373
Tectonophysics, 197 (1991) 373-389 Elsevier Science Publishers B.V., Amsterdam
Thermal and kinematic model of the southern Rio Grande rift: inferences from crustal and mantle xenoliths from Kilboume Hole, New Mexico Gilles Y.A. Bussod ’ and David R. Williams Department
’
of Earth and Space Sciences, University of California, Los Angeles, CA 90024, USA
(Received March 9.1990; revised version accepted May 14.1990)
ABSTRACT Bussod, G.Y.A. and Williams, D.R., 1991. Thermal and kinematic model of the southern Rio Grande rift: inferences from crustal and mantle xenoliths from Kilboume Hole, New Mexico. In: A.F. Gangi (Editor), World Rift Systems. Tectonophysics,
197: 373-389.
A simple thermal conduction model constrained by piezothermometric data obtained from coexisting minerals in both crustal and mantle xenoliths from Kilboume Hole maar, New Mexico, is used to understand the development of rifting and the thermal evolution of the lithosphere in the southern Rio Grande rift. The one-dimensional finite difference code models crustal thinning and underplating by magmatic injection into the lower crust and uppermost mantle. The model is in good agreement with crustal and mantle thermometric arrays assuming a transient, pm-rift gradient superimposed on a stable geotherm and yielding a net gradient of 20°C km-’ in the 40-km-thick crust and 3’C km-’ in the upper mantle 30 million years ago. This is consistent with pm-rift volcanic activity in the area possibly related to Laramide erogenic events. A massive injection of basaltic dikes and sills (= 70 vol.%) in the uppermost mantle and at the base of the crust probably occurred in the early development of the southern Rio Grande rift, between 20 and 30 Ma. The heat contribution due to the injection of magma is treated as a temperaturedependent heat source distributed uniformly throughout the intruded region which extends from 25 to 45 km depth. Textural observations along with major and trace element compositions of the mantle and crustal xenoliths indicate that metasomatism and recrystallization may have occurred during liquid infiltration and partial melting 20 to 30 Ma. This is consistent with early rift bimodal volcanism associated with crustal contamination and linked to approximately 25% total NE-SW extension. Die injection and CNStd thinning ceased 20 Ma associated with the mid-Miocene lull in volcanic activity. The total volume of material injected, 100 kms per km length of rift per million years, is approximately an order of magnitude less than for a typical mid-ocean ridge. Conductive cooling of an uppermost mantle (40-45 km depth) from temperatures of over 1200°C (22 Ma) leads to the formation of a granulite metamorphic crustal complex and generation of a surface heat flow of 82 mW m-*, in accordance with observed present-day values in the southern Rio Grande rift. Overall, the results from the dike injection model are consistent with both the crustal and mantle piexothermometric arrays and present-day geochemical and geophysical data from the southern Rio Grande rift.
Intmduction The southern Rio Grande rift (SRGR) is located between two main physiographic and crustal provinces; the Colorado Plateau and southern Basin
’ Present
address: Bayerisches Geoinstitut, Universitgt Bayreuth. Postfach 10 12 51. 8580 Bayreuth, Germany. 2 Presentadd-: Carnegie Institution of Washington, DTM, 5241 Broad Branch Rd., Washington, D.C. 20015, U.S.A. 0040-1951/91/$03.50
and Range to the west, modified by late Cenozoic tectonism, and the unmodified North American craton to the east. The SRGR merges with the Basin and Range in southern New Mexico and extends southeast into west Texas and Chihuahua, Mexico (Seager and Morgan, 1979; Olsen et al., 1987). Kilboume Hole is one of several late Pleistocene basaltic maars found in the central part of the Potrillo volcanic field and contains a large variety of crustal and mantle xenoliths associated with basanitic lavas dated at 80 (f 10) Ka by apatite fission track (V. Harder, pers. commun.,
8 1991 - Elsevier Science Publishers B.V. All rights reserved
374
1989). The maar is located of the SRGR north-south road”
nearly “axial”
refraction
and constitutes
along
the central
at the intersection and the east-west
profiles a natural
axis
of the “pipeline
tion based and
gradient
from Sinno et al. (1986),
spheric
structures
bore-hole
an extended
servations
sections,
on the Kilbourne
on current
geophysical
is coupled
into the mod-
P-T-time
evolution
informa-
Hole xenolith and petrological
with a comprehensive
mal model in order to understand tectonic
et al., 1986; Morgan
perature
ern lower crust and upper mantle. In the following
1984; Aldrich
lead us to conclude
the thermal
of the continental
suite obtherand
lithosphere
were substantially
period
A number
of pre-rift
of sources
a 10 m.y. rhyolite
tem-
and litho-
modified
magmatism.
in our thermal point
Ma as representing
stage extension
stable
as well as the crustal
modification is included an initial condition. 27-32
et al.. 1986).
that the original
by This
model
as
to approximately
the beginning
in the Rio Grande
of early
rift followed
period
of massive
bimodal
volcanism
(e.g.,
Chamberlin,
by
basaltic1978;
beneath the SRGR. The 1-D finite difference thermal model incorporates crustal thinning and dike
Chapin, 1978, 1979; Eaton, 1979; Seager et al., 1984; Aldrich et al., 1986; Morgan et al., 1986;
injection, and is constrained at depth by thermobarometric data obtained from lower crustal and
Olsen et al., 1987). This stage of rifting
upper mantle
xenoliths.
comprises
nearly 80% of all southern rift volcanics and is intimately associated with low-angle extensional faulting (Morgan, 1986). In our thermal model we
Tectonic setting
assume
The base of the crust in the SRGR consists of intraoceanic island arc and marginal basin litholo-
jection at 30 Ma, and a 10 m.y. duration stage rifting in the southern Rio Grande ding 20 Ma. Most details of our results
gies accreted to the North American continental margin between 1.8 and 1.6 Ga (Condie, 1982). Subsequent periods of talc-alkaline arc magmatism and eastward migration of volcanism from the continental margin 75 to 60 Ma are associated with subduction/accretion events (Lipman et al., 1986). Increased magmatism between 75 and 50 Ma appears to be related to an abrupt change in plate convergence rates (Engebretson et al., 1984) and a possible flattening of the subducted plate (Coney and Reynolds, 1977). Middle Tertiary volcanism in the Rio Grande rift and in the Basin and Range to the west corresponds to a period of transition from convergent margin/continentalinterior arc magmatism to extensional continental tectonics, and may have resulted from the collision of the East Pacific Rise with the North American plate approximately 30 Ma (Atwater, 1970). We assume that the region prior to rifting ( = 50 Ma) was analogous to the present-day Great Plains (see Cook et al., 1979; Morgan et al., 1986). From this time until about 30 Ma, this “Great Plains” structure was modified by magmatism and crustal thinning (Elston and Bornhorst, 1979; Olsen et al., 1987). Intense magmatic activity in the area, ranging from 45 to 30 Ma (Seager et al.,
sensitive
the initiation
of rifting
to the time chosen
and magmatic
in-
of early rift, enare not
for the initiation
of
rifting and magmatism. The initial stage of extensional tectonics was followed by a period of quiescence, or a “midMiocene lull”, of 10 million years duration, possibly related to the early development of a transform boundary along the west margin of North America (proto-San Andreas fault) and extreme extension in the Basin and Range. We infer that after 20 Ma, only stable mantle heat flow, radiogenie heating, and conduction occurred until the present time. This may be an oversimplification as late stage upper the mid-Miocene
crustal magmatic events followed lull (Chapin, 1979; Eaton, 1979).
However, the effect of these events on the development of the southern rift is assumed to be minor. Limited extension characterized by highangle normal faulting and relatively minor lower Miocene to Holocene alkalic basaltic volcanism distinguishes the SRGR from the Central and Northern rift provinces, where tholeiitic basalts erupted from central volcanic complexes from 13 Ma to the present (Baldridge et al., 1989). However, the SRGR may be more representative of the early rift history as the tholeiites from the central rift are generally associated with an anomalous
THERMAL
AND
KINEMATIC
MODEL
OF THE
SOUTHERN
RIO GRANDE
north-northeast trending linear volcanic array Jemez Lineament, located at the boundary of Colorado Plateau core and a transition zone to southeast (Baldridge et al., 1989; Aldrich et 1986).
375
RIFT
TEMPERATURE (‘C)
or the the al.,
,‘I”
q
,yl
,“T”
,8tfO ,ly,,,,l4/
L
Vp (Km/s)
I PLAGIOCLASE
20 -
Thermobarometric
estimates
Details of the thermobarometric determinations on the Kilboume Hole xenolith suite are presented elsewhere (Bussod and Irving, 1990). Estimation of equilibration temperatures and pressures recorded by mineral assemblages in xenoliths is fraught with difficulty because of inaccuracies in their calibration, differing P-T blocking conditions for different mineral reactions, and matrix effects in their application. Nevertheless, based on the careful evaluation and application of available thermobarometric techniques suitable to the xenolith assemblages from Kilboume Hole, determination of the important constraints on the thermal history and composition of the lithosphere beneath this area is possible. Shown in Fig. 1 are the piezothermometric arrays determined from the crustal granulites and mantle spine1 lherzolites from Kilboume Hole. This data along with the occurrence of upper crustal assemblages within the maar indicate that the lithosphere has been sampled continuously from a depth of 70 km. The xenolith suite from Kilboume Hole is therefore assumed to be a representative sample of the lithologies which characterize the central SRGR.
z
40 -
s
RE%E PRECISION
E 4 o
60-
80 -
0
CRUSTAL GRANULITES LHERZOLITES
0
FINE
0
COARSE COMPOSITES
-
OUVINE - SPINEL TEYPERATLIRE RANGE
(99)
100 x (h@Mg+Fe) 1
Fig. 1. Kilboume
I
\ *=\ w
@E
100 -
\
I
I
OLMNE
I
I
I
granuloblastic
i
I
I
I
I
granulites (circles)
data. P-T
and protogramtlar
mantle lherxolites
Composite
represent
olivine
lherxolite xenoliths
bols). Arrows
connected
the estimated
sample
olivine-spine1
using
and
are from Padovani
Mg numbers
Numbers
in lhetzolites.
are also shown (half open sym-
to thermobarometric
tines represent
estimates
(squares)
(1977) and Bussod and Irving (in prep.), respectively. in parentheses
I
Hole lower crustal (solid circles) and upper
mantle (open symbols) piezothermometric for the crustal
I ]
data points by
range in temperatures thermometry
for each
(Fabribs,
1979).
