Abundance and Distribution of Ultramafic Microbreccia in Moses … · 2013. 7. 25. · metamorphic rocks comprise about 3% of the dike, while dense ultramafic mantle-derived fragments

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 92, NO. B10, PAGES 10,376-10,390, SEPTEMBER 10, 1987

Abundance and Distribution of Ultramafic Microbreccia in Moses Rock Dike'

Quantitative Application of Mapping Spectroscopy

JOHN F. MUSTARD AND CARLE M. PIETERS

Department of Geological Sciences, Brown University, Providence, Rhode Island

Moses Rock dike is a Tertiary diatreme containing serpentinized ultramafic microbreccia (SUM) located on the Colorado Plateau in Utah. Field evidence indicates that SUM was emplaced first followed by breccias derived from the Permian strata exposed in the walls of the diatreme and finally by complex breccias containing basement and mantle-derived rocks. SUM is primarily found dispersed throughout the matrix of the diatreme. We examined the nature of SUM abundance and spatial distribution using data collected remotely by the Airborne Imaging Spectometer (AIS). The minerals serpentine, gypsum, and illite as well as desert varnish and the lithologies SUM and various sandstones were identified from the AIS data. Six end-members (SUM, two types of sandstone, gypsiferous soil, clay-rich soil, desert varnish) were chosen to represent the dominant lithologies of the surface in Moses Rock dike region. Spectra of these end-members were used in an intimate mixing model to deconvolve the AIS spectral data into surface abundance coefficients for each component using a nonnegative least squares inverse algorithm. The results of this calculation of surface composition are consistent with field observations and investigations. SUM distribution and abundance in the matfix of the diatreme were examined in detail, and two distinct styles of SUM dispersion were observed. One style is characterized by high to moderate SUM abundance surrounded by halos of lesser SUM abundance. The dispersion halos grade steeply into regions of the matrix with little to no SUM. The second style is characterized by moderate SUM abundances and broad dispersion halos with SUM distributed across the entire width of the diatreme. These dispersion styles are consistent with eraplacement of the dike as a fluidized solid-volatile system where SUM, which is emplaced early, becomes reincorporated into the turbulent flow of the diatreme by abrasion and comminution by particulate matter carried along in the eruption. This eroded material is then dispersed throughout the matrix by eddy diffusion. The first style represents an early arrested phase of the eruption sequence, while the second style represents a more mature phase. Distri- butions of the second style probably indicate the location of channels where flow was concentrated during later stages of eruption. Since both styles are observed at the same level of erosion, this indicates that the duration of eruption varied along strike at Moses Rock dike.

INTRODUCTION

Moses Rock dike is one of several Tertiary diatremes containing mantle-dedhved matedhal located in the Navajo volcanic field on the Colorado Plateau. Although these diatremes are minor geologic features volumetrically and in extent of outcrop, they contain abundant geologic information as they are believed to erupt from volatile-rich regions deep in the lithosphere. During the rapid ascent of these gas-charged magmas, mantle-derived and deep crustal rocks are commonly entrained along with upper crustal rocks in the diatreme breccias. The xenoliths and

ultramafic mantle-derived components provide insight into the composition of otherwise inaccessible regions of the lithosphere.

The inferred sequence of events for the emplacement of the plateau diatremes involves the propagation of a dike from the source region and eruption of a complex multiphase medium. At Moses Rock dike the flow is believed to have eventually stabi- lized into a few channels which were manifested at the surface

as maar-type eruption craters [Shoemaker, 1962; McGetchin, 1968]. The highly comminuted and brecciated textures of the blocks and particles in diatremes are an indication of the extremely violent nature of their formation, and mixing between the wide range of components in the dike, from upper level sedimentary lithologies to ultramafic mantle rocks, occurs on all scales [McGetchin and Silver, 1972]. The primary mantle-derived component in Moses Rock dike was originally described as kimberlite by McGetchin [1968]. However this ultramafic material lacks the minerals perovskite and phlogopite, rare earth

Copyfight 1987 by the American Geophysical Union.

Paper number 7B 1012. 0148-0227/87/007B- 1012505.00

element systematics typical of kimberlite, and evidence of ever being a true magma [Mitchel, 1986]. Roden [1981] has sug- gested serpentinized ultramafic microbreccia (SUM) as a more appropriate term to describe the ultramafic tuffs of the Navajo volcanic field, and we have adopted this usage for Moses Rock dike.

At Moses Rock dike, SUM is found throughout the dike and is thought to be involved in all stages of the eruption. Although massive SUM is found as dikes and sills injected into the host rocks, the majority of this component (>90%) is found dispersed throughout the matrix of the dike [McGetchin, 1968]. The SUM became dispersed in the diatreme during eruption, and therefore the nature of this dispersion may provide insight into mechanisms for material redistribution in the vent of explosive, volatile-rich eruptions. The best approach to determining the nature of SUM dispersion is to map accurately the abundance and distribution of SUM in the matrix of Moses Rock

dike. However, the detailed mapping of such a component using classical field techniques would be impractical and an inefficient use of resources.

Recent advances in remote sensing technology have led to the development of imaging or mapping spectrometry [Goetz et al., 1985] which can be used to map regional mineralogical variations. Mapping spectrometers measure both spatial and spectral information simultaneously with sufficient wavelength resolution such that a near-infrared reflectance spectrum is measured for each picture element (pixel) in the image. Direct mineralogical information is derived from the position, strength, and shape of absorption bands in the spectra. The absorptions are primarily due to electronic processes involving transition metal ions ( Fe3+,Fe 2+, Cr 2+, etc), and vibrational processes involving H20, OH-, and CO3-, where the position, strength, and shape

10,376

MUSTARD AND PIETERS: ULTRAMAfiC MICROBRECCIA IN MOSES ROCK DIKE 10,377

I

co I

Qs- Quaternary deposits Mrd- Moses Rock Dike

Pcd- De Chelly Sandstone Pco- Organ Rock Member

Pcm- Cedar Mesa Sandstone

Pcha- Halgiro Member

Cutler

Formation

Contact, dashed where concealed

Limit of quaternary deposits Feeler dike

o

Fig. 1. Generalized geologic map showing the contact between Moses Rock dike and the sediments of the Permian Cutler formation which it intrudes [after O'Sullivan, 1965]. The position of the diateme on the Colorado Plateau is shown in the location map in the upper right.

of the absorption bands are controlled by the particular crystal structure in which the absorbing species are contained [Burns, 1970; Hunt and Salisbury, 1970; Adams, 1975]. Although most natural surfaces are complex mixtures of soil, rock, and vegetation, reflectance spectra of mixtures are systematic combinations of the reflectance spectra of the surface components. By deconvolving reflectance spectra into component abundances, the high spatial resolution in imaging spectrometers can be used to map the distribution and abundance of specific mineral components on the surface.

In this paper, data from the Airborne Imaging Spectrometer (AIS) are used to map the distribution and abundance of the SUM component in Moses Rock dike. The geologic setting and composition of Moses Rock dike are reviewed, followed by a discussion of the nature of the surface materials (composition, texture, and relationship to bedrock). These observations along with laboratory spectroscopic data are used to interpret surface mineralogy of the dike and surrounding regions from the imaging spectrometer data. The spatial distribution and abundance of primary surface components are calculated using a nonlinear model for the mixing of spectra from multicomponent surfaces. The nature of the derived distribution and abundance of SUM is

examined in detail and supports McGetchin's [1968] model for the emplacement of Moses Rock dike as a fluidized system.

GEOLOGIC SETTING

Moses Rock dike is located on the Colorado Plateau in the

extreme southeastern corner of Utah (Figure 1). It occurs in a cluster of diatremes east of the Monument Uplift along the axis of Comb Ridge Monocline. The dike is hook-shaped in plan, about 6 km long and 100-300 m in width. It intrudes unde- formed and unmetamorphosed sediments of the Permian Cutler Formation, and the contact between the dike and the sediments

is sharp and the vent walls are steep. At the present level of exposure, approximately 1500 m below the surface at the time of eruption [McGetchin, 1968], the dike is in contact with three members of the Cutler Formation: the Organ Rock Tongue, the Cedar Mesa Sandstone, and the Halgito Tongue. The Organ Rock member consists of dark reddish-brown siltstone, sandy siltstone, and very fine grained sandstone. The lithologies of the Cedar Mesa Sandstone in this region include variegated siltstone and shale, buff sandstone, gray limestone, and gypsum. The Halgito Tongue is composed dominantly of reddish-brown siltstone and very fine grained silty sandstone with thin beds of gray limestone [O'Sullivan, 1965]. The dike was mapped in detail by McGetchin [1968], a summary of which appears in the work by McGetchin and Silver [1970, 1972]. The following account of the geology and eruption of the diatreme is derived from these sources.

