit 4 is an exception: the Bessel Ray, originating from Menelaus crater (Campbell et al. 1989) crosses this region. It was necessary to avoid sampling areas having an apparent clustering of secondary craters. Units were not sampled near their edges because the exact position of their borders was not always clear.

Most of unit 10 is in Mare Tranquillitatis but is included because it extends into the Serenitatis basin. The crater-counting sampling area for this unit was entirely in Mare Tranquillitatis because the portion in Mare Serenitatis contains a high density of secondary craters. These are presumed to have come from Dawes crater (17.2°N/26.4°E) and could distort the crater density if any were mistakenly included in the total.

The unit crater densities found in this work have been used in conjunction with those of Schultz and Spudis (1983) and BVTP (Table 8.8.1) in Fig. 2 to infer the ages of the basaltic units. It was important therefore to ensure that systematic differences did not exist in crater densities produced in this work and these other sources. To check this, crater densities of the Apollo 12 and 15 basaltic units were calculated using the method used here. It was confirmed that no significant difference existed between these results and those of the other sources. There was therefore no need to incorporate a “calibration” to the densities found here before using them with the graph in Fig. 2.

Figure 2.

 Ages of returned Apollo samples plotted against the crater density at the sites from which they were collected. Inferred ages of Copernicus and Tycho are plotted against the crater density in their ejecta blankets. Crater densities are based on craters of diameter >500m. The diagram is modified from Fig. 3 of Schultz and Spudis (1983). The Apollo 11 and Apollo 17 data have been added and the commonly accepted age of 109 ± 4Ma (Drozd et al. 1977) used in preference to 96 ± 5Ma quoted by Schultz and Spudis.

Derivation of Inferred Ages of Individual Basaltic Units

Many Apollo returned samples have been dated radiometrically (BVTP 1981; Stöffler et al. 2006). Schultz and Spudis (1983) plotted the ages of radiometrically dated samples from Apollo 12 and Apollo 15 against the >500 m diameter crater densities of the basaltic units from which they were collected. Silver (1971) deduced the age of Copernicus crater to be 0.85 Ga ± 0.1 Ga from similar exotic material found in both Apollo 11 and 12 regolith samples. Arvidson et al. (1976) noted that grooves among secondary crater clusters at the Apollo 17 landing aligned with the great circle path from Tycho. They interpreted these grooves as being created by ejecta from the Tycho impact. Cosmic-ray exposure dating of returned samples of these ejecta by Drozd et al. (1977) were found to have a mean age of 109 ± 4 Ma indicating that this is the age of the Tycho impact. The crater densities of Copernicus and Tycho were inferred from their ejecta blankets.

The densities of >500 m diameter craters at the Apollo 11 and 17 landing sites were deduced from data in BVTP (Table 8.8.1) and added to the Schultz and Spudis (1983) graph in Fig. 2. Schultz and Spudis (1983) show the crater density of >500 m diameter craters at the Apollo 15 site to be 5 × 10−2 km−2. The BVTP table quotes the Apollo 15 landing site as having a crater density of 0.5 relative to the lunar mare average; therefore, the average lunar mare crater density for >500 m craters is 10 × 10−2 km−2. The table quotes the Apollo 11 and 17 crater densities as 1.3 and 1.2 × the lunar mare average, respectively. It follows that the >500 m diameter crater densities at these sites are 13 × 10−2 km−2 and 12 × 10−2 km−2, respectively. The Apollo 12 crater density relative to the average lunar maria quoted in BVTP is incorrect (Spudis, pers. comm.). Spudis has resurveyed the area and found the density to be 4.8 × 10−2 km−2.

Straight lines can be drawn almost perfectly through the data points in Fig. 2 suggesting that a correlation exists between crater densities and absolute ages. The sharp change in direction at the Apollo 15 point is unrealistic, but it has been retained because the crater density and age data produce it. A more realistic curve can be derived within the margins of error that are included in Table 1. The steep decline in crater densities before the Apollo 15 point is attributed to the rapid decline in late heavy bombardment activity.