Curves I and 2 delineate the stability field of spine1 lherzolite. Crustal compressional
wave velocities
et al. (1986). Sample numbers xenoliths (open circles). T-P
are derived from Sinno
refer to granuloblastic relative precision
mantle
of the therm*
barometric estimates for all samples is also shown.
The crustal suite
The lower crustal xenoliths from Kilboume Hole are mafic and include garnet-bearing paragneisses, two-pyroxene-bearing orthogneisses, charnockite, and anorthosite (Padovani and Carter, 1977). The crustal piezothermometric array is based on garnet granulites consisting of strongly foliated rocks (granuloblastites) with almandinepyrope garnet, sanidine, quartz and silhmanite (Padovani and Carter, 1977). These assemblages are believed to represent a Proterozoic metamorphosed pelitic crust (Padovani and Carter, 1977), subsequently reheated by a later thermal
event related to rifting (Bussod and Irving, 1981; Padovani and Reid, 1989). Piezothermometric estimates are from Padovani and Carter (1977), and are based on Ca-exchange equilibria between garnet and plagioclase (Ghent, 1976), and on K, Na, and Ca equilibria between potassium feldspar and plagioclase (Stormer, 1975). More recent thermobarometric calibrations applied to these samples are in good agreement with these estimates (Bohlen et al., 1983). Ahhough P-T uncertainties are difficult to ascertain, the relative precision for the piezothermometric data is approximately f 0.1
376
GPa pressure and f50°C, based on mineral assemblage (Padovani, 1977). The crustal velocity structure (Fig. 1) is also consistent with the P-T error estimates (Keller et al., 1979; Olsen et al., 1979; Sinno et al., 1986). The crustal thermobarometric data form an array (Fig. 1). which has been interpreted to represent present-day in-situ conditions (Padovani and Carter, 1977; Olsen et al., 1979; Padovani and Reid, 1989). This implies an average present-day geothermal gradient of 30-35°C km-’ for the SRGR and a partially molten crust-mantle boundary. This interpretation is principally based on the homogeneity of coexisting mineral compositions, a high present-day average regional heat flow (100 mW mv2) and anomalous P,, upper mantle seismic velocities of 7.6-7.8 km s-’ (Olsen et al., 1979). However, several observations suggest that the crustal array represents peak metamorphic- and not present-day conditions: (1) The anomalous MOHO velocity in the SRGR is not correlated with high seismic attenuation (Davis, 1987), commonly associated with the presence of a liquid phase. (2) The present-day heat flow along the rift axis and in the vicinity of Kilbourne Hole (80 + 10 mW me2) is lower than the average regional value which may include hydrothermal circulation in the rift flanks (Reiter et al., 1986). (3) Garnet terrains throu~out the world exhibit similar average geothermal gradients (dT/dZ ~22%35°C km-‘), regardless of their emplacement history. This is p~ncip~ly due to the low cation diffusivities and reaction kinetics in Iithologies of this composition (Elphick et al., 1985). The garnets from these assemblages commonly exhibit broad, homogeneous core compositions and thin zoned rims (Bohlen, 1987). However, garnet rims from Kilboume Hole granulites have been replaced by glass and quench crystals formed by decompression melting (Padovani, 1977) such that their cooling history has been obliterated. Conversely, exsolution features are common in feldspar and pyroxene mineral “cores”, indicating that higher temperatures and/or possibly higher pressures predated the “rim” equilibration conditions (Padovani and Carter, 1977). Although local equilibrium can be maintained dur-
G.Y.A. BUSSOD AND
D.K. WIl.l.IAMS
ing initial retrograde thermal events (Bohlen, 1987), for relatively rapid cooling rates crustal granulite assemblages do not necessarily record the modem cooling history as slow cation diffusivities and reaction kinetics do not allow mineral equilibration (Elphick et al., 1985; Bohlen, 1987). Based on these observations, the crustal geothermometric array is interpreted as representing relatively recent (d 20 Ma) “peak” metamorphic thermal conditions, and not present-day conditions. The mantle suite The upper mantle xenoliths from Kilbourne Hole are dominantly spine1 lherzolites. Although commonly regarded as “typical” upper mantle samples, they span a wide range of mineral modes, textures, isotopic-, major- and trace-element compositions (Irving, 1980; BVSP, 1981; Roden et al., 1988; Bussod and Irving in prep.). The spine1 lherzolites used in the determination of the mantle piezothermometric array are divided into two textural types (Harte, 1977): granuloblastic tabular (fine grained) and protogranular (coarse). Both are occasionally cut by spine1 pyroxenite veins (in composite xenoliths), interpreted as the products of polybaric fractionation from basaltic intrusives at depth (Irving, 1980). Equilibration temperatures estimated for the mantle samples are based on two independent calibrations using independent phase assemblages: the temperature-dependence of the pyroxene miscibility gap (Wells, 1977), and the partitioning of Mg-Fe between coexisting olivine and spine1 pairs (Fabrib, 1979). Equilibration pressures were estimated using the two-pyroxene thermometer {Wells, 1977) in conjunction with a thermobarometer based on the solubility of Ca in olivine coexisting with ortho- and clino-pyroxene (Finnerty and Boyd, 1978; FiMerty and Rigden, 1981; Adams and Bishop, 1982, 1986; Finnerty, 1989; Kiihler and Brey, 1990) that is empirically calibrated to the natural system (Bussod and Irving, in prep.). The P-T estimates (Fig. 1) generally agree with previous determinations for the same locality (Reid, 1976; Mercier, 1977, 1980), and based on an overview of synthetic and natural systems are within the stability field limits for
THERMAL
AND
KINEMATIC
spine1 lherzolites
MODEL
at f 0.15 GPa and
determinations
f 30°C (Bussod
prep.). Based on thermobarometric two-pyroxene 1990),
(Bussod
the granuloblastic are
spatially
shallower
depths
(34-54
xenoliths
and
more, the granuloblastic
and Irving,
using the
and and
in
samples
similar
protogranular
and lherzolites disThe
generally the more xenoliths in prep.). also show their con-
described by Mercier (1977, 1980) for the adjoining Basin and Range and Colorado Plateau provinces, which suggests some lateral continuity to (Reid, 1976;
Mercier, 1977) the mantle array (within error) predicts temperatures lOO-200°C lower than the crustal array at a depth of 30 km (Fig. 1). However, this apparent discrepancy is resolved when considering the thermal history recorded in the mantle xenolith assemblages. The temperatures derived from two-pyroxene and olivine-spine1 mineral
pairs (Fig.
cores of spine1 grains
1) are in good agreement
for
the protogranular xenoliths (AT = 50°C) and yield apparent equilibration temperatures between 950°C and 1060°C (Bussod and Irving, in prep.), whereas large temperature discrepancies are found using the two methods in the granuloblastic and composite rocks. The temperature estimates using olivine-spine1 pairs in the composites and three of the four granuloblastic rocks, are systematically higher than temperature estimates using coexisting pyroxene pairs. In addition, the oxide phase is demonstrably out of equilibrium with the silicate phases for these samples (Bussod and Irving, 1981). However, neither this disequilibrium nor the inherent differences between the thermometers can
temperature zoning
range
in spine1
sent in all xenoliths,
grain
sizes.
and transport
reaction
Lobate are pre-
the zoning
perturbations
liths by the host basanite. and silicate mineral
even though both samples show a
indicating
with the incorporation
in the
to the granu-
with local melting
due to late stage thermal
cate different
differences
patterns
are restricted
loblastic and composite xenoliths protogranular and granuloblastic
mantle are from
counterparts.
investigators
large
comparable
stituent phases, represented by the olivine compositions in Fig. 1. Similar relationships have been
the west. As noted by previous
such
spine1 edges associated
are compositionally
former are compositionally fertile and have higher bulk Fe and LREE than barren MORB-related protogranular (Irving, 1980; Bussod and Irving, Granuloblastic and composite samples a wider range in the Mg/Fe ratios of
for
Irving,
km) (Fig. 1). Further-
lherzolites
from composite to their
account
(AT = 3OO’C). Retrograde
km) than the coarse pro-
(50-67
371
RIFT
The relative
composite
associated
RIO GRANDE
is estimated
estimates
thermometer
xenoliths togranular
SOUTHERN
of this composition.
of the P-T
precision
OF THE
is not
associated of the xeno-
These observations
kinetics
between
indi-
the oxide
phases and are consistent
with
an earlier thermal event affecting only the granuloblastic and composite rocks. This thermal perturbation reconciles the crustal and mantle piezothermometric arrays (Fig. 1) and indicates that an inverted
thermal
gradient
once existed
in the up-
per mantle beneath Kilbourne Hole. Consistent with mineral homogenization times (Bussod and Irving, in prep.), the inverted gradient is assumed to be associated with a relatively recent
thermal
restricted
to
perturbation the
(< 25 Ma), spatially
crust-mantle
boundary,
and
therefore is inconsistent with a simple model of lithospheric thinning by asthenospheric upwelling. Assuming
that the thermal
event
recorded
in the
xenolith mineral assemblages from Kilbourne Hole is the result of a regional metamorphic process associated with rifting, we have developed a numerical model which incorporates crustal thinning and injection of magma into the lower crust and upper mantle of the SRGR. The concept large intrusive complex in the lower crust
of a and
upper mantle of the SRGR is further supported by the unusually widespread occurrence of intrusive xenolith lithologies (composite xenoliths and mafic megacrysts) this area (Padovani Baldridge, Thermal
found in many localities and Carter, 1977; Fodor,
1979; Irving,
1980; Dromgoole,
from 1978; 1984).
model
The thermal evolution of the center of the southern Rio Grande rift is calculated using the following equation:
378
G.Y.A.
where T is temperature. f is time, z is depth, K is thermal diffusivity (= k/pC,, where k =
estimated difference
conductivity.
lateral
and
right-hand advection heat
p = density,
o is advective
side represent
certain injection
and
heat
terms
respectively sources
contexts
hereafter
of magma
effects.
a one-dimensional
ference
scheme.