The breccia fragments filling the dike range in size from microscopic particles to blocks up to a hundred meters across. The large blocks, derived almost exclusively from the units exposed in the vent walls, constitute the majority of the breccia. The interstices between the blocks are filled with a matrix con-

taining fragments from the large blocks, limestone fragments, crystalline rock fragments, and mafic and ultramafic constitu- ents. Basement-derived igneous, altered igneous, and foliated metamorphic rocks comprise about 3% of the dike, while dense ultramafic mantle-derived fragments of eclogite, lherzolite, and websterite are rare.

Outcrops of SUM constitute about 1% of the dike and are found primarily as small dikes and sills in the wall rocks and large breccia blocks in the dike. SUM is also found as small bodies plastered against the walls of the dike and as pods within the dike. Mineralogically, SUM is a highly serpentinized ultramafic microbreccia containing remnants of olivine, pyrox- ene, garnet, and spinel set in a fine-grained serpentine matrix.

10,378 MUSTARD AND PIETERS: ULTRAMAfiC MICROBRECCIA IN MOSES ROCK DIKE

Petrologic evidence indicates that the source region for the SUM was between 50 and 110 kin, but there was no appreciable melt component associated with this eruption. Instead, the minerals in SUM are thought to be derived from the dissagregation of garnet- and spinel-lherzolite [McGetchin, 1968; Hunter and Smith, 1981]. Dissagregation has also been proposed as a possible mechanism for deriving the ultramafic material in the Red Mesa and Garnet Ridge diatremes which are related texturally and petrologically to Moses Rock dike [Hunter and Smith, 1981]. The fact that the microbreccias of the Colorado Plateau

never had an appreciable melt component implies that models for formation of kimberlite diatremes, which have well-docu-

mented melt components, may not be appropriate for the diatremes of the Colorado Plateau.

Although massive SUM constitutes only 1% of the dike, a further 12% of SUM derived material is found dispersed in varying concentrations throughout the matrix of the dike. The contribution of SUM in the matrix varies from >90% in zones of

diluted SUM to almost zero in regions consisting primarily of Cutler rubble. Complex breccias, identified by the presence of basement- and mantle-derived xenoliths, contain SUM concen-

trations which vary from 50% to 10%. The following observations made by McGetchin [1968] con-

strain the mechanism of dike eraplacement: (1) the absence of metamorphism and deformation in the host rocks indicate that the dike was eraplaced rapidly, (2) no evidence of a silicate melt component in the SUM is observed, (3) the breccias are particulate on all scales with particle size distributions like those produced by comminution, (4) the breccias are intricately mixed on all scales, and (5) stratigraphic and superposition relationships between the breccia units in the dike indicate that SUM dikes were eraplaced early, followed by rubble consisting of large blocks of Cutler rocks and later by units of complex breccias containing progressively larger crystalline rock fragments. These field observations and hydrodynamic modeling led McGetchin [1968] to conclude that the diatreme was formed as a fluidized solid-volatile system probably driven by H20 and CO2.

SURFACE CHARACTERISTICS

The near-infrared reflectance spectra measured by AIS contain two primary components of radiation scattered by the surface: reflection and volume scattering. Diagnostic absorptions are present in the volume-scattered component of this radiation caused by the absorption of photons during transmission through mineral phases. Typically, the penetration of this component of radiation into the surface is limited to the outermost 50-100 [tm [Buckingham and Sommer, 1983] but could extend up to a mil- limeter [Pieters, 1983]. Therefore, in order to understand and

interpret geologically the remote observations of a surface, a thorough characterization of the properties of the topmost layer, the relationship of the surface to the underlying bedrock, and an assessment of the vegetation cover are essential.

Moses Rock dike is located in the arid central region of the Colorado Plateau at about 1600 m elevation. The vegetation in this area consists primarily of small shrubs and bushes with radii of 10-30 cm and heights of less than 0.5 m. Vegetation counts from several regions around Moses Rock dike were obtained for representative 20 by 20 m areas over the sedimentary units and the diatreme in July 1985. Results of these measurements show that the total vegetation cover does not exceed

5% and averages around 2.5%. Also, of the total vegetation, 50% is unfoliated. Two ancillary broadband spectral data sets for this region (infrared color photographs and thematic mapper simulator) chosen to emphasize vegetation indicate that significant vegetation is not detectable with resolutions of 2-10 m except in some wash channels and in the vicinity of springs.

Outcrops of massive bedrock are sparse and restricted mainly to cliff exposures and local topographic highs. More commonly the surface is composed of the local products of weathering, and in this region the dominant weathering processes are mechanical, namely, exfoliation, abrasion, and crystal growth. Chemical erosion and alteration, due primarily to infrequent rainfalls and morning dews, are less important. The primary products of this weathering are fragments of the original lithologies, and these generally accumulate in situ on the surface. Particle sizes range from platey fragments 3-7 cm across to microscopic mineral grains and rock fragments. The dominant particle size in any given area is loosely correlated to lithology. For example, limestone beds commonly weather to platey fragments, while sandstone and siltstone lithologies typically weather to fine-grained particles. Transport of the weathering products is primarily by wind and rare rainfalls sufficient to generate surface runoff.

Surfaces of unaltered crystalline rock fragments, found in the dike, are commonly covered with a dark shiny mineral coating referred to as desert varnish. The coating is only rarely observed on sedimentary rock surfaces and soft, friable lithologies such as the microbrecciated SUM. It has been shown that the physical properties of varnish from widely separated localities are consistent, and the composition is independent of the host rock [Perry and Adams, 1978]. It is typically up to 100 [tm thick and consists of ferro-manganese oxides deposited within a clay matrix [Potter and Rossman, 1977]. Due to the strongly absorbing nature of this coating it, masks the composition of the host rock. Varnish has preferentially developed on the crystalline rock fragments such as meta-basalt and gabbro because the surfaces of these lithologies are stable in the local weathering environment.

The nature of the surface materials (i.e., particle size, texture), relationships to the underlying geologic formations, and the stratigraphy of the upper 10-15 cm were examined at many locations both in the dike and sedimentary units. In general, there is a stratigraphy consisting of three layers. The top layer is composed of erosional lag deposits of locally derived resistant rock fragments 0.5-5 cm across. This layer is most commonly observed on the tail section of the dike and on exposed limestone beds where the fragments are large (>3 cm) and can comprise greater than 60% of the surface. Elsewhere the fragments of the lag deposits are sparse or absent. The middle layer is composed of medium to very fine grained particles predomi- nantly derived from the local substratum or the fragments lying on the surface. It is not uncommon to observe a vesicular tex-

ture in the upper 1-2 cm of the middle layer. The vesicules form from periodic absorption and desiccation of water in clays such as montmorillionite and illite. The vesicular textured mid-

dle layer is found in regions of mature soil profiles and in regions of the dike composed of very fine grained material. The third layer is composed of angular fragments derived from bedrock which typically lies 5-15 cm below the surface.

Although there is no major mixing between geologic units in the weathering process, locally significant deposits of foreign, dominantly wind transported, material accumulate. These deposits are generally composed of very fine grained (<40 [tm) parti-


0.60 [

0.20

0.00 1.1 o

0.40

Uncalibrated AIS Data

_

I , , I , , I , , I

1.40 1.70 2.00 2.30

Wavelength In Microns

Fig. 2. Unprocessed raw spectroscopic data measured by AIS for a surface covered by sandstone near Moses Rock dike. Note the overall decrease in reflectance toward longer wavelengths caused by decreasing solar irradiance and the prominent atmospheric absorption bands centered near 1.4 and 1.9 gm

cles derived from the reddish-brown silt and sandstones of the

Cutler Formation. The deposits occur in wind-deposited sedi- ment traps such as in the lee of small ridges and, on a smaller scale, in the interstices between large surface rock fragments. Examination of the stratigraphy in the areas of interstitial depo- sition shows that mixing between the foreign and local components occurs in the upper 1-2 cm and contamination by this foreign component can be as much as 25-30%.

