composition of the lunar highland crust from near&infrared ... · intimate relative of the...

REVIEWS OF GEOPHYSICS, VOL. 24, NO. 3, PAGES 557-578, AUGUST 1986

Composition of the Lunar Highland Crust From Near-Infrared Spectroscopy

CARLE M. PIETERS

Department of Geolo#iCal Sciences, Brown University, Providence, Rhode Island

Near-infrared reflectance spectra of the lunar highlands are synthesized to obtain global information about the composition of the lunar near-side crust. Immature exposed surfaces from small fresh craters, mountains and massifs, and large craters with central peaks are included in a survey of rock types across the lunar near side. Compositions recognized in these data include anorthosite, noritic, gabbric, and troctolitic mineral assemblages. Three quarters of the areas studied that represent samples from the upper 1-2 km of lunar crust exhibit noritic compositions with different amounts of pyroxene and/or brecciation alteration. Norites are ubiquitous across the near side and show no spatial clustering associated with any of the major basins. In contrast to this noritic megaregolith of the upper crust, material representing stratigraphically deeper zones (5-15 km) of the lunar crust is dominated by gabbros, anorthosites, and troctolites, with less than a quarter of the areas studied exhibiting a noritic composition. The origin of the noritic megategolith and its relation to deeper crustal material is thus somewhat of an enigma. The current limited statistics for areas identified as anorthosites and troctolites are insufficient to derive global spatial distributions. Gabbroic areas, however, are notably concentrated in the western hemisphere. Although incomplete for a global assessment of crustal materials, these data demonstrate that the lunar near-side crust is clearly mineralogically heterogeneous, both laterally and vertically, and not well mixed below 1-2 km.

1. INTRODUCTION

Our current understanding of the global composition o( the lunar crust comes largely from samples obtained during six Apollo and three Luna missions to the moon and limited remote sensing information from Apollo 15 and 16 spacecraft and earth-based telescopes. Although the first lunar landing occurred in 1969 on a basaltic mare, anorthositic fragments found in the returned fines were correctly interpreted as being derived from the nearby highlands [Wood et al., 1970]. With subsequent lunar missions the general nature of the two primary lunar rock types (basaltic mare and feldspathic highland crust) became apparent (summarized by Taylor [1975, 1982]). Both were found to be very old and compositionally related in a differentiation event very early in lunar history. Understand- ing the global composition of the lunar crust, together with its lateral and vertical variations, is fundamental to ufiderstand- ing the structure and evolution of the moon, since the present highland crust, however complex, represents the results of crustal formation that occurred largely during the first 800 million years of lunar history.

The dark mare basalts prominent on the lunar near side were derived from a lunar mantle, probably formed con- currently with the feldspathic crust and later remelted. These mantle-derived low-viscosity lavas were extruded as extensive deposits between 4.0 and 3.0 b.y. ago, filling the lowlands with a variety of basalt types [e.g., Papike et al., 1976; Pieters, 1978]. The mare basalts constitute about 17% of the surface area of the moon, the vast majority of which occur on the lunar near side [Head, 1976]. The remaining 83% of exposed crust is usually referred to as the highlands, owing to the natural topographic difference between the maria and the surrounding terrain.

Most of the crustal rocks returned from the highlands are extremely brecciated, being both texturally and compositionally mixed by multiple impact events. This complex

Copyright 1986 by the American Geophysical Union.

Paper number 6R0243. 8755-1209/86/006R-0243515.00

557

family of breccias are evidence of a continued heavy bombardment early in the history of the moon after crustal formation. Bombardment decreased dramatically with time during the emplacement of the basaltic maria, allowing the cratering statistics for each mare to be used to place the maria in a relative time sequence [e.g., Boyce, 1976]. For the last few billion years the surfaces of both the highland crust and the maria have been little modified except by local cratering events and soil formation involving micrometeorite impacts and interactions with the solar wind. These broken rock fragments and soils are the major clues that lunar scientists have to piece together a description of the composition of the lunar crusi.

The purpose of this review is to synthesize a growing amount of new information about the composition of the lunar crust derived from near-infrared reflection spectroscopy. This remote sensing technique allows new mineralogical information to be obtained for unsampled areas on the lunar near side, currently using spectroscopic instruments on earth-based telescopes. Some of the most basic questions about the global composition of the lunar crust that such data can address are, What proportion of lunar rock types do the carefully studied returned samples represent? What are the mineral compositions of unsampled rock types ? What is the spatial extent of known (and unknown) rock types? What is the degree and nature of compositional heterogeneity in the lunar crust? Is the local lunar crust homogeneous with depth? Ultimately, we would like to know full details about the global composition of the moon, both to satisfy our compelling curiosity about the nature and geologic diversity of our nearest neighbor, an intimate relative of the earth, as well as to be prepared for eventual pragmatic uses of lunar resources. Some of the most recent new data about the moon are examined here and provide deeper insight into the composition of the lunar crust; this review should be considered an early framework, however, to be substantially expanded when more complete data are available from orbiting satellites.

2. BACKGROUND FROM THE LUNAR SAMPLES

All the returned lunar samples were taken from within a narrow swath of area around the equator on the lunar near

558 PIETERS.' COMPOSITION OF THE LUNAR CRUST

*'"•"?i ';..,' ?" ....... ..--:*";•i:•:;:.:•:;:' •::'.J;i½!::: * II ..: ....--. •: ......, . • , .,.:•;..::•.-::-';T•..•;;•...,i..-'. '%;-• :.'*. ' ::• .:..'?•:* • .. ;;:•; , .,....-;,•,.,•.• •. :':?:;t • ..... .:,.,.::•.• ..... ,...•. . ....... • .... •:•%:• ;

' ..,,:-.:.:....,s..:.:-:-,. : -- .•......... , ,; . .• ... ß - .:..- ..e z:...½½..:.. ..... : ...... •,.:, .• ..... -,::,. s,-,,-:-:.:..•,.•,:..;. * Z*'*".. "' "' ... :..:.. . ..!;j•-..* .... '•{•:.½:;%-•::.: .4.•?:;:, . '•: .j .:.,;:•:: ...:½':: •.:•.:.'!B-** .:(•...'.•;(- ß ½' •.-

.... }.;?.... '•*•%;.....4.-;,:.,.. '.-:;::•:--: .......... "..':::;X*}**;:",.::::;:...* :,: ':':*;• .,.•;:½ -,.•w.•:%::---?•,, :'.* ' :':b:<': *:*":'** ' ..•:" ':- ,.$;: "':'::;;'.:•-','...**" :::,'.' ß '. ,;:-- . :..*':- :':.';' %?-,-*:,;.. *'::"'* .• **:t.. *' ...*;.":•*.;-<'.'**;':'""•¾;': %*":'•:. • *' :'•* ,;,

::, ;:'•..?:•.' '* .... *:;:.;: %'b:?'::': ". '*';:-•*;-;?-:.' *,..•." :;'-:-.':'•:;;, ,:;..$..:.¾ ' F?: ...,:j•*,X-'?.: .... ;5•:, '"'"':';•?.:; :.,,...,..'.:':;" .:,:. .';;'.. "•:.-'i:; /-":•.'.7 :•-:-::?.'";"'...'

.... ;:**;:..::•;":':: .... . ½-•-. ':(: ' ........ ':;;•...;; ...... ':.½; .... ".-:-:-.:..,..';:•;;Z,; • ..%. ';"- '. ' ::*" -,% "';:-' -'- -• .';:;b::;"' ..:,'4*-v':;:.; .,.',. ' . .... -"-:•:::* 3'*.: '. ..• a-q:. . :

,,; ..... .... ,.:'.' .::;:. , .... .::-:??." .?'4•.'?:" ..•';; • 7' , . :•;•'/",--':'?. 'b:(;" .;% ..... ;...:;•; ..'.,'

...... •:•:..

....! .

- .... :•:•.:.:.•:. . :: .:.....• :.:

..

...::?{.' , ';:3'::: '

..

Fig. 1. Full moon photograph of the lunar near side. The locations of Apollo and Luna sampling sites are indicated with straight and curved arrows, respectively.

PIETERS' COMPOSITION OF THE LUNAR CRUST 559

side (30øN to -10øS; 30øW to 60øE). Three sampled areas were entirely within highland terrain (Apollo 14, Apollo 16, and Luna 20) and two were on a mare-highland boundary at the mare-filled edge of two large impact basins (Apollo 15 and Apollo 17). The location of the landing sites are indicated on the lunar near-side image shown in Figure 1. The hemisphere of the lunar far side has not been sampled nor is it accessible to earth-based remote sensing techniques.

The most appropriate classification of returned lunar high- .land rocks was not immediately apparent to all lunar chem- ists, petrologists, and geologists and resulted in an expanding profusion of partially synonymous terms. The complex brecciated form of the majority of returned highland rocks required a classification that could convey the normally multicomponent nature of these samples. An Apollo 16 breccia which exhibits typical mineralogic diversity from multiple impact events is shown in Figure 2. An extensive literature exists on the details of breccia types, although usage in the literature has not been uniform until recently. A good summary of the common terms used to describe the structure and texture of highland breccias has been compiled by Taylor [1982].

Currently, terms preferred by most lunar geochemists to describe the bulk mineralogy of highland rocks and breccias are those summarized by $toffier et al. [1980], which are defined by the amount of the major mafic mineral present in a mixture with plagioclase. In this classification the less dominant species is used as an adjective to describe the rock type: >90% plagioclase (anorthosite); between 10% and 22.5% mafic (orthopyroxene, clinopyroxene, or olivine), with the remainder plagioclase (noritic anorthosite, gabbroic anorthosite, and troctolitic anorthosite, respectively); between 22.5% and 40% mafic, with the remainder plagioclase (anorthositic norite, anorthositic gabbro, and anorthositic troctolite, respectively); more than 40% mafic (norite, gabbro, and troctolite, respectively). Additional adjectives may be used to describe the relative proportions of important incompatible elements (for example, enrichment or depletion of potassium and rare earth elements.

Many current characterizations of the composition of the lunar crust discuss more idealized suites of rock types that are presumed to represent the primary constituents of the lunar crust independent of the effects of the intense early bombardment and brecciation. A number of such "pristine" lunar samples have been identified as fragments within the collection, providing important constraints on the nature of the lunar crust. Three components of this primordial lunar crust, each with a different possible origin, are summarized by Taylor [1982], largely from the detailed sample work of P. H. Warren et al., O. B. James et al., and others: First are the anorthosites (or ferroan anorthosites), composed almost entirely of high-Ca plagioclase feldspar (> 90%). These rocks would "float" in the hypothesized cooling magma ocean and could represent the earlier crust [Warren and Wasson, 1980, 1977]. Second is the Mg-rich suite, including significant components of norites, troctolites, dunite, and anorthositic gabbros. These more mafic rocks exhibit a range of plagioclase content, generally < 75%. They are often referred to as plutonic in nature and described as originally occurring as layered plutons within the anorthositic crust [James, 1980; Warren and Wasson, 1980]. Alter- natively, both the ferroan anorthosites and the Mg-rich suite could have been produced by massive amounts of serial mag- matism during early crustal formation [Walker, 1983]. Third is KREEP (or Fra Mauro Basalt), evidenced by high REE, Th,

•i...' ,..:.i:%..½::.:.{' :-'"'%:...

..

..

.•.:.:

:::

Fig. 2. Apollo 16 breccia 60019. This 14-cm cut surface exhibits a variety of clasts several centimeters in size down to the scale of the minerals in the darker matrix.

U, K, etc., a widespread but elusive component that occurs in varing amounts in many rocks, although few plutonic fragments exist to define a rock type. This component is commonly thought to be a (premare) residual liquid of crustal/mantle differentiation.