The
scheme
is equivalent
magma although
by
of material
an
spreading
crustal
and
solved on an Eulerian
is solved dif-
the
finite
through
thinning.
difference
through the the material. the grid due to
Equation
grid. The grid spacing
and time steps (At)
were chosen
effect of third-order
terms in the conduction
(1) is ( AZ)
to minimize
the solu-
tion. The values used in this calculation are AZ = 1 km and Ar = 5 ka. The value K is 1.11 X 10 --’ m”
that the amount
in the crust and 1.06 x 10ph m* s- ’ in the mantie. The advective term in eqn. (1) represents the upward movement of material due to spreading concurrent with dike injection and the attendant S
symmetric about its axis. Calculating the characteristic time, T = d2/tc, for conduction over a d for geologic materials, shows this to be assumption. We have also numerically
km
is transported
and movement
of heat conducted laterally out of the region of dike injection is small, and that the rift itself is
distance a good
lO”C, which
to these calculations.
in
use of a one-dimensional to assuming
loss is approximately
scheme in three ways: conduction material, injection of dikes within
form will have forward
thermal
Heat
AN13 D K W,,_I.,AMS
the effect using a two-dimensional finite grid and found that the maximum
is not significant
to magma
The equation explicit
the
radioactive
due
as dikes,
in any other
on
conduction,
thinning,
We will refer to the intruded
the same thermal using
The
due to lithospheric
sources,
injection.
and Cr = heat capacity),
velocity.
HUSSOD
’
thinning
of the crust.
It is assumed
that all thin-
TEIIP
a
KH
0
mantle
j
dike Ton
t----22
km
j
: 800-c km---i
33
1050T
60
Fig. 2. Finite difference thermal model. (a) Initial conditions 30 Ma: no extension or intrusive% and the crust-mantle boundary is at 40 km depth. The box represents the initial intrusive area. (b) Final configuration after 10 m.y. inlrusive history (20 Ma): 25% extension
and 70% intrusives.
The original crust-mantle
boundary
(banded box) is at 26 km depth. Kilbourne Hole location (KH),
is at 31 km depth and the top horizon model densities and approximate
of underplated
temperatures
region
are also shown.
THERMAL
AND
KINEMATIC
MODEL
OF THE
SOUTHERN
RIO GRANDE
ning takes place above the dike injection region, and that upward movement of the dike injection region itself as well as the area below the dikes is simply due to this thinning (Fig. 2). The effect of advection on the temperature distribution is evaluated using an upstream differencing method. We have limited the total volume of dikes injected per km of distance along the rift to 1000 km3, evenly distributed within a predefined dike injection region. The heat due to dike injection, H, is represented by a temperature-dependent heat source:
HZ+ P
[
L+
(Th.4
-
TLWPL
RIFT
379
heating due to radioactive decay in the crust and mantle. This term is represented in the model as a stable temperature distribution, upon which the other temperature contributions are superimposed. A plausible heat production distribution is used to calculate an equilibrium temperature profile, which is assumed to be maintained throughout the history of the rift. The model also includes the effect of pre-rift heating. This is represented as a transient temperature profile superimposed on the stable profile before the actual initiation of dike injection and rifting. This pre-rift temperature profile is simply added to the time-dependent portion of the temperature as an initial condition. Parameters
+(r,-T,)(
ci$L)+(=s-T)Cp]
(2)
where y is the proportion of material injected per unit time step, C, and CpL are the solid and liquid heat capacities, respectively, L is the latent heat of fusion, and TM, TL, and T, are the injection, liquidus, and solidus temperatures respectively. The total heat input from the dikes is a function of the ratio of mass of injected material to mass of surrounding rock, the temperature at which the dikes are injected, the temperature of the adjacent country rock, and the thermal parameters representing the various materials involved. The dikes cool from their initial injection temperature, releasing heat to their surroundings. Between the liquidus and solidus temperatures, latent heat of crystallization is also released as the dikes solidify. The material continues to cool, releasing heat to its surrounding until the temperatures equilibrate. The dikes are assumed to be emplaced rapidly, with negligible loss of heat from the material to the mantle until the predefined dike injection region is reached and the magma halts its ascent. Therefore, the dike injection term in the equation is only applied within the dike region. Note, however, that spreading and crustal thinning will cause this region to move upward with time. The heat from the dike injection is assumed to be distributed over the entire dike injection region, and the dikes are assumed to reach thermal equilibrium with the country rock in one time step. The remaining term (A) in eqn. (1) represents
In order to apply our thermal model to the evolution of the SRGR, we must determine or estimate values for the relevant parameters (Table 1). In the following discussion, we will attempt to justify the values chosen, and discuss the effects of changing these parameters on the final results. The timing of events used in our model is principally based on the tectonic history of the region. The geologic structure is modelled at 30 Ma (Fig. 2a) and at 20 Ma (Fig. 2b), before and after magmatic injection, respectively. Our thermal model has its beginning at 30 Ma, the initiation of rifting. At this time, the crust is assumed to be 40 km thick, consisting of a 30 km upper crust, and a 10 km thick lower crust. This value is intermediate to that for the present-day Colorado Plateau to the west and the Great Plains to the east, initially approximately 50 km thick (Stewart and Pakiser, 1962; Keller et al., 1979; Cook et al., 1979), and thinned by about 10 km due to pre-rift activity. Our lowest crustal layer is equated with the southern Great Plains layer determined by Keller et al. (1979) using surface wave dispersion, which after crustal thinning up to 30 Ma occupies the depth range from 30 to 40 km. The density for the upper crust is taken to be 2700 kg rnT3 (cf. Keller et al., 1979; Cordell, 1982; Daggett et al., 1986). Measured densities of lower crustal granulites from Kilboume Hole range from 2900 to 3100 kg rnp3 (Padovani and Reid, 1989). Keller et al. (1979) using surface wave dispersion estimate a
3RO
Ci Y A BUSSOD AND
I, K WILLIAMS
TABLE 1
value of 2900 kg m-3
Parameter values used in the thermal model
Great Plains, and we take it to represent the lower 10 km of our modified crust. We assume a mantle
Model parameter K/t
Value
References
history
Inception of rifting Cessation of rifting Total extension Percentage dikes initial
30 Ma 20 Ma 50 km 70%
a, e, f, 1. u. z a, e, f. 1. u, 2 u
40km 30 km 10 km
g. 4
studies
initial
Reiter
“C-’
(e.g., Decker
et al., 1979;
1988), although
the values
cluster
3.0 W m-
in the most recent
about
a mean
of ap-
’ “C -‘. which we choose as
S
b (p.1153)
8x10-.” W kg -’ 10-i’ W kg-’ IO-” W kg-’ 10 km 32 mW me2
c c p c c,j, t
r, v. x
u. x
u, x
value for the thermal
conductivity
aa, k, m, n aa. n, 0
of
upper crustal and mantle values. This allows us to use a constant thermal diffusivity for the whole crust. For the mantle, a slightly higher thermal conductivity of 3.5 W m-’ OC-’ is assumed, whereas assumed
a heat capacity for all (solid)
of 1000 J kg-’ “C- ’ is material in this model
(BVSP, 1981, pp. 1153 and 1135, respectively). We assume that the initial stable geotherm remains constant throughout the 30 m.y. represented by the model. We have assumed a surface heat production of 8 X lo-” W kgg’ with a characteristic depth of 10 km, in accordance with Blackwell (1971). Decker and Smithson
(1975) and
Cook et al. (1978). A constant heat production of lo-” W kg-’ is used throughout the lower crust
1450°c 13cQ”C 1200°C 8x10s J kg-’
0
(Blackwell,
d
1980). The upper mantle heat production is taken to be constant to 120 km with a value of lo-” W
IGO0J kg-’ “C-’ 1.500J kg-’ “C-’
b (p.1135) y
kg-’ (Kaula, 1968). A boundary condition
Heat capacities
Solid Liquid
Kelley
1986; Reiter et al., 1986; Nathenson
the upper crust in our model. For convenience, we assume a conductivity of approximately 3.2 W m - ’ “C ~ 1 for the lower crust. part way between
Dike material (dry hasanite)
Injection temperature Liquidus (1.5 GPa) Solidus (1.5 GPa) Latent heat
conductivi-
a wide range of
a reasonable
3.0 W m-’ “C-’ 3.2 W m-’ Y-’ 3.5 W m-’ “C-l
20°C km-’ 3°C km- ’
1975;
thermal
rift exhibit
et
(Keller
h, i, q
transient gradients
Crust Mantle
crustal
of the
9. w h. q
Heat produciion
Surface Lower crust Mantle Characteristic depth Mantle heat flow
and Duncan,
proximately 27OOkgm-’ 2900 kg m-s 3300 kgm-’ 3000kgme3
of upper
Smithson,
and Guffanti.