In summary, surface components are primarily derived from local geologic units. Since the dominant weathering processes are mechanical, there is little or no chemical erosion and altera-

tion. Particles sizes on the surface range from silt to cobble, and the dominant size varies with the lithology from which it was derived. Vegetation cover is sparse and homogeneously distributed. To first order, the surface components imaged by AIS are an accurate reflection of local bedrock geology, although contamination by foreign, wind-blown material is a factor in some localities. These variables need to be considered when interpret- ing the spectra geologically and when applying mixing models.

DATA ACQUISITION AND CALIBRATION

Spectroscopic data were acquired over Moses Rock dike with the Airborne Imaging Spectrometer (AIS). AIS is an infrared mapping spectrometer that images 32 cross-track pixels simultaneously in 128 contiguous spectral bands [Goetz et al., 1985]. The detector for this instrument is a 32 by 32 element array which is stepped through four grating positions to obtain the 128 spectral channels. In order to maintain signal dynamic range, the analog data are digitized with a 12-bit analogue to digital converter (ADC). The ADC switches among four gain states and is then recorded as 8-bit data in a 16-bit word. The

particular configuration of gratings used in the flights over Moses Rock dike recorded data from 1.155 to 2.337 [xm with a spectral resolution of 9.3 nm/channel and a ground resolution of about 10 m/pixel. The swath width is therefore approximately 320 m. The AIS data were collected in late July 1984 under clear and cloudless conditions.

The uncorrected data for AIS, expressed as intensity in Figure 2, are dominated by two broad atmospheric water absorption bands centered at 1.4 and 1.9 [xm and a falloff in intensity

toward longer wavelengths caused by decreasing solar irradiance. In order to identify mineral absorption features it is neces- sary to correct not only for the atmospheric and solar effects shown in Figure 2 but also for overall detector and instrumental response. Currently, the most efficient and satisfactory technique toward this purpose is to calibrate the data using an internal standard such as a homogeneous low-relief region contained within the AIS scene. For optimum results the standard area must either have known spectral reflectance characteristics or a featureless spectrum.

The area selected for standardizing the Moses Rock data is located south of the dike in quaternary sand deposits (Qs in Fig- ure 1). These consist of wind-deposited sand of uniform texture and color with an average particle size of 150 [tm. The surface of the deposit is characterized by low undulating swells and hol- lows with a local relief of less than a meter. Twenty continuous lines of data were selected from a homogeneous region in the data and averaged in the along-track direction only. This was done to preserve systematic across-track detector variations in the calibration data. Each line in the AIS scene was then

divided by the along-track average for the 20 lines from the internal standard. Reflectance data processed in this manner are referred to as relative reflectance.

The results from this first-order calibration technique were sufficient to identify several mineral species [Mustard and Piet- ers, 1985] that were later verified by field investigations. How- ever, a laboratory reflectance spectrum of a surface sample from the standard area, shown in Figure 3, revealed several spectral features which bias the relative reflectance AIS spectra. In the wavelength region of the AIS data there is an overall small increase in reflectance toward longer wavelengths, OH_ and H20 overtone bands at 1.4 and 1.9 •tm, and several small but distinct absorption bands longward of 2.1 [tm in the laboratory spectrum. Thin section and XRD analysis of the standard indicates that it is composed primarily of rounded quartz grains with thin iron oxide coatings, 10% clay, and less than 1% opaques. To correct for these features, the AIS relative reflectance data

are multiplied by the laboratory spectrum of the standard material obtained under viewing conditions similar to that for AIS

0.70

0.60

0.50

0.40

0.30

, I I '

Standard

0.20 ' ' • ' ' 0.60 0.90

• AIS Coverage Almospherm Water Bands

/, '1 I ".'.'.L.v..'..'.'-.t. ' ' 'l.':..V.'•:.:.'..'::¾.il ' '

I r.-;?..'-..'-.'.-'.:• r;::'.'.:;i'-':'.:i?;i'-'.:. i E:.::':: :.. ß --:. ß ß :..'..:..'..:..': :::..'::..'::'.'::

I ?.:.2..'.:;..'-.'..'.:½• t.':.:Z ;':"':':"':':':':1

':½.'5:...

::'.'.'::'.':::5 '.':::('" :'-'::?-'"::1 I ,' :' '"' .'.::i!l i '-...-':',..-':'.. I i"::- 'i:i.il I ':' .. ß ."'... '........

I •.•o •.•o •.ao z•o

Wavelength In Microns 2.40

Fig. 3. Laboratory reflectance spectrum of soil from the standard area which consists of quartz grains with Fe-oxide coatings and clay. The bands in the visible and at 0.9 [tm are due to Fe 3+ and Fe 2+, while the OH- bands near 1.4 and longward of 2.0 [tm and the H20 bands at 1.9 [tm are characteristic of the clay illite.


0.80

0.70

(:v 0.60

o 0.50

rT- 0.40

0.30

0.20 0.60

,• AIS Coverage ß Atmospherm Water Bands

0.90 1.20 1.50 1.80 2.10 2.40


Fig. 4. Laboratory spectra (solid lines) of illitc and the standard together with AIS spectra (asterisks) of soils interpreted to contain illitc. The wavelength interval covered by AIS is indicated as well as regions of poor data quality in the atmospheric water bands. See caption for Figure 3 for band assignments.

data. This cancels the effects of the slope and absorption features of the standard area and calibrates the AIS data using the absolute calibration of laboratory standards.

The quality of the resulting calibrated AIS data is generally good, and absorption features can be interpreted in terms of mineralogy. However, distinct line to line spatial banding in the AIS data near the broad atmospheric water bands and a very low signal to noise ratio in the center of these bands result in a notable decrease in the precision and accuracy of the data for the wavelength regions 1.32-1.52 rim. and 1.76-2.03 rim. Because of the overall low quality of the data in these wavelength regions, these data were not used in these analyses and for purposes of clarity have been omitted from the figures of AIS spectra.

Other spectral data important to these analyses include laboratory and field spectral measurements. Laboratory spectra of samples returned from the field were measured in the RELAB, a high-resolution bidirectional reflectance laboratory [Pieters, 1983], using approximately the same viewing geometry as the AIS reflectance data with an incidence angle of 30 ø and an emergence or detection angle of 0 ø. Most samples were prepared as bulk soils to simulate field conditions. Selected soils were sieved to determine the range of dominant particle sizes on the surface.

Reflectance spectra measured with RELAB are obtained relative to the standard halon. Halon is an inert fluorocarbon and is

spectrally featureless in visible and near-infrared wavelengths. It has an average reflectance of >96% and can be approximated as a Lambertian reflector (scatters light equally in all directions) [Pieters, 1983]. Reflectance values measured relative to halon are then calibrated to the absolute reflectance of halon as deter-

mined by the National Bureau of Standards (1975 National Bureau of Standards test 232.04/213908). The laboratory spectral measurements are to first order true or absolute reflectances.

Field spectra of soils and vegetation were measured with a Geophysical Environmental Research infrared interferometer spectroradiometer (IRIS), a portable field spectrometer. These spectra are an important cross reference between the AIS data and the laboratory spectra. The field spectra were obtained in

July 1985, an apparently wetter season than July 1984. IRIS is a dual-beam spectroradiometer which simultaneously measures the reflectance spectra of a sample and the standard used for calibration. The instrument has a spectral resolution of 2-4 nm and samples a surface area of approximately 10 cm 2 The standard used in the field for these measurements is Fiberfax. The

raw reflectance data are initially calibrated by dividing by the simultaneously measured spectrum of Fiberfax. To maintain consistency in our measurements, a sample of Fiberfax was measured in RELAB, and then the IRIS relative reflectance

spectra were multiplied by the RELAB spectrum of Fiberfax. This then calibrates the laboratory, field, and AIS spectral data to the same standard. Although no independent wavelength calibration was available for the IRIS data, comparisions to laboratory data indicate that a 0.012-0.014 rim correction is required in the IRIS data for wavelengths longer than about 1.6 rim. However, this does not affect the results discussed here.