Table 1.   Margins of error for data points in Fig. 2.
Apollo 113760 ± 50Apollo 123170 ± 60
Apollo 173700 ± 50Copernicus 850 ± 100
Apollo 153290 ± 50Tycho 109 ± 4

Estimation of the Minimum Volume of Basalt within Mare Serenitatis

Small, bowl-shaped craters (<20 km diameter) have a diameter to depth ratio of about 10:1 and a similar morphology (Croft 1980; Melosh 1989; Hörz et al. 1991). If an impact penetrates through to the highland, then the crater ejecta will be a mixture of highland material and basalt. A small amount of impact melt created during the basin impact may be present but is exceedingly difficult to distinguish using remote sensing data and so has not been considered in these calculations. If the highland/basalt ratio of the ejecta can be estimated, and the shape of the crater is known, it should be possible to calculate the depth of the highland/basalt contact that existed before the impact and thus the depth of the basalt in the area close to the crater. Each pixel on the FeO wt% map represents 200 × 200 m of the lunar surface. This scale can be used to represent the diameter of craters.

The ejecta around a crater that exposes the highland beneath have two endmembers: basalt and highland. The ejecta mixture can be expressed as:


where: E = percentage of FeO in the ejecta. It is recognized that the ejecta from a crater forms layers close to the crater (Stöffler et al. 1975; Melosh 1989; Hörz et al. 1991). Also, Oberbeck (1975) demonstrated that considerable gardening of ejecta with existing surface material occurs in the distal area of the blanket. The outer and inner 20% radius of the ejecta blanket were avoided during sampling to reduce skewing of the results by these effects; M = FeO wt% in the mare (measured from nearby fresh basalts in the same unit as the crater); H = FeO wt% in highland material (measured from nearby highlands using samples that have been freshly exposed by impacts); X = the proportion of basalt in the ejecta (expressed as between 0 and 1.0).

The formula can be rearranged to:


A spherical cap is a geometrical shape that is a portion of a sphere cut off by a plane. An inverted spherical cap having a 10:1 diameter to depth ratio may be regarded as being analogous to the shape of a simple crater. Hackwill (2006) developed a mathematical method to calculate the depth of basalt ejected from simple, spherical cap-shaped craters. This was used for the six craters in Mare Serenitatis where the impact has penetrated through the basalt to the highland.

Where a crater does not expose the underlying lunar highland, the 10:1 diameter/depth ratio can be used to infer the minimum depth of basalt. It has been assumed that Mare Serenitatis is symmetrical around its center. In Fig. 3, a profile is suggested for basin floor based on number of data points. The depths of basalt for the six craters that have penetrated through to the highland are plotted as red triangles. It was not possible to estimate the depth of basalt nearer the center of the basin using this method because no craters have pierced through to the highland here.

Figure 3.

 Crater depths and ALSE data point displayed as radial distances from the center of Mare Serenitatis (27°N, 17°E). Blue diamonds represent minimum basalt depths; red triangles show the estimated depth of the basalt-highland boundary inside craters; the red circle indicates the depth of basalt 50km south of Bessel as inferred from the ALSE data; the blue square shows the minimum depth of basalt at the Bessel pre-impact target site. The black line indicates the probable minimum depth of the basin floor.

The Apollo 17 lunar sounding experiment (ALSE) returned sub-surface radar data along a path across southern Mare Serenitatis. This passed about 50 km south of Bessel. Peeples et al. (1978) interpreted the basalt-highland contact to be about 1.5 km deep and provides another basalt-depth data point.

In Fig. 3, the minimum depths of craters that have not exposed highland are plotted as blue diamonds. The black line is a probable minimum-depth profile of the basin suggested by the crater data points. This is based on the radius of Mare Serenitatis being ∼390 km (Kodama and Yamaguchi 2002), the “actual” basalt depth data points, and the minimum depth point for Bessel.


Based on the basin profile in Fig. 3, the minimum volume of basalt in Mare Serenitatis is estimated to be ≈500,000 km3.

Figure 4 shows the basaltic units that were identified within Mare Serenitatis. The ages are given for those units that have sufficient craters from which a meaningful age can be inferred. Each of these has informally been assigned a number for the purpose of this discussion and the same numbers are used in Table 2. A few units are shown without ages. These have too few >500 m diameter craters from which to infer an age because it was found that the addition or subtraction of a few craters from the count made huge difference to its age. For example, a reduction of two craters in a small 150 km2 unit would reduce its inferred age by ∼1 Ga. Areas of dark mantling deposit and extensive crater ejecta have been annotated “dmd” and “e,” respectively. The Apollo 17 landing site (A17) is also indicated.