Thermal conductiuities
Upper crust Lower crust Mantle
Estimates
ties for the Rio Grande
values from 2.8 to 3.3 W m-’
DPP?Si
Upper crust Lower crust Mantle injected magma
density of 3300 kg m- ’ for our model al., 1979; Cordell, 1982).
and
thickness
Crustal thickness upper crust Lower crust
for the lower crust
References: a = Aldrich et al., 1986; aa = Arculus, 1975: b = Basaltic Volcanism Study Project, 1981; c = Blackwell, 1971: d = Bottinga and Allegre, 1978; e = Chamberlin, 1978; f = Chapin, 1978; g = Cook et al., 1979; h = Cordell, 1982; i = Daggett et al., 1986; j = Decker and Smithson, 1975; k = DePaolo, 1979; I= Baton, 1979; m = Gbiorso and Carmichael, 1980; n = Green, 1973; o= Irving, 1974; p = Kaula, 1968; q = Keller et al.. 1979: r = Kelley and Duncan, 1986: s = Kushiro, 1980; t = Lachenbruch and Sass, 1978; u = Morgan et al., 1986; v = Nathenson and G&anti, 1988; w = Padovani and Reid, 1989; x = Reiter et al., 1986; y = Richet and Bottinga, 1986: z = Seager et al., 1984.
1971; Cook et al., 1978; Sclater
for heat
et al..
flow into
the
lithosphere from the mantle at depth must also be assumed. Blackwell (1971) uses a value of 30.5 mW m-* for his stable continental model which Cook et al. (1978) apply to a Great Plains geoDecker and Smithson (1975) and therm. Lachenbruch and Sass (1978) each estimate a reduced heat flux for the southern Rio Grande rift of 33.4 mW rnm2. For our thermal model, we take a value of 32 mW m-’ for the heat flux coming into the base of the thermal lithosphere at 120 km. Because the heat is being efficiently removed to
THERMAL
AND
KINEMATIC
MODEL
OF THE SOUTHERN
RIO GRANDE
Tempmtufe("C) 0
200
400
600
600
1000
1200
1400
'I"'I"'I"'I"""'I"'
381
RIFT
greater than 3°C km- ‘. We have assumed transient temperatures superimposed on the stable geotherm
yielding
gradients km-’ this
more
of 20°C
conservative
km-’
in the uppermost exact
transient
mantle profile
match
the piezothermometric
dicate
that
required.
a pre-rift
of material
of the crust and upper initial
Great
gradients
,oo’,,,‘,,.‘...‘...‘...‘~.. Fig. 3. Assumed stable and transient southern Rio Grande
the crust-mantle flow
remains
’
model geotherms for the
rift at the start of the model (30 Ma).
boundary, constant
we assume
throughout
the
this heat 30 m.y.
history of the rift. The upper boundary condition is a constant temperature of 0°C at the surface. The values chosen here give us a stable geotherm (Fig. 3) on which changes due to dike injection and crustal spreading are superimposed. The surface heat flow obtained from the stable geotherm model is 56.5 mW m-*, which seems to be a reasonable value, considering that the original Great Plains heat flow has been modified. Although the pre-rift history of crustal extension and magmatism has important consequences
events.
Attempts
thermal
record
by later
have been made to approximate
this transient temperature gradient. Reiter et al. (1986) assumed that the long period of pre-rift magmatism led to an equilibrium geotherm of 23.4’C from the surface to the base of a 35 km crust, and a mantle geotherm of 4.5’C, reaching 1200°C at 120 km depth. Morgan et al. (1986) apply a similar approach to pre-rift heating, assuming the base of the crust to be at the melting point, somewhat greater than 1100°C. Depending on the assumed crustal thickness, the gradients vary from about 25’C km-’ to 35’C km-’ and the mantle is assumed to have a gradient slightly
to this is
model shows that 15 into the lower 25 km of an
will give temperature
those we have assumed
in
anomaly. The pre-rift transients could have formed by lithospheric thinning as well as by magmatic injection. Within the rifting period, extension and magma injection is assumed to occur with a half-sinusoidal history, peaking ages on basaltic
at 25 Ma. A study of radiometric rocks in New Mexico (Aldrich et
al., 1986) shows an increase in the occurrence of intrusives up to 24-25 Ma, followed by an abrupt decline, consistent with the dike emplacement history assumed in our model (see e.g., Cook et al., 1979). The primitive magma intruded in the base
approximately
of the event has been erased
similar
prior to the inception of rifting, no inferences are made about the pre-rift history leading to this
much of the memory
and
to in-
less than 5% of the volume of the region. Although our model requires a thermal lithospheric anomaly
of the crust
the geologic
necessary
data, our results
our model. The pre-rift crustal thinning is 10 km, and the total amount of material injected is much
for the present-day thermal structure of the rift, this transient effect is difficult to quantify because from
3°C
25 km of the mantle
Plains structure
very similar.to
and
(Fig. 3). Although
is not
anomaly
A simple numerical
m.y. of injection
temperature
in the crust
(Irving
and
and Frey,
the upper
mantle
1984), assumed
150 km depth
on the intersection
and
of our initial
model
basanite
liquidus
1975;
DePaolo,
1979;
(Green, Ghiorso
is a basanite to originate 1450°C
geotherm 1973; and
at
based and a
Arculus,
Carmichael,
1980). The liquid is rapidly emplaced at the crust-mantle boundary at a superliquidus temperature (3 1450°C) by fracture propagation (Shaw, 1980) with minimal heat loss to the surrounding rock during
its ascent.
After emplacement,
as the
basanite rapidly loses heat to the surrounding country rock, liquidus phases pyroxene and spine1 begin to fractionate and latent heat is released. The crystallization is complete at the solidus temperature roxenite
(5 1200°C), and the newly formed dikes continue cooling conductively.
py-
382
G.Y.A.
The density crystallization (Kushiro,
to be 1500 J kg-’
1980)
a model
mineralogy
approximately
a latent
gg ’ (Bottinga
800 J
1978). Based on observations liths (Irving,
heat
for these
composites,
400
600 ‘,‘/
D.R
W1LLlAMS
(“C)
800 ,,I.
1000 1200 .,..,,I,_
1400
of the liquid and Bottinga, of 85% clino(Irving,
of fusion and P-T
of
Allegre,
of composite
1980) and the narrow
200
AND
’ based on heat
15% spine1 for the pyroxenites
we calculate
occurrence
"C-
for silicate liquids (Richet
1986). Using pyroxene,
Temperature
dike material upon be 3000 kg m-’
1980) and the heat capacity
is assumed capacities
of the injected is taken to
BUSSOD
xeno-
range of
the magma
is
modelled as being injected in the form of narrow dikes extending from a depth of 35 km to 55 km. The heat flow due to the injection is assumed to be distributed evenly within the region in one time step. The heat then conducts its surroundings.
out of the region
As the dikes are being
to
injected,
100
1
Fig. 4. Kilbourne
’
I
‘,
/
’ AfiJI
Hole thermal evolution
1
model-geotherm
re-
suits for 22 Ma, 20 Ma, 10 Ma, and the present, starting with the transient geotherm at 29 Ma (Fig. 3). The P-T-time of the original crust-mantle
path
boundary is represented by arrows.
the region is simultaneously extending by an amount equal to the volume of the dikes. The thermal history is not affected by whether dike injection
or lithospheric
extension
is assumed
to
be responsible for rifting. The crust directly above the region and for some distance to the sides is assumed
to extend
an equivalent
amount,
material to move upwards, carrying the geotherm with it, and providing the advection term in eqn. (1). The total extension in the intruded region is taken to be 50 km, so that the final width of the extended region is 200 km. This value for the present-day size of the extended and thinned area agrees well with the surficial fault distribution delineating the southern Rio Grande rift (see e.g., Chapin, 1979; Aldrich et al., 1986) and the seismic cross-section for this area (Sinno et al., 1986). The 200 km width represents 25% extension, in accord with the estimated value of 30% for the southern Rio Grande rift (Morgan et al., 1986). From our model, the mean extension velocity of the rift is 0.5 cm yr-‘, and the total resulting amount of thinning is approximately 9 km. the completion of extension and dike injection (20 Ma), the heat sources in the model comprise radiogenic heating and the heat flux from the mantle, and all heat is transported by conduction. At
it has reached
the present-day
structure.
and
therefore become progressively thinner. We have assumed the crustal thinning takes place over a region initially 150 km wide. This causes the
crustal
The geotherm is allowed to evolve under these conditions for 20 m.y., at which point we assume
Results Representative geotherms calculated by our finite difference model for various times using the parameters described above are shown in Fig. 4. Note that the maximum amplitude of the thermal “bulge” due to dike injection occurs at approximately 22 Ma. The bulge then relaxes conductively to the present-day feature of this model,
geotherm. An important distinguishing it from a
straightforward lithospheric thinning/asthenospheric upwelling model, is the development of an inverted temperature gradient in the upper mantle, as required by the piezothermometric data. The temperature variations with time below 60 km are very small, less than 100°C over the 25 m.y. duration shown. The maximum temperature reached at 60 km depth is 1069% 18 Ma. For comparison, the temperature at this depth 20 Ma was 1056*C, and the present-day temperature is 972°C. Conversely, the range in temperatures between 20 Ma and the present around 40 km depth is approximately 360% which corresponds to an
THERMAL
AND
KINEMATIC
average
cooling
complexes
present-day
estimated
OF THE
rate of 18”C/m.y.,
pine metamorphic The
MODEL
model
SOUTHERN
typical
(Dodson, surface
to be 82 mW m-*
and
is approximately
from
a maximum Note that km-‘.
is
the maximum at 19 Ma.