At the close of the Apollo program in 1972 some direct information about the global distribution for these components was provided by the Apollo 15 and 16 X-ray and gamma ray spectrometers which measured bulk elemental compositions (AI/Si, Mg/Si, Th, U, K, and Fe + Ti) from orbit for the 10-20% of the lunar surface along the Apollo ground tracks. A recent compilation and summary of these orbital geochemical data can be found in Basaltic Volcanism [Basaltic Volcanism Study Project, 1981, chap. 2]. Two very important results were immediately apparent: (1) The distinct chemical difference between the basaltic maria (Fe and Mg rich) and the feldspathic highland crust (A1 rich) was observed regionally for the 10% of the lunar surface measured by the X ray spectrometer. The basaltic nature of the mare and the feldspathic nature of the highland crust were thus confirmed beyond the areas actually sampled. (2) The distribution of radiogenic elements (Th, U, and K) was distinctly asymmetric and appar- ently not associated with the distribution of mantle-derived basaltic units. Of the 20% of the lunar surface measured with

the gamma ray spectrometer, the least radiogenic was the highland crust on the far side, and localized regions on the western near side exhibited the highest concentration of radiogenic elements. If these variations are largely due to a KREEP component, ubiquitous throughout the sampled sites, KREEP clearly occurs in local concentrations, none of which have been sampled.

Since there are strong correlations between many chemical elements for returned highland rocks, Taylor [1975; 1982, chap. 5] has used the results from the orbital geochemical experiments, however limited in coverage, and constraints from geophysical observations to estimate an average crustal composition and ultimately an average composition for the moon. Such estimates require broad assumptions about representative lateral sampling and vertical homogeneity, neither of which are well constrained. Thus as additional compositional information becomes available about the global distribution

560 PIETERS: COMPOSITION OF THE LUNAR CRUST

Lighl-Molrix Breccios I Powders

• ,43,8 36(3, • -

0.5 1.0 1.5 2.0 2.5 WAVELENGTH (•.m)

'-I "' ' ' I .... ' .... ' .... J •>Anorthosite

o• Noritic

• '" Anorthosite

• "•Troctolitic c)

-'• Anorthosite

z Anorthositic •- Gabbro

,, Norite

• Troctolite

0.5 1.0 1.5 2.0 2.5 WAVELENGTH (•m)

w

z

- Dark-Matrix Breccias

_ 6881• -

64435,16 (•C)

7221•

1.0 1.5 2.0 2.5 WAVELENGTH ( I. Lm )

- Non-Mare Melt Rocks 14310,15• ' 20 - -

•'"'---'"'"------• s 15205,43(•c) - _

KREEP

"' •:• 60315,30(•c) _ • Z --

0.5 1.0 1.5 2.0 2.5

WAVELENGTH ( Iz, m)

Basalts

Poikilitic Rocks

Fig. 3. Diffuse reflectance spectra of returned lunar highland rock interior chips (IC) and powders (P) measured in the laboratory [from Chafetie and Adams, 1977]. The spectra are offset vertically with the short horizontal bar indicating the reflectance at that wavelength. The classification scheme referenced here has since been updated by Stoffler et al. [1980] (see text).

of rock types, values for the average composition of the lunar crust will become better known.

3. SPECTRAL REFLECTANCE STUDIES

Throughout the Apollo era, spectral reflectance measurements were being investigated as an additional tool to determine surface composition remotely through identification of mineralogical components from the spectral analysis of reflected light. Absorption features characteristic of specific rock- forming minerals occur in reflected light from 0.35 to 2.5 #m and are documented and discussed by Adams [-1974, 1975] and by G. R. Hunt and J. W. Salisbury (summarized by Hunt •1977, 1982]). Bidirectional reflectance measurements were made of the lunar surface using telescopic instruments, and diffuse reflectance measurements were made of returned lunar

samples using laboratory spectrometers. At that time, laboratory reflectance measurements were more precise and were possible for broader spectral coverage and resolution than were the telescopic measurements for small areas on the moon about 20 km in diameter. The visible and near-infrared reflec-

tance properties of returned lunar samples were analyzed and compared to extended visible telescopic spectra of small lunar areas in a series of papers by Adams and McCord [1970, 1971a, b, 1972, 1973]. A fundamental result of these explora- tory studies involving all landing sites was that the telescopic reflectance measurements of undisturbed surface areas corre-

spond directly (within the precision of measurement) to laboratory measurements of returned mature soils. This "ground truth" documentation provided essential verification that reflectance measurements obtained remotely are comparable to

PIETERS.' COMPOSITION OF THE LUNAR CRUST 561

laboratory measurements and hence can be interpreted in the same context.

From these early reflectance studies of returned lunar samples it was readily recognized that the spectral properties of lunar rocks or crushed lithic fragments were notably different from the spectral properties of mature surface soils [Adams and McCord, 1972, 1973]. Mature soils contain mineral and lithic fragments from local rock types, but they also contain up to about 70% dark absorbing agglutinates, complex glass- welded aggregates thought to be accumulated by the repeated bombardment and gardening of the surface by micrometeorites [e.g., Heiken, 1975]. The amount of agglutinates increases as a soil matures in the lunar surface environment, with a freshly exposed surface containing the fewest. Agglutin- ates were found to dominate the reflectance properties of returned lunar soils [e.g., Adams and McCord, 1971a, b, 1973]. The spectral effects of increasing agglutinate content of a soil lowers the albedo, steepens the reflectance continuum slope toward the near infrared, and lowers the spectral contrast of absorption features from mineral constituents. Since the surfaces of all areas except fresh craters consist of mature soils, most of the early lunar reflectance studies concentrated on deriving information from surface soils. A large fraction of reflectance data available for lunar material (collection of J. B. Adams) are for lunar soils of varying degrees of maturation.

An early telescopic survey of the reflectance properties of many highland and mare areas (using detectors limited from 0.3 to 1.1 /•m) clearly distinguished the spectral properties of mare and highland soils and craters [McCord et al., 1972a, b]. The material excavated by the large crater Copernicus, located in a near-side mare area, was shown to be of highland composition rather than mare, thus providing one of the early clues that the mare was not very thick for that part of the moon. Mare soils exhibited abundant variation in the visible part of the spectrum, due largely to their differences in TiO,_ content (summarized and mapped in a more detailed study of mare basalt types by Pieters [1978]). The notable spectral variations of basaltic mare soils in the visible spectrum allowed their spatial extent to be easily defined with multispectral imaging techniques [Whitaker, 1972; McCord et al., 1976, 1979; E. A. Whitaker, personal communication, 1978]. With only a few exceptions, however, the highland soils appeared remarkably similar to each other in spectral characteristics with these early telescopic studies. Small spectral differences were detected for highland soil surfaces, but since these differences were often smaller than variations due to viewing geometry, no global maps of highland units comparable to those produced for the maria were possible at that time. Highland craters, on the other hand, exhibited sufficient variation to allow an initial classification to be developed on the basis of spectral properties [Pieters, 1977]. These limited extended visible spectra were primarily used to classify surface types, but only a minor amount of mineralogical information could be derived directly from the telescopic spectra of highland surfaces.

Simultaneously, laboratory studies that extended into the near infrared were being pursued for selected returned lunar rocks and breccias [e.g., Adams and Charette, 1975]. A preliminary summary of the near-infrared reflectance characteristics of returned lunar highland rocks was prepared by Charette and Adams [1977] and is shown in Figure 3. The classification scheme used for this initial study has since been replaced by the more widely used terms of Stoffier et al. [1980]. For example, most "light-matrix breccias" are now

O •1 PTROXENE {HTPERSTHENE) Z 0LIVINE • 3-

<I• 3 F'L $ R ifR 0 'r IT )1 I I I I I I I I I o 1.o •--

n' 10• •

o 1.o•

c3 0.6 -a

O 0.2-• 0.5 1.0 1.5 2.0 2.5

WAVELENGTH (microns)

Fig. 4. Laboratory diffuse reflectance spectra of minerals commonly found in lunar highland rocks. Lunar pyroxenes exhibit two well-defined absorption bands near 1 and 2 /•m, the wavelength of which depends on the composition of the pyroxene [Adams, 1974]. The pyroxene shown here is a low-Ca orthopyroxene with band centers near 0.90 and 1.90/•m. Olivines exhibit a single broad multiple band centered just beyond 1 /•m. Iron-bearing (>0.1% FeO) plagioclase exhibits a broad band centered near 1.25/•m.

referred to as "fragmental breccias" with additional information that describes their general mineralogy. Similarily, most "dark-matrix breccias" are "regolith breccias." In addition, some components of the previous "anorthosite, norite, and troctolite (ANT) suite" are now classified as either feroan anorthosites or part of the Mg-rich suite, depending on the amount of modal plagioclase and the Mg* (=Mg/(Mg + Fe 2 +)) of their mafic minerals. Regardless of what classifi-

cation scheme is used, the lunar sample spectra of Figure 3 provide excellent examples of the diagnostic absorption features that can be observed for lunar material. Laboratory spectra for common mineral components in lunar highland rocks are shown for comparison in Figure 4.

As is discussed by Charette and Adams, the dominant mafic mineral noted in almost all lunar highland rock spectra is low-Ca orthopyroxene, as is evidenced by the two pyroxene bands near 0.90 and 1.90 /•m (due to Fe 2+) seen in most spectra of Figure 3 and the orthopyroxene spectrum in Figure 4. With increasing calcium and iron content the pyroxene band centers shift in a regular manner to longer wavelengths [Adams, 1974]; the high-Ca clinopyroxenes of mare basalts typically have band centers near 0.98 and 2.15 /•m. A weak Fe-bearing feldspar band near 1.25/•m can be seen influencing the spectra of many samples in Figure 3 and dominates the spectrum of the anorthosite 15415. Since the overall absorption coefficient of plagioclase is relatively weak, the Fe •- + feldspar band (seen more clearly in Figure 4) is usually only a minor contribution to a rock spectrum even though plagioclase may be a major mineral component fAdams and McCord, 1972]. In addition, if the feldspar has undergone shock in excess of 150 kbar, the crystal structure may be sufficiently altered to eliminate any characteristic absorption bands due to the presence of Fe 2+ [Adams et al., 1979; Bru- kenthal and Pieters, 1984]. Olivine exhibits a broad multiple absorption band near 1.05/•m but no features at longer wavelengths (Figure 4). The two troctolitic samples in Figure 3 (79215,14 and 76535,17) exhibit a broadening and distortion of the 1-/•m pyroxene absorption feature due to a major component of olivine and a minor component of pyroxene in these samples. Multicomponent samples are now commonly ana-

562 PIETERS' COMPOSITION OF THE LUNAR CRUST

1.8

LLI

o 1.4 z <

C3 10 LLI '

o 1.0

ß a MS2/SUN / " b APOLLO lqB/SUN Soils ß C APOLLO 16 [62231.1]

ß A MSR/SUN ß B FAR HRURO 6/SUN Craters

ß C DESCARTES 2/SUN

5"" --'"

- j

, I , i , i , i , i , , , i , , ,

0.7 1.0 1.3 1.6 1.9 2.2

0 I-- < n-

Z

Z o (..3

1.0

1.0

1.0

0.9

0.8

Ap 16

Ap 14

Mare S.

0.7 1.0 1.3 1.6 1.9 2.2

WAVELENGTH (MICRONS) WAVELENGTH (MICRONS)

Fig. 5. Lunar reflectance spectra for soils and craters near Apollo 16, near Apollo 14, and in Mare Serenitatis. The spectrum for Apollo 16 highland soil was measured in the laboratory (soil sample 62231) and plotted with the same resolution as the telescopic spectra; all other spectra were measured with earth-based telescopes. (Left) Spectra scaled to unity at 1.02 #m and offset vertically. (Right) Residual absorption of spectra after division by a straight line continuum fit tangent at 0.73 and 1.60 #m. Vertical lines are drawn at 0.90, 1.00, and 1.30 #m for ease in noting spectral variations. Freshly exposed surfaces (craters) exhibit greater spectral contrast than mature soils. The average composition of the pyroxene component in these surface materials can be inferred from the wavelength of absorption bands near 1 and 2 #m: Apollo 16 contains low-Ca orthopyroxenes and Mare Serenitatis contains abundant high-Ca clinopyroxenes.

lyzed with curve-fitting routines that extract specific information about individual absorption bands. Since all samples studied by Charette and Adams [1977] had not been prepared under the same physical conditions (some are chips, and some are powders), the spectral contrast (band strength) and continuum slope for the spectra of Figure 3 are not directly comparable.