Figure 4.

 False color image of Mare Serenitatis and the adjacent maria (5°E – 35°E, 15°N – 40°N). The units are delineated and numbered. Those for which an area >500km2 could be crater-counted have their inferred age shown (margins of errors are given in Table 2). White areas are highlands. “e” indicates ejecta. Estimated minimum thicknesses (in meters) of basalt inside craters that do not expose the underlying highland are shown in yellow. Estimated thicknesses (in meters) of basalt inside craters where the impact has exposed the underlying highland are shown in yellow and are underlined. Areas of dark mantling deposits are labeled as “dmd.” The approximate path of the Apollo 17 ALSE is indicated.

Table 2.   Average FeO, TiO2 wt% (including the margin of error associated with the Lucey et al. [2000] algorithms), and inferred ages of the 14 major basaltic units in Fig. 4.
UnitNo. of craters >500 m diameterArea/km2 sampledCraters/km2Inferred age/GaFeO ± 1.3% wt%TiO2 ± 1% wt%
  1. aUnit 10 is part of Mare Tranquillitatis.

  2. bUnit 4 contains a very large number of secondary craters; see the Discussion section.

1422068010.032 ± 0.0022.1 ±
13298210.035 ± 0.0072.3 ±
12358300.042 ± 0.0072.9 ±
11459560.047 ± 0.0073.2 ±
10a357080.049 ± 0.0083.3 ± 0.119.313.3
927355580.049 ± 0.0033.3 ±
828454090.053 ± 0.0033.3 ±
79919780.050 ± 0.0053.3 ±
67213340.054 ± 0.0063.3 ±
515428510.054 ± 0.0043.3 ±
4b53188050.060 ± 0.0033.3 ±
39616650.058 ± 0.0063.3 ±
224241780.058 ± 0.0043.3 ±
131141510.075 ± 0.0043.4 ±

Table 2 lists (for the units that could be dated) the number of craters counted in each sample area, the size of this area, the crater density, its inferred age and average FeO and TiO2 wt%.


All the units dated in this project, except units 13 and 14, are of Upper Imbrian age although unit 7 is on the Upper Imbrian-Eratosthenian boundary (3.2 Ga). Units 13 and 14 are Eratosthenian. The ages that have been inferred for these units are dependent on the six data points in Fig. 2 and the two straight lines that have been drawn through (or very close) to them being correct. This graph has been used to summarize 2.8 Ga of the Moon’s impact history. It is possible that the actual trend varied considerably from these straight lines, particularly during the 2.4 Ga between the Apollo 12 and Copernicus data points. This is unlikely to be resolved until samples are returned that can be used to constrain the crater density/age relationship during this period.

In this work, adjacent units have been accepted as being discrete by statistical differences. The Lucey et al. (2000) TiO2 wt% data has a margin of error of ±1%. A few units, e.g., 5, 11, and 12 are not discrete if the error is included. However, in each case, the difference in ages indicates that they are distinct. Units 7 and 8 have <1% difference in TiO2 wt% and are of the same age (3.3 Ga) but are not adjacent on the surface. It is possible they are the same unit if unit 4 has flowed over parts of them.

McCauley and Wilhelms (1971) mapped the lunar nearside using Lunar Orbiter and Apollo imagery. They note an area on the eastern rim that corresponds approximately with unit 14 and also deduce that it is a young unit. The outlines of their other units broadly match units identified here, but, in this work, they have been subdivided and so comparison of ages is not straightforward. Wilhelms (1987) also noted the unit in the east to be young but again subdivision of his major units in this work makes comparison of ages difficult. Howard et al. (1973) and Pieters (1978) mapped major basaltic units in Mare Serenitatis with Earth-based observations. In this work, several of their units have been sub-divided and smaller ones found using Clementine data. Hiesinger et al. (2000) used data from Galileo to map units in Mare Serenitatis. Many of the unit boundaries that they noted correspond with the ones identified in this work. They have subdivided some of these into smaller units, but evidence for a number of them is not apparent in the four Clementine images used here. For example, they subdivide unit 4 (in this work) into their units S15 and S28, but the Clementine false color image indicates their unit S15 to be an area of extensive secondary cratering, part of which is the Bessel Ray. In this work, no significant difference in TiO2 wt% was found between these areas and so they have been interpreted as a single unit. The unit that Hiesinger et al. (2000) identified as their unit S2 is interpreted as an area of dark mantling material in this work and by Weitz et al. (1998).