SURFACE MINERALOGY

Several surface units with distinct compositions were identified from the calibrated AIS data. Interpretations of surface mineralogy were aided by field and laboratory spectra and from samples returned from the field. In the wavelength region of AIS (1.15-2.34 rim), clays and other minerals containing structural water and hydroxyl groups commonly exhibit characteristic absorptions between 2.0 and 2.34 rim. Absorp- tions associated with the transition metal ions are less prominent in this wavelength region, although the presence of these absorptions may be inferred from the overall slope of the spectra.

Two AIS spectra of surface soils interpreted to contain clay minerals are shown in Figure 4 along with laboratory spectra of illite and a sample from the standard area. Both AIS spectra are characterized by a slight increase in reflectance toward longer wavelengths (positive slope), a reflectivity maximum near 2.14 rim, a 5-15% absorption near 2.22 rim followed by a local reflectivity maximum near 2.28 rim. The 2.14- and 2.28-rim maxima are asymmetric about the 2.22-rim absorption where the 2.14-rim peak is brighter. The asymmetric reflectance peaks about the 2.22-rim absorption band are characteristic of several clays including muscovite, montmorillionite, and illite. How- ever, laboratory reflectance spectra of field samples (Figures 3 and 4) show a small absorption at 2.34 rim and a reflectance peak at 2.37 rim which are more diagnostic of illire than other clays. The positive slope is due primarily to absorptions associated with reddish-brown iron oxide bearing soils or with Fe 3+ in the clay structure.

Field spectra and laboratory spectra of SUM are shown in Figure 5 along with a calibrated AIS spectrum of a surface interpreted to contain SUM. This AIS spectrum is characterized by a reflectivity maximum near 2.2 rim, a 15-30% absorption centered at 2.33 rim, and a strong positive slope. The features longwards of 2.0 rim are due to a combination of a MgOH- bending mode with OH-stretching fundamentals and overtones [Hunt and Evarts, 1981]. The positive slope is due primarily to electronic absorptions in Fe 2 + and Fe 3 + ions at 0.72, 0.91, and 1.1 rim observed in both the laboratory and field spectra [Adams, 1974]. X-ray diffraction patterns of field samples collected from the surface units containing the strongest serpentine spectral features indicate that the primary mineral in this unit is antigorite.


0.80

0.70

•o 0.60

"• 0.50

rr' 0.40

0.30

0.20 0.60

' ' • ' '1 • ' :'-:. ::.'..':..t .... t.':.:..':.'..'.'.::':..:[• ' '1 • I

SUM I I:::ø::'::;'"'=":;'::'";I ½':i'"':'"::':'='"::'::":::":ti ] t iD e s e r t V a r n i s h i I:;i:..'.::i:.i• '?:.:."1 [".:i'"-:."'=:'.:) I ,i:ii.:i::jo:..::::i½ I J ,:.:.;-:.--;i,' 5i r?:::.-.:'i::j.:!

i'h ':"'::;'; •0 0.30 -• AIS

0.20 ........ 0.90 1.20 1.50 1.80 2.1 0 2.40 0.60 0.90 1.20 1.50 1.80


AIS Coverage • • AIS Coverage ß Atmospheric Water Bands Atmospheric Water Bands

.ii::.':i:•

2.1 0 2.40


Fig. 5. Laboratory (solid line), field (crosses), and AIS (asterisks) reflectance spectra of SUM-bearing soils. The wavelength interval covered by AIS is indicated as well as regions of poor data quality in the atmospheric water bands. The 2.33-[zm absorption bands, 2.2-[zm reflectance peaks, and minor 2.15-[zm absorption bands seen in all three spectra are due to Mg-OH stretching mode absorptions. The AIS spectrum is offset by 10% toward lower reflectance for clarity.

The AIS spectrum is from the same region as the field measurements and would be expected to exhibit similar features. It is evident from Figure 5, however, that the AIS spectrum has the same overall characteristics as the laboratory and field spectra except that the 2.33-•tm absorption band in the AIS spectrum is 20% less strong. Also the maximum reflectance in the 2.04-2.34 •tm wavelength region of the AIS spectrum is 5-10% less than is expected from the lab and field spectra. The most likely source for the observed discrepancy is a recently detected instrumental problem with the AIS detector concerning second order overlap.

In grating spectrometers, second-order spectra are generally eliminated with long pass filters so that only first-order spectra are measured. This was not done for much of the AIS observa-

0.80

CD 0.60

(!.) 0.40

0.20

0.00 0.60

• AIS Coverage . • Atmospheric Water Bands

' ' I ' ' ' I ' "''"/"t* ' ' '•

:....-::i:::ii :...........-.:..¾.:.:....::......., ':,..'):..'.:•;i)

. ...-.:::::½.?;.....(:;,

,i::i'..":ii½.'i½

i:' ."-.-. .' .'. . .-.?, , , I, I 0.90 1.20 1.50 1.80 2.1 0 2.40


Fig. 6. Laboratory (solid lines) and AIS (asterisks) reflectance spectra of gypsum and gypsiferous soils. The gypsiferous component is identified in the AIS spectrum by the broad 2.2-•m absorption, the prominent 1.68-[tm reflectance peak, and an overall decrease in reflectance toward longer wavelengths.

Fig. 7. Laboratory (solid line), field (crosses) and AIS (asterisks) spectra of surfaces containing desert varnish characterized by very low albedo and a small 2.33-[tm absorption band.

tional program, including the Moses Rock flights. It is suspected that this design error caused a component of radiation to be included in the measurements between 1.6 and 2.34 •tm proportional to the brightness of the spectrum between 0.8 and 1.17 •tm. The contribution to the measured spectra between 1.6 and 2.4 •tm may be of the order of several tens of percent [Vane, 1986]. Since an internal standard was used to calibrate the Moses Rock data and the standard region has a somewhat higher than average albedo for the AIS scene, the apparent strengths of absorption bands in the calibrated data are reduced, and regions which have strong positive slopes display an apparent drop in reflectance in the 2.04-2.34 •tm wavelength region as illustrated by the AIS spectrum of SUM in Figure 5. Although at present we are not able to correct fully for this instrumental effect, the diagnostic features of surface materials are clear.

Laboratory spectra of pure gypsum and gypsiferous soils, common in the Cedar Mesa member of the Cutler formation, are shown in Figure 6 along with an AIS spectrum from a surface interpreted to be gypsiferous. The prominent spectral features in the AIS spectrum are an overall negative slope in the spectrum, a distinct 1.68-•tm reflectance peak and a 2.21-•tm absorption band centered between asymmetric reflectance peaks, the higher one at 2.08 •tm and the lower one near 2.3 •tm. The absorption bands, shown more strongly in the spectrum of pure gypsum, are due to overtones and combination overtones of molecular water in the crystal structure of this hydrated calcium sulfate [Hunt et al. , 1971 ].

The spectra shown in Figure 6 clearly illustrate the effects of mixing between different geologic materials on reflectance spectra. The strong absorptions shown in the spectrum of pure gypsum are much reduced when gypsum occurs as a component in reddish-brown soils derived from local sandstone and gypsiferous sandstone. This effect is observed both in the laboratory spectra and AIS spectra, and the degree of attenuation is proportional to the amount of reddish-brown soil mixed with the gypsum.

AIS spectra from the tail region of Moses Rock Dike are shown in Figure 7 along with IRIS and laboratory spectra of desert varnish. As discussed earlier, desert varnish is common

as a coating on stable surfaces such as the crystalline rock frag-


ments in the dike. The laboratory and field spectra of the desert varnish in Figure 7 are characterized by a positive slope, low reflectance (10-20%), a reflectivity maximum near 2.2 •tm and a 7-15% absorption near 2.32 •tm. AIS spectra from the same area as the field samples, however, show a reflectivity maximum near 1.7 •tm and a 10% absorption at 2.32 •tm, and the reflectances are in general >20%. The higher reflectance relative to the spectra of pure desert varnish indicates that mixing between desert varnish and units of the dike and Cutler For- mation is occurring in pixels of AIS. The drop in reflectance in the last grating position of the AIS data is due to second-order overlap.

APPROACH TO ABUNDANCE ESTIMATES

In addition to detecting the presence of individual mineral species, spectral reflectance data can be used to derive direct information about the abundance of components on the surface. Since the AIS spectral data contain spatial information, the spatial distribution of the surface components can also be examined. Reflectance spectra of mineral mixtures, however, are not a simple linear combination of the spectra of the mineral components in the mixture. An analytical model that accurately describes the observed spectrum of multicomponent mixtures is required to deconvolve AIS reflectance into abundances of individual surface components. SUM abundance and distribution in Moses Rock dike can then be determined and analyzed quantita- tively.