21’C
383
RIFT
Nathenson
and
noticeably
flow
average crustal temperature
ent for this model down
of al-
1976).
heat
model surface heat flow is 117 mW m-* The present-day
RIO GRANDE
gradikm-‘,
20 to 22 Ma of about the present-day upper-
m-*.
Guffanti,
This value
1988).
is
higher than our model result of 82 mW
Examination
of heat flow measurements
for
the SRGR (Reiter et al., 1986) reveals that the highest values (> 100 mW m-*) are concentrated along
the edges of the rift, while
surements vicinity
range
near-axis
mea-
from 70 to 90 mW m-*.
of Kilbourne
In the
Hole, a value of 80 mW m-*
is indicated
(Reiter
crustal thermal gradient is much steeper (dT/dZ = 25°C km-‘) and, as expected, near-surface
hydrothermal cause higher
circulation at the rift flanks would effective heat flow, but would not
gradients
have much effect on the deep thermal
35°C
are remnants
tures at depth.
Simply
of earlier extending
high
tempera-
this gradient
give misleading results for the geotherm and the time lag for thermal diffusivity taken into account when interpreting temperature gradients. The position
will
at depth, must be
near-surface and tempera-
ture of the original crust-mantle boundary through time is shown by the arrows (Fig. 4), the upward component
representing
crustal
thinning.
the rift (Lachenbruch. Morgan
be controlled
This model has significant implications for the present-day geologic structure as well as for the thermal structure of the rift. The final southern rift axial crustal (Fig. 2) consists
structure predicted by this model of a 22 km thick upper crust, a 4
km thick lower crust, tered by the magmatic composed of granulites,
which
is structurally
unal-
intrusives and principally and a 5 km thick lower
crust that is composed of 70% mafic intrusives 30% granulites. The mantle is divided into
and two
parts, an upper intruded region extending from 31 to 46 km composed of 70% spine1 pyroxenites (basaltic intrusive fractionates) and 30% compositionally re-enriched fertile lherzolitic mantle, and the region below 46 km, which consists of MORB lithospheric mantle material (Bussod and Irving, in prep.). The densities of these regions are also shown (Fig. 2). Studies of the present-day largescale structure of the rift provide an opportunity to test this model result against data.
actual geophysical
It is generally concluded that the average heat flow for the southern Rio Grande rift falls in the range of about 100 f 10 mW m-* (Sass et al., 1971; Blackwell, 1978; Reiter et al., 1979, 1986;
Concentrated
structure
of
and Sass, 1978). Seager and
also note
that
the variations
in
as well.
by the location
This
indicates
of the flank fractures
that
82 mW
m-*
may
represent a good estimate for the heat flow from the axis of the rift, the site of both Kilboume Hole our thermal
represent
structure
(1979)
1986).
heat flow within the rift may be due to localized near-surface magma bodies, an effect which may
and Present-day
et al.,
model.
a minimum
This
however
value
is taken
as the model
to
does
not take into account late-stage magmatism in the upper crust which would presumably cause an increase in surface heat flow at the present time. A large number of seismic studies have been done on the southern and Sanford (1976)
Rio Grande rift. Toppozada studied the refraction data
along a N-S line extending from the Gasbuggy shotpoint in northern New Mexico to the Texas border. They concluded that the crust at Albuquerque Applying buquerque
could
be
no
their
estimated
and
Kilboume
thinner 2”
than
dip
Hole
30
km.
between
Al-
constrains
the
crust of the southern Rio Grande rift to be thicker than about 20 km. The seismic evidence across the southern
Rio Grande
rift (McCullar
and
Smith-
son, 1977; Cook et al., 1979; Keller et al., 1979; Olsen et al., 1979; Sinno et al., 1986) is generally interpreted as consistent with asthenospheric welling beneath the Rio Grande rift, and
upthe
“anomalous mantle” P, velocities as evidence for a partially molten crust-mantle boundary. However, no evidence for high seismic attenuation is found at these depths (Davis, 1987), and an asthenosphere upwelling model cannot explain the piezothermometric data. We propose an alternate explanation, based on the similarities between
384
G Y.A. HUSSODAND
these southern Rio Grande cross-rift seismic results and the final state of our model. A discontinuity
at 20 to 27 km depth
(Cook
et al., 1979;
Keller et al., 1979; Sinno et al., 1986) is similar either
the lower-
model
to upper-crust
at 22 km,
and/or
assumed
to represent
Although
we have
matic
intrusion
physical crust, intruded
the 26 km boundary
constrained
crustal
altered
level, the actual
much more complex,
less distinction base.
zone.
the top of mag-
to be this single
a thermally
in our
the top of the intruded
case is undoubtedly
with somewhat
boundary
to
between lower
Using
the upper
crust,
seismic
and
an
reflection
studies of the central Rio Grande rift, Brown et al. (1979) concluded that the crust-mantle transition beneath the rift represents a complex zone of discontinuous reflector units. The boundary discovered seismically at 30-40 km depth is congruent with the interface between intruded lower
reviewed by Seager and Morgan (1979). The general consensus is that a high conductivity zone is present between about 20 and 30 km beneath the southern models
Rio
Grande
geomagnetic suggested
induction a N-S
the deep crust Colorado melting
rift,
have been proposed.
profile,
and upper
model.
intrusives
The
to 50 km
the axial line (7.95 km s-l)
and the cross-line
(7.7
km ss’) found by Sinno et al. (1986) may be due at least in part to anisotropy caused by the presence of dikes in the upper mantle oriented subparallel to the rift axis. Daggett et al. (1986)
studied
a large
set of
gravity measurements covering the southern Rio Grande rift. From these measurements, they constructed density distribution models for the structure beneath the rift. They interpreted these models as being consistent with anomalous mantle,
extending
these anomalies, and mantle
for from
with a simple
of N-S
assemblages
Hole lower crustal
trending
in the lower as Kilboume
xenoliths
are un-
and amorphous carbon and in intra- and inter-
granular fractures (Mathez et al., 1984; Dromgoole and Pasteris, 1987) which could result in anomalous electrical conductivities (Duba land, 1982; Frost et al.. 1989).
Results of the thermal on the piezothermometric
from 32 km (7.7 km s-‘)
(1970)
structure
crust could explain
usually rich in sulfides along grain boundaries
melting
based on a
Schmucker
mantle
presence
and high-grade
intruded area would not be sharply consistent with the observed gradual P-wave velocity
partial
to west Texas, inconsistent
Discussion
(8.3 km s-‘) (Sinno et al., 1986). Finally, the observed difference in P-wave velocity between
and
However,
two-dimensional
crust and intruded mantle at 31 km depth in our model. We assume the base of this intruded mantle layer to be at 46 km, but again the actual delineated, increase in
I, K WII.I.IAhlS
and
Shank-
model are superimposed arrays obtained from
both crustal and mantle xenoliths from Kilboume Hole presented earlier (Fig. 5). The apparent mismatch
between
the
crustal
and
mantle
arrays
strongly constrains the thermal evolutionary history beneath Kilboume Hole. The intrusive model fits the data well, given our set of geophysically reasonable parameter values. The crustal array temperatures representing “ peak” metamorphic matches the 20 Ma model temperatures. This result is consistent with the relatively slow cation diffusivities and reaction kinetics typical of these crustal rock mineral assemblages (Elphick et al.. 1985). Homogenization times for average size garnets from Kilboume Hole granulites are greater
elled by Daggett et al. (1986) is equal to our model column mass, indicating a similar gravity measure-
than 3 m.y. at 1000°C and 30 m.y. at 900°C. Given the range of model cooling rates (3 45 10°C/m.y.). closure temperatures are within the range of temperature estimates for the crustal garnet grant&es (78O”C-1000°C; Bussod and Irving, in prep.). Within error the mantle array
ment would be obtained. Keshet and Hermance (1986) have modelled the conductivity and magnetotelluric data for the SRGR. The electrical conductivity data is also
matches the present-day model geotherm well. Our calculations, however, do not take into account late stage ( < 20 Ma) crustal magmatism as proposed for this region (Morgan et al., 1986) which
related to magmatic activity at or near the base of the crust. This zone is described as transitional between normal upper mantle and lower crust. The total column mass to the 50 km limit mod-
THERMAL
AND
KINEMATIC
MODEL
OF THE
SOUTHERN
RIO GRANDE
TEMPERATURE (‘C)
RELATIVE PRECISION
80
Fig. 5. Model geotherm line), 10 Ma (stippled) on piezothermometric Model
basanite
results from Fig. 4, 20 Ma (dashed and present (solid line). superimposed
estimates (Fig. 1). Symbols as in Fig. 1.
liquidus
modified
from Green
(1973)
and
Arculus (1975).