Both Howard et al. (1973) and Hiesinger et al. (2000) interpret the area adjacent to Dawes as being a discrete unit although Howard et al. only note it to the northwest of this crater. Staid et al. (1996) analyzed Galileo UV/VIS data to identify basaltic units in Mare Tranquillitatis and interpreted an area corresponding to unit 10 in this work and extending down to the southern and eastern edges of Tranquillitatis as a single unit. The Clementine TiO2 wt% image reveals this area to be extensively covered with ejecta from Dawes; it extends to unit 4 and the highland in the east. Clementine images do not indicate a contact between two units and so the interpretation here is that unit 10 extends around Dawes and is overlain with ejecta from it rather than there being another unit.

A large area of dark mantling material exists around the Apollo 17 landing site (Hawke et al. 1990). Dark mantling consists of unconsolidated material and craters within it degrade more rapidly than those in the maria (Lucchitta and Sanchez 1975). It will therefore probably have fewer craters than maria of similar age and so it can not be dated using Fig. 2. This figure shows that unit 10 is estimated to be 3.4 Ga. Returned samples of basalt from the Apollo 17 mission have crystallization ages of 3.75 Ga (Stöffler and Ryder 2001) and so if the inferred age of unit 10 in this work is correct, the mantling material must obscure a boundary between unit 10 and the unit in which the Apollo 17 site is located.

Where two units are of the same age but geographically separated by a younger unit between them, it is possible that they are the same unit but have been overlain by the younger unit. All units of the same age were tested against each to test if their TiO2 wt% was significantly different from the others using the method described above. All were found to have significantly different TiO2 wt% and so it is concluded that each unit in Fig. 4 is discrete.

The ages of eruptions do not suggest a clear lateral trend in any direction across the mare surface with time. Units 6, 7, and 14 in the east are younger than many of the others but do not indicate a trend. The average FeO and TiO2 wt% of units suggest a trend in any lateral direction either. This result supports the findings of Hiesinger et al. (2001) who carried out a similar investigation of more than ∼220 units in 9 mare regions.

Figure 5a shows the average FeO wt% of each unit plotted against its age. There is no clear trend in FeO wt% with age. However, there is a significant reduction in variation between the lowest and highest FeO wt% with age: the greatest variation being at 3.4 Ga (15.9 to 19.3 wt%). In a study of Mare Humorum and SE Procellarum, Hackwill et al. (2006) found that here too the widest variation in FeO wt% occurs in the earliest eruptions (12.5 to 18.6 wt%). Subsequent eruptions had relatively high FeO (∼18.5 wt%). Later eruptions in Mare Serenitatis also have relatively high FeO wt%.

Figure 5.

 a and b) Evolution of FeO and TiO2 wt% for the 14 units that have been dated. c) FeO plotted against TiO2 wt% for all 14 units.

Figure 5b shows the average TiO2 wt% of each unit plotted against its age. The data point at 3.3 Ga and 13.3 wt% represents unit 10, which has an exceptionally high TiO2 wt% value. Most of its area is within Mare Tranquillitatis and, as explained earlier, was included because a small part of its area extends into Mare Serenitatis. There is little evidence for any trend in TiO2 wt% with time. However, if unit 10 is excluded, there is a similarity between these results and those found by Hackwill et al. (2006) for basaltic units in Mare Humorum and southeast Oceanus Procellarum. In both cases, there is wide variation in TiO2 wt% before 3.2 Ga and this variation is reduced with later eruptions. There are only two data points after 3.2 Ga and so this observation of Mare Serenitatis may not be significant.

Figure 5c shows that a correlation exists between FeO plotted against TiO2 wt% for all 14 units. Similar characteristics were found by Hackwill (2006) for units in Mare Humorum and southeast Procellarum using Clementine data and also by Haskin and Warren (1991) from Apollo returned samples.

The six “actual” basalt depth and ALSE data points in Fig. 3 generally slope toward the center of the basin. This is realistic and gives confidence that they represent the outer slope of the basin. From this profile, it has been estimated that the minimum volume of basalt in Mare Serenitatis is ≈500,000 km3. No previous estimates have been found in the literature. However, the estimation made here rests very heavily on the minimum basalt depth data point of Bessel because, unfortunately, there are no impacts or other features nearer the center of the basin to provide additional data.

Other researchers have employed various methods to estimate the thickness of basalt in Mare Serenitatis. DeHon and Waskom (1976) estimated the thickness of basalts in southern Serenitatis from measurements of partially buried craters to be at least 500 m. Hörz (1978) disputed these results, pointing out that crater degradation had not been considered and that the thicknesses suggested by DeHon and Waskom should be halved. Williams and Zuber (1998) used an empirical relationship to indicate that the depth of basalt in Mare Serenitatis is 4.30 ± 0.33 km, which is considerably greater than the estimation in this paper of a minimum depth of ∼2.4 km. However, their result does not conflict with depth suggested in this work because this is an estimation of minimum depth while Williams and Zuber estimate the actual depth. Arkani-Hamed (1998) used Clementine gravity data to suggest the depth of basalt is ∼3 km (although it is difficult to estimate the depth accurately because of the scale of the cross section through the basin on his diagram). Bratt and Solomon (1982) developed a crustal structural model for lunar basins from which they produced a cross section of the Mare Serenitatis basin. The model assumed the pre-mare basin to be isostatically compensated and having a crustal thickness based on Apollo seismic data. Measurement from their diagram (although the scale is small) suggests the basalt in the center of the basin to be ∼3 km thick. This is compatible with the minimum depth estimation in this work of ∼2.4 km.

There are two potential statistical difficulties with this work. First, all the unit ages have been inferred from only six data points (Fig. 2) and two of these, Copernicus and Tycho, are based on assumptions made from rock fragments in returned samples (Arvidson et al. 1976; Silver 1971). Second, there are no dated samples between the Apollo 12 and Copernicus data points. This represents ∼2.3 Ga for which there are no dated samples. This is a period of about half the Moon’s existence and so the assumption that the impact rate declined at a linear rate between these two data points may not be valid.


This project has shown that volcanism occurred at least from 3.4 to 2.2 Ga in Mare Serenitatis. The oldest unit identified in this work is younger than the Apollo 17 returned samples. This implies that the Apollo 17 samples came from different basaltic units to any unit identified in this work.

This is the first time craters as small as 500 m have been used to estimate the ages of basalts in Mare Serenitatis. Previous crater density surveys have used larger craters, e.g., Hiesinger et al. (2000) and Neukum et al. (1975) used 1 km and 0.8 km, respectively. The greater number of craters counted should produce a more accurate result. The area sampled in this study for each unit was considerably larger than those used in previous studies in an attempt to reduce the effects of inhomogeneities of crater densities.

The sequence of eruptions does not appear to have taken any lateral direction across the lunar surface. There is no evidence of any increasing or decreasing trends in FeO and TiO2 wt% with time although the results show that variation in FeO and TiO2 wt% is much greater between earlier eruptions. These results therefore do not provide any evidence of systematic lateral variation in FeO or TiO2 wt% in the mantle source region. They also imply that, in the Serenitatis region, they either erupted from a source that had an FeO and TiO2 content that was not evolving in a particular direction or the eruption came from a number of sources having different FeO and TiO2 content.

The minimum volume of basalt within Mare Serenitatis has been estimated to be ≈500,000 km3 and this appears to be the first time an estimation of the minimum volume has been made.

Hackwill et al. (2005) suggested the minimum volume of basalt within Mare Humorum. The intention is to extend this project to the other maria. It is hoped that suitable methods will be employed in future work to estimate the amount of heat required to produce the total volume of lunar basalt and this may infer the amount of radioactive material in the young Moon.

A similar project to this has also been completed for Mare Humorum and SE Procellarum (Hackwill et al. 2006) and it is intended to extend this work to the other lunar maria.


The author thanks Brian Fessler at the Lunar and Planetary Institute, Houston, for preparing the images and also the anonymous reviewers who provided invaluable help with the script.

Editorial Handling

Timothy Jull