The systematics of spectral mixing between components in a mixture can be described as belonging to one of two broad classes, macroscopic and microscopic, where these classes are defined largely by the scale of mixing between components in the mixture. Macroscopic mixing occurs when the components of the surface are arranged in discrete areas or patches that are large relative to the average path length of a photon. In this case the spectrum of the surface is a linear combination of the spectra of the surface components [Singer and McCord, 1979]. Microscopic mixing refers to fine-grained, intimate mixing of the surface components where photons often interact with more than one type of material. In this case the mixing systematics are nonlinear [Nash and Conel, 1974; Singer, 1981]. Different analytical models have been developed for both macroscopic [e.g., Singer and McCord, 1979] and microscopic [Hapke, 1981] mixtures. Therefore, in order to calculate accurately the abundance of surface components from the AIS data, the dominant type of mixing first needs to be determined.

The general physical nature of Moses Rock dike can be characterized as large blocks of locally derived wall rock set in a fine-grained, fragmented matrix which contains the dispersed component of SUM. Although wind transport and splash tend to mix intimately and to homogenize locally the surface, at the scale of the AIS pixels (10 m by 10 m), the type of mixing between erosion surfaces developed over blocks in the matrix and surfaces of the matrix constitutes macroscopic mixing. Mix- ing in the matrix between SUM and particles derived from comminution of xenoliths in the vent, however, is primarily microscopic. Since microscopic mixing of SUM with other breccia components is expected to dominate the spectral reflectance measurements of the matrix, a model for intimate mixing is used in determining SUM distribution on the surface.

Mustard and Pieters [1987] have modified Hapke' s [1981] equations for bidirectional reflectance such that the abundance

coefficients of components in intimate mixtures can be accurately calculated from reflectance spectra measured with a fixed viewing geometry if spectra are available for each component or end-member in the mixture. In this approach, reflectance spectra of the end-member components and the mixture spectra are con- verted to single-scattering albedo (SSA). The proportion of each end-member is calculated by a least squares fit of the end-member SSA spectra to the mixture SSA spectrum. For each component an abundance coefficient is calculated that is referred to as

the relative geometric cross-section, or F parameter. The F parameter is defined by

F, = (M, I @,d,) I Y.(M,,/@nd,,)

where F,, M,, @,, and d, are the F parameter, mass fraction, density, and particle diameter of component i, respectively, and n is the number of components in the mixture. For a surface containing components having similar effective particle size and densities, the F parameters calculated are roughly equivalent to mass Ëaction. This is the case for the intimately mixed matrix regions of the diatreme, and in the discussion below, the term abundance is used interchangeablly for F parameter in the discussion of SUM distribution and abundance.

The F parameters of end-member components in surface material at Moses Rock dike are calculated from the AIS data

using an advanced mixture program developed by the Jet Pro- pulsion Laboratory as part of the Spectral Analysis Manager (SPAM) software package. The program was modified at Brown University to include the method of conversion from bidirectional reflectance to single-scattering albedo developed by Mus- tard and Pieters [1987]. The advanced mixing program uses a constrained, nonnegative least squares inverse algorithm to calculate F parameters for each component in every pixel. The numerical results are quantitative and can be displayed as a series of images, one for each end-member, showing the abundance coefficient of that mineral or end-member in gray level intensities. A residual is also calculated for each pixel to determine the quality of the fit of the model spectra to AIS spectra.

Determination of appropriate end-member reflectance spectra to be used in deconvolving the AIS data into surface abundances of individual components is critical. The end-members must be representative of materials in the dike and be unique with respect to one another. The end-member spectra need not be of pure minerals but may be of mineral mixtures [Mustard and Pieters, 1987]. Therefore common homogeneous lithologies, like the reddish-brown sandstone found throughout the field area, can be accurately represented by a single end- member spectrum.

The end-member reflectance spectra used in this analysis were selected on the basis of field observations, laboratory data, and interactive analysis of the AIS images. The end-member spectra chosen are shown in Figure 8 and include SUM, standard, dark sandstone soil, gypsiferous soil, clay-rich soil (contains more illite than typical soils), and desert varnish. The SUM and dark sandstone soil spectra were carefully selected from areas of known composition in the AIS data, while the rest of the end- members are laboratory reflectance spectra of field samples. The spectra selected from the AIS data minimize the problem of second-order overlap. Although replacing laboratory spectra of SUM with spectra derived from the AIS data does not significantly alter the derived distribution patterns discussed below, the calculated F parameter values are more accurate with the internally derived SUM spectrum. The SUM end-member

MUSTARD AND PIETERS' ULTRAMAfiC MICROBRECCIA IN MOSES ROCK DIKE 10,383

0.70

0.60

0.50

c• 0.40 -

•_ 0.30 -

0.20

0.1 0

Endmembers

0.00 ' , I , , I , , I I 1.1 0 1.40 1.70 2.00 2.30


Figure 8. Endmember reflectance spectra used to derive surface abundances from the AIS reflectance spectra. These are gypsiferous soil (crosses), clay rich soil (squares), standard (circles), SUM (solid line), dark sandstone (triangles), and desert varnish (inverted triangles)

contains >90% serpentine, the gypsiferous soil contains approximately 60% gypsum and 40% fine-grained quartz with iron oxide coatings, and the dark sandstone soil is similar to the standard in composition except it contains a greater proportion of opaques and iron oxide materials.

Carbonate, although a component of the breccias of Moses Rock dike and host rock sediments, is not included in the suite

of end-members because the spectra of SUM and carbonate have similar overall albedo and absorption band positions and strengths in the wavelength range of the AIS data. In the least squares approach to calculating surface abundances, serpentine and carbonate cannot be distinguished with the present spectral range and signal to noise ratio of AIS. Given improved detector response and a greater spectral range, however, these components should be readily differentiated with the more advanced sensors being developed (e.g., AVIRIS, HIRIS)

RESULTS

Figure 9 shows the areas for which abundance coefficients of surface components have been calculated. The abundance and distribution for all end-members will be discussed for the flight line area outlined by the dashed line. Discussion will then focus on the SUM abundance and distribution for the regions of the dike where AIS data are available.

Abundance and distribution for all six end-members

calculated for the the area outlined by the dashed line in Figure 9 are shown in Plate 1. The left strip is a black and white image showing the surface features imaged at approximately 1.2 gm. The strip crosses, from top to bottom, the Halgito Tongue, Moses Rock dike, the Cedar Mesa Sandstone, and the Organ Rock Tongue and ends in the Quaternary deposits. The seven colored strips show the calculated F parameters for the six end- members (standard, dark sandstone, clay, gypsum, desert varnish, and SUM) as well as an image showing the residual from the least squares calculation. The values of the F parameters are color coded according to the scale along the bottom of Plate 1 where blue colors represent low values, yellows are intermedi- ate, and red to purples are high.

The calculated F parameter values for the standard and reddish-brown sandstone components (the strips labeled STD and SST) indicate these are the primary surface components in this

region. The standard component is most heavily concentrated in the Quaternary deposits at the bottom of the strip. The sandstone component exhibits higher F parameter values in regions containing lower albedo sandstones. The clay-rich component (CLAY) displays the greatest concentration in thin beds of the Organ Rock Tongue and indicate that these sandstones contain a greater component of clay than the other sandstones. Field observations and analysis of samples returned from the area support these results. Reddish-brown sandstone soils are a pervasive component of almost all surfaces, while the sediments of the Organ Rock Tongue contain a greater than average component of clay.

The distribution of abundances coefficients calculated for gypsum (strip labeled GYPS in Plate 1) are for the complex end- member that contains gypsum in a mixture with fine-grained reddish-brown soil. The calculated distribution is confined

almost entirely to the Cedar Mesa Sandstone of the Cutler For- mation and shows peak F parameter values near the top of the Cedar Mesa unit. O'Sullivan [1965] reported that gypsum is a characteristic component of the Cedar Mesa unit in this area, while detailed stratigraphic sections by McGetchin [1968] show the thickest gypsum beds to be near the top of the Cedar Mesa unit. Therefore the distribution and abundances calculated for

the gypsiferous soil are consistent with previous field investigations and clearly show the distribution pattern.