may explain why our model present-day geotherm falls below the uppermost mantle thermobarometric estimates (Fig. 5). An initial transient crustal geotherm of 25°C km-’ (30 Ma) also improves the fit; however, the data does not warrant these modifications. In addition, the olivine-spine1 thermometric estimates for the granuloblastic rocks (Fig. 5), which represent a relic thermal perturbation (2 10 Ma), are in agreement with a 20-22 Ma model geothermal perturbation at the crustmantle boundary. Maximum thermal perturbations within this zone occurred 22 Ma at a depth of 41 km, within the modified upper mantle. This thermal maximum is within 30°C of the “dry” melting temperature of the MORB-related Kilboume Hole mantle spine1 lherzolites (Takahashi, 1986; Bussod, 1991), and is above the solidus temperature of the lowermost crustal assemblage (Irving, 1974; Vielzeuf and Holloway, 1988). This implies pervasive recrystallization (T/T, > 0.95), as well as possible melting and chemical re-equilibration at the crust-mantle boundary at that time (Bussod and Irving, in prep.). The narrow P-T range for composite xenoliths and the
R1l-V
385
deepest protogranular mantle samples which show no thermal perturbation are also consistent with our thermal model and thus limit the zone of 70% magma injection to a present-day depth of approximately 26 to 46 km. The granuloblastic textures of the crustal and uppermost mantle xenoliths are interpreted as textural evidence for our model which predicts pervasive recrystallization at and near the crustmantle boundary during the first 10 Ma of rifting due to a thermal perturbation. Isotopic data on one of the shallowest granuloblastic mantle xenoliths (KH77-7) indicates this sample is in isotopic equilibrium with the pyroxenites found in composite mantle xenoliths (Roden et al., 1988). Furthermore, this sample has been shown to be chemically re-enriched in a basaltic component (Irving, 1980; BVSP, 1981; Irving and Frey, 1984; Roden et al., 1988). These observations, along with the modelled maximum temperatures, are taken to imply that the uppermost mantle and lowermost crust were re-enriched in a basaltic component between 20 Ma and 30 Ma due to magmatic intrusion. Mineral isotopic data of a lower crustal granulite from Kilboume Hole yields a re-equilibration age of 34 f 10 Ma, whereas bulk isotopic model ages of the lower crust indicate 1.6 f 0.1 Ga (Padovani and Reid, 1989) again consistent with our model. The deepest granuloblastic mantle xenoliths (KH77-15, KH77-11, KH77-12; open circles, Fig. 5) fall in the region where near-solidus temperatures are never attained in our model. These xenoliths do not show the extreme compositional re-enrichment of the shallowest granuloblastic and composite mantle samples, and are in isotopic disequilibrium with the pyroxenite intrusives (Roden et al., 1988). Unlike the granuloblastic samples, the protogranular mantle xenoliths are compositionally more barren and do not show the progressive bulk Fe enrichment or increasing LREE with decreasing depth (Bussod, 1982; Bussod and Irving, in prep.). This implies that the degree of basaltic re-enrichment of the uppermost mantle may be temporally linked to the intrusive event 20-30 Ma as in the case of the textural re-equilibration. Evidence for partial melting of the wall rock can be observed in some composite xenoliths
386
BUSSOI) AND
ti.Y.A.
(Bussod upper
and Irving, mantle
although
this
pyroxenite the
is contingent mineral
fabric
pyroxenite
shape
along
within
the
age of the
flow continued amphibole,
anisotropy
beyond
15) and rare mica-bearing
possibly
and
deforma-
dikes indicate
the solidification
The presence
amphibole
in
wall rock to the
with plastic
of dissemi-
in veinlets
samples
granuloblastic and composite that a hydrous liquid phase recrystallization, complex.
Continuity
the pyroxenite
time for the intrusives. nated
to 30 Ma,
on
in the lherzolitic
interface
tion features plastic
that a fertile
prior
dikes, which is unknown.
observed
tectonic
1990) suggesting
may have existed
(KH77-
present
only in
xenoliths, suggests was present during
related
to the intrusive
lower crust and upper thinning (25% extension) SRGR
characteristics.
a process, position
and
reasonably profile
It seems plausible could
type
of this
constrained.
However, spheric
lithospheric
cause of the thermal with the southern magmatic
thinning anomaly
Rio Grande
intrusion
event
are
temperature and/or
for the pre-rift
are precluded
to a
magnitude,
intrusive
thinning
is required
upwelling
The
A transient
due to lithospheric
matic intrusion
that such
lead naturally
structure.
timing
WILLIAMS
mantle produces crustal and matches present-day
if continued,
spreading-center
L).R
and
maghistory.
astheno-
as the primary
currently
associated
rift. The timing
at the crust-mantle
for
boundary
beneath Kilbourne Hole is consistent with a regional event associated with the origination of a
These observations are all consistent with our model and further support the concept that the basaltic re-enrichment is correlated with a crust-
continental extension zone and the formation of a granulite metamorphic complex in the lower crust. The modelled present-day heat flow of 82 mW
mantle
mm2 is in agreement with present-day heat flow measurements in the vicinity of Kilbourne Hole. The recent emplacement age of the xenolith suite
boundary
intrusive
complex.
Other
than
the mantle composites, candidates for such a complex in the crust and mantle are the ortho- and clino-pyroxenites
described
by Padovani
(1977)
Baldridge (1979) Irving (1980), Dromgoole and others. Although barometric estimates
(1984) cannot
at Kilbourne Hole maar ( < 100 ka) precludes other modern regional thermal or structural turbation
of the lower
crust
and
upper
any per-
mantle.
their place
This model predicts that the anomalous compressional wave velocities measured in this region are
them in the range of 25 to 50 km depth 20 Ma ago, using our thermal model or within the present depth range of the intrusive zone in our model
not the result of a partially molten crust-mantle boundary but are due to a relatively young (< 20 Ma) N-S trending mafic intrusive complex. This is consistent with observed seismic anisotropies in
be made accurately closure temperatures
(26-46
for these assemblages, (900-1050°C) would
km).
the vicinity of the crust-mantle boundary in the SRGR (Sinno et al., 1986) and the unusually large
Conclusions Evaluation
of the P-T-time
mantle boundary complex thermal last
30 m.y..
at Kilbourne and kinematic
The
path of the crustHole reveals a history over the
piezothermometric
data
Kilbourne Hole xenoliths support the idea inverted temperature gradient in the upper existed within the past 20 m.y.. This is sistent with a simple lithospheric thinning
for
that an mantle inconmodel
by asthenospheric upwelling, but rather requires a rapid transfer of heat directly to the crust-mantle boundary from a deeper mantle source. Numerical thermal calculations constrained by piezothermometric data at depth show that a model of regional crustal underplating by intrusion of magma in the
proportion of composite and mafic lithologies found in the xenolith popuiations of this area (Padovani and Carter, 1977; Fodor, 1978; Baldridge,
1979; Irving,
others). The extent ble to the northern
1980; Dromgoole,
1984; and
to which this model is applicaand central Rio Grande rift is
uncertain at this time as the original rift structure in these regions may have been modified by late stage volcanism associated with large central volcanic centers unrelated to rift development. Acknowledgements The authors would like to thank J.-C. Mareschal, L. Hirsch and an anonymous reviewer
THERMAL
AND
KINEMATIC
for their careful
MODEL
reviews
G.B. acknowledges
OF THE
SOUTHERN
and critical
RIO GRANDE
Brown,
comments.
of Geophysics
R.L.,
Krumhansi,
P.A., Chapin,
Cook, F.A., Kaufman,
the support of T. Shankland
and the Institute
387
RIFT
and Planetary
Physics at Los Alamos.
CGCORP
seismic reflection
In: R.E. Riecker Magmatism. Bussod,
References
1982.
mantle/lower
G.E. and Bishop,
tion of Ca-Mg and
F.C.. 1982. Experimental
exchange
clinopyroxene:
Planet. Adams,
between
potential
for
geobarometry.
G.E. and Bishop,
Earth
Contrib.
Aldrich,
M.J.,
Jr., Chapin,
Stress history
results in the CaO-FeG-
Mineral
C.E.
and tectonic
and
R.J., 1975. Melting lo-35
opside
A.W.,
1986.
at high
of two basanites
pressures.
Carnegie
in the
T.,
Inst.
Cenozoic
Wash.
Implications
of plate
tectonics
tectonic
evolution
for Western
North
for
the
America.
W.S., 1979. Mafic
from the Rio Grande on the composition Volcanol. Baldridge, Leavy,
Perry,
F.V.,
extension:
rift, New
Basaltic
Resour.
state of the lithosphere.
Vaniman,
Nealy,
A.W., Kyle, P., Bartov, Middle
J.
and
L.D.,
V., Steinz,
associated
to Late Cenozoic
Colorado
Mexico
Plateau
with
magma-
and central
Arizona.
N.M.
Volcanism
Study Project, Planets.
D.D., 1971. The thermal
crust.
In: J.G. Heacock
cal Properties Geophys. Blackwell,
of the Earth’s
Monogr.,
structure
(Editor),
Rio
N.Y.,
on 1286
of the continental
The Structure
Crust.
Cenozoic
Tectonic
S.R.,
tectonic
Union,
14: 169-184. loss in the west-
Geophysics
1987. Pressure-temperature-time of granuhtes.
of the West-
S.R., Wall, V.J. and Boettcher, and geological
paths
and
a
J. Geol., 95:
A.L., 1983. Experimenapplications
in the system Fe&TiOr-AlsOs-SiOs-HsO. Y. and
tures.
spreading 501-525.
ridges.
Phil. Trans.
1978.
Partial
R. Sot. London,
Contrib.
segments and
Rep., LA-7498-C.
of the Rio Grande and
Sci. Lab.
rift: compari-
the role of transverse 1978 Rift.
struc-
International
Los Alamos
SymSci. Lab.
pp. 24-27.
C.E., 1979. Evolution (Editor),
of the Rio Grande
Rio Grande Union
accretion
rift. In: R.E.
Rift: Tectonics
and Magma-
Spec. Publ., pp. l-5.
K.C., 1982. Plate-tectonics
model
for Proterozoic
in the southwestern
United
con-
States. Geol-
ogy, 10: 37-42. Coney, P.J. and Reynolds, Cook,
S.J., 1977. Cordilleran
Benioff zones.
270: 403-406.
F.A.,
Decker,
liminary
transient
E.R.
and
Smithson,
heat flow model
New
Mexico.
Earth
Cook, F.A., McCullar,
D.B., Decker,
structure
S.B.,
1978.
Pre-
of the Rio Grande
rift
Planet.
40:
Sci.
Lett.,
rift. In: R.E. Riecker
Tectonics
and
E.R. and Smithson,
and evolution
Magmatism.
S.B.,
of the southern
(Editor), Am.
Cordell,
L., 1982. Extension
Rio G&de
Geophys.
Rio Rift:
Union
in the Rio Grande
Daggett,
P.H.,
Structure
Keller
Davis,
G.R.,
Morgan,
of the southern J. Geophys.
Spec.
rift. J. Geo-
P.M., 1987. Geophysical
Am. Mineral.,
Economics
of the Pacific
Ser. A, 288:
Decker,
Gold Coast,
interpretation
across
studies Structure, Rim,
Australia,
E.R. and Smithson,
C., 1986.
from
gravity
of rift structures Mineralization
Proceedings
Pacific
and and Rim
87: 99-104.
S.B., 1975. Heat flow and gravity
the Rio Grande
Mexico and west Texas.
Wen, rift
Res., 91: 6157-6167.
In: The Geology,
under
P. and
Rio Grande
dynamics.
melting
central
1978 International
Rift. Los Alamos
Abstracts,
tism. Am. Geophys. Condie,
and Abstracts,
on the Rio Grande
Riecker
of the Lemiuplift,
pp. 22-24.
Program
Congress, AB&gre, C.J.,
Program
of equilibria
68: 1049-1058. Bottinga,
of
Institute,
development
on the Rio Grande
interpretation.
tal investigations
In:
on the Processes
phys. Res., 87: 8561-8569.
617-632. Bohlen,
Rio
Publ., pp. 195-208.
Geol. Sot. Am. ‘Mem., 152: 175-208.
model for the evolution
rheologic
Hole xenoliths.
tilted fault-block
C.E., 1978. Evolution
Grande
and Regional
and
the southern
and Planetary
an intrarift
1979. Crustal
States, In: R.B. Smith and C.P. Eaton (Editors),
em Cordillera.
Los
316-326.
and Physi-
Am. Geophys.
D.D., 1978. Heat flow and energy
em United
beneath
from Kilboume
Lunar
Rep., LA-7498-C,
in southern
PP. Blackwell,
at Los Angeles,
to the Conference
Rifting.
New Mexico.
tinental
Bur. Min.
Volcanism
New York,
Spine1
Conditions.
1981. Thermal
R.M., 1978. Structural
Nature,
1981. Basaltic
Pergamon,
A.J., mantle
Presented
Chamber&
Chapin, D.T.,
of
Hypersolidus
of California
upper
rift; evidence
posium
Mem., 46: 187-202.
the Terrestrial
Bohlen,
suites
and their bearing
E.S., 1989. Magmatism
tism of the southeastern Miner.
inclusion
Res.. 6: 319-351.
B.D., Laughlin,
Grande
ultramafic
and thermal
G. and Gladney, lithospheric
and
Deformation
and
Irving,
Grande
sons between
rift (New Mexico)
Geotherm. W.S.,
and
of the
Chapin,
Geol. Sot. Am. Bull., 81: 3513-3536. Baldridge,
G.Y.
Symposium
1970.
upperHole, New
2: 266-267.
University
tar Mountains,
74: 512-515.
Atwater,
continental Kilboume
457, pp. 145-148.
Res., 91: 6199-6211.
behavior
the
Experimental
history Papers
and
Calif., 185 pp. (unpublished).
Planetary
of the Rio Grande
kb and the effect of TiOr on the olivine-di-
reactions
Yearb.,
94: 230-237.
Laughlin,
development
rift, New Mexico. J. Geophys. range
Petrol.,
of
beneath
Subsolidus
Ph.D. Thesis, Bussod,
F.C., 1986. The olivine-clinopyro-
MgO-SiO,
1991. at
Angeles,
experimental
system.
G.Y.,
rift.
Rift: Tectonics
Union Spec. Publ., pp. 169-184.
crust transition
Lherzolite
orthopyroxene,
Sci. Lett., 57: 241-250.
xene geobarometer:
Arculus,
olivine,
Bussod,
investiga-
Rio Grande
Nature
Mexico. Terra Cognita, Adams,
A.R.,
studies of the Rio Grande
(Editor),
Am. Geophys.
G.Y.,
C.E.. Sanford,
S., Oliver, J.E. and Schilt, F.S., 1979.
J. Geophys.
rift in southern
New
Res., 80: 2542-2552.
388
DePaolo,
D-J.. 1979. Estimation
magmas:
a modified
parison
of the depth of origin of basic
thermodynamic
with experimental
approach
and a com-
melting studies. Contrib.
Mineral.
Petrol., 69: 265-278. Dodson,
Dromgoole.
processes
solids. Nature,
from Kiiboume
Washington Dromgoole,
history
of
1973. Conditions
Harte.
University,
and Chemistry
of Sulfides in
St. Louis, 199 pp. (unpublished). J.D., 1987. Interpretation
in a suite of xenoliths
B., 1977. Rock nomenclature
xenoliths. studies
of garnet
liths from the Delegate Irving,
A.J.,
processes
and electriRes. Lett.,
9: 1271-1274. Eaton.
G.P.,
1979. A plate-tectonic
crustal
spreading
Riecker
(Editor),
Rio Grande
tism. Am. Geophys. Elphick,
model
in the western
Rift: Tectonics
J. and Loomis. of cation
1. Experimental
methods
Contrib.
Mineral.
90: 36-44.
Petrol..
W.E. and Bomhorst.
context
In: R.E.
and Magma-
of regional
events. In: R.E. Riecker (Editor), ics and Magmatism.
volcanic
and
Rio Grande
Am. Geophys.
Union
rift in
Spec. Publ., pp.
lation
of plate
motions
mide to basin-range. Fabrib.
with continental
Tectonics,
J., 1979. Spinel-olivine
from
ultramafic
tectonics:
Contrib.
A.A., 1989. Inflected
liths
are
real:
Kimberlites
and
mantle
Petrol.,
g-therms
from
olivine
Related
Rocks.
Geol.
69:
from xeno-
barometry. Sot.
In:
Aust.
Spec.
A.A. and Boyd, F.R., 1978. Pressure~e~dent
bility of calcium enstatite. Finnerty,
Carnegie
A.A.
Application rocks.
In: Lunar Institute,
Rigden, and
S.M., estimation
Planetary
Houston,
R.V., 1978. Ultramafic
erysts in Pliocene
coexisting
Inst. Wash. Yearb.,
to pressure
Planetary Fodor.
and
in forsterite
basalt,
with diopside
1981. Olivine
graphite
trical conductivity Ghent,
E.D.,
potential
1976.
XII.
71: 323-342.
Magmatism.
Am.
Rio
Geophys. to middle
Rio Grande
rift,
J.F.. 1986. A new regional
electrical
section
Rift and
of the Rio Grande
and Great
Lunar
peridotites Cosmochim.
Kushiro,
Plains.
J. Geophys.
melts at high pressures,
for high elec-
Plagioclase-garnet-Al,SiO,-quartz:
Mathez,
61:
McCullar, crustal
I.S., 1980. A regular Mineral.
solution Petrol.,
and and
heat
flow
Regional
in the Basin
Geophysics
H.H. and Naeser,
and
Range
(Editors),
Ceno-
of the Western
of carbon
V.J. and in mantle
New Mexico,
rift, as indicated
and fission track dating.
E.A., Dietrich.
C.W., 1986. Evolu-
field, Northern
to the Rio Grande
Irving,
J. Geophys. A.J.,
peridot&es.
and
by potasRes., 91:
1984. The geoGeochim.
Cos-
Acta, 48: 1849-1859. D.B. and Smithson, refraction
rift. Eos, Trans. Mercier,
Processes.
N.J., pp. 93-120.
Geol. Sot. Am. Mem., 152: 209-250.
P.W., Mehnert,
mochim.
Physics of Magmatic
In: R.B. Smith and G.P. Eaton
chemistry a
Am. Mineral.,
(Editor),
of silicate applications.
A.H. and Sass, J.H., 1978. Models of an extend-
sium-argon
340: 134-136.
and structure
and their petrologic
tion of the Latir volcanic
and mega-
between
Acta, 54: 2375-2388.
density,
Univ. Press, Princeton,
Cordillera. Lipman,
exchange
as a geothermobarom-
from 2 to 60 kb with applica-
I., 1980. Viscosity,
zoic Tectonics
and
calibrated
tions. Geochim.
6329-6365.
Contrib.
wave
(Editor),
Res., 91: 6246-6262.
eter for natural
K. and Chan, T.. 1989. Grain-
liquids.
The
1979. Regional
Riecker
T.P. and Brey, G.P., 1990. Calcium
its relation
in rocks and implications
M.S. and Carmichael,
J.W.,
rift from surface
of the northern
olivine and clinopyroxene
Geol.
in the lower crnst. Nature,
model for metaluminous
Schlue,
LJ., 1986. Late Cretaceous
history
Basin and Range
province.
and lunar
710-714. Ghiorso,
Physics:
pp. 115-126.
Y. and Hermance,
ing lithosphere
New Mexico.
geobarometer-g-thermometer.
to Planetary
and
New Mexico. J. Geophys.
Lachenbruch,
Texas, pp. 279-281. and mafic inclusions
B.R., Fyfe, W.S., Tazaki,
boundary
tectonic
and
barometry:
for terrestrial
Black Range,
Cosmochim.
In: R.E.
Tectonics
Union Spec. Publ.,
Princeton
Sot. Am. Bull., 89: 451-459. Frost,
L.W. and
measurements. Rift:
solu-
77: 713-717.
Science,
Geochim.
of the Rio Grande
In: R.B. Hargraves
Pub]., 14 (2): 883-900. Finnerty,
in
on partition
Res.. 91: 6359-6366.
Mineral.
evidence
abundances
Wiley, New York, N.Y.. 490 pp.
Braile,
structure
dispersion
Kahler,
in peridotites
329-336. Finnerty,
G.R.,
adjacent
Lara-
3: 115-119.
geothermometry
complexes.
Planets.
model for the southern
G.A., 1984. Corre-
for
Am. J. Sci., 280-A:
constraints
genesis.
W.M.. 1968. An Introduction
Tertiary
Rift: Tecton-
of composite
and implications
the mantle.
and megacryst
Keshet, D.C., Cox, A. and Thompson,
xeno-
J. Petrol.,
Acta, 48: 1201-1221.
Kelley. S.A. and Duncan,
tectonic
416-438. Engebretson,
within
coefficients
Grande
m.y.
geochemistry basalts
and their host basalts:
crustal
data.
and
in alkalic
megacrysts
Keller,
in aluminosilicate
T.J., 1979. The Rio Grande
post-40
pipes, Australia.
389-426.
Terrestrial
and interdiffusion
to
experimental
granulite
Irving. A.J. and Frey, F.A., 1984. Trace element
Kaula,
T.P.. 1985. Experimen-
diffusivities
garnets, Elston,
States.
Union Spec. Publ., pp. 7-32.
S.C., Ganguly,
tal determination
for late Cenozoic
United
and pyroxene
basaltic
1980. Petrology
magmatic
Geophys.
relation
in ol~vine-bearing
15: l-40.
of the
Hole, New Mexico. Geol. Sot. Am. Spec. Pap., 215: 25-46. T.J.. 1982. Free carbon
textures
and high pressure
pyroxenite
xenoliths
in the Earth’s mantle.
magma
J. Geol., 85: 279-288.
ultramafic
cal conductivity
of basanrte
with particular
and r~~stallization
from Kilboume
Duba, A.G and Shankland,
of melting
Earth Planet. Sci. Lett., 17: 456-465.
Irving, A.J.. 1974. Geochemical
Hole, New Mexico. M.A. Thesis.
E.L. and Pasteris,
sulfide assemblages
and thermal
259: 551-553.
E.L., 1984. Petrology
Xenoliths
D.H.,
from garnet peridotite. deformation
M.H.. 1976. Kinetic
slowly cooling
Green,
J.-C.C.,
Rheological
profile
S.B., 1977. Unreversed across
Am. Geophys. 1977.
Natural
Heterogeneity
the southern Union,
58: 1184.
Peridotites:
of the Upper
seismic
Rio Grande Chemical
Mantle.
and
Ph.D The-
THERMAL
AND
KINEMATIC
MODEL
OF THE
SOUTHERN
RIO GRANDE
389
RIFT
sis, State Univ. of New York at Stony Brook, Stony Brook,
of silicate glasses and liquids:
N.Y., 669 pp.
l-25.
Mercier,
J.C.C.,
1980.
Tectonophysics, Morgan,
Single
pyroxene
accretionary
xenoliths:
history.
clues to the Earth’s
J. Geophys.
Res.,
91: 12,287-
P.. Seager,
W.R.
and Golombek,
zoic thermal
and tectonic
J. Gwphys.
Res., 91: 6263-6276.
Nathenson,
evolution
M. and Guffanti,
in the coterminous
M.P., 1986. Ceno-
of the Rio Grande
States.
gradients
J. Geophys.
Res., 93:
Keller,
G.R.
and
Stewart,
structure
along the Rio Grande
profiles.
In:
Tectonics
R.E.
and
Riecker
J.N.,
Crustal
rift from seismic refraction (Editor),
Magmatism.
1979.
Am.
Rio
Grande
Rift:
Geophys.
Union
Spec.
Olsen,
K.H.,
Grande
Baldridge,
sky and G. Qvale and
W.S. and Callender,
rift: an overview.
Regional
J.F.,
In: LB. Ramberg,
(Editors),
Continental
Characteristics.
1987. Rio
E.E. Milanov-
Tectonophysics,
143:
119-
Padovani,
E.R.,
1977.
Granulite
Facies
Kilboume
Hole Maar
Evolution.
Ph.D. Thesis, University
Xenoliths
and Their Bearing
from
on Deep Crustal
of Texas, Dallas, Texas,
Padovani,
E.R. and Carter,
crustal J.G.
evolution
Heacock
Physical
J.L.,
beneath
(Editor),
1977, Aspects
south
Am.
Crust:
Geophys.
of the deep
New Mexico.
beneath
Hole, New
R.J.,
Greene,
G.W.
the southwest
United
Res., 76: 6376-6413
United
of geomagnetic
States.
Bull. Scripps
variations
in
Inst. Gceanogr.,
13: 165 pp. J.G., Jaupert,
through
oceanic
the earth.
C. and Galson,
Rev. Geophys.,
em New Mexico,
P., 1979. Rio Grande Rio
Grande
Am. Geophys.
W.R.,
and the heat loss of
Shafiquillah,
tion of the southern
Union
Rift:
Tectonics
J.W.
and
from basahs
Rio Grande
In: and
Spec. PubI., pp. 87-106.
M.. Hawley. dates
rift in southChihuahua.
west Texas and northern
(Editor),
Magmatism. Seager,
cmst
Space Phys., 18: 269-311
Seager, W.R. and Morgan, Riecker
D., 1980. The heat flow
and continental
J.R.,
gradient
Its Nature
Marvin,
and the evolu-
rift. Geol. Sot. Am. Bull.,
Rocky Grande Union
and
lower
the southern
M., Mansure,
New Mexico. limits
Rift:
Complex.
N.M. Bur.
rift. Geol. Sot.
Tectonics
In: R.E. Riecker
and
Magmatism.
the southern (Editor),
Am.
Rio
R.E.,
W.E.
Geophys.
1986. Estimates
of terrestrial
leum tests along
the Rio Grande
ern New Mexico. J. Geophys.
B.R. and
heat flow from rift in central
Mimer
J.,
deep petroand south-
Res., 91: 6225-6245.
Y., 1986. Thermochemical
Takahashi,
Univer-
Morgan,
P. and
of the southern
Rio
profiling.
J.
J. Geophys.
structure
in
the Gnome
two-feldspar
explo-
geothermometer.
of a dry peridotite
on the origin
A.R., 1976. Crustal
New Mexico interpreted
sion. Bull. Seismol. and
termination
Contrib.
systems.
structure
in
explo-
Sot. Am., 66: 877-886.
Holloway,
1977.
up to upper
from the Gasbuggy
J.R.,
of the fluid-absent
pehtic system.
KLB-1
of peridotitic
Res., 91: 9367-9382.
T.R. and Sanford,
P.R.A.,
from
60: 667-674.
implications
D.
1962. Crustal
Sot. Am., 52: 1017-1030.
E., 1986. Melting
Toppozada,
complex properties
L.C.,
interpreted
J.C., 1975. A practical
Am. Mineral.,
Wells,
G.R.,
structure
from seismic refraction
Pakiser,
Mexico
sion. Bull. Seismol.
Vielzeuf,
Broadwell,
and
New
centrat
Princeton
Res., 91: 6143-6156.
eastern
mantle.
C., 1979. Geothermal
Spec. Publ., pp. 253-267.
P. and Bottinga,
Geophys.
trans-
pp. 201-264.
Keller,
rift determined
to the geothermal
rift within
P.H.,
of magma
In: R.B. Hargraves
Processes.
New Jersey,
Daggett,
14 GPa:
A.J. and Shearer,
M., EggIeson,
Y.A.,
Stormer,
Rio Grande
of the Rio Grande
Mountain
Physics of Magmatic
Grande
Geophys.
Union,
Sinno,
mechanisms
to the surface.
and
Prog., 8: 621.
characteristics
(Editor),
the mantle
S.H., 1986. Crustal
Mem., 46: 174-179.
1976. Upper beneath
1980. The fracture
from
Stewart,
Dona An - a county,
Am., Abstr.
port
Harder,
20: 19-55.
Hole Maar,
Shaw, H.R.,
In:
E.R. and Reid, M.R., 1989. Field guide to Kilboume
Min. Miner. Resour.
Richet,
central
The Earth’s
Properties.
Monogr., Padovani,
Reiter.
1988. Isotopic
Acta, 52: 461-473. Munroe,
U., 1970. Anomalies
sity Press, Princeton,
158 pp. (unpublished).
Reiter,
A.H.,
24:
95: 87-99.
139.
Reid,
Cosmochim.
Lachenbruