NASA Ames Research Center, Space Science Division, M/S 245-3,

Moffett Field, CA 94035-1000, and

Hawaii Institute of Geology and Planetology, School of Ocean, Earth, and Atmospheric Sciences,

University of Hawaii, 2525 Correa Rd., Honolulu, HI 96822

phone: 415-604-0324; fax: 415-604-6779




Hawaii Institute of Geology and Planetology, School of Ocean, Earth, and Atmospheric Sciences,

University of Hawaii, 2525 Correa Rd., Honolulu, HI 96822

phone: 808-956-6488; fax: 808-956-6322

Submitted to Icarus: September 15, 1994.

Revised: .

Pages: 30

Figures: 6

Tables: 2

1Now at: Center for Radiophysics and Space Research, Cornell University, Ithaca, NY 14853

Icarus MS# 6193 (revised)

Corresponding author:

Before June 15, 1995:

Dr. James F. Bell III

NASA Ames Research Center

Space Science Division, M/S 245-3,

Moffett Field, CA 94035-1000

phone: 415-604-0324; fax: 415-604-6779


After June 15, 1995:

Dr. James F. Bell III

Center for Radiophysics and Space Research

Cornell University

Ithaca, NY 14853

phone: 607-255-8542; fax: 607-255-9002


Proposed running head:



We present results from a combined CCD multispectral imaging and near-IR spectroscopic study of a region near the border between the Mare Serenitatis and Mare Tranquillitatis lunar impact basins. We have detected and spatially mapped variations in mare TiO2 abundance, in the degree of spectral maturity of impact craters and their ejecta blankets, and in the level of fresh highlands contamination within mare units. Techniques employed include color ratio images, image-oriented linear mixing models, and spectral band analyses. Specific findings include the detection of relatively fresh highlands materials in the craters Tacquet and Sulpicius Gallus and in the central peak and floor of the crater Plinius, the detection of low-TiO2 Serenitatis-like mare materials in a confined region of the ejecta from the crater Dawes, and the determination of immaturity rather than composition as the origin of the high albedo in the rectangular deposit surrounding Dawes. The combination of high spatial resolution imaging and high spectral resolution spectroscopy and the use of synthesizing techniques such as mixture modeling will provide valuable tools to optimize the scientific return from telescopic, Galileo, Clementine, and other future lunar data sets.


The region of the Moon near the border between Mare Serenitatis and Mare Tranquillitatis is one of the most geologically and compositionally complex areas of the nearside (Figure 1). The geologic history of this region has been shaped by impacts of widely-varying spatial scale and temporal occurrence, by volcanism of variable style and composition with time, and by limited tectonism. We have been studying this region as part of a larger multi-technique effort to understand the composition, morphology, geology, and stratigraphy of the Moon at spatial scales of 2 km or less (cf. Bell and Hawke 1991,1992; Hawke et al. 1992,1993; Campbell et al. 1992) and to provide additional scientific input towards the development of future scientific- and resource-driven exploration of the Moon. We have been aided in this effort by the proximity of this area to the Apollo 11, 15, and 17 landing sites and by the occurrence of one of the primary lunar spectroscopic "standard areas" within our scene. Here, we report the findings from the multispectral imaging and spectroscopy part of this effort.


Photogeologic Mapping and Other Remote Sensing Studies

Initial geologic mapping of the Serenitatis/Tranquillitatis border region was conducted by Carr (1966) and Morris and Wilhelms (1967) using telescopic observations and photographs. These authors divided the mare regions into four albedo units and suggested that the lower the albedo, the younger the unit. Materials in the Haemus Mountains were mapped as regional assemblages of Fra Mauro Formation, crater ejecta, and dark Serenitatis-like mare materials. Several pyroclastic units were proposed, including the Tacquet and Sulpicius Gallus Formations in Serenitatis and the Cayley formation (a smooth, light plains unit) in the highlands west of Tranquillitatis (Fig. 1).

The first post-Apollo effort to synthesize spectral, albedo, compositional, and radar data for this region was made by Thompson et al. (1973), who mapped 5 distinct surface types ranging from relatively young, relatively high albedo, red [high ultraviolet (UV) to visible (VIS) spectral slope], radar-bright, low TiO2 mare in eastern Mare Serenitatis to old, blue (lower UV to VIS slope), radar-dark, high TiO2 mare in Mare Tranquillitatis. Howard et al. (1973) used Apollo orbital photography to establish stratigraphic age relationships among the many units they defined in this region. The dark ring of outer Serenitatis mare material was interpreted as the unburied portion of an older, faulted mare surface, effectively reversing the albedo vs. age correlation postulated by previous workers. More recently, Wilhelms (1980) mapped much of the Serenitatis/Tranquillitatis region using mostly Apollo 15 mapping camera frames. He defined six mare units and several dark mantling units in this area based on color, stratigraphy, and cratering age. Although there is broad agreement between this map and previous efforts, the proposed units and their spatial boundaries are mapped in much greater detail owing to the superior spatial resolution of the mapping camera frames as compared to the earlier groundbased photographs.

Multispectral Imaging and Reflectance Spectroscopy

Johnson et al. (1975, 1977) used telescopic observations with a vidicon camera and four narrowband filters to define six spectral units along the Serenitatis/Tranquillitatis border region. Their filter selection allowed them to make inferences on the TiO2 content of mature mare soils using the spectral ratio technique developed by Charette et al. (1974), on the relative maturity of mare regions based on the relative strengths of the 1-um absorption feature (Adams and McCord 1971), on the distribution of dark mantling materials surrounding Serenitatis, and on the spatial identification of spectrally anomalous units (e.g., Sulpicius Gallus crater). McCord et al. (1976) used a similar multispectral imaging system to search for compositionally distinct color units over a much larger region of the nearside, and they derived results for Serenitatis and Tranquillitatis that were in excellent agreement with those of Johnson et al. (1975).

McCord et al. (1972) provide a comprehensive review and analysis of VIS to near-IR lunar reflectance spectra for many different regions on the nearside. Their classification scheme is broken down into four primary categories: mature and immature mare, and mature and immature highlands. Each class has distinct spectral characteristics related to variations in Ti, glassy agglutinates, and pyroxene content. This classification scheme was extended and enhanced by data farther into the near-IR by Pieters (1978, 1986) and McCord et al. (1981). Pieters (1978) used all imaging and spectroscopic data available to date to identify and map the distribution of mare basalt types on the entire nearside. This map depicts three mare units in the Serenitatis/Tranquillitatis region: (1) "standard" central Serenitatis mare of intermediate albedo and exhibiting strong 1-um and 2-um features; (2) low albedo, 2 to 5% higher UV/VIS ratio (relative to the standard unit) mare exhibiting a weak 1-um band and located in a dark partial ring around the outer portion of Mare Serenitatis; and (3) low albedo, > 5% higher UV/VIS ratio mare located in Tranquillitatis and the dark SE Serenitatis ring and exhibiting a weak 1-um band and a weak or no 2-um band.

More recently, high precision telescopic and spacecraft CCD cameras have been used to obtain images of both large and small lunar regions through narrowband filters in the 0.35 to 1.1 um region. Many of these data sets have focused specifically on the complex Serenitatis/Tranquillitatis region. Jaumann and Neukum (1989) and Jaumann (1989, 1991) used CCD multispectral imaging along with higher spectral resolution point spectra and comparisons to returned Apollo samples to spatially map compositional variations in Serenitatis and Tranquillitatis. Johnson et al. (1991a) and Melendrez et al. (1994) used multispectral imaging to create spectacular near-side TiO2 abundance and soil maturity maps at spatial scales between 1.2 and 5 km/pixel, paying particular attention to the high-TiO2 mare basalts in Tranquillitatis. Bell and Hawke (1991) and Campbell et al. (1992) used a synthesis of high spatial resolution radar and multispectral CCD data sets to examine the compositional variations and physical nature of lunar crater rays, including the so-called Bessel ray passing through Serenitatis. Hawke et al. (1992) used multispectral images and reflectance spectra as the basis for a general spectral survey of the Serenitatis basin region. Hiesinger et al. (1993) used a combination of telescopic and Galileo spacecraft multispectral images in the Serenitatis/Tranquillitatis region to define mare units near and around the Apollo 17 landing site. Finally, Staid et al. (1994, 1995) have also used the Galileo images combined with new Clementine imaging data and a mixing model analysis to assess spectral variations and stratigraphic implications within Tranquillitatis and its border near Serenitatis. Although all of the observations for other regions of the Moon cannot be summarized here, it is clear that major advances have been made in the past five years in the collection and analysis of high precision lunar multispectral imaging data, and that the Serenitatis/Tranquillitatis region has been a particularly interesting target of many of these studies.


Multispectral CCD Imaging

We obtained telescopic multispectral CCD images of the Serenitatis/Tranquillitatis border region (Fig. 1) using a 384 x 576 pixel Photometrics cooled CCD camera. The camera and an uncooled narrowband filter wheel were mounted to the University of Hawaii Planetary Patrol 61-cm telescope at Mauna Kea Observatory. Images were obtained with 8 narrowband transmission filters, chosen to maximize the potential spectral differences between the various units in the region as well as to provide details on the strength and shape of the 1-um lunar absorption band. The central wavelengths (in um) of these filters were 0.370, 0.400, 0.733, 0.866, 0.900, 0.933, 0.966, and 1.000, and all had a 0.03 um bandwidth. Exposure times (in sec) for the lunar images through these filters were: 8.0, 2.0, 0.8, 0.4, 0.4, 0.6, 1.0, and 2.0. Observations were carried out on August 7 UT, 1990, at a lunar phase angle of 10deg.. The Planetary Patrol telescope was configured at f/13.5, yielding a plate scale of 0.573 arcsec per 23 um CCD pixel. This translates to ~1.9 km/pixel resolution in our study region and a 200 x 310 arc sec field of view. The boundaries of our study region are shown in Figure 1.

Reflectance Spectroscopy

The reflectance spectra presented here were obtained using the University of Hawaii Planetary Geosciences Division near-IR circular variable filter (CVF) spectrometer with a cooled InSb detector. The data were obtained at a spectral resolution of [[Delta]][[lambda]]/[[lambda]] ~ 1.25% over the wavelength range 0.65 to 2.6 um using the University of Hawaii 2.24-m telescope at Mauna Kea Observatory. The CVF was stepped through 120 wavelengths over the course of about 90 seconds, and anywhere from 2 to 6 independent measurements were made of each region. Observations were carried out on October 29 and 30 UT, 1985, at a lunar phase angle of 10deg. and 20deg. respectively. The 2.24-m telescope was configured at f/35, yielding at spot size of 1.3 km diameter on the Moon using the smallest aperture and assuming 0.65 arcsec seeing. Spectra were obtained of 20 spots in the Serenitatis/Tranquillitatis region (Table 1 and Fig. 1; see also Table 37 in Jaumann 1989). Most of the spectra were obtained through the smallest aperture (1.3 km), but a few were obtained using a larger aperture having an effective ground spot size of 2.9 km. Additional spectra of selected regions in the Serenitatis/Tranquillitatis study area (obtained earlier with the same instrumentation; Table 1, spots 21-26) have been obtained from the University of Hawaii/NASA Pacific Regional Planetary Data Center lunar spectral archives.


Multispectral Imaging

Multiple CCD exposures of our study region were made through each filter in order to "freeze" the atmospheric seeing, and the images exhibiting the highest spatial resolution were saved for analysis. Numerous calibration frames were obtained in association with each of these images. These included: (1) dark current frames recorded at the same exposure times as the Moon images in order to correct for bias and dark current offset; (2) flatfield images using both the lighted interior of the dome and the twilight sky in order to correct for pixel-to-pixel nonuniformities on the chip; and (3) images of the sky just off the lunar limb recorded just after the surface imaging data were obtained in order to correct for any possible scattered or stray moonlight in the images.

The data reduction process included subtraction of the dark (bias + dark current) values from all images and then division of a normalized (mean = 1.0) flatfield frame from each Moon and sky image at the appropriate wavelength. The sky images showed signal levels of only a few DN above the bias level and no spatial structure, as the night was clear and the telescope and instrument were well baffled. Nonetheless, the residual sky image values were subtracted from the Moon image for each filter. Finally, the multispectral images were spatially co-registered into an image cube. This was achieved using a flux-conserving procedure whereby each image is convolved with a sinc2 function having a width consistent with the estimated point spread function of the telescope. We estimate that this procedure yields a registration accuracy of 0.25 pixel.

The resulting pixel values are thus a scaled representation of the lunar flux times the product of the Earth's atmospheric transmission and the instrument (filter + detector + camera and telescope optics) efficiency. All of the factors besides the lunar flux are effectively removed by ratioing all the values in the image by the value of a "standard" area in the same scene obtained at the same time. For this analysis, we chose the well-studied lunar standard area in southeast Mare Serenitatis known as MS2 (18deg. 40' N, 21deg. 25' E), so that we could compare spectra generated from these multispectral images with previously-obtained spectroscopic data sets also presented as spot/MS2 ratios (e.g., McCord et al. 1972).

Analysis techniques used on these images included (1) generation of simple ratio images between different wavelengths in order to examine spectral slope and absorption band variations; (2) extraction of 8 channel spectra ("pseudospectra") from the image cube for comparison with other spectroscopic data sets; and (3) creation of simple linear mixing models using a small number of endmembers. We followed the mixing technique of Adams et al. (1986, 1993) and Possolo et al. (1994) for spatial mapping of spectral heterogeneities and for the analysis of compositional variations focusing on specific processes (e.g., soil maturation and highlands-mare mixing).


Multiple independent measurements of the near-IR spectrum of each region were averaged to produce a final "raw" spectrum with associated error bars representing a combination of instrumental noise and positional accuracy. The spectra were converted into scaled reflectance units by taking a ratio of each region relative to the Apollo 16 region and then multiplying by the ratio of Apollo 16 to the Sun derived from spectral studies of returned Apollo 16 lunar samples (Adams and McCord 1973). For many of these spectra, linear continua were removed by fitting straight lines to the data at local tangent points near 0.7-0.8 and 1.4-1.6 um. Alternately, for comparison with pseudospectra derived from the multispectral images of this region, many of the spectra were converted to relative reflectance after dividing out the spectrum of the standard MS2 mare region within Serenitatis. This technique amplifies the subtle differences between spectra of different units in this region.

The primary focus of our spectral interpretations is an absorption feature near 1.0 um that occurs in spectra of many regions of the Moon. This particular absorption band, which is one of the primary spectral benchmarks in lunar remote sensing, can provide diagnostic information on the pyroxene and olivine mineralogy of lunar basalts, on the abundance of certain ferrous glasses and/or glassy agglutinates, and on the maturity or spectral "freshness" of lunar soils. Determination of the band minimum positions, depths, and widths for the 1-um absorption feature in these spectra was performed using standard algorithms (e.g., McCord et al. 1981; Clark and Roush 1984; Lucey et al. 1986). We define continuum slope as the slope of a linear segment between the continuum points used to define the minimum, depth, and width parameters of the band. We define band skew as the ratio of the half-width at half-maximum (HWHM) values from the left and right sides of each band. Thus, a symmetric band has skew = 1, while a band with more area at its longest wavelengths relative to its shortest wavelengths will have skew < 1. Because the spectra do not extend to long enough wavelength and also because thermal effects begin to complicate the lunar spectrum longward of 2.0 um, the accuracy of our determinations of parameters for the very broad lunar 2-um absorption band is questionable and thus we do not use the 2-um band values in our analysis.


Multispectral Imaging

Color Ratio Images. We have attempted to define spectral and compositional units using ratios of registered images at different wavelengths following techniques described in many of the references cited earlier. Figure 2 shows examples of four ratio images generated.

The 0.40 / 0.73 um ratio image (which we call [[Rfraktur]]1; Figure 2a) shows a substantial degree of heterogeneity related to variations in mare basalt TiO2 content (e.g., Charette et al. 1974; Johnson et al. 1991a,b). Serenitatis and Tranquillitatis mare regions are divided into three main units: (1) a large low ratio region ([[Rfraktur]]1 = 0.95 to 1.00) in central Serenitatis; (2) an irregularly-shaped moderate ratio region ([[Rfraktur]]1 = 1.05 to 1.10) in southwest Serenitatis and along the boundary between Mare Serenitatis and the Haemus mountains; and (3) a large high ratio region ([[Rfraktur]]1 = 1.10 to 1.20) in northwest Mare Tranquillitatis.

The 1.00 / 0.40 um ratio image ([[Rfraktur]]2; Figure 2b) shows a very different kind of heterogeneity related to differences in the strength of the 1-um band, changes in "blueness" or 0.40 um reflectance caused by mare TiO2 variations, and changes in continuum slope across the VIS-NIR. The lowest ratio values ([[Rfraktur]]2 = 0.68 to 0.89) are associated with both large and small impact craters. The lowest non-crater ratio values ([[Rfraktur]]2 = 0.86 to 0.91) are spatially confined to the relatively high-albedo, rectangularly-shaped deposit that surrounds the crater Dawes. The highest ratio values ([[Rfraktur]]2 = 1.07 to 1.13) occur in the highlands south of the Haemus mountains and in a N-S trending band within Mare Serenitatis just west of the crater Bessel. The remainder of Serenitatis and Tranquillitatis are characterized by relatively uniform, moderate values of [[Rfraktur]]2.

The 1.00 / 0.73 um ratio image ([[Rfraktur]]3; Figure 2c) is sensitive to the VIS to near-IR continuum slope as well as the strength of the 1-um band. This ratio shows the best contrast among the various rays and splotches that occur within Mare Serenitatis. The most conspicuous such feature is associated with the high albedo ray that passes through the crater Bessel. The value of [[Rfraktur]]3 for the ray ranges from 0.96 to 0.97, as compared to a background Serenitatis value of [[Rfraktur]]3 = 1.00 to 1.02.

Finally, the 1.00 / 0.90 um ratio image ([[Rfraktur]]4; Figure 2d) shows yet another style of spatial heterogeneity related to minor variations in the continuum and the width and strength of the 1-um feature. Most of the smaller craters are not visible in this ratio image. The largest contrasts are between the Haemus mountains ([[Rfraktur]]4 = 0.95 to 0.99) and the background Serenitatis and Tranquillitatis mare units ([[Rfraktur]]4 = 1.00 to 1.02), and between the larger craters (Menelaus, Plinius, and Dawes) and their surroundings. Many of the rays seen in the [[Rfraktur]]3 image can also be easily discerned.

Spectral Mixture Modeling. In order to try to isolate better the spatial distribution and the spectral character of the heterogeneities in these imaging data, we used a linear spectral mixing model designed specifically for image-oriented applications (Adams et al. 1986, 1993; Possolo et al. 1994). Many different mixing model analyses were performed, and each one worked best using a small number of endmembers (< 5) chosen to address a specific scientific question. For example, models were developed using image endmembers representing mature Serenitatis and Tranquillitatis mare units in order to map immaturity (which shows up as a residual spectral effect) within mare materials. Models using mature and immature Serenitatis mare endmembers were developed in order to examine compositional variations among Serenitatis and Tranquillitatis mare materials. Other models were developed using both mature and immature highlands and mare endmembers in order to map and quantify highlands-mare mixing (e.g., Campbell et al. 1992).

Results from one of these models are shown in Figure 3. This model used four endmembers in six spectral bands chosen with the goal of distinguishing and mapping the distribution of mature northwestern Tranquillitatis mare (EM1; Fig. 3a), fresh Haemus-like highlands material (EM2; Fig 3b), immature mare crater materials (EM3; Fig. 3c), and mature central Serenitatis mare (EM4; Fig. 3d). Pixels that could not be characterized well by the model as linear mixtures of any of these classes were stored in a residual root mean square (RMS) error image (Fig. 3e). Endmember spectra used for this model are shown in Figure 4, and the locations from which the endmembers were selected are shown in Fig. 1c and Fig. 3.

The spectral mixing model technique has proven to be quite adept at picking apart the often subtle spectral variations among the different materials in our study area. Tranquillitatis mare units, characterized in Fig. 4 by a negative 0.40-0.73 um spectral slope relative to MS2 (indicating a higher TiO2 content), are accurately mapped at high fractional abundances in the EM1 fraction image (Fig. 3a). Also apparent in this image is the fact that the moderate [[Rfraktur]]1 ratio unit in Fig. 2a in southwest Serenitatis is modeled with an intermediate "Tranquillitatis-like" fractional contribution of 45 to 65%. [For clarification, we stress that the fractional endmember percentages derived from this technique are not to be taken literally as the abundances of materials in each region. The many assumptions that are folded into the modeling (e.g., the mixing is linear, the endmembers are pure) must all be tested and verified, the data must be more accurately calibrated to reflectance, and the results directly compared to well-calibrated laboratory spectral measurements, before true mineralogic abundances can be estimated from such model results.]

The fresh highlands endmember image (Fig. 3b) provides a map of the fractional abundance of materials that, relative to MS2, exhibit high albedo but which do not exhibit a strong 1-um absorption feature (Fig. 4). The Haemus mountains highlands regions in and around Menelaus are the most prominent feature in this image, but other features also display substantial amounts of this endmember. First, the central peak and part of the floor of the crater Plinius (Fig. 1) exhibit elevated fresh highlands-like fractional abundances (15 to 35%). Second, the smaller craters Tacquet and Sulpicius Gallus exhibit a substantial fresh highlands-like fractional spectral signature (20 to 35%) that is unlike that seen in other craters of comparable size in this region. Finally, the ray passing through the crater Bessel (Fig. 1) exhibits a distinct fresh highlands-like component (10 to 15%). The ray model results are consistent with results derived from a different model that was developed by Campbell et al. (1992) to focus specifically on the problem of highlands-mare mixing in and along the Bessel ray.

The immature mare crater endmember image (Fig. 3c) provides a spectacular map of materials that exhibit a strong 1-um absorption feature relative to MS2 (Fig. 4). Mare and highlands craters of all sizes are seen in this fraction image, including Plinius (steep walls), Tacquet, and Sulpicius Gallus, which were shown to also exhibit a substantial fresh highlands-like component to their spectra (Fig. 3b). The highest fractional abundance values (corresponding to the deepest 1-um band depths) come from the crater Dawes and its associated ejecta blanket (Fig. 1). Other regions of enhanced immature material fractional abundance can be seen as diffuse patches in east and southeast Serenitatis and northwest Tranquillitatis, and around Bessel crater. Most of the Copernican age craters in our study area can be easily identified in this image by their bright ejecta blankets.

The mature Serenitatis endmember fraction image (Fig. 3d) is a map of materials that exhibit spectra very similar to that of the MS2 standard area. Central and eastern Serenitatis mare units dominate this fractional abundance map; the largest anomalies are the occurrence of an elevated abundance of MS2-like mare (45 to 65%) on the floor and in the ejecta of Plinius, and a high MS2-like fractional abundance in the Haemus mountains and in the eastern Serenitatis rim highlands material near the crater Vitruvius (Fig. 1). The highlands anomalies result from the model's inability to classify unambiguously the mature highlands materials: they have high albedo and thus must have high fractions of the brightest endmember (Fig. 3b), but they are also relatively flat spectrally (relative to mature MS2 mare) and thus can be mathematically mixed with a relatively "gray" material like EM4 and still satisfy the mixing criteria. It is possible that small mare patches in the Haemus mountains are influencing the spectral behavior enough at our spatial resolution to be detected in the model, but this does not dominate the effect observed here.

Limitations on the modeling imposed by the lack of strong, diagnostic highlands absorption features at these wavelengths make it not an optimal tool for highlands material characterization, yet it still does a fair job of discriminating the freshest highlands units while providing important information about variability within and between the mare. The quality of the model can be assessed by examining the RMS error image (Fig. 3e), which provides a quantitative estimate of the errors in the fractional abundance derivations discussed above. The maximum RMS error value summed across all bands ranges from 5 to 15% and occurs in the highlands regions (e.g., Haemus Mountains). Other spatially coherent regions of relatively high residual error occur at Plinius (2 to 5%), in Serenitatis around and southwest of Bessel (~ 1.5%), and in northeastern Tranquillitatis around Dawes (~ 2.5%). Most of the mare units modeled in this region show a residual RMS error of less than 1%.


We have assembled a group of VIS to near-IR reflectance spectra of regions within and near our study area in order to verify and augment the imaging results by providing data suitable for more detailed compositional and mineralogic interpretations. Many of these spectra are shown in Figure 5 and spectral parameters for these data are compiled in Table 2 and shown as variation diagrams (e.g., Lucey et al. 1986) in Figure 6.

For the 1-um band, spectra of mature central Serenitatis mare units are characterized by relatively low values of band width (0.28 to 0.29 um), medium values of band depth (8 to 11%), medium to high values of spectral slope (0.71 to 0.77 relative reflectance units/um), and high values of band minimum position (0.97 to 1.01 um) and band skew (0.83 to 0.97). Spectra obtained for small areas in the dark outer Serenitatis annulus (labeled "A" in Fig. 1c; spots 10, 16, and 26) are slightly different, exhibiting generally lower values of band depth, minimum, and slope. Mature Tranquillitatis mare spectra are characterized by lower values of band depth and slope than central Serenitatis mare, and comparable values of skew, width, and band minimum position. Immature mare craters exhibit the highest 1-um band depths (13 to 18%) and the lowest spectral slopes (0.51 to 0.65 R/um) but otherwise exhibit comparable values of center, width, and skew as central Serenitatis mare units. Highlands region spectra have low band center positions (0.91 to 0.95 um), band depths (4 to 7%), and skew (0.61 to 0.81), spectral slopes intermediate between central Serenitatis and Tranquillitatis mare values, and high values of band width (0.31 to 0.35 um).


Compositional Variability of the Study Region

Mare TiO2 Content. The most direct assessment of variations in mare TiO2 content comes from the [[Rfraktur]]1 ratio image (Fig. 2a). This ratio, through comparisons with returned lunar sample analysis (e.g., Charette et al. 1974), reveals that the highest TiO2 content units in our study region occur in northwestern Tranquillitatis, with decreasingly lower TiO2 abundances in northern Tranquillitatis, southwestern Serenitatis, the southern Serenitatis annulus, and central Serenitatis. These results are consistent with previous imaging and spectroscopy studies of TiO2 variations in this region (e.g., McCord et al. 1976; Pieters 1978; Johnson et al. 1991a,b; Melendrez et al. 1994), and they support the interpretations by those and other workers that the Tranquillitatis region is a prime target for future resource-driven lunar exploration initiatives. The distribution of TiO2-rich materials was also determined by our mixing model analysis, as the fractional abundance of EM1 (northwest Tranquillitatis mare; Fig. 3a) shows there to be a higher TiO2 mare unit in southern and southwestern Serenitatis. This result arises because the mixing model primarily relies upon differences in the 0.73 um filter (Fig. 4) as an indicator of the EM1 endmember abundance. Thus the Tranquillitatis and south-southwestern Serenitatis mare units have spectroscopic similarities, although there is not necessarily any requirement of a genetic link.

Highlands Compositions. A limited number of near-IR reflectance spectra were collected for highlands features in the Serenitatis/Tranquillitatis region as part of our study (Table 1; Figs. 1c, 5). The results of an analysis of these spectra (Table 2) were combined with those presented by Hawke et al. (1992) in order to investigate the composition of highlands units in the region. Spectra obtained for various portions of the North and South Massifs at the Apollo 17 landing site exhibit 1-um absorption bands with minima shortward of 0.95 um and band strengths of ~ 6%. These characteristics indicate a feldspar-bearing mineral assemblage with a mafic component dominated by low-Ca pyroxene. The areas for which these spectra were obtained contain abundant anorthositic norite. Spectra were also collected for various areas of Littrow crater. These include two mature portions of the crater floor and a small, fresh crater on the north rim of Littrow. The dominant lithology is anorthositic norite but the highlands-rich soil in the floor of Littrow may contain minor amounts of dark mantle material of pyroclastic origin (Hawke et al. 1992).

The Haemus Mountains form one segment of the mare-bounding ring of Serenitatis basin. Menelaus crater (Fig. 1c) straddles the mare-highlands border, and its deposits are heterogeneous in composition. Noritic anorthosite has been identified in the interior of Menelaus (Pieters 1986). However, more pyroxene-rich highlands debris also occurs in the Menelaus area (Hawke et al. 1992). To date, no deposits of pure anorthosite have been identified in the Serenitatis basin region. The most plagioclase-rich material in our study region was exposed by Menelaus crater.

Highlands Oases. Our study reveals several regions of exposed or excavated highlands materials surrounded by mare units. The most prominent such "highlands oasis" occurs in the central peak and parts of the floor and wall of Plinius crater (Figs. 1c and 3b). The highlands composition of this material is postulated from the analysis of the mixing model results, and is confirmed by the analysis of high resolution near-IR spectra (Fig. 5), discussed next.

Several near-IR reflectance spectra were obtained for Plinius and analysis of their spectra (Fig. 5, Table 2) has yielded important results. The spectrum of spot 11, collected from Plinius ejecta immediately southwest of the crater, exhibits a rather symmetric absorption band with a minimum at 0.98 um and a band depth of 4.7%. The area from which the spectrum was obtained is dominated by mature mare basalt. The spectrum of the lower southwest wall and floor of Plinius (spot 12) is very similar but has a slightly deeper band (minimum=0.98 um; depth=6.7%) due to the presence of fresh material on the steeper surfaces of the lower wall. A small amount of highlands material may be present in the area.

The 1-um bands of the small-aperture Plinius central peak and northern wall spectra (spots 13 and 14) appear to be composed of two features: a short-wavelength component centered near 0.92 um and a longer-wavelength component centered near 1.0 um. The 0.92 um feature is weaker than the 1.0 um band in the northern wall spectrum (spot 14), suggesting that this is primarily high-Ca pyroxene-rich mare material with a minor highlands component. However, the 0.92 um band is much more prominent in the central peak spectrum (spot 13), suggesting that the peak is primarily noritic highlands material with minor mare contamination. Our linear mixing model studies of the six-color CCD images verify and extend these spectral interpretations. The Plinius central peak and several regions of the floor clearly have a substantial highlands signature (Fig. 3b).

Table 2 also presents the results of an analysis of a spectrum of the central peak of Plinius obtained with the larger aperture (spot 24). The area for which this spectrum was collected includes not only the central peak but also portions of the crater floor. This spectrum exhibits a 1-um band with a 5% depth and a minimum at 0.97 um. These spectral parameters might be interpreted to indicate the presence of a gabbroic highlands lithology with a mafic assemblage dominated by high-Ca pyroxene. However, based on the analysis of the small aperture spectrum of the Plinius central peak (spot 13) as well as the results of the mixing model studies, we know that both noritic highlands material and mare basalt are present in the area for which the large aperture spectrum was obtained. This example demonstrates the potential and necessity for both high spatial and high spectral resolution observations of the Moon.

An interesting observation from our spectra and mixing model results is that there is apparently very little fresh highlands material in the Plinius ejecta blanket. A crater of this size (43 km diameter) should have penetrated the relatively thin mare deposits in the Plinius pre-impact target site (DeHon 1974; DeHon and Waskom 1976; Hörz 1978) and excavated highlands debris. In fact, Fig. 3d shows an elevated level of central Serenitatis-like material in the Plinius ejecta blanket. The lack of evidence for significant amounts of fresh highlands material in the Plinius ejecta may be due to inadequate spatial resolution (the highlands component of the ejecta is undetectable at the signal-to-noise of our data) or possibly to the effects of vertical mixing, where the spectral signature of the pre-existing mare materials now completely dominates the mixed highlands-mare signal. In addition, the mare-rich impact melt deposits derived from the upper portion of the target site may have been emplaced on top of more highlands-rich material. Still, the available evidence indicates that less highlands material is present in the exterior deposits of Plinius than would be predicted based on the currently-available cratering models.

Other possible highlands oasis regions detected in the mixing analysis include the craters Tacquet and Sulpicius Gallus. The mixing model results (Fig. 3) indicate that both Sulpicius Gallus crater and Tacquet have substantial fresh highlands-like components. Although we have not yet substantiated this assertion with high resolution spectra, it is likely that our mixing model results are caused by a combination of high albedo and a broad, shallow 1-um band that can be well modeled as a mixture of the immature highlands endmember spectrum and the immature mare crater spectrum (Fig. 4). Despite the lack of other spectroscopic evidence for highlands debris, we believe, based on their size and location and on our mixing model results, that these craters have in fact excavated highlands materials as well as mare material and, possibly (based on previous studies of this region), pyroclastic debris (see Wilhelms 1980; Gaddis et al. 1985).

The putative identification of fresh highlands material in the Sulpicius Gallus and Tacquet crater spectra in Fig. 3b can be used to demonstrate the limitations of the spectral mixing technique for compositional interpretations. The model that we used makes the implicit assumption that every pixel can and must be fit as a mixture of the input endmembers. In some cases, there may be no single unique linear combination of endmembers that fit the spectra of certain regions, so the model chooses one set of fractions from among many which may be statistically the same. Simply increasing the number of endmembers is not a solution, as there are mathematical limitations to the number of endmembers that can be used, and the potential for the determination of non-unique mixing solutions increases as the number of endmembers increases (Possolo et al. 1994). Without accompanying high-resolution spectra such as those in Fig. 5, it is often not possible to determine whether the mixing model has arrived at a unique solution. This caveat to the otherwise powerful potential of mixing model analysis must be considered in the analysis of any relatively low spectral resolution or coarse spectral sampling data set.

Mixing Across Geologic Boundaries. The results of our multispectral imaging and mixing model studies can be used to assess the degree of mixing by lateral transport along geologic contacts in the Haemus-Serenitatis-Tranquillitatis boundary region and within bright rays passing through Mare Serenitatis. We find that the major highlands contamination in the mare comes from exposure of highlands materials by a few isolated craters (Plinius, Sulpicius Gallus, Tacquet) and from bright rays such as the one passing through Bessel. In a previous study, Campbell et al. (1992) were able to map semi-quantitatively the distribution of highlands materials along the Bessel ray and conclude that the origin of the ray is not Bessel crater itself but either Menelaus crater or a distant highlands impact crater such as Tycho. There has apparently been little spectrally-detectable highlands-mare mixing along the Serenitatis-Haemus boundary. This situation is in contrast to the apparent high degree of highlands-mare mixing noted by Staid et al. (1994, 1995) along the western Tranquillitatis margin. This difference may be related to age differences between Serenitatis and Tranquillitatis mare units, or possibly to the presence of pyroclastic units along portions of the southern Serenitatis boundary that have effectively diluted the spectral signature of locally-mixed Haemus highlands materials. Our data also show very little evidence of mixing across various compositionally-distinct mare units. Most of the mare boundaries within Mare Serenitatis and between Serenitatis and Mare Tranquillitatis mare units are sharp and well-defined in our mixing model studies. This observed lack of mixing across compositionally-distinct mare units and across other geologic contacts has long been noted by some authors (e.g., Johnson et al. 1977; Hörz 1978) as qualitative evidence against extensive lateral transport of materials by impact processes.

Spectroscopic Distinction Between Maturity and Composition. An important capability that can be obtained by combining imaging and spectroscopic techniques is the ability to distinguish between spectral variations caused by maturation (e.g., Adams and McCord 1971, 1973) and those due to true compositional differences. Examples of this ability can be seen in our data in association with the craters Dawes and Bessel.

Dawes is located within a relatively high-albedo, rectangular unit (~ 80x120 km) of uncertain origin (labeled "R" in Fig. 1c). Our results can be used to determine the probable origin of this bright feature between the two most likely possibilities: immaturity or the presence of highlands material. Both are valid possibilities because of the immature nature of the spectra of Dawes itself (Fig. 5b) and the nearby occurrence of fresh highlands materials in the interior of Plinius. In addition, some studies of the Apollo X-ray fluorescence data have indicated that highlands material was exposed by the Dawes impact event. For example, Schonfeld (1980) noted than an enhanced Mg/Al ratio image showed a geochemical anomaly at the center and in the north and northwest parts of the crater rim and suggested that Dawes had excavated some highlands material from beneath mare basalt flows. Two near-IR reflectance spectra for the interior of Dawes (spots 4 and 5) are shown in Fig. 5b. The spectral parameters derived from these spectra (Table 2) indicate that the interior of Dawes is composed of immature mare basalt. The mixing model results (Fig. 3) show that Dawes and its ejecta deposit overall have little or no fresh highlands-like material and only a minor, localized central-Serenitatis-like spectral component. Instead, this region can be modeled largely as a combination of Tranquillitatis mare and "immature mare crater" material. This result implies that the brightness of the Dawes rectangular deposit is a result of immaturity rather than compositional variability, raising the possibility that this enigmatic deposit is actually the extended ejecta blanket of Dawes itself.

Some support for this interpretation can be found in the polarization images of this region obtained by Dollfus (1990), who noted that the rectangular Dawes deposit exhibits an anomalously high value of linear polarization normal to the scattering plane. We note that a high-albedo lobe of material extending ~ 40 km directly northwest of Dawes (labeled "L" in Fig. 1c) has different spectral properties, characterized by a low Tranquillitatis mare content and the existence of a distinct central Serenitatis mare component (Fig. 3d). Melendrez et al. (1994) have interpreted this feature as evidence of oblique impact and the excavation of lower-TiO2 material from beneath the higher-TiO2 Tranquillitatis mare deposits. Our results are consistent with this interpretation, which is also supported by more recent Clementine multispectral image analyses of this region (Staid et al. 1995).

The crater Bessel is also surrounded by moderately high albedo materials arranged both linearly and in splotches. Our results and those of Campbell et al. (1992) indicate that different origins exist for the various portions of these units surrounding Bessel. Most of the spectral signature can be modeled using the central Serenitatis endmember. However, Fig. 3b clearly shows that the linear ray system extending through Serenitatis has a fresh highlands-like spectral component, and Fig. 3c also shows that both the ray system and the rest of the unit around Bessel have elevated immature mare signatures. Thus, the higher albedo of this region is a result of both compositional and maturation effects, acting together in some places, and independently in others.

Another possible example of the way in which data such as these can discriminate between maturation and compositional effects comes from the [[Rfraktur]]4 ratio image (Fig. 2d). The fact that most of the fresh mare and highlands craters seen in the fresh crater endmember image (Fig. 3c), do not show up in the 1.0/0.9 um ratio map suggests that maturity may not be the driving factor in 0.9-1.0 um spectral slope changes in this image. Instead, the spatial patterns seen in Fig. 2d indicate that this ratio may be sensitive to differences between the spectra of low-Ca pyroxene-rich highlands material and high-Ca pyroxene-rich mare material. This interpretation remains uncertain without including a more complete set of higher spectral resolution spectroscopic measurements (such as those in Fig. 5) that are essential in order to make proper mineralogic inferences. However, it is useful to realize that imaging at even a few wavelengths, if carefully selected, can provide certain types of important information on specific questions in lunar mineralogy (see also Lucey et al. 1991; Pinet et al. 1993).

Stratigraphy and Topography

Results presented here and by Johnson et al. (1991), Melendrez et al. (1994), and Staid et al. (1994, 1995) allow interesting information to be inferred about the depth of mare units and the subsurface distribution of highlands and mare materials. This is primarily achieved by using craters as indicators of local stratigraphy (e.g., DeHon 1974; Hörz 1978). For example, our results and those of Bell and Hawke (1992) and Melendrez et al. (1994) for the Dawes and Plinius ejecta deposits indicate that along the Serenitatis-Tranquillitatis boundary the high-TiO2 mare units are thin and underlain by lower-TiO2 mare material. Other large craters in Tranquillitatis that have not excavated highlands or low-TiO2 mare materials (such as Ross and Maclear; Fig. 1c) indicate that the high-TiO2 mare is not uniformly shallow. Apparently, the high-TiO2 Mare Tranquillitatis basalts in our study region have buried a compositionally distinct surface composed of lower-TiO2 basalts and highlands materials and which likely has a highly variable topography. Some portion of the buried topography in this border region must be related to the presence of major Serenitatis and Tranquillitatis basin rings (e.g., Wilhelms 1980). Another example is Bessel crater in central Serenitatis (Fig. 1c), which shows evidence in our mixing model results of excavating higher-TiO2 mare materials (Figs. 3a, 3d). If the interpretation of these model results is correct, then this implies that the pre-impact stratigraphy in the region around Bessel is composed of lower-TiO2 central Serenitatis mare materials overlying a higher-TiO2 mare surface. This result and interpretation is consistent with the observation and analysis by Campbell et al. (1992) of a lower radar return unit surrounding parts of Bessel. Because Bessel is a small crater (16 km diameter), it is probable that the lower-TiO2 surface mare materials in this region are rather thin.

The apparent large-scale differences in mare TiO2 content within Serenitatis (Figs. 2a and 3a) form distinct patterns, with perhaps the starkest contrast occurring near Sulpicius Gallus crater, where the border between the lower-TiO2 and higher-TiO2 mare units forms a nearly perfect square edge. Howard et al. (1973) interpreted the darker mare units in Serenitatis as the unburied portions of an older, faulted mare surface. In order to form this nearly perfect square edge, there must have been topographic controls on the flow of low-TiO2 Serenitatis mare materials that prevented them from forming a more irregular, "natural" flow margin. Future analysis of Clementine and other lunar topography data may help to prove or disprove this. It is interesting to note that the TiO2 abundance difference is the only expression of this boundary in the mixing results: there is no difference in highlands contamination or in the distribution of immature craters and rays.

Mixing model results for the immature mare crater materials in this region (Fig. 3c) show an apparent difference between eastern and western Mare Serenitatis units. While the number of resolved immature craters is roughly constant across the mare, eastern Serenitatis displays an overall increase in the fractional abundance of unresolved or mixed immature materials (appearing as diffuse splotches of higher fractional abundance). One possible interpretation of this dichotomy is that the western Serenitatis mare materials are substantially older than the eastern ones. Boyce (1976) provides evidence for an age difference among Serenitatis mare units consistent with this hypothesis: far western Serenitatis mare units are mapped having an age of 3.0+/-0.1 Ga, central Serenitatis mare units (which correspond to the westernmost regions of Serenitatis in our study area) are mapped at 3.6+/-0.1 Ga, and eastern Serenitatis mare units are mapped at 3.4+/-0.1 Ga. However, the small differences in ages should have no measurable effect on spectral parameters of these ancient units. A more likely possibility is that the dichotomy is a result of the preferential occurrence of rays and secondary crater clusters in eastern Serenitatis. These large-scale spectral differences between eastern and western Serenitatis deserve additional study via a combination of photogeologic and spectroscopic techniques.


We have conducted a study of the spectral variability of the Serenitatis-Tranquillitatis border region of the Moon using a variety of multispectral imaging and spectroscopic techniques. Our results indicate that a combination of these techniques provides a powerful set of tools with which to understand the composition, mineralogy, and stratigraphy of the lunar crust. In the particular region that we focused on, we were able to detect and map variations in mare basalt TiO2 abundance, in the abundance of fresh highlands materials emplaced on mare surfaces, and in the degree of spectral maturity of craters, ejecta, and mare. We present specific evidence for the exposure of immature highlands materials by the craters Plinius, Tacquet, and Sulpicius Gallus and of lower-TiO2 mare by the craters Dawes and Plinius and we discuss the implications of the apparent immaturity of a large rectangular unit surrounding Dawes.

The techniques discussed here, especially supervised image-oriented mixture modeling, can be applied to any image-based lunar multispectral data set. Examples of currently-available data sets include additional telescopic images of other nearside regions, the Galileo Earth-Moon flyby multispectral images, and the Clementine mission global multispectral images (e.g., McEwen et al. 1994; Pieters et al. 1994; Staid et al. 1995). The latter data set has particularly strong potential because the images are free of terrestrial atmospheric artifacts, the filter wavelengths were optimized for lunar mineralogy, topographic information associated with the images is available, and excellent image calibration and registration seem to be possible (Nozette et al. 1994). In concert with moderate to high spectral resolution spectroscopy and techniques such as those outlined here, much of the inherent compositional non-uniqueness present when dealing with such relatively sparsely wavelength-sampled multispectral imaging data sets can be removed, and the information content of the image data sets themselves can be substantially enhanced.


We thank Keith Horton and Paul Lucey for acquisition and assistance with the reduction of the CCD imaging data. Dave Blewett, Jeff Bosel, Bruce Campbell, Beth Clark, Ralf Jaumann, Gerhard Neukum, Pam Owensby, and Greg Smith provided valuable comments and/or assistance with the acquisition and reduction of the near-IR spectroscopic data. John Spencer kindly helped with the image registration software. John Adams and Milton Smith provided substantial training and resources towards our understanding and application of the image-oriented mixing model for lunar applications. We sincerely appreciate the greatly detailed reviews by Steve Kadel and Carlé Pieters that resulted in substantial improvements to an earlier version of this paper. This work was supported by grants from the NASA Planetary Astronomy and Planetary Geology and Geophysics programs.


Adams, J.B. 1974. Visible and near-infrared diffuse reflectance spectra of pyroxenes as applied to remote sensing of solid objects in the solar system. J. Geophys. Res. 79, 4829-4836.

Adams, J.B. and T.B. McCord 1971. Alteration of lunar optical properties: Age and composition effects. Science 171, 567-571.

Adams, J.B. and T.B. McCord 1973. Vitrification darkening in the lunar highlands and identification of Descartes material at the Apollo 16 site. Proc. Lunar Planet. Sci. Conf. 4th, 163-177.

Adams, J.B., M.O. Smith, and P.E. Johnson 1986. Spectral mixture modeling: A new analysis of rock and soil types at the Viking Lander I site. J. Geophys. Res. 91, 8098-8112.

Adams, J.B., M.O. Smith, and A.R. Gillespie 1993. Imaging spectroscopy: Interpretation based on spectral mixture analysis. In Remote Geochemical Analysis: Elemental and Mineralogical Composition, ed. C. Pieters and P. Englert, pp. 145-166. Cambridge Univ. Press.

Bell, J.F., III, and B.R. Hawke 1991. CCD Narrowband Filter Imaging of Lunar Crater Rays. Lunar and Planetary Science XXII, 75-76. [Abstract]

Bell, J.F., III, and B.R. Hawke 1992. Composition and Stratigraphy of the Plinius-Dawes Region of the Moon. Lunar and Planetary Science XXIII, 77-78. [Abstract]

Boyce, J.M. 1976. Ages of flow units in the lunar nearside maria based on Lunar Orbiter IV photographs. Proc. Lunar Planet. Sci. Conf. 7th, 2717-2728.

Campbell, B.A., J.F. Bell III, S.H. Zisk, B. Ray Hawke and K.A. Horton 1992. A High Resolution Radar and CCD Imaging Study of Crater Rays in Mare Serenitatis and Mare Nectaris. Proc. Lunar Planet. Sci. XXII, 259-274.

Carr, M.H. 1966. Geologic map of the Mare Serenitatis region of the Moon. U.S. Geological Survey Map I-489, U.S.G.S., Washington, D.C.

Charette, M.P., T.B. McCord, C. Pieters, and J.B. Adams 1974. Application of remote spectral reflectance measurements to lunar geology: Classification and determination of titanium content of lunar soils. J. Geophys. Res. 79, 1605-1613.

Clark, R.N. and T.L. Roush 1984. Reflectance spectroscopy: Quantitative analysis techniques for remote sensing applications. J. Geophys. Res. 89, 6329-6340.

DeHon, R.A. 1974. Thickness of mare material in the Tranquillitatis and Nectaris basins. Proc. Lunar Planet. Sci. Conf. 5th, 53-59.

DeHon, R.A. and J.D. Waskom 1976. Geologic structure of the eastern mare basins. Proc. Lunar Planet. Sci. Conf. 7th, 2729-2746.

Dollfus, A. 1990. Video-polarimetry of the lunar regolithic surface. Lunar and Planetary Science XXI, 296-297. [Abstract]

Gaddis, L.R., C.M. Pieters, and B.R. Hawke 1985. Remote sensing of lunar pyroclastic mantling deposits. Icarus 61, 461-489.

Hawke, B.R., C.A. Peterson, P.G. Lucey, G.J. Taylor, D.T. Blewett, B.A. Campbell, C.R. Coombs, and P.D. Spudis 1993. Remote sensing studies of the terrain northwest of Humorum basin. Geophys. Res. Lett. 20, 419-422.

Hawke, B.R., C.A. Peterson, P.G. Lucey, D.T. Blewett, J.F. Bell III, and P.D. Spudis 1992. A spectral survey of the Serenitatis Basin region of the Moon. In, Workshop on Geology of the Apollo 17 Landing Site (G. Ryder, H.H. Schmitt, and P.D. Spudis, eds.), LPI Technical Report 92-09, Part 1, 14-15.

Hiesinger, H., R. Jaumann, G. Neukum, and the GLL Imaging Team 1993. Earth-based and Galileo SSI multispectral observations of eastern Mare Serenitatis and the Apollo 17 landing site. Lunar and Planetary Science XXIV, 655-656. [Abstract]

Hörz, F. 1978. How thick are lunar mare basalts? Proc. Lunar Planet. Sci. Conf. 9th, 3311-3331.

Howard, K.A., M.H. Carr, and W.R. Muehlberger 1973. Basalt stratigraphy of southern Mare Serenitatis. In Apollo 17 Preliminary Science Report, pp. 29-1 to 29-12. NASA SP-330, NASA, Washington, D.C.

Jaumann, R. 1989. Spectrophotometric analysis of the chemical and mineralogical composition of lunar surface materials. Ph.D. Thesis, Universität München, 284 pp.

Jaumann, R. 1991. Spectral-chemical analysis of lunar surface materials. J. Geophys. Res. 96, 22793-22807.

Jaumann, R. and G. Neukum 1989. Spectrophotometric analysis of the lunar Plinius/Apollo 17 region. Lunar and Planetary Science XX, 456-457. [Abstract]

Johnson, T.V., D.L. Matson, R.J. Phillips, and R.S. Saunders 1975. Vidicon spectral imaging: Color enhancement and digital maps. Proc. Lunar Planet. Sci. Conf. 6th, 2677-2688.

Johnson, T.V., J.A. Mosher, and D.L. Matson 1977. Lunar spectral units: A northern hemispheric mosaic. Proc. Lunar Planet. Sci. Conf. 8th, 1013-1028.

Johnson, J.R., S.M. Larson, and R.B. Singer 1991a. Remote sensing of potential lunar resources: I. Near-side compositional properties. J. Geophys. Res. 96, 18861-18882.

Johnson, J.R., S.M. Larson, and R.B. Singer 1991b. A re-evaluation of spectral ratios for lunar mare TiO2 mapping. Geophys. Res. Lett. 18, 2153-2156.

Larson, S.M., J.R. Johnson, and R.B. Singer 1991. Evaluation of the sensitivity of reflectance ratios to mafic minerals in the lunar regolith. Geophys. Res. Lett. 18, 2149-2152.

Lucey, P.G., B.R. Hawke, C.M. Pieters, J.W. Head, and T.B. McCord 1986. A compositional study of the Aristarchus region of the Moon using near-infrared reflectance spectroscopy. J. Geophys. Res. 91, D344-D354.

Lucey, P.G., B.R. Hawke, and K. Horton 1991. The distribution of olivine in the crater Copernicus. Geophys. Res. Lett. 18, 2133-2136.

McCord, T.B., M.P. Charette, T.V. Johnson, L.A. Lebofsky, C. Pieters, and J.B. Adams 1972. Lunar spectral types. J. Geophys. Res. 77, 1349-1359.

McCord, T.B., C. Pieters, and M.A. Feierberg 1976. Multispectral mapping of the lunar surface using ground-based telescopes. Icarus 29, 1-34.

McCord, T.B., R.N. Clark, B.R. Hawke, L.A. McFadden, P.D. Owensby, and C.M Pieters 1981. Moon: near-infrared spectral reflectance, a first good look. J. Geophys. Res. 86, 10883-10892.

McEwen, A.S., M.S. Robinson, E.M. Eliason, P.G. Lucey, T.C. Duxbury, and P.D. Spudis 1994. Clementine observations of the Aristarchus region of the Moon. Science 266, 1858-1861.

Melendrez D.E., J.R. Johnson, S.M. Larson, and R.B. Singer 1994. Remote sensing of potential lunar resources, 2, High spatial resolution mapping of spectra reflectance ratios and implications for nearside mare TiO2 content. J. Geophys. Res. 99, 5601-5620.

Morris E.C., and D.E. Wilhelms 1967. Geologic map of the Julius Caesar quadrangle of the Moon. U.G. Geological Survey Map I-510, U.S.G.S., Washington, D.C.

Nozette, S. and 33 others 1994. The Clementine mission to the Moon: Scientific overview. Science 266, 1835-1839.

Pieters, C.M. 1978. Mare basalt types on the front side of the Moon: A summary of spectral reflectance data. Proc. Lunar Planet. Sci. Conf. 9th, 2825-2849.

Pieters, C.M. 1986. Composition of the lunar highland crust from near-infrared spectroscopy. Rev. Geophys. 24, 557-578.

Pieters, C.M., M.I. Staid, E.M. Fischer, S. Tompkins, and G. He 1994. A sharper view of impact craters from Clementine data. Science 266, 1844-1848.

Pinet, P.C., S.D. Chevrel, and P. Martin 1993. Copernicus: A regional probe of the lunar interior. Science 260, 797-801.

Possolo A., J.B. Adams, and M.O. Smith 1994. Mixture models for multispectral images, J. Geophys. Res., in press.

Schonfeld, E. 1980. Enhanced orbital geochemical images by the Laplacian subtraction method. Proc. Lunar Planet. Sci. Conf. 11th, 677-688.

Staid, M.I., C.M. Pieters, and J.W. Head 1994. Spectral variation within Mare Tranquillitatis: Implications for stratigraphy and mixing mechanisms. Lunar and Planet. Sci. Conf. XXV, 1329-1330. [Abstract]

Staid, M.I., C.M. Pieters, and J.W. Head 1995. A multispectral view of stratigraphy in Mare Tranquillitatis. Lunar and Planet. Sci. Conf. XXVI, 1345-1346. [Abstract]

Thompson, T.W., K.A. Howard, R.W. Shorthill, G.L. Tyler, S.H. Zisk, E.A. Whitaker, G.G. Schaber, and H.J. Moore 1973. Remote sensing of Mare Serenitatis. In Apollo 17 Preliminary Science Report, pp. 33-3 to 33-10. NASA SP-330, NASA, Washington, D.C.

Whitaker, E.A. 1972. Lunar color boundaries and their relationship to topographic features: A preliminary survey. Moon 4, 348-355.

Wilhelms, D.E. 1980. Stratigraphy of Part of the Lunar Near Side. U.S. Geological Survey Professional Paper 1046-A, U.S. Gov't Printing Office, Washington D.C., 71 pp.

Table 1. Lunar IRCVF Spot Names and Locations.

Spot Classa No.b Lat. Lon. Name/Description

1 MT 407/660 18.5deg. 29.6deg. Mare Tranquillitatis west of Vitruvius

2 MT 409/660 16.8 29.0 Mare Tranquillitatis between Dawes and Jansen C

3 MT 1338/660 18.7 27.5 Region along Serenitatis/Tranquillitatis border NE of Dawes

4 C 1340/660 17.7 26.5 Dawes Crater Center

5 C 1342/660 17.1 26.4 Dawes Crater SSE Wall

6 MSC 1344/660 19.9 22.5 Region in Mare Serenitatis NE of MS2

7 MSC 1346/660 20.2 24.0 Region in Mare Serenitatis ENE of MS2

8 HL 1348/660 21.6 31.2 Smooth floor in center of Littrow crater

9 HL 1350/660 18.5 31.6 High-albedo highlands area on N rim of Vitruvius crater

10 MSA 1352/660 20.6 28.1 Region in dark Serenitatis outer annulus N of Mons Argaeus

11 MT 1354/660 14.8 23.1 Plinius crater ejecta, just SW of crater rim

12 MIX 1356/660 14.9 23.4 Lower SW wall and some of the floor of Plinius crater

13 HL 1358/660 15.5 23.5 Plinius crater central peak, small aperture

14 C 1360/660 15.9 23.7 Plinius crater N wall

15 MT 1362/660 17.5 24.7 Region along Serenitatis/Tranquillitatis border NE of Plinius

16 MSA 1364/660 16.9 21.5 Region in dark Serenitatis outer annulus N of Prom. Acherusia

17 MT 1370/660 14.1 21.1 Region of Tranquillitatis E of crater Tacquet A

18 C 1372/660 21.8 18.0 Bessel crater center

19 C 1374/660 21.4 18.1 Bessel crater SE rim

20 MSC 1376/660 22.3 27.0 Border of bright and dark Serenitatis mare E of LeMonnier C

21 MT 1156/594 0.7 24.3 Apollo 11 Landing Site in Mare Tranquillitatis

22 SG 314/544 20.4 9.7 Sulpicius Gallus dark mantle

23 MSC 275/615 18.7 21.4 MS2 Standard Area in Mare Serenitatis

24 MIX 220/614 15.5 23.5 Plinius crater central peak, large aperture

25 HL 817/589 21.6 31.2 Littrow crater

26 MSA 796/589 20.2 30.8 Apollo 17 Landing Site in Taurus-Littrow Valley

aSpectral type classification, determined from analyses in this paper. MT=mature Mare Tranquillitatis; C=fresh mare crater; MSC=mature Mare Serenitatis Central; MSA=mature Mare Serenitatis Annulus; HL=highlands; SG=Sulpicius Gallus dark mantle deposit; MIX=mixed mare/highlands spectrum.

bSpectrum number followed by U. Hawaii Planetary Geosciences Lunar Spectral Archive tape number.

Table 2. 1 um Band Spectral Parameters

for Lunar CVF Spectra Used in This Study.

Minimum Depth Width Slope

Spot Class (um) (%) (um) Skew (R/um)

1 MT 0.990 4.5 0.315 0.87 0.707

2 MT 0.989 6.8 0.259 0.98 0.598

3 MT 0.990 7.8 0.275 0.98 0.633

4 C 0.986 16.0 0.271 0.95 0.566

5 C 0.975 18.5 0.330 0.71 0.508

6 MSC 0.984 9.3 0.280 0.92 0.713

7 MSC 1.007 8.9 0.289 0.97 0.755

8 HL 0.932 3.8 0.349 0.62 0.701

9 HL 0.952 7.3 0.311 0.81 0.655

10 MSA 0.978 4.7 0.306 0.95 0.636

11 MT 0.976 4.7 0.279 0.96 0.610

12 MIX 0.977 6.7 0.323 0.85 0.618

13 HL 0.944 5.9 0.340 0.67 0.627

14 C 0.985 13.5 0.304 0.90 0.653

15 MT 0.990 7.2 0.281 0.96 0.615

16 MSA 0.974 5.1 0.265 0.97 0.648

17 MT 0.973 3.4 0.269 1.00 0.618

18 C 0.994 13.4 0.272 0.97 0.575

19 C 0.992 13.4 0.300 0.93 0.626

20 MSC 0.985 10.5 0.286 0.94 0.752

21 MT 0.971 5.6 0.292 0.92 0.615

22 SG 0.995 5.6 0.428 0.71 0.767

23 MSC 0.975 10.9 0.279 0.83 0.766

24 MIX 0.974 5.0 0.351 0.83 0.614

25 HL 0.912 3.7 0.312 0.61 0.803

26 MSA 0.971 5.6 0.307 0.90 0.650



(a) Full-Moon image showing location of our study region. Image from Johnson et al. (1991a).

(b) 0.73 um image of our study region, showing major albedo feature boundaries in this region.

(c) Map of the study region showing major features discussed in the text. Base map is NASA Lunar Earthside Chart LMP-1 (1970). Locations of point spectra in Table 1 are indicated by numbers and crosses centered on the aperture locations. Locations of endmember spectra from Figures 3 and 4 are indicated by small squares. Other features discussed in the text include: A=dark Serenitatis outer annulus; R=rectangular deposit surrounding Dawes; L=bright Dawes ejecta lobe.


(a) 400 nm / 730 nm ratio image. High and low image brightness values are indicated on the scalebar to the right. Each image that went into these ratios was first scaled to 1.0 in the MS2 standard region (Fig. 1c).

(b) 1000 nm / 400 nm ratio image.

(c) 1000 nm / 730 nm ratio image.

(d) 1000 nm / 900 nm ratio image.


(a) Results of linear spectral mixture model analysis. This image shows the modeled fractional abundance of mature Mare Tranquillitatis materials (EM1 in Fig. 4). Brighter pixels correspond to higher modeled fractional abundances, and the scale on the right ranges from <=10% (black) to >=90% (white). The approximate region that the endmember spectrum (Fig. 4) was chosen from is outlined in black here in in Figs. 3b, 3c, and 3d.

(b) Fractional abundance of fresh highlands endmember (EM2 in Fig. 4). Scale = 0 to 50%. Note the occurrence of a highlands signature in Plinius, Tacquet, and Sulpicius Gallus craters, and within the bright ray to the west of Bessel.

(c) Fractional abundance of fresh crater endmember (EM3 in Fig. 4). Scale = 0 to 30%. Note the enhanced fresh crater signature in the rectangular deposit surrounding Dawes and in the ray units around Bessel.

(d) Fractional abundance of mature Mare Serenitatis endmember (EM4 in Fig. 4). Scale = 30 to 100%. Note the small amount of mature Serenitatis signature in and near Plinius and Dawes.

(e) RMS error image associated with the endmembers in (a) through (d). The scale is 0 to 10% total error, measured as the root mean square of the errors in each band. The maximum errors occur in the highlands, indicating that this particular model was not optimized for analysis of those areas. Subtle errors associated with image misregistration and shadowing can also be seen.

FIGURE 4: Endmember spectra used in the mixture models for Fig. 3. The locations of the regions that these endmembers were chosen from are indicated in Fig. 1c and in the individual endmember images of Fig. 3.


(a) CVF spectra for regions shown in Fig. 1c and listed in Table 1. The spectra have had a linear continuum removed and all are plotted at the same contrast level. Spectral parameters for these data are compiled in Table 2. The spectra are organized in order of increasing 1-um band depth, and this figure plots spectra with 1-um band depths ranging from 3.4% to 6.7%.

(b) Continuation of Figure 5a. This figure plots spectra with band depths from 7.2% to 18.5%.

FIGURE 6: Variation diagrams showing relationships between spectral parameters compiled in Table 2. The classes shown in the key are: MT=mature Mare Tranquillitatis; C=fresh mare crater; MSC=mature Mare Serenitatis Central; MSA=mature Mare Serenitatis Annulus; HL=highlands; SG=Sulpicius Gallus dark mantle deposit; MIX=mixed mare/highlands spectrum.