Desert varnish is virtually absent as a component of these soils, as shown in the strip labeled VARN. This is an expected result since desert varnish is found primarily as a coating on fragments of crystalline rock fragments on the tail region of the dike which is not covered by this flight line.

Pixels containing SUM, shown in the strip labeled SUM, are confined primarily to the matrix regions of the dike. Peak SUM abundance values are greater than 95% and the location of the peak values correspond with SUM mapped by McGetchin [19681. The spatial distribution and abundance of SUM throughout the diatreme is discussed in more detail below.

Several thin bands of low "SUM" values observed in the

Cedar Mesa and Organ Rock sedimentary units are clearly not due to SUM but rather correlate with surfaces which contain

limestone. This illustrates an ambiguity that results from the limited spectral range and signal to noise ratio of the imaging spectrometer used in this study in differentiating between SUM and carbonate. However, since carbonate does not constitute a

significant proportion of the dike in this region, the inability to discriminate between SUM and carbonate does not seriously compromise these results.

In the residual image the goodness of fit between the end- member spectra and the AIS data is shown qualitatively where pixels colored dark blue represent the best fit. In quantitative terms, residuals represented by the light blue color are of the order of 1-2%. The homogeneity of the residuals in the image indicates that the suite of end-member spectra used does explain most of the spectral variation encountered in this region. The areas where the residuals are higher than average correlate with drainage channels and shadowed regions seen in the black and white image. This is expected since viewing geometry varies rapidly in drainage channels and no spectral components were included in the suite of end-members for shadowed areas

[Adams e! al., 1986]. Details of SUM abundance and distribution within the dike

are shown in Figure 10 for the areas labeled a-e in Figure 9. Figure 10a is an enlarged view of SUM abundance in the dike

10,384 MUS'•ARD AND PIE'•ERS: UL'•RAMAfiC MICROBRECCIA IN MOSES Roche DIKE

LOC ST CLAY GYPS VARN u RESIDUAL

0.5 1.0

Plate 1. F parameter abundances calculated from the AIS data for the six end-member spectra shown in Figure 9. The strip labeled LOC is a black and white image at 1.2 •tm for location purposes. The approximate boundaries of the dike are shown by the yellow lines and the contact between the Cedar Mesa Sandstone and Organ Rock Tongue is shown by the blue line in the LOC strip. The strip labeled redidual shows qualitatively the goodness of fit of the end-member spectra to the AIS spectra See text for discussion of individual strips of end-member concentration patterns.


•.. "z::i;:':.,-':.:::,...•.. -:-•.•, . :<'• =====================

..:..: ß ........::..-..:..-.•...:.:...: .......... :::•.•:::•:;.•..:•:•;...?*::•:F:•::.•::.•.•.•.•.•..:.:•:..•:•.•.*:.•.s..•:.•.•`.: ... :.....:.:.:.:.::.....::...:....,,.,., ,•x.,.•.•...::.:::..:.:.;:....:-:.: ..:r,•,.........•.,..,.:..•...,. ........... -.....,-,-....:-.•.... :...?.-•,;•tv•?.:•:5.:,:::,.: .... ;**:•,.<..::.,,. ::. ,w: :.::, ;::;. :-;. - .: ......... ':?, ...... ...h.

Fig. 9. Black and white aerial photograph of Moses Rock dike showing areas for which AIS data have been analyzed to obtain surface abundances of endmember components. The results for the area outlined by the dashed line are shown in Plate 1, while the results for the segments labeled a-e are shown in Figure 10. The field of view is approximately 7 km by 11 km.

of the line shown in Plate 1. For each outlined area, or inde-

pendent segment of AIS data, an image of the calculated SUM F parameters is shown where high values are bright. For each region a contour plot of the same data with values between 0.0 and 1.0 is also shown to illustrate better the systematics of SUM distribution. In the contour plots the lowest contour encloses pixels with calculated F parameters >0.2 or 20%. Although values less that 20% are calculated in the images, the distribution of these lower values do not correlate with field

observations and are believed to be within the noise of the data.

The 0.2 contour therefore marks the lower limit of unambiguous detection of SUM. The contour interval is 0.1 and the bold con-

tours mark the 0.6 contour line.

In Figure 10a the large elliptical zone of moderate to high SUM abundance contains values >90% near the top of the zone and shows strong concentration gradients in the contour plot in all directions from the peak values. The location of these peak values correlates with an outcrop of SUM mapped by McGetchin [1968]. Bulk analysis of a sample from this outcrop shows it is composed of approximately 95% SUM and 5% rock fragments [McGetchin, 1968]. The eastern edge of the zone is defined by the walls of the dike, while the western edge abuts against several large blocks of Cedar Mesa Sandstone entrained in the dike (see Figure 9). The north and south limits of this elliptical zone grade into regions of the matrix where SUM abundance is below detection. The matrix here is composed pri-

10,386 MUSTARD AND PIETERS' ULTRAMAfiC MICROBRECCIA IN MOSES ROCK DIKE

Fig. 10. SUM abundances calculated for AIS data segments a-e shown in Figure 9. For each segment a black and white image showing the pixels and abundances in gray tones is displayed along with a contour plot of the same data with a contour interval of 0.1. The lowest contour represents an F parameter of 0.2 and the bold contour is 0.6. The approximate outline of the dike is shown by the dashed line in the contour plots. In the contour plot in Figure 10e the cross-ha. tched pattern shows the extent of limestone outcrop.


...

:.:

.

(e) :

(f) Fig. 10. (continued)

marily of fragments from the Cutler Formation. Note the circu- lar depression contour near the southwestern edge of the elliptical zone which is due to a block of wall rock in this zone of

relatively high SUM abundance (Figure 9). Another region of relatively high SUM abundance in Figure

10a occurs near the western edge of the dike. This occurrence is set between the western wall of the dike and a block of wall

rock in the matrix. A diffuse halo of lower abundance values is

observed northeast and southwest from the peak concentrations. The location of this halo correlates with complex breccia mapped by McGetchin [1968] which contains a higher proportion of matrix SUM than the Cutler breccias, although concentrations may vary in these breccias between 10 and 60% [McGetchin, 1968]. Elsewhere in the dike in Figure 10a SUM values of the


matrix are low; much of the area showing no SUM abundance is occupied by large blocks of wall rock in this region of the dike.

The region shown in Figure 10b (segment b in Figure 9) crosses the lower part of the diluted SUM zone shown in Figure 10a. Although the distribution patterns vary in detail between Figures 10a and 10b, the maximum SUM abundances in both areas are >0.7, and the locations of the peak values are well correlated. The same block of wall rock in the diluted SUM

noted earlier is also evident in Figure 10b. The small differences in distribution patterns can be attributed to the fact that the flight paths for the data acquisitions are perpendicular and slight variation in air speed as well as roll and pitch of the aircraft during flight result in minor changes in the physical size of the pixels on the ground. The calculated abundances of SUM from line to line across the same areas are nevertheless very consistent.

The southern region of the dike, covered by segment c contains fewer mappable blocks of wall rock compared to segments a and b and is composed primarily of fine-grained comminuted material. Analysis of near-surface stratigraphy in this region indicates that the surface material is comparable to the subsur- face material in composition and contamination by wind-transported material is not significant. Although redistribution of material by rain splash and downslope movement may homogenize the surface composition on a fine scale, field observations indicate that this is not important on the scale of the AIS pixels. Taking these factors into consideration, the F parameter values of SUM calculated for segment c in Figure 10c show a funda- mentally different distribution than observed in Figures 10a and 10b. The SUM is more homogeneously distributed across the full width of the dike and peak values (0.6) are less than in Fig- ures 10a and 10b. Bulk analysis of a sample taken from near the center of the dike in the area of segment c indicated a SUM abundance of approxi•nately 50% [McGetchin, 1968], which correlates well with the F parameter value derived from the A|S data. The abundance distributions do not show the strong concentration gradients observed in Figures 10a and b. Instead, the concentration gradients are shallow giving a more diffuse appearance to the halos. The weak linear bands crossing the dike evident in the black and white image are believed to be due to the presence of carbonate-bearing sediments crossing this segment of the dike.

The northern region of the dike, covered in part by segment d, is characterized by many meter sized blocks set in a fine-grained fragmented matrix. The distribution and abundance of SUM in this segment shown in Figure 10d, reflect these characteristics in that the low to moderate abundances are distributed

in the matrix surrounding the blocks in a meshlike pattern. The abundances are also nonuniform and show large variations in abundance from pixel to pixel.

The "head" region of the dike, a portion of which is covered by segment e, the most complex region of Moses Rock dike. It contains a wide variety of lithologies varying from basement rock fragments to fragments derived from stratigraphic levels above those exposed in the walls of the dike. Also, this is the only region of the dike where significant, mappable deposits of carbonate occur. Regions of carbonate as mapped by McGetchin [1968] have been identified by a cross-hatched pattern in the contour plot in Figure 10e. The complexity of the dike and presence of carbonate make it difficult to interpret the distribution and abundance of SUM shown in Figure 10e. Distributions similar to those observed in Figure 10d can be seen in the west-

em regions of the dike. The highest SUM abundances (0.7) occur within the dike along the southern border of this segment and are associated with relatively broad halos of SUM. These distributions however, lack both the uniformity observed in Fig- ure 10c and the regular halo systematics in Figure 10a.

IMPLICATIONS FOR ERUPTION STYLE AND PROCESSES

From this detailed examination of SUM abundance and distri-

bution in Moses Rock dike, two primary distribution styles are observed. Style A shows high peak abundances surrounded by tight halos of lesser abundance, steep concentration gradients, and regions of the matrix with SUM abundance below detection. In the dike this style is illustrated in Figures 10a and 10b. Style B is characterized by moderate peak abundances surrounded by diffuse halos, shallow concentration gradients and a cross-dike distribution of SUM (all regions of the matrix contain detectable SUM concentrations). This style is well illustrated by the abundance map of SUM in Figure 10c.

The implications of these distribution styles for diatreme eruption relate to the mechanism by which the erupting system propagated through the crust and the nature of the eruption in the vent after the system has breached the surface. The dike probably rose through the crust in a manner similar to Clement's [1979, 1982] model for root zone emplacement of kimberlite. In this model, CO2 is exsolved from the rising magma, and this high-pressure volatile phase penetrates fractures and joints (hydraulic ramming) in the wall rock. Pressure reductions due to the opening of new pathways and fluctuations in the rate of magma rise cause implosion and shattering of the country rock and the formation of a brecciation front. Continued rise of the

system advances the front of brecciation toward the surface. The situation at Moses Rock dike differs from Clement's [1979, 1982] model in that the volatiles were not exsolved from a liq- uid magma but were most likely a fundamental part of the erupting solid-volatile material and there was probably only one phase of intrusion. Nevertheless, in either case it is the high- pressure volatiles which do most of the work of brecciation and propagation.

Two competing models that describe diatreme formation after a system has breached the surface are fluidization and hydrovolcanism. Hydrovolcanism is an explosive interaction between magma and some external source of water [Sheridan and Woh- letz, 1983] and has been strongly advocated by Lorenz [1973, 1975, 1986] as the fundamental mechanism for diatreme forma-

tion. In this model a hydrostatic pressure barrier of about 20-30 bars controls the maximum depth of explosive interactions between magma and groundwater. Ejection of groundwater as steam, and wall rock fragments, leads to a decrease in pressure above the explosion interface and therefore the depth at which explosive magma-water interactions occur can propagate downward. The maximum depth of penetration is controlled by the availability of groundwater but may reach depths of 2000-2500 m. A diatreme is formed between the explosion front and the surface and grows laterally as the front propagates downward by collapse and slumping of wall rock [Lorenz, 1986]. If hydrovolcanic processes were responsible for the distribution of SUM in Moses Rock dike, a central conduit of well-mixed pyroclastic rocks surrounded by a collar of subsided wall rock would be expected [Hearn, 1968; Lorenz, 1986]. This is not observed except in Figure 10c which has a central region of homogeneous SUM abundance, although it lacks a collar of subsided wall rock.


A

0 100 0 50

I I .'.•.'F :•:.'. '.

- -'::5,': f..: .' .'.':

z-Z-_-•_-C• Z Z- - -

Early Late

Fig. 11. Schematic diagram illustrating the concepts of SUM erosion and dispersion by eddy diffusion during eruption as discussed in the text. The arrows indicate the direction of flow, the solid bar represents a dike of SUM, and the dots represent eroded SUM reincorporated into the vent flow. The numbers along the top illustrate typical distribution gradients. Figures 1 l a and 1 lb refer to early and late stages of eruption, respectively.

Fluidization as a geologic process refers to the circulation and transport of solids by a high-velocity gas-solid flow [Reynolds, 1954] and has been advocated by Dawson [1962, 1971], McGetchin [1968], and Wyllie [1980] as a principal transport mechanism for ultramafic material from the mantle and as a process to explain upper crustal structures. In this model the erupting medium propagates through the crust as a mixture of solids and gas under high pressure. After the dike breaches the surface a rarefraction wave passes downward through the vent accelerat- ing dike materials across the wave and greatly increasing flow velocities of the materials. Fragments of wall rock are incorpo- rated into the flow by spalling, slumping, and tensile fracturing. The turbulent flow is continuously fed by the solid-volatile system below the accelleration front and xenoliths and wall rock

fragments rise or fall in the vent depending on their density relative to the apparent density of the system.

If fiuidization were the primary process occurring in the vent, material exposed in the vent would be expected to be eroded by blocks and particles carried by the flow. This flow is almost certainly turbulent, and the eroded materials will be redistributed in the vent by eddy diffusion. In a fluidized system as complex as a diatreme, the locations of flow regimes would not be restricted to any particular part of the vent but would fol- low paths of least resistance. The distribution of SUM observed in Figures lea and 10b are consistent with distributions resulting from a fluidized eruption. The elliptical zone of moderate to high SUM abundance can be explained as erosion of SUM from a coherent source and redistribution of the eroded

material by turbulent flow. Also the location of this high SUM zone against the east wall is permitted in a fluidized system but is inconsistent with the hydrovolcanic model which predicts this area should be occupied by slumped wall rock. The systematics observed in Figure l ec are also consistent with the fluidized model if this region has experienced a longer period of flow or a more intense flow.

Following from the arguements above, the two primary distribution styles observed at Moses Rock dike are interpreted as representing different stages of erosion and redistribution of SUM in the vent by turbulent flow in a fluidized system. This is illustrated schematically in Figure 11. In Figure 1 la fragments of SUM are eroded from a dike in the wall rock and redistri-

buted into the central regions of the vent by eddy diffusion in the turbulent flow. Extended eruption or increased flow velocities would result in a homogeneous distribution of SUM across the width of the vent as illustrated in Figure I lb.

In McGetchin's [1968] discussion of the eruption, he con- cludes that the main flow in the diatreme was eventually stabi- lized into two or three primary channels and identified these channels based on field evidence. Figure lea covers one of the areas McGetchin [1968] identified as a possible channel. The interpretation that the homogeneous distribution of SUM observed in the remote measurements in Figure l ec indicates a more mature stage of mixing (i.e., this region has been sub- jected to a more prolonged period of fragmentation and mixing than the regions covered by Figures lea, 10b, and led) is consistent with the field evidence for a proposed channel in the dike.

The "head" region of the dike, covered in part in Figure lee, has also been identified as a late stage channel by McGetchin [1968]. Although some aspects of SUM distribution and abundance in this region are consistent with this interpretation, the chaotic nature of the head region and the presence of carbonate complicate interpretations from the incomplete remote measurements. Several broad halos of moderate SUM abundance are

nevertheless observed, apparently set between large blocks of wall rock in matrix. These regions exhibit peak values of SUM close to 0.7. These distributions suggest a relatively mature state of mixing that would also be consistent with being located in a late stage channel.

SUMMARY

AIS data obtained for Moses Rock dike in Utah contain

important information concerning the distribution and abundance of components in the diatreme. Interactive analysis of the spectroscopic data in conjunction with field observations and reflectance spectra measured in the field and laboratory resulted in the recognition of the primary lithologies and mineral species on the surface. Spectra representative of these primary surface units (SUM, two types of sandstone, gypsum, desert varnish, and clay-rich soil) were selected for use in an intimate mixing model to derive surface abundances of these components from

10,390 MUSTARD AND PIETERS.' ULTRAMAfiC MICROBRECCIA IN MOSES ROCK DIKE

the AIS reflectance spectra. The calculated abundances and distributions for each component are entirely consistent with field observations of this study as well as previous investigations.