This early laboratory and telescopic experience with lunar materials indicated that identification and quantification of specific mineral components for an unknown surface from a reflectance spectrum is possible and has two implications for measurement strategy: (1) The measurement instrument should have broad spectral coverage (0.3 to 2.5 #m) and high spectral resolution (,-• 10 nm per channel, continuously). (2) Target areas for analysis should be chosen to have sufficient spectral contrast to allow characteristic mineral absorption features to be identified within instrumental precision. The development of sensitive near-infrared detectors in the last decade allowed the instrument requirement to be met, and programs to measure and analyze the near-infrared spectral reflectance properties of the lunar near side have been under-

way since 1976 (summarized by McCord et al. [1981]). Lunar science applications resulting from these new data include those by Pieters et al. [1980-1, Bell and Hawke [1981, 1984], Hawke and Bell [1981], Pieters [1982], Pieters et al. [1983], Pieters and Wilhelms [1985], Spudis et al. [1984], Gaddis et al. [1985], Lucey et al. [1986], Pieters et al. [1985], Smrekar and Pieters [1985], and numerous abstracts of work in progress by various permutations of these researchers. The instrument and observational techniques used for these studies and the data discussed in the following sections concerning lunar crustal composition are described in detail in the appendix.

Selective, but typical, near-infrared spectra of small lunar areas (5-10 km in diameter) are shown in Figure 5. All data are presented as scaled reflectance spectra (scaled to unity at 1.02 #m) since absolute albedo information is essentially lost during the calibration processes (see the appendix). The scaled reflectance spectra on the left exhibit the typical steep continuum slope (overall increase of reflectance toward the infrared) observed for lunar particulate material. To examine the nature of the weak absorption bands near 1 #m, the continuum slope was estimated as a straight-line tangent to the

PIETERS.' COMPOSITION OF THE LUNAR CRUST 563

spectrum (usually near 0.73 and 1.6 #m); the ratio of the spectrum/continuum allows the superimposed features to be studied separately.

Each crater and soil in Figure 5 was chosen to be from the same geologic region to represent freshly exposed material and mature soil from the Apollo 16 area, the Apollo 14 area, and central Mare Serenitatis. The Apollo 16 spectrum of mature soil was measured in the laboratory by J. B. Adams; the spectrum has been smoothed and sampled at the same wavelengths as the other spectra obtained with an earth-based telescope. The Apollo 16 mature soil is an essential part of calibration for the telescopic spectra (see the appendix).

The spectra of Figure 5 demonstrate some of the common systematics observed for lunar reflectance spectra (both laboratory and telescopic): (1) Absorption bands are stronger for rocks and fresh craters than for mature soils. (2) Mature soils generally exhibit a steeper continuum than rocks or fresh material exposed by craters. (3) Although the continuum slope and band strengths are different for soils and their nearby craters, the band centers (and shape) are essentially the same for the two areas from the same geologic unit, since it is the same bulk mineralogy contributing absorption features. (4) The nature of a pyroxene absorption band near 1 #m is distinctly different for the mare than for the highland areas. The band center near 0.90-0.93 #m for these two highland examples (Apollo 14 and 16) is indicative of low-Ca orthopyroxenes as the dominant mafic component, while the band center near 0.98 #m for the mare in Serenitatis is indicative of the high-Ca clinopyroxene mineralogy of the basalts. (The slightly longer band center for Apollo 14 in relation to Apollo 16 indicates an additional clinopyroxene component in typical Apollo 14 material.) (5) The highland spectra exhibit a change of continuum slope between 1.40 and 1.60 #m. (A less steep slope should be used for analysis of the long-wavelength pyroxene band; this was not done in Figure 5). This change of slope is associated with the feldspathic nature of highland material (see sections 4 and 5 concerning rock type A). On the other hand, the in- herently steeper continuum slope around the 1-#m as well as the 2-#m pyroxene absorption bands of the mare regions appears to be associated with the large abundance of high-Ca clinopyroxenes in the basaltic mare.

For both the highlands and the maria the greatest spectral contrast occurs for spectra of freshly exposed surfaces. Since mature lunar soils are dominated by the dark agglutinitic alteration products, the absorption bands of soils are substantially weaker and thus more difficult to accurately measure. Although distinct differences in band strength and band center can be noted between the soil spectra of Apollo 14 and Apollo 16, these characteristics (which are controlled by differences in mineralogy) are more reliably measured and compared from the crater spectra. With data derived from earth-based instruments the most readily detectable mineralogical information applicable to the study of crustal composition thus comes from areas where rocks or rock powders are exposed or abundant. Examples include fresh craters, since they are areas where mature soils have not had time to develop [McCord et al., 1972b; Pieters, 1977], and mountains or massifs, since they also have not accumulated mature soils because of their steep topography. The remaining sections of this review will concentrate on the analysis of such telescopic data accumulated over the last several years.

An important cautionary note concerning the difference of scale between laboratory and telescopic measurements was made by Adams and Charette [1975] and should be carefully

considered when comparing the remote measurements with laboratory materials. The remote measurements are derived from at least a few kilometers of surface area, integrating the properties of all components into one measurement. The laboratory measurements concentrate on a single lithology or chip on the centimeter scale. The only way the two can be comparable is if one lithology strongly dominates the region measured by the remote observations and the physical form of material for the two measurements is similar. There has been

no problem comparing laboratory and telescope lunar soil spectra, since the soil formation process is itself a homogen- ization process. The returned samples of soil are representative of large regions observed remotely. The multicomponent breccias from the highlands, such as 60019 shown in Figure 2, present additional complexities, since it is not only difficult to describe the bulk mineralogy of the whole rock, but it is equally difficult to obtain a representative spectrum of the sample. The use of laboratory spectra of highland rock types as ground truth for the spectra of fresh highland craters will require a more systematic measurement of highland rock type end-member components, an estimation of the effects of physical properties for surface materials, and the development of appropriate mixing models to be used with remote reflectance measurements. The importance of such detailed studies has been recognized over the last several years, and specific topics are being pursued by research groups involved in geological applications of spectral reflectance measurements.

4. NEAR-INFRARED SPECTRA OF EXPOSED

CRUSTAL MATERIAL

Discussion of near-infrared spectra for craters and mountains of the lunar highland crust is divided into two parts. This section includes a summary of the data available, a description of the spectral characteristics of distinct highland near-surface rock types, and the mineralogical interpretation of these observed rock types. The following section contains a discussion of the implications of these data and an analysis of the vertical and lateral distribution of crustal rock types represented by these data.

The areas examined in this survey, grouped according to compositional properties (see below), are listed in Tables 1 and 2 and include small fresh craters in the highlands, steeply sloped highland mountains or massifs, and central peaks and walls of large craters. Many of these areas have been discussed in earlier publications (references in section 3). All areas have a higher albedo than surrounding material and are thus inferred to contain immature soils and/or freshly exposed rocks and lithic fragments. The largest class of immature surfaces studied are fresh craters 5 to 15 km in diameter (indicated as "cr" in the seventh column of Table 1). Since the telescopic measurements cover an area 4-10 km in diameter, usually only the rim of craters 15 to 40 km in diameter was measured (indicated as "rm"). Mountains or massifs are a separate class of surfaces and are indicated as "rot" in Table 1. The central peaks of large craters expose material from a deeper stratigraphic layer than do the small craters and mountains (see below) and are thus analyzed separately (Table 2). Interior walls associated with these large craters (which could be a mix of upper and lower stratigraphic units) are indicated as "w" in Table 1.

Near-surface rock types identified from these telescopic spectra are discussed using some of the same mineralogical descriptions as those used for the returned lunar samples (section 2), although the proportion of mineral constituents are currently only roughly constrained. Identification of specific


TABLE 1. Highland Near-Surface Rock Types

Type* Characteristics Examples

Band

Continuum Strength Center Class Latitude Longitude

N-1

N-2

N-3

N-O

Continuum: 0.35 to 0.50; strength: weak, _< 6%; shape: asymmetric, long; center: 0.90-0.94 #m.

Interpretation: feldspathic, minor low-Ca pyroxene (orthopyroxene) component; highly brecciated with secondary matrix.

Continuum: 0.53-0.65; strength: 5-9%; shape: asymmetric, long; center: 0.90-0.94 #m.

Interpretation: noritic composition (notable orthopyroxene); high breccia content.

Continuum: 0.40-0.58; strength: 10-14%; shape: symmetric; center: 0.93-0.95 #m; plagioclase band present often.

Interpretation: noritic composition; minor Ca pyroxene; brecciated.

Strong Band: > 15%; shape: asymmetric, long; shallow continuum.

Interpretation: feldspathic, pyroxene plus olivine-rich.

Continuum: 0.30-0.55; strength: 7-10% shape:fat, wide center: 0.95-0.99; plagioclase band present.

Interpretation:two pyroxenes, gabbroic anorthosite composition.

Broad band, centered near or beyond 1.0 #m.

Interpretation troctolite (olivine plus plagioclase minor pyroxene if any).

Anaxagoras A center, Anaxagoresd B rim, Descartes 2, Descartes Crater 3,• Descartes Crater 4,• Descartes Crater 6,• Eichstadt H, Gassendi Crater, M. Spumans Crater,õ Menelaus Center, Proclus E Rim, S Ray Crater.

Apennine Front, Conon S Wall, Fra Mauro B,õ Fra Mauro Ridge,õ Hadley East, Hadley Mountain, J Hershel Crater,õ Mosting Rim,õ North Massif, Ptolemaeus Crater, South Massif, S. High Crater A,:!: Sulpicius Gallus Crater, Arzachel Wall,õ Copernicus Wall 1, Copernicus Wall 3, Langrenus Wall, Theophilus Rim.õ

Aratus, Central Crater Hi,õ Descartes 22, Eichstadt G, Fre Mauro 6, Fra Mauro C, Joy, Littrow NR, Mosting A, Piton, Plato B,õ Plato Crater 1.

Eimmart A.

Archimedes A,õ Byrgius, Censorinus, Dark Halo Crater 1, Dark Halo Crater 2, Kepler Center, LaLande Rim, Theoph Wall Crater (7/12), Aristarchus E Wall, Aristarchus N Rim, Aristarchus SW W, Tycho Wall.

Aristarchus P1 Peak,õ Aristarchus S Rim.

0.44 5 0.94 cr9 73.5 N 11.0 W 0.50 5 0.95, rm 74.0 N 13.0 W 0.36 4 0.9.2 cr 10.5 S 16.0 E 0.41 7•' 0.92 cr 10.0 S 15.9 E 0.54•' 4 9 cr 10.5 S 14.4 E 0.40 6 0.91 cr 17.5 S 11.5 E 0.51 5 0.92 cr 19.2 S 79.8 W 0.46 4 0.90 cr 16.5 S 42.7 W 0.42 cr 2.2 N 63.5 E 0.53•' 3 0.91 cry' 16.1 N 15.9 E 0.44 7•' 0.95•' rm 16.1 N 47.3 E 0.42 4 0.94 cr 9.2 S 15.5 E

0.53 7 0.91 mt 23.1 N 1.3 E 0.58 6 0.93 rm 21.4 N 2.2 E 0.62 6 0.93 cr 7.5 S 16.8 W 0.61 8 0.94 mt 2.3 S 15.6 W 0.56 7 0.92 mt 26.9 N 4.8 E 0.53 8 0.93 mt 26.5 N 4.0 E 0.57 9 0.94 cr 62.8 N 39.1 W 0.55 9 0.94 rm 1.1 S 6.0 W 0.55 5 0.93 mt 19.8 N 30.1 E 0.54 115 0.92 cr 8.5 S 0.8 W 0.54 6 0.92 mt 19.8 N 30.1 E 0.45• 8 0.94 cr 42.4 S 23.5 E 0.64 6 0.90 cr 21.7 N 8.9 E 0.65 5 w 17.6 S 3.2 W 0.60 4• 0.92 w 9.3 N 21.2 W 0.55 4• 0.92 w 10.8 N 19.7 W 0.57 8 0.92 w 9.8 S 62.5 E 0.65 5 w 9.7 S 27.1 E

0.50 10 0.93 cr 23.6 N 4.5 E 0.44 10 0.94 cr 16.7 S 10.1 E 0.46 13 0.93 cr 12.8 S 12.2 E 0.38 8•' 0.93 cr 22.4 S 80.8 W 0.40 12 0.95 cr 8.6 S 15.8 W 0.36 19•' 0.95 cr 4.1 S 15.5 W 0.55 13 0.94 cr 25.0 N 6.6 E 0.52 10 0.94 mt 22.3 N 31.2 E 0.38 18•' 0.93 cr 3.1 S 3.1 W 0.58 12 0.93 mt 40.0 N 1.2 W 0.25•' 14 0.95 cr 53.1 N 15.5 W 0.49 14 0.94 cr 53.1 N 0.9 W

0.30 21 0.97 cr 24.2 N 65.6 E

0.51 8 0.96 cr 28.0 N 6.4 W 0.31 9 0.95 cr 24.5 S 65.8 W 0.30 9 0.95 cr 0.5 S 32.6 E 0.60•' 7 0.96 dhc 44.1 S 53.6 W 0.64•' 9 0.97 dhc 44.4 S 67.2 W 0.55 10 0.98 cry' 8.1 N 37.9 W 0.45 9 0.99 rm 4.7 S 9.0 W 0.50 7 0.95 w 10.8 S 25.1 E 0.25•' 10 0.99 w 23.8 N 46.8 W 0.21•' 14•' 0.97 w 24.4 N 47.6 W 6130 13•' 0.97 w 23.9 N 47.9 W 0.38 10 0.97 w 42.9 S 12.6 W

0.52 12 1.04 mt 23.7 N 47.5 W 0.36 6 1.01 w 23.3 N 47.7 W

PIETERS: COMPOSITION OF THE LUNAR CRUST 565

TABLE 1. (continued)

Type* Characteristics Exam pies

Band

Continuum Strength Center Class Latitude Longitude

A Featureless red spectrum; Oriental Interior 2 change of slope near 1.6 #m. (see central peaks II).

Interpretation shocked anorthosite.

0.50 mt 22.5 S 87.0 W

Classes are cr, small highland crater (< 15 km); mt, mountain, massif, ridge; rm, rim of medium crater (15-40 km); w, wall of large crater (>40 km); dhc, dark halo crater (possible highland basalts) [see Hawke and Bell, 1981; Bell and Hawke, 1984].

*See text for description and discussion of types. •'Parameter beyond general type values. $Pre-1980 data, low resolution. õPoor quality of data, tentative assignment.

mineral species from absorption features in the telescopic spectra is similar to that discussed for lunar samples in section 3. Pyroxene is identified by the existence of paired absorption bands near 1 and 2 pm. The average composition of the pyroxene is estimated from the wavelength of the 1-pm band center [e.g., Adams, 1974; Hazen et al., 1978]. Low-Ca orthopyroxenes exhibit a band near 0.91 pm; as the iron and calcium content increase, the band center moves to longer wavelengths. High-Ca, relatively Fe-rich clinopyroxenes exhibit a

band near 0.99 #m and normally exhibit a steeper continuum slope. Band centers near 0.95 #m could indicate a pyroxene of an intermediate composition but more likely indicate a two- pyroxene mixture. Accurate measurement of the pyroxene 2-pm band center helps resolve an ambiguity between the iron and calcium content [Adams, 1974; Hazen et al., 1978]. Al- though the 2-#m absorption band of pyroxene can be detected in these spectra, measurement of the band center is only possible for a few areas since many of these near-infrared spectra

TABLE 2. Central Peaks

Group Characteristics Examples

Band

Rock

Continuum Strength Center Type Latitude Longitude

III

IV

Center: 0.90-0.95 #m; steep continuum.

Interpretation: noritic composition.

No 1.0 #m Fe 2' band; steep continuum.

Interpretation: shocked anorthosite.

Olivine band only; No pyroxene bands.

Interpretation: troctolite.

High-Ca pyroxene bands plus other unidentified components; steep continuum.

Interpretation: gabbroic breccia (plus alteration ?).

Strong high-Ca clinopyroxene bands; shallow continuum.

Interpretation: gabbro.

Aristillus (plus Ca pyroxene), Arzachel, Langrenus.

Alphonsus, Petavius, Theophilus, Piccolomini.

Copernicus Peak 1, Copernicus Peak 2,* Copernicus Peak 3.

Eratosthenes, Alpetragius, Plinius, Bullialdus.*

Aristarchus, Tycho.

0.54 14 0.94 N3 33.7 N 1.0 E 0.62 6 0.93 N2 18.3 S 2.2 W 0.48 5 0.90 N 1 8.7 S 61.0 E

0.51 < 1 A 13.4 S 2.7 W 0.50 < 1 A 25.2 S 60.5 E 0.42 < 1 A 11.3 S 26.4 E 0.60 < 1 A 29.7 S 32.1 E

0.56 5 1.04 O 9.6 N 20.3 W 0.53 7 1.01 O 9.7 N 19.8 W 0.47 12 1.05 O 9.7 N 20.0 W

0.65 7 1.00 G + 14.6 N 11.4 W 0.65 6 0.95 G + 16.0 S 4.5 W 0.58 6 0.96 Gq- 15.3 N 23.6 E 0.60 10 0.97 G 20.7 S 22.3 W

0.20 8 1.00 G 23.7 N 47.5 W 0.28 17 0.99 G 43.2 S 11.4 W

*Poor quality data, tentative assignments.


1.8

1.4

1.0

1.0

10

ß ' d ' ß DF'SCRRTF'S 2/SUN b

ß PROCLUS F' RIH/SUN ½

N-1

0'7 1•0' '113' '1•6' '1'9' '2'2'

WAVELENGTH (MICRONS)

1.0

1.0

1.0

0.9

0.7 1.0 1.3 1.6 1.9 2.2


Fig. 6. Reflectance spectra for rock type N-1 (noritic 1) crustal material: (left) spectra scaled to unity at 1.02/zm and offset vertically; (right) residual absorption after a single straight-line continuum has been removed.

are not considered sufficiently accurate beyond 2.2 /zm, because of a minor thermal component (see the appendix). The detection of the 2-/zm pyroxene band is, however, a key measurement in determination of olivine content: olivine is detected by a broad, multiple band centered slightly longward of 1 /zm but has no prominent feature at 2/zm. Iron-bearing glass, which might be present in large-impact melt sheets, has a similar broad band near 1 /zm but generally can be distinguished by a lower albedo than olivine-bearing assemblages. The identification of the primary mafic mineralogy of surface material is thus coupled to the detection and analysis of the nature of absorption features near 1 and 2 #m. In the discussion below the terms "noritic," "gabbroic," and "troctolitic" refer to the detected presence of orthopyroxenes, clinopyroxenes, and olivine, respectively.

The abundance of plagioclase, a key parameter in the Stof- .tier et al. [ 1980] classification of highland rocks, is more difficult to determine with the currently incomplete laboratory spectroscopy calibrations. As was mentioned in section 3, Fe- beating crystalline plagioclase exhibits a broad weak band near 1.25/zm (Figure 4) and a relatively flat continuum slope. This diagnostic band, however, is easily lost as the result of an impact event or regolith formation. Furthermore, because it is a weak absorption, this band rarely occurs as a well-defined feature. In mixtures with pyroxene the effect of significant amounts of plagioclase (_> 50%) is to weaken or flatten the reflectance peak between the two pyroxene bands [Crown and Pieters, 1985]. Areas of anorthosites 5-10 km in diameter are thus normally identified spectroscopically by a high albedo and the lack of mafic mineral features.

An estimation of mafic mineral abundances comparable to that used for lunar samples classification is possible, but laboratory calibration work with lunar and lunar analog material has only recently been started. Relative mafic mineral abun-

dances can be reliably discussed, and a lower limit to the actual abundances can currently be determined on the basis of mineral mixing experiments for analog lunar soils (D. Crown and C. M. Pieters, unpublished manuscript, 1986). The pyroxene abundances mentioned below are thus 'underestimated, probably in a regular manner. Until the additional laboratory and modeling work which will provide the desired calibration for lunar mineral abundances is completed, the rock type classification used here for the telescopic spectra is restricted to a somewhat more general scheme than the one used to describe returned lunar samples.

This discussion of highland crustal rock types is based on the analysis of observed spectral absorption features for individual lunar areas. It should be emphasized that the spectal reflectance of each of these lunar areas, however, is not neces- sarily comparable to the reflectance of individual laboratory samples of the same nominal composition. Although the nature of spectral features for specific mineral assemblages can certainly be easily recognized, the telescopic and laboratory spectra of Figures 6-15 and 3, respectively, are not expected to "match" in the strictest sense because of significant differences in physical characteristics (particle size, distribution of parti- cles, etc.), probable differences in proportions of components in a complex mixed sample, and possible effects from differences in exposure to the lunar surface environment (solar wind and micrometeorites). Furthermore, most of the telescopic spectra are for areas (impact craters and basin massifs) that have undergone substantial shock brecciation. The brecciation alteration effects are only partially understood [e.g., Adams et al., 1979; Pieters and Horz, 1985] but are expected to affect the strength and symmetry of absorption features by an effective change in particle size (more interfaces), the disordering of feldspar crystals, and the possible addition of impact melt or a reprocessed absorbing matrix material.


LU

z < i--

<

1.8

1.4

N-2

1.0

1.0'

1.0

1.0

O I-- < rc

O

1.0

1.0'

1.0

1.0

0.9

0.7 1.0 1.3 1.6 1.9 2.2 0.7


1.0 1.3 1.6 1.9


Fig. 7. Reflectance spectra for rock type N-2 (noritic 2) crustal material: (left) spectra scaled to unity at 1.02 #m and offset vertically' (right) residual absorption after a single straight-line continuum has been removed.

2.2

Small Craters and Mountains

The variety of distinct compositional groups of highland crustal material derived from near-infrared spectra is summarized in Table 1. Lunar areas are organized according to spectral features that indicate similar mineral assemblages for members in a group. Each near-surface rock type identified in these data is discussed separately below; representative spectra for these near-surface rock types are presented in Figures 6 through 10. Although some areas have been measured more than once, the observational errors associated with each spectrum (see the appendix) always provides some degree of uncer- tainty for each area. When spacecraft spectral measurements of the lunar surface become available, the major results of this initial analysis of crustal rock types are not expected to change significantly, but individual areas studied may be classified with more accurate compositional parameters.

Type N-1 (noritic rock type 1). Many craters near the Apollo 16 landing site, including South Ray Crater, are part of this spectral group (Figure 6). The bulk of returned Apollo 16 breccias should be representative of this compositional type. The spectra exhibit a relatively shallow continuum slope; in the visible spectrum, this shallow continuum slope causes them to appear as relatively "blue" craters in multispectral images. Weak pyroxene absorption bands occur near 0.91 and 1.90 #m, indicating a low-Ca orthopyroxene composition as a minor mafic component. The short-wavelength pyroxene band appears distorted toward longer wavelengths. The weakness

and distortion of the pyroxene bands can arise from two situations: (1) There is a low abundance of pyroxene (5-8%, minimum) in the bulk rock type present and/or (2) the exposed material is highly brecciated and contains components (such as dark or glassy matrix) that reduce spectral contrast and dilute the effects of the mafic minerals present. The cause of the observed weak distorted bands is most likely a combination of both situations. The major component that imparts a high albedo but no absorption features is inferred to be plagioclase; the low abundance of orthopyroxene observed from the spectra implies a noritic anorthositic composition for type N-1 areas, in general agreement with that observed for the Apollo 16 samples.

Type N-2 (noritic rock type 2). Most, but not all, mountains and massifs associated with Mare Serenitatis and Mare

Imbrium are included in this spectral group along with a few craters (Figure 7). Continuum slopes for this group are relatively steep, almost comparable to that observed for regional soils. Pyroxene absorption bands are centered near or slightly beyond 0.91 and 1.90 #m, indicating an orthopyroxene composition. The pyroxene bands of type N-2, stronger than those for type N-I, indicate a noritic composition with an appar- ently larger pyroxene component (7-10% minimum, if unaltered). Many areas associated with the Imbrium basin have stronger bands than the few areas associated with the Ser- enitatis basin.

The short-wavelength band of type N-2 is also distorted toward longer wavelengths, implying brecciation or a compo-


UA

o 18 Z '

u.. 14 U,.I '

o 1.0

1.0

1.0

N-3

o

< cr

o

1.0

1.0

1.0

0.9

0.8

0.7 1.0 1.3 1.6 1.9 2.2 0.7 1.0 1.3 1.6 1.9


Fig. 8. Reflectance spectra for rock type N-3 (noritic 3) crustal material' (left) spectral scaled to unity at 1.02 #m and offset vertically' (right) residual absorption after a single straight-line continuum has been removed.

2.2

nent that lowers spectral contrast. The band distortion observed in types N-1 and N-2 is distinct from that observed when two pyroxene compositions are mixed or when one pyroxene is mixed with a material with a longer wavelength band such as olivine [e.g., Singer, 1981]. The distorted band observed here has a rather sharp or pointed minimum (de-

scribed as a "check" by Hawke et al. [1984]), whereas the mineral mixture distortion produces a broad and fatter band (examples of such mineral mixtures are discussed under the heading of type G, below).

Type N-3 (noritic rock type 3). Small fresh craters from a wide variety of localities are included in this spectral group

z <

<

1.8

1.4

1.0

, ] [ [

ß EIMMRRT R/SUN

N-O

! ' ! , [ , i [ [ [ ! ! • "

, [ , i ! i , ! , ! , ! , ! ,

0.7 I 0 1 3 16 I 9 22


! !

07 1.0

, , [ [ ! ! ! ! ! , ,

! ] [ ] ! ! ! , ! , i ! 1.3 1.6 1.9 2.2


Fig. 9. Reflectance spectra for rock type N-O (norite plus olivine) crustal material' (left) spectra scaled to unity at 1.02 #m and offset vertically' (right) residual absorption after a single straight-line continuum has been removed.

PIETERS.' COMPOSITION OF THE LUNAR CRUST : 569

ß B'mOIUS •./SU. b .

. c,sox.us,su. c .

• 1.4 a 1.0

1.0

0.7 1.0 1.3 1.6 1.9 2.2


1.0

0.7 1.0 1.3 1.6 1.9 2.2 •'


Fig. 10. Reflectance spectra for rock type G (gabbroic) crustal material' (left) spectra scaled to unity at 1.02 •tm and offset vertically' (right) residual absorption after a single straight-line continuum has been removed.

(Figure 8). They often occur near other craters of a different type (for example Descartes 22 is about 140 km from Apollo 16, but most other craters of the region are of type N-I). All type N-3 craters exhibit pyroxene absorption bands that are prominent and symmetric, while the continuum slope is only moderately steep. Assuming that no absorbing or other alteration products are present, the minimum pyroxene abundance would be 10-18%. The band centers for the short-wavelength pyroxene band range from near 0.92 #m for Aratus (Imbrium basin) to about 0.95 #m for Fra Mauro 6, indicating a range

of average pyroxene compositions. Most of these areas exhibit spectra with an inflection around 1.25 #m, indicating the'Pres- ence of crystalline Fe-bearing plagioclase. The composition of these areas is thus generally noritic, but those areas with longer-wavelength band centers also contain a clinopyrøxene component, such as augite. The well-defined feldspar and pyroxene features of type N-3 areas indicate that the mineral components have not been as heavily altered as those for types N-1 and N-2, suggesting a less brecciated rock type or one with fewer absorbing alteration products.

o z

I- o

o

1.8

1.4

1.0

1.0

1.0

ß R'RZS•TZ!'LU'S P•R•/SL•N ' a' ß RRRZRCHEL PERK/SUN b

ß LRNGEENUS PERK/SUN C

z

1.0

1.0

0.9

0.7 1.0 1.3 1.6 1.9 2.2 0.7


1.0 1.3 1.6 1.9


2.2

Fig. 11. Reflectance spectra for noritic central peaks I crustal material' (left) spectra scaled to unity at 1.02 #m and offset vertically; (right) residual absorption after a single straight-line continuum has been removed.


ß RLPHONSUS PERK/SUN 8

PETRVIUS PERK/SUN b

ß THEOPHILUS PERK/SUN C

1.8 PICCOLOMINI PERK/SUN d

-,.4

ß

1 1.0

1.0'

1.0

1.0

1.0

1.0

0.9

0.7 1.0 1.3 1.6 1.9 2.2 0.7 1.0 1.3 1.6 1.9


Fig. 12. Reflectance spectra for anorthosite central peaks l! crustal material' (left) spectra scaled to unity at 1.02 #m and offset vertically; (right) residual absorption after a single straight-line continuum has bccn removed.

2.2

Two exceptional N-3 highland craters exhibit very strong pyroxene absorption bands (comparable in strength to those observed for many fresh craters in the maria). The strength of the pyroxene bands for Fra Mauro C and Mosting A implies a pyroxene component of at least 20-25%.

Type N-O (norite plus olivine bearinfl). One highland crater on the eastern edge of Mare Crisium is currently unique (Figure 10). Eimmart A has the strongest absorption bands observed for any highland crater (21%). Both pyroxene band centers for Eimmart A can be measured, but they are not

consistent with pyroxene being the only mafic mineral present. The long-wavelength band centered near 2.0 #m indicates an average pyroxene composition of Ca/(Mg + Fe + Ca)= 15 + 5% [Adams, 1974], but the short-wavelength band is notably asymmetric and centered at a wavelength (0.97 #m) inconsistent with such a composition. A significant component of olivine (_> 20%) would produce the observed broadening and band center shift (to longer wavelengths), consistent with laboratory spectra of olivine/pyroxene mixtures [Adams, 1974; Singer, 1981]. Eimmart A was included in the observational

z

' COPERNICUS PK1/SUN a

1.8 'COPERNICUS PK3/SUN b

1.4

1.0

1.0

! ! !

a

1.0

J b

1.0

0.7 1.0

! ! i ! i i ! !

0.7 1.0 1.3 1.6 1.9 2.2 1.3 1.6 1'.9 2'.2 WAVELENGTH (MICRONS) WAVELENGTH (MICRONS)

Fig. 13. Reflectance spectra for troctolitic central peaks III crustal material' (left) spectra scaled to unity at 1.02 #m and offset vertically; (right) residual absorption after a single straight-line continuum has been removed.

PIETERS' COMPOSITION OF THE LUNA}t C}tUST 571

1.8

o 1.4 z

o

u_ 10 UJ '

1.0

1.0

ß ERRTOSTHENES PERK/SUN a

ß RLPETRROIUS PERK/SUN b

ß PLINIUS PERK/SUN C

ß BULLIRLDUS PERK/SUN d

1.0

1.0

1.0

1.0

0.9

0.7 1.0 1.3 1.6 1.9 2.2 0.7 1.0 1.3 1.6 1.9


Fig. 14. Reflectance spectra for gabbroic (melt breccias?) central peaks IV crustal material' (left) spectra scaled to unity at 1.02 ttm and offset vertically' (right) residual absorption after a single straight-line continuum has been removed.

2.2

program at the suggestion of R. Strom and A. Treiman (pri- vate communication, 1983) as a possible source area for the lunar meteorite ALHA81005 [Pieters et al., 1983]. These near- infrared reflectance data indicate that this crater area on the

near-side limb is the best candidate measured to date for a

source area of the lunar meteorite.

Type G (gabbroic). Fresh craters of this compositional type (Figure 9) are in the minority for areas associated with the upper lunar crust, but occur across the entire lunar near side from Byrgius (near Oriental) to Censorinus (east equa- torial highlands). Their spectra exhibit characteristic pyroxene absorption bands, but the band centers are at distinctly longer wavelengths, and the pyroxene band shape is broader (larger width/strength) than that observed for the noritic areas of similar band strength. These characteristics are typical of mineral mixtures involving more than one composition of pyroxene [Singer, 1981]. Because of longer-wavelength band centers, the pyroxene composition for type G areas clearly contains a clinopyroxene component that is more iron and calcium rich than that for other highland areas. Most type G areas also exhibit a notable inflection near 1.25/•m, indicating the presence of a significant crystalline Fe-bearing plagioclase component. The composition of type G areas is thus gabbroic in general nature, although the pyroxene abundance is not well constrained beyond a minimum 10-12%.

Type 0 (olivine bearing). A few small lunar areas associated with the Aristarchus plateau and Copernicus crater have been identified as containing only olivine as the major mafic mineral. These areas exhibit relatively unusual features: a broad asymmetric absorption band centered longward of 1 /•m with no obvious band near 2/•m (see central peaks III).

These spectral properties are characteristic of olivine-bearing rocks, or troctolitic material, with no pyroxene component (< 5%). The areas on the Aristarchus plateau (a mountain and a localized area on the south wall of Aristarchus crater) are interpreted as exposed olivine-rich crustal rocks [Lucey et al., 1986]. Fe-bearing pyroclastic glass (which exhibits a band at similar wavelengths) exists in abundance on the Aristarchus plateau [Gaddis et al., 1985], but the areas mentioned here are believed to be uncontaminated by such glass because of their very high albedo. Although poorer quality telescopic spectra suggest there may be additional troctolitic areas on the lunar near side, the Aristarchus mountain is currently the only area clearly identified that is not associated with materials excavated by a large impact event.

Type A (anorthosite). One crater and a massif associated with the Inner Rook Mountains of the Oriental basin exhibit

spectra with no observable mafic absorption bands [Hawke et al, 1984; Spudis et al., 1984]. Such featureless spectra were first observed for the central peaks of a few large craters on the near side and have been interpreted as shocked anorthosite. They are discussed in more detail below (see central peaks II).

Central Peaks

The central peaks of large craters (> 50 km in diameter) are derived from a deeper stratigraphic zone than most material excavated by the small craters discussed above [e.g., Grieve et al., 1981]. Central peaks of 14 large near-side craters have been measured to date. The inferred composition of each of the central peaks studied falls into one of the seven rock types discussed above. Because of their importance to lunar crustal stratigraphy, data for all central peaks studied to date are


ß RRiSTRRCHUS PERK/SUN a .... ' .. 1.8 . TTCHO PERK/SUN b .

0 V ' z

o 14 LU '

LU -

LU 1.0

1.0

,

0.7 1.0 1.3 I 6 I 9 2 2

WAVELENGTH (MIORONS)

0009

0.7 1.0 1.3 1.6 1.9 2.2


Fig. 15. Reflectance spectra for gabbroic central peaks V crustal material' (left) spectra scaled to unity at 1.02 #m and offset vertically' (right) residual absorption after a single straight-line continuum has been removed.

presented here. These spectra are grouped according to their observed spectral features itemized in Table 2 and are presented in Figures 11-15.

Central peaks I. Only a few central peaks exhibit spectral features indicating a noritic composition comparable to that observed for the majority of small fresh areas discussed above (types N-l, N-2, and N-3). Pyroxene band centers near 0.90 #m for Langrenus and Arzachel (Figure 11) indicate a low-Ca orthopyroxene component, whereas the band center near 0.94 #m for Aristillus implies a slightly more Ca-rich component.

Central peaks II. The spectral characteristics of these areas (Figure 12) are distinct: no mafic absorption bands are detected, placing upper limits on possible mafic abundances (estimate < 5% pyroxene and < 15% olivine). A distinct change of continuum slope is observed for these areas near 1.6 #m. The preferred interpretation is a surface composition of shocked plagioclase (anorthosite). Material uplifted to form ihe central peaks of large craters is expected to have undergone the most severe shock pressures during the impact event [Grieve et al., 1981]. Since these lunar areas are high-albedo features, they cannot contain hidden absorption bands that have been masked by dark material.

Central peaks III. The three Central peaks of the crater Copernicus (Figure 13) are in a class by themselves and have been discussed by Pieters [1982] and Pieters and Wilhelms [1985]. All three central peaks exhibit a broad multiple absorption band centered beyond 1 #m, indicating the presence of olivine as the major mafic component. Since no pyroxene band is detected near 2 #m, any pyroxene component is estimated to be less than about 5%. The band strength varies by more than a factor of 2 among the peaks (measured with the same observing conditions and resolution), indicating that the modal abundance of olivine varies in a similar manner. The

strongest olivine component is observed for the small peak in the middle (peak 3). Owing to their observed high albedo the second major mineral component is inferred to be plagioclase.

Central peaks IV. Although the central peaks included in this group clearly contain pyroxenes, they are diverse in character (Figure 14) and exhibit bands at relatively long wavelengths. The nature of the short-wavelength band is not symmetric, however, implying an additional Fe-bearing com-

ponent contributing to the feature near 1.0 #m. The mineral species causing the unfamiliar distortion of the 1.0-#m band for these peaks is not well defined from labortory work but could be either a contaminant produced by impact melt (Fe- bearing glass, recrystalized glass, or metamorphic minerals) or a gabbroic component mixed with but spectrally dominating other mineral components. Since these central peak areas do not exhibit the characteristics of (Fe bearing) impact melt glass measured for areas of known impact melt at large craters [Smrekar and Pieters, 1985], the tentative interpretation for these areas is a mineral assemblage containing a significant gabbroic component. These four large craters have excavated material from depth and have probably incorporated additional alteration components in their mineral assemblages as a result of the impact process. The amount of gabbro and the remaining mineral components in this assemblage are unknown. •

ß

Central peaks V•. This group includes the central peaks of two of the most prominert[ fresh large craters on the lunar near side, Aristarchus and Tycho (Figure 15). Both exhibit strong, well-defined pyroxCne bands at long wavelengths (near 0.98-1.0 and 2.1-2.2 #m). These band centers imply a relatively Ca-rich average clinopyroxene composition of Ca/ (Ca + Mg + Fe) = 35 +• 10%. These are the only central peaks that exhibit a well-defined crystalline Fe • + plagioclase feature near 1.25 om, especially prominent for Tycho. Al- though the crystalline material exposed by these two large craters is gabhrOic in composition, it is readily distinguished from the basaltic material excavated by fresh craters in the maria (for example, Figure 5)' the flatter continuum for spectra from these large craters implies a high proportion of plagioclase mixed with the pyroxene. A minimum pyroxene abundance estimate would be 10% for Aristarchus peak and 20% for Tycho peak.

5. DISCUSSION

From the preceding information it is clear that a wide variety of mineralogical rock types exists throughout the near-side crust of the moon. Major rock types identified from reflectance spectra include noritic, anorthositic, gabbric, and troctolitic compositions. These mineral assemblages mimic rock


types studied from the lunar sample collection. Although the proportions may not be the same, the returned lunar samples contain components of all the major rock types noted for the near-side crust of the moon with these remote sensing techniques. Within the limited number of areas studied in this project, no areas of a totally unexperienced mineralogy were recognized, although some components of reworked material within craters have not been identified.

It is appropriate to reiterate the relation of existing laboratory near-infrared reflectance measurements of returned highland samples (Figure 3) and these remotely obtained near- infrared spectra of small exposed crustal areas (Figures 6-15). Even though the scale of the measurements made for the laboratory and telescopic data is different by a factor of almost 10 a, both contain compositional information that can be readily identified (orthopyroxenes, clinopyroxenes, Fe-bearing plagioclase, olivine, etc.). On the basis of the characteristic absorption bands for known minerals, distinct rock types have been identified here for surface areas using the telescopic spectra. Specific rock types of the returned lunar samples similarly display a combination of absorption bands characteristic of the minerals present. Two of the most prominent and regular spectral parameters of the telescopic data, continuum slope and strength of absorption bands, are strongly affect6d by the physical form of the surface material (particle size, size distribution, compaction, etc.) and viewing geometry of the measurement. Since neither of these parameters were controlled in the initial laboratory studies of lunar samples by J. B. Adams, the infrequent correlations, or direct matches, be- {ween the sample and the telescopic spectra of similar composition are not surprising. Fortunately, identification of characteristic mineral absorption bands is relatively independent of these physical parameters. In addition to detecting absorption bands, and inferring the presence of specific minerals, absorption strength (sensitive to modal mineralogy) can be measured precisely for the telescopic spectra. A more precise quantification of modal abundances from measured spectral parameters, however, requires more extended and controlled data from laboratory measurements and coordinated analytical techniques that extract abundance information about individual components. Such studies are currently being instigated by a variety of investigators in preparation for anticipated geochemical missions to the moon and planets.

It was unexpected, nevertheless, to find lunar areas a few kilometers in extent that are completely dominated by rock types which often, occur only as minor fragments a few centimeters in size in the sample collection (the troctolite, for example, forming Copernicus' three central peaks). The early intense bombardment of the lunar crust and the multiple phases of brecciation frequently observed in the sample collection had suggested the upper lunar crust was well mixed and possibly homogeneous with depth. The mineralogical data presented here for the near-side hemisphere show that this cannot be the case on a lateral nor a vertical scale of tens of kilometers.

Most small a.reas of freshly exposed crustal material not associated with large craters on the lunar near side exhibit a remarkably similar composition. Three quarters of the small craters and mounta{ns studied (Table 1) are generally noritic in composition with varying degrees of brecciation and secondary matrix (types N-l, N-2, and N-3). If the depth of exca- vation is estimated to be _<•0 the crater diameter, these rock types represent the dominant composition of the upper 1 km of thenear-side lunar crust. Only about a quarter of the areas studied representing this upper kilometer (Table 1) contain

significant gabbroic components and infrequent olivine (types G and O). Thus the upper kilometer of near-side lunar crust appears to be compositionally homogeneous in mafic mineral content to a first order, with type of brecciation and variations in pyroxene abundance being the primary differences between areas, and only pockets of significantly different composition occurring sporadically across the surface. This upper compositional zone corresponds approximately to the megaregolith, a 1- to 2-km brecciated and mixed zone thought to result from the early intense bombardment [e.g., Aggarwal and Oberbeck, 1979].

The central peaks of the large craters contain material uplifted from deeper stratigraphic layers, 5 to 10 km in depth (inferred from studies of the central uplifts for terrestrial craters, for example, that of Grieve et al. ['1981] and Grieve and Head [1983]). The variety of rock types observed for central peaks document that the lunar crust is clearly heterogeneous just below the surface kilometer or two of megaregolith. In addition, the central peaks of these lunar near-side craters exhibit a distinctly different distribution of rock t•pes than that observed for the upper kilometer of crust, or megaregolith. Very few of the central peaks studied, only the three group I peaks, are of a noritic composition comparable to that observed for the upper kilometer of lunar crust. The compositions observed for these deeper crustal materials of the central peaks exhibit an array of rock types, many of which are good candidates for outcrops of the more pristine (unmixed, or compositionally unaltered since emplacement) crustal material. The anorthosites (peak group II), the troctolites (peak group III), and the more crystalline gabbros (peak group V) are examples of possible pristine lithologies that occur as mountains tens of kilometers in scale. Numerous fragments of pristine anorthositic and troctolitic compositions have been noticed and studied by persistent lunar sample geochemists •e.g., Warren and Wasson, 1980], but they are only very minor components of the sample collection. Although gabbros comparable to peak group V material have not been abundantly recognized in the lunar collection as potential pristine crustal samples, they may represent a significant pristine rock type for the moon.

In order to address additional questions raised by these data, the distribution of these various near-surface rock types must be examined. A sketch map showing the locations of all the areas studied (coordinates listed in Tables 1 and 2) is presented in Figure 16. The locations of major basins are outlined according to their stratigraphic age [Wilhelms, 1985; Whitaker, 1981; Wilhelms and McCauley, 1971]. Although the number of freshly exposed surfaces that can be measured with earth-based telescopes is somewhat limited, the total distribution encompasses many of the major highland areas on the lunar near side. In Figure 17 the distribution of individual rock types has been separated. Areas included in both Table 1 and Table 2 are combined in these figures with the shape of the symbol coded to distinguish small craters from peaks and walls associated with the large craters.

Crustal material of noritic composition (N-I, N-2, N-3, and N-O) is distributed throughout the highlands (Figure 17, top). There is no apparent, clustering associated with any basin. Neither does there seem to be any apparent pattern to distinguish the three major different types of noritic material, one of which (N-3) appears to be not as severely brecciated or reworked as the others. If there are physical properties that do indeed distinguish these three near-surface rock types, additional clues will have to be sought in morphological studies


ß -o•'"-....... ß sin. craters, rots. •"...... • ,C. peaks

•_,.• •. Ig. craters

'. "" -....., ,•, ;,'""•, ß ß ' ' • ' '•' ' ' ' .......

Fig. 16. Sketch map of the lunar near side showing the location of the areas listed in Tables 1 and 2. The oversized circles indicate

small craters or mountains included in this study; oversized squares indicate walls associated with large craters, and oversized triangles indicate central peaks. The major basins are outlined according to their approximate age: Imbrium (solid lines, youngest), Nectarian (dashed lines), and the hypothesized Pre-nectarian Procellarum (dotted lines). The large 'T' and "P" indicate the proposed center for the Imbrium and Procellarum basins, respectively.

of these craters and their surroundings in order to account for why some areas (for example, N-3) have been spared disrup- tion more than others.

The distribution of crustal material containing or being dominated by gabbroic components is shown in the lower left of Figure 17. In spite of the small number measured, the distribution of crustal material with gabbroic composition is quite different from the global distribution of norites in Figure 17: gabbroic compositions occur preferentially in the western hemisphere. If gabbro is a significant pristine crustal rock type, as suggested by the peak V data, its lack of emphasis in lunar sample discussions may simply be a sampling limitation due to the fact that most samples were collected from the eastern hemisphere (see Figure 1). A parallel limitation is known to occur in studies of lunar basalt samples where less

than a half of identified basalt types have been sampled directly in significant quantities and the majority of unsampled lunar basalts occur in the western maria [Pieters, 1978].

The distribution of both anorthosites and troctolites includ-

ed in the lower right of Figure 17. Additional data is clearly needed to draw any solid inferences from their distributions. If these data are not misleading, however, most of the olivine- rich troctolites (those at Aristarchus and Copernicus) may owe their current surface existence to one of the earliest and largest (possible) basin-forming events in lunar history, namely, that which is thought to have created the Procellarum basin and perhaps thinned the lunar crust permanently in that region (see 'discussions by Wilhelms [1985] and by Pieters and Wil- helms [1985]). •.

These distributions of compositional rock types document the complexity and lateral heterogeneity of the lunar crust. They do not, however, illuminate what is in fact still a major observational paradox: On the other hand, the composition of the megaregolith is relatively homogeneous with noritic material being the dominant near-surface rock type. The deeper (5 to 15 km) crustal material, on the other hand, is not only compositionally very heterogeneous (gabbros, anorthosites, minor norites, and troctolites) but because of the observed major differences in proportion of rock types, cannot be the direct parent of the megaregolith.

Given this dilemma, a few possible solutions should be proposed to be tested. They fall into two categories: one empha- sizes our limited knowledge about the early evolution of the moon, the other our limited technical capabilities in mea- suring and interpreting new data. Acceptance of the implications of these data indicates that contrary to common assumptions the upper lunar crust is clearly stratified, as was argued by Ryder and Wood [1977]. The compositional stratification of the highland crust observed in these reflectance data is an uppermost zone (megaregolith) dominated by noritic breccias overlying a very spatially heterogeneous middle crust containing regions (many kilometers in dimension) of anorthosites, gabbros, norites, and troctolites. This compositional stratification could have occurred (1) during crustal formation (crystallization sequence in a possible magma ocean), (2) just

ß-NI ß-N3

Fig. 17. Sketch maps of the lunar near side indicating the location with oversized symbols for small lunar areas listed in Tables 1 and 2. The top three maps indicate the distribution of areas with a noritic composition (N-I, N-2, and N-3). The bottom left map shows the distribution of observed areas of gabbroic composition (G and G +). The bottom right map indicates lunar areas of anorthosite composition with open symbols (A) and areas of troctolitic composition (O) with solid symbols.

PIETERS: COMPOSITION OF THE LUNAR CRUST 575

after crustal formation (plutonic intrusions and late stage accretion), or (3) subsequent to crustal formation (big basin global redistribution of surface material). Alternatively, (4) the data discussed here may be grossly incomplete (probably) and unrepresentative (possibly) of the lunar crust, or (5) further laboratory and theoretical analysis of lunar materials and (mega)regolith formation may be able to show that what is observed here are noritic breccias can in fact be made from

the anorthosites, gabbros, and troctolites if the are battered and put together in the right manner. The biggest difficulty is the observed pyroxene compositions: there does not appear to be sufficient low-Ca pyroxene in lower crustal stratigraphy to account for that observed in breccias on the surface. All of

these'hypotheses probably play a partial role in accounting for the observations (except perhaps late stage accretion, which would be inconsistent with the ages of the known suite of lunar samples). Which hypothesis is the dominant explanation for the difference between the upper megaregolith and the rest of the lunar crust has yet to be determined and will certainly require more complete global geochemical data.

6. SUMMARY AND CONCLUSIONS

Remotely sensed spectral reflectance data used to infer the mineralogical composition of the near-side lunar highland crust have been synthesized and discussed in the previous sections. New compositional information is derived that sup- plements and expands the knowledge obtained from returned lunar samples. A few highlights are worth reiterating.

1. Characteristic absorption features for spatially extensive surface compositions can be identified in spectra for both soils and freshly exposed material. The most readily detected minerals include orthopyroxene (lo-Ca), clinopyroxene (hi-Ca), olivines, plagioclase, and Fe-bearing glass. The strength of absorption bands, an essential measurement for mineral abundance estimates, is most prominent for freshly exposed material. For mature soils the strength of absorption features is greatly reduced because of the presence of absorbing agglutinates created during regolith formation.

2. For unsampled areas of exposed immature crustal material, largely craters, the average mineral compositions observed are generally comparable to rock types that can be identified in the lunar sample collections. A variety of noritic, gabbroic, anorthositic, and troctolitic compositions, some of which are only found as minor fragments in the lunar sample collection, are all found as distinct near-surface rock types occurring as compositional units that are spatially at least as extensive as the scale of the observations (5-15 km).

3. The mineral assemblages observed for three quarters of the areas studied that sample the upper 1-2 km of the near- side lunar highland crust (the megaregolith) are noritic in composition, the major mafic mineral being low-Ca pyroxene. Three subgroups are identified that are distinguished from each other by pyroxene abundance or characteristics that are commonly dependent on physical properties (such as degree of brecciation, amount and type of matrix, and effective particle size). All three noritic subgroups are intermixed spatially across the near-side highland crust.

4. The composition observed for the noritic upper few kilometers of crust (megaregolith) is distinctly different from the stratigraphically deeper (5-10 km) crustal material. Less than a quarter of the areas measured that sample deep-seated material are noritic in composition. The dominant mineral assemblages of these deeper, less disrupted, and perhaps more pristine, crustal materials are (listed in decreasing number of

areas observed) gabbroic, anorthosite, noritic, and troctolitic compositions.

5. Noritic compositions occur across the entire lunar near- side crust with no apparent clustering associated with any of the major basins. The spatial distribution of gabbroic compositions, however, exhibits a concentration in the western hemisphere.

6. If the statistics of these studies are not misleading, the noritic megaregolith cannot have been derived directly from the average lunar crustal rock types found at 5-10 km depth.

7. The lunar near-side crust is thus both laterally and vertically distinctly heterogeneous in composition. Although the composition of the uppermost portion of the highland crust, the megaregolith, is dominated by one mineralogical rock type, the lunar crust is not well mixed below 1-2 km in spite of the extensive impact record.

As new information for the lunar crust has become avail-

able, lunar scientists have relearned a familiar truth concerning exploration of the unknown: the complexity of an- swers to a geological, geochemical, or geophysical question is dependent on the amount of information about the subject. The simple magma ocean models originally proposed for the origin and composition of the lunar crust are clearly inad- equate to describe the diversity of lunar materials observed across the lunar near side. The more complex models invoking layered plutons within an anorthositic crust [e.g., James, 1980; Warren and Wasson, 1980] are probably more realistic. To be consistent with the mineralogical information discussed above, the scale of the plutons would have to be at least several kilometers, and an origin for the plutons that accounts for the range of compositions observed would need to be developed. Similarly, the relation of the megaregolith to the complex un- derlying crust and the early impact record needs to be clar- ified. Resolution of these unknown relationships are dependent on a global assessment of lunar materials. They underline some of the fundamental unanswered questions about the formation and early evolution of the moon. As the moon is explored in more detail over the next decade by spacecraft with sophisticated sensors, a more complete and thorough understanding of earth's nearest neighbor will emerge. When the unknown becomes familiar and commonplace, the current questions will have been answered or found to be unimportant and will certainly be replaced by new questions that may then be directed more toward utilization than exploration.

APPENDIX: DATA ACQUISITION TECHNIQUES

An ongoing data acquisition program has been underway for the last several years to obtain near-infrared reflectance spectra from 0.65 to 2.5 #m for small (3 to 10 km diameter) areas on the lunar surface. About 350 independent reflectance spectra have been acquired and processed for about 210 areas (some areas were measured on different nights or different lunations). All data were acquired using telescopes at Mauna Kea Observatory (MKO) on Hawaii, which is located at an elevation of almost 14,000 feet (4200 m). This is one of the few earth-based observatories that provides two essential conditions for this type of data: (1) sufficiently low atmospheric absorption, especially in near-infrared water bands, to allow an adequate signal for a continuous spectral measurement to 2.5 #m (see, for example, McCord and Clark [1979]) and (2) good to excellent "seeing" (a term in astronomy used to describe the general turbulence of the earth's atmosphere that limits the clarity and resolution of an image).

Most of the lunar reflectance data were taken with the Uni-


versity of Hawaii/NASA 2.2-m (88-inch) telescope, using either a floor f35 secondary mirror. The highest spatial resolution (,,,3km) required exceptionally still atmospheric conditions,

ß

using a small spectrometer aperture at f35 near full moon when the reflected flux is highest. Only about 5% of the data were obtained under these optimal conditions. About a third of the spectra were measured using a 60-cm (24-inch) telescope. These lower spatial resolution data, obtained early in the program, concentrated on soils in the maria which are relatively homogeneous at the observed scale.

The near-infrared spectrometer used for these measurements was developed by T. B. McCord in the mid-1970s explicitly for planetary spectroscopy needs and has been described by McCord et al. [1981]. The detector used is a liquid nitrogen- cooled InSb (indium antimonide) single detector. Spectral coverage is obtained by sweeping through a circular variable filter in 120 steps, providing a spectral resolution of about 100 A from 0.6 to 1.4 #m and 200 A from 1.35 to 2.5 #m. Each such sweep, or "data run," takes a little less than 2 min and must be repeated many times on the object being measured as well as on a calibration standard (star or lunar landing site). For each data measurement a dark background, or sky, measurement is obtained. During data acquisition a photograph is obtained of the lunar area, as it is seen on the mir- rored aperture, to be used as documentation for area location. Raw data are viewed on a monitor during data acquisition and:•Stored on magnetic tape with observational parameters for later processing. Most processing routines are contained in an interactive software package (SPECPR) initially developed by Clark [1980] for efficient manipulation of spectroscopy data.

For lunar data the calibration standard used is normally the Apollo 16 landing site. Prior to 1980 an area in Mare Ser- enitatis (MS2) was used as a lunar standard with earlier instruments [McCord et al., 1972a, b; Pieters, 1978] because there are no nearby albedo features that can add unwanted spurious data if the area is not properly located through the telescope. With this infrared spectrometer on the MKO 2.2-m telescope, the spatial resolution is sufficiently high to allow accurately repeated measurements of a homogeneous area of Cayley Formation continuous with, but located about 10 km to the west of, the Apollo 16 site. This Apollo 16 standard is measured every 15 to 25 min to monitor atmospheric extinction and calculate appropriate signal adjustments with time.

The raw data are converted to reflectance measurements

through a data processing sequence that includes the use of laboratory reflectance measurements of a carefully chosen returned mature soil from Apollo 16 (62231). The use of lunar soil .as a calibration standard has been found to be more accurate for telescopic reflectance measurements than using "solarlike" stars or stellar models [Nygard, 1972]. Data processing to derive the spectral reflectance of area X can be summarized as

area X Apollo 16 •. halon ApOllo 16telescop e laboratory

ß

• measurements measurement

area X ) = • = reflectancearea x

sun

The first flux ratio of this sequence, a relative reflectance measurement, provides atmospheric and instrumental calibration but still includes observational and statistical errors of

repeated measurements. The precision of the reflectance data

depends on this ratio, which in turn also depends on the observing conditions and the number of times an independent measurement is made. The second ratio, laboratory measurements of 62231, is quite precise (<«%). The laboratory standard, halon, has been calibrated by the National Bureau of Standards (NBS) as an absolute reflectance standard with reflectance values of 97% to 99% in the near infrared (1975 NBS test 232.04/213908). Since albedo data are not derived in this process, lunar reflectance data are scaled to unity at an appropriate wavelength, usually at 1.02 #m for the near- infrared data since there are no major atmospheric absorp- tions at that wavelength.

There are three major sources of error that are not easy to separate for these reflectance measurements. The first is statistical and is dependent on the sensitivity and stability of the detector. Normally, this source of error is the smallest and is less than 1% after averaging four to six measurements. The second is atmospheric transparency and stability. This is not only often the largest source of error, it is also the least pre- dictable and controllable. The procedure described above eliminates most of the systematic atmospheric variations, but nothing can effectively be done about the smaller-scale atmospheric turbulence that earth-based telescopes must tolerate. The third source of error arises from the difficulty in repo- sitioning the telescope at precisely the same lunar location for each observation and keepng it pointed at that same location (within a fraction of an arcsecond) throughout the measurement. For surfaces that contain significant small-scale albedo variations these pointing errors can create signal variations that mimic spectral variations (since they are measured sequentially). The error bars normally plotted with a measured reflectance spectrum are the statistical variation (standard deviation) between repeated measurements of the object area and the standard area during the night. The error bars thus only measure the random nature of the errors and are not sensitive to any slightly systematic or discontinuous variations that occurred over the period of the measurements (about 1.5 hours). For complex surface areas it is thus important to have more than one completely independent measurement of the spectrum.

Data analysis utilizes the scaled reflectance values from this processing. Virtually all lunar telescopic spectra exhibit a continuum that increases toward longer wavelengths. Since most lunar mineral absorption bands are weak and superimposed on this continuum, an early step in data analysis includes the estimation and removal of the overall continuum slope, allowing the residual absorption features to be examined in more detail. For all the data presented here (and in publications since 1980) the continuum around the 1-#m absorption features is estimated as a straight-line tangent to the spectrum (usually near 0.73 and 1.60 #m). The ratio of the reflectance spectrum to this continuum provides a residual reflectance spectrum that enhances the mineral absorption features near 1.0 #m.

Most lunar spectra from the highlands exhibit a change of continuum slope near 1.6 #m. A second less steep continuum normally should be estimated to examine the nature of features near 2.0 #m. Although this can easily be done for laboratory samples, the telescopic measurements may include a minor (< 10%) thermal component of radiation increasing beyond about 2.2 #m when the lunar area and the standard area are observed at greatly different illumination (heating conditions). This thermal component can mimic variations on the weak 2.0-#m pyroxene absorption band and must be pre-


cisely and accurately determined before the band center can be estimated. Removal of the thermal component, which varies with position of the subsolar point on the lunar surface, is theoretically possible [e.g., Clark, 1979], but reliable esti- mations of the proportion of thermal flux at 2.5 #m require accurate determination of the temperature of lunar areas when reflectance data were obtained. Preferably, the temperature of an area should be measured at the same time the reflectance is

measured.

Acknowledgments. NASA support (NAGW-28, NAGW-37) for this research is gratefully acknowledged. The data were obtained while the principal investigator was a visiting astronomer using the University of Hawaii/NASA 2.2-m telescope of Mauna Kea Observa- tory. The assistance of Suzanne Smrekar and David Crown in orga- nizing these data is much appreciated. Careful reviews by G. J. Taylor and an anonymous reviewer were very helpful and allowed the discussion to be significantly improved. This paper is dedicated to Rita Paquette, who, as a planetary geology intern, initially explored the important crustal stratigraphy at the large crater Copernicus.

REFERENCES

Adams, J. B., Interpretation of visible and near-infrared diffuse reflectance spectra of pyroxenes and other rock forming minerals, in: Infrared and Raman Spectroscopy of Lunar and Terrestrial Materi- als, edited by C. Karr, pp. 91-116, Academic, New York, 1975.

Adams, J. B., 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, 4839-4836, 1974.

Adams, J. B., and M.P. Charette, Spectral evidence of highland rock types at Apollo 17: Evidence from Boulder 1, Station 2, Moon, 14, 483-489, 1975.

Adams, J. B., and T. B. McCord, Remote sensing of lunar surface mineralogy: Implications from visible and near-infrared reflectivity of Apollo 11 samples, Geochim. Cosmochim. Acta, Suppl., 1, 1937- 1945, 1970.

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

Adams, J. B., and T. B. McCord, Optical properties of mineral sepa- rates, glass, and anorthositic fragments from Apollo mare samples, Proc. Lunar Sci. Conf, 2nd, 2183-2195, 1971b.

Adams, J. B., and T. B. McCord, Electronic spectra of pyroxenes and interpretation of telescopic spectral reflectivity curves of the moon, Proc. Lunar Sci. Conf, 3rd, 3024-3034, 1972.

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

Adams, J. B., F. Horz, and R. V. Gibbons, Effects of shock-loading on the reflectance spectra of plagioclase, pyroxene and glass (abstract), Lunar and Planet Sci., 10, 1-3, 1979.

Aggarwal, H. R., and V. R. Oberbeck, Monte-Carlo simulation of lunar megaregolith and implications, Proc. Lunar Planet. Sci. Conf., loth, 2689-2705, 1979.

Basaltic Volcanism Study Project, Basaltic Volcanism on the Terres- trial Planets, 1286 pp., Pergamon, New York, 1981.

Bell, J. F., and B. R. Hawke, The Reiner Gamma Formation: Compo- sition and origin as derived from remote sensing observations, Proc. Lunar Planet. Sci. Conf., 12B, 679-694, 1981.

Bell, J. F., and B. R. Hawke, Lunar dark-haloed impact craters: Origin and implications for early mare volcanism, J. Geophys. Res., 89(B8), 6899-6910, 1984.

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

Brukenthal, E. A., and C. M. Pieters, Spectral effects of natural shock on plagioclase feldspar (abstract), Lunar Planet. Sci., 15, 96-97, 1984.

Charette, M.P., and J. B. Adams, Spectral reflectance of lunar highland rocks (abstract), Lunar Planet. Sci., 8, 172-174, 1977.

Clark, R. N., Planetary reflectance measurements in the region of planetary thermal emission, Icarus, 40, 94-103, 1979.

Clark, R. N., A large scale interactive one dimensional array processing system, Publ. Astron. Soc. Pac., 92, 221-224, 1980.

Crown, D., and C. M. Pieters, Spectral properties of plagioclase and pyroxene mixtures (abstract), Lunar Planet. Sci., 16, 158-159, 1985.

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

Grieve, R. A. F., and J. W. Head, The Manicougan impact structure: An analysis of its original dimensions and form, Proc. Lunar and Planet. Sci. Conf 14th, J. Geophys. Res., 88, suppl. A807-A818, 1983.

Grieve, R. A. F., P. B. Robertson, and M. R. Dence, Constraints on the formation of ring impact structures, based on terrestrial data, Proc. Lunar Planet. Sci. Conf., 12A, 37-57, 1981.

Hawke, B. R., and J. F. Bell, Remote sening studies of lunar dark-halo impact craters' Preliminary results and implications for early volcanism, Proc. Lunar Planet. Sci. Conf., 12B, 665-678, 1981.

Hawke, B. R., E. Cloutis, P. Owensby, P. Lucey, J. F. Bell, and P. D. Spudis, Spectral reflectance studies of the Oriental region of the moon' Preliminary results (abstract), Lunar Planet. Sci., X, 352- 353, 1984.

Hazen, R. M., P.M. Bell, and H. K. Mao, Effects of compositional variation on absorption spectra of lunar pyroxenes, Proc. Lunar Planet. Sci. Conf., 9th, 2919-2934, 1978.

Head, J. W., Lunar volcanism in space and time, Rev. Geophys., 14, 265-300, 1976.

Heiken, G., Petrology of lunar soils, Rev. Geophys., •3(4), 567-588, 1975.

Hunt, G. R., Spectral signatures of particulate minerals in the visible and near-infrared, Geophysics, 42, 501-513, 1977.

Hunt, G. R., Spectroscopic properties of rocks and minerals, in Hand- book of Physical Properties of Rocks, vol. 1, edited by R. S. Carmi- chael, pp. 295-385, CRC Press, Boca Raton, Fla., 1982.

James, O. B., Rocks of the early lunar crust, Proc. Lunar Planet. Sci. Conf., 11th, 365-393, 1980.

Lucey, P. G., B. R. Hawke, C. M. Pieters, J. W. Head, and T. B. McCord, A compositional study of the Aristarchus region of the moon using near-infrared reflectance spectroscopy, Proc. Lunar Planet. Sci. Conf. 16th, J. Geophys. Res., 91(B4), D344-D354, 1986.

McCord, T. B., and R. N. Clark, Atmospheric extinction 0.65-2.5 #m above Mauna Kea, Publ. Astron. Soc. Pac., 91, 571-576, 1979.

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

McCord, T. B., M.P. Charette, T. V. Johnson, L. A. Lebrofsky, C. Pieters, Spectrophotometry (0.3 to 1.1 /•m) of visited and proposed Apollo lunar landing sites, Moon, 5, 52-89, 1972b.

McCord, T. B., C. Pieters, and M. A. Feierberg, Multi-spectral map- ping of the lunar surface using groundbased telescopes, Icarus, 29, 1-34, 1976.

McCord, T. B., M. Grabow, M. A. Feierberg, D. MacLaskey, and C. Pieters, Lunar multispectral maps' Part II of the lunar nearside, Icarus, 37, 1-28, 1979.

McCord, T. B., R. N. Clark, B. R. Hawke, L. A. McFadden, P. D. owensby, C. M. Pieters, and J. B. Adams, Moon: Near-infrared spectral reflectance, A first good look, J. Geophys. Res., 86(B11), 10,883-10,892, 1981.

Nygard, S., Alpha Lyrae/Sun flux ratios for use in standard star calibrations' Results from three techniqueS, M.S. thesis, 172 pp., Mass. Inst. of Technol., Cambridge, 1972.

Papike, J. J., F. N. Hodges, and A. E. Bence, Mare basalts' Crystal chemistry, mineralogy, and petrology, Rev. Geophys., 14, 475-540, 1976.

Pieters, C., Characterization of lunar mare basalt types-II' Spectral classification of fresh mare craters, Proc. Lunar Planet. Sci. Conf., 8th, 1037-1048, 1977.

Pieters, C. M., Mare basalt types on the front side of the moon: A summary of spectral reflectance data, Proc. Lunar Planet. Sci. Conf., 9th, 2825-2850, 1978.

Pieters, C. M., Copernicus Crater central peak' Lunar mountain of unique composition, Science, 215, 59-61, 1982.

Pieters, C. M., and F. Horz, Spectral characteristics of experimental regoliths (abstract), Lunar Planet. Sci., XVI, 661-662, 1985.

Pieters, C. M., and D. E. Wilhelms, Origin of olivine at Copernicus, Proc. Lunar Planet. Sci. Conf. 15th, J. Geophys. Res., 90, suppl., C415-C420, 1985.

Pieters, C. M., J. W. Head, J. B. Adams, T. B. McCord, S. H. Zisk, and J. L. Whitford-Stark, Late high-titanium basalts of the Western Maria' Geology of the Flamsteed Region of Oceanus Procellarum, J. Geophys. Res., 85(B7), 3913-3938, 1980.

Pieters, C. M., B. R. Hawke, M. J. Gaffey, and L. A. McFadden,


Possible lunar source areas of meteorite ALHA81005: Geochemical

remote sensing information, Geophys. Res. Lett., •0(9), 813-816, 1983.

Pieters, C. M., J. B. Adams, P. Mouginis-Mark, S. H. Zisk, M. O. Smith, J. W. Head, and T. B. McCord, The nature of crater rays: The Copernicus example, J. Geophys. Res., 90(B14), 12,393-12,413, 1985.

Ryder, G. H., and J. A. Wood, Serenitatis and Imbrium impact melts: Implications for large-scale layering in the lunar crust, Proc. Lunar Sci. Conf., 8th, 655-668, 1977.

Singer, R. B., Near-infrared spectral reflectance of mineral mixtures: Systematic combinations of pyroxenes, olivine, and iron oxides, J. Geophys. Res., 86, 7967-7982, 1981.

Smrekar, S., and C. M. Pieters, Near-infrared spectroscopy of probable impact melt from three large lunar highland craters, Icarus, 63, 442-452, 1985.

Spudis, P. D., B. R. Hawke, and P. Lucey, Composition of Oriental Basin deposits and implications for the lunar basin-forming process, Proc. Lunar and Planet Sci. Conf. 15th, d. Geophys. Res., 89, suppl., C 197-C210, 1984.

Stoffler, D., H. D. Knoll, U. B. Marvin, C. H. Simonds, and P. H. Warren, Recommended classification and nomenclature of lunar highland rocks: A committee report, in Proceedings of the Confer- ence on the Lunar Highland Crust, pp. 51-70, Pergamon, New York, 1980.

Taylor, S. R., Lunar Science: A Post-Apollo View, 372 pp., Pergamon, New York, 1975.

Taylor, S. R., Planetary Science: A Lunar Perspective, 481 pp., Lunar and Planetary Institute, Houston, Tex., 1982.

Walker, D., Lunar and terrestrial crust formation, Proc. Lunar Planet. Sci. Conf. 14th, J. Geophys. Res., 88, suppl., B17-B25, 1983.

Warren, P. H., and J. T. Wasson, Pristine nonmare rocks and the nature of the lunar crust, Proc. Lunar Sci. Conf., 8th, 2215-2235, 1977.

Warren, P. H., and J. T. Wasson, Early lunar petrogenesis, oceanic and extraoceanic, in Proceedings of the Conference on the Lunar Highland Crust, pp. 81-99, Pergamon, New York, 1980.

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

Whitaker, E. A., The lunar Procellarum basin, Proc. Lunar Planet. Sci. Conf, 12A, 105-111, 1981.

Wilhelms, D. E., The geologic history of the moon, U.S. Geol. Surv. Prof. Pap., in press, 1986.

Wilhelms, D. E., and J. McCauley, Geologic map of the near-side of the moon, U.S.G.S. Map 1-703, U.S. Geol. Surv., Washington, D. C., 1971.

Wood, J. A., J. S. Dickey, U. B. Marvin, and B. N. Powell, Lunar anorthosites and a geophysical model of the moon, Geochirn. Cos- rnochirn. Acta, Suppl., 1, 965-988, 1970.

C. M. Pieters, Department of Geological Sciences, Brown Univer- sity, Providence, RI 02912.

(Received June 19, 1985; accepted May 6, 1986.)

composition of the lunar highland crust from near&infrared ... · intimate relative of the...

Documents