The distribution and abundance of SUM in the dike was

examined in detail using the AIS, field, and laboratory spectral data. Two distinct styles of SUM concentration and dispersion were observed in the dike. Style A is characterized by moderate to high abundances of SUM surrounded by well-defined halos of lesser SUM abundance and relative steep concentration gradients into regions of the matrix where SUM abundance is below detection. Style B is characterized by moderate SUM abundance surrounded by broad diffuse halos of lower SUM abundance and SUM is distributed in the matrix across the entire width of the

dike. These styles are consistent with a fluidized environment in the vent [McGetchin, 1968] and are interpreted to be due to erosion of SUM by abrasion and comminution in the vent of the erupting diatreme. The eroded material is then dispersed throughout the matrix by eddy diffusion. Regions of the dike which exhibit SUM distribution in the matrix similar to the first

style indicate that the eruption was arrested at a relatively early phase, while regions of the dike which show SUM distribution more characteristic of the second style represent regions of the dike where the eruption proceeded for a longer period of time and probably indicate the location of mature channels in the erupting diatreme.

Acknowledgments. Many thanks to Lee Silver (California Institute of Technology) for leadership and discussion in the field, to ARCO for the use of their IRIS field spectrometer, to the U.S. Geological Survey in Flagstaff for field support, to the Navajo Nation for permission to con- duct field work on their land, and to Paul Fisher (Brown University) for invaluable programming assistance. Helpful reviews by D. Smith, K Wohletz, F. Kruse, and M. Kingston are greatly appreciated and helped to strengthen this manuscript. Support from NASA grants NASW-4048. NAGW-748, and an NSERC post graduate scholarship is gratefully acknowledged.

REFERENCES

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, 4829-4836, 1974.

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

Adams J. B., M. O. Smith, and P. E. Johnson, Spectral mixture modeling: A new analysis of rock and soil types at Viking Lander 1, J. Geophys. Res., 91, 8113-8125, 1986.

Buckingham, W. F., and S. E. Sommer, Mineralogical characterization of rock surfaces formed by hydrothermal alteration and weathering- Application to remote sensing, Econ. Geol., 78, 664-674, 1983.

Bums, R. G., Mineralogical Application to Crystal Field Theory, 224 pp., Cambridge University Press, London, 1970.

Clement, C. R., The origin and infilling of kimberlite pipes, in Kimber- lite Symposium H (extended abstracts), University Press, Cambridge, 1979.

Clement, C. R., A comparative geological study of some major kimberlite pipes in the Northern Cape and Orange Free State, Ph.D. thesis, 312 pp., Univ. of Cape Town, South Africa, 1982.

Dawson, J. B., Basutoland kimberlites, Geol. Soc. Am. Bull., 73, 545-560, 1962.

Dawson, J. B., Advances in kimberlite geology Earth Sci. Rev., 7, 187-214, 1971.

Goetz, F. H., G. Vane, J. E. Solomon, and B. N. Rock, Imaging spectrometry for earth remote sensing, Science, 288, 1147-1153, 1985.

Hapke, B., Bidirectional reflectance spectroscopy, 1, Theory, J. Geo- phys. Res., 86, 3039-3054, 1981.

Heam, B. C., Diatremes with kimberlite affinities in north-central Mon- tana, Science, 159, 622-625, 1968.

Hunt, G. R., and R. C. Evarts, The use of near-infrared spectroscopy to

determine the degree of serpentinization of ultramafic rocks, Geophy- sics, 46, 316-321, 1981.

Hunt, G. R., and J. W. Salisbury, Visible and near infrared reflectance spectra of minerals and rocks, I, Silicate minerals, Mod. Geol., 1, 219-228, 1970.

Hunt, G. R., J. W. Salisbury, and C. J. Lenhoff, Visible and near-infrared spectra of minerals and rocks, IV, Sulphides and sulphates, Mod. Geol., 3, 1-14, 1971.

Hunter, W. C., and D. Smith, Garnet peridotite from Colorado Plateau ultramafic diatremes: Hydrates, carbonates, and comparative geother- mometery, Contrib. Mineral. Petrol., 76, 312-320, 1981.

Lorenz, V., On the formation of maars, Bull. Volcano!., 37, 183-204, 1973

Lorenz, V., Formation of phreatomagmatic maar-diatreme volcanoes and its relevance to kimberlite diatremes, Phys. Chem. Earth, 9, 17-27, 1975.

Lorenz, V., On the growth of maars and diatremes and its relevance to the formation of tuff rings, Bull. Volcanol., 48,265-274, 1986.

McGetchin, T. R., The Moses Rock dike: Geology, petrology, and mode of emplacement of kimberlite-bearing breccia dike, San Juan County, Utah, Ph.D. thesis, 405 pp., Calif. Inst. of Technol. Pasa- dena, 1968.

McGetchin, T. R., and L. T. Silver, Compositional relations in minerals from kimberlite and related rocks in Moses Rock dike, San Juan County, Utah, Am. Mineral., 55, 1737-1771, 1970.

McGetchin, T. R., and L. T. Silver, A crustal-upper mantle model for the Colorado Plateau based on observations of crystalline rock fragments in Moses Rock dike, J. Geophys. Res., 77, 7022-7037, 1972.

Mitchel, R. H., Kimberlites: Mineralogy, Geochemistry, and Petrology, 442 pp., Plenum Press, New York, 1986.

Mustard, J. F., and C. M. Pieters, Spectroscopy of Moses Rock dike using remote sensing, in 1985 IEEE International Geoscience and Remote Sensing Symposium, Digest, edited by K. R. Carver, IEEE, New York, pp. 147-152, 1985.

Mustard, J. F., and C. M. Pieters, Quantitative abundance estimates from bidirectional reflectance measurements, Proc. Lunar Planet. Sci.

Conf. 17th, part 2, J. Geophys. Res., 92, suppl., E617-E626, 1987. Nash, D. B., and J. E. Conel, Spectral reflectance systematics for mix-

tures of powdered hypersthene, labradorite, and ilmenite, J. Geophys. Res., 79, 1615-1621, 1974.

O'Sullivan, R. B., Geology of the Cedar Mesa-Boundary Butte area, San Juan County, Utah, U.S. Geol. Surv. Bull., 1186, 128 pp., 1965.

Perry, R. S., and J. B. Adams, Desert varnish: Evidence for cyclic dep- osition of manganese, Nature, 276,489-491, 1978.

Pieters, C. M., Strength of mineral absorption features in the transmitted component of near-infrared reflected light: First results from RELAB, J. Geophys. Res., 88, 9534-9544, 1983.

Potter, R. M., and G. R. Rossman, Desert varnish: The importance of clay minerals, Science 196, 1446-1448, 1977.

Reynolds, D. L., Fluidization as a geologic process and its bearing on the problem of intrusive granites, Am. J. Sci., 25, 577-613, 1954.

Roden, M. F., Origin of coexisting minette and ultramafic breccia, Navajo Volcanic Field, Contrib. Mineral. Petrol., 77, 195-206, 1981.

Sheridan, M. F., and K. H. Wohletz, Hydrovolcansim: Basic considera- tions and review, J. Volcanol. Geotherm. Res., 17, 1-29, 1983.

Shoemaker, E. M., Interpretations of lunar craters, in Physics and Astronomy of the Moon, chapter 8, edited by Z. Kopal, pp. 283-360, Academic Press, New York, 1962.

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.

Singer, R. B., and T. B. McCord, Mars: Large scale mixing of bright and dark surface materials and implications for analysis of spectral reflectance, Proc. Lunar Planet. Sci. Conf., loth, 1835-1848, 1979.

Vane, G., Introduction to Proceedings of the Second Airborne Imaging Spectrometer (AIS) Data Analysis Workshop, JPL Publ. 86-35, 1-16, 1986.

Wyllie, P. J., The origin of kimberlite, J. Geophys. Res., 85, 6902-6910, 1980.

J. F. Mustard and C. M. Pieters, Department of Geological Sciences, Brown University, Providence, R102912.

(Received December 12, 1986; revised February 5, 1987; accepted May 12, 1987.)

Documents

Abundance and Distribution of Ultramafic Microbreccia in Moses … · 2013. 7. 25. · metamorphic rocks comprise about 3% of the dike, while dense ultramafic mantle-derived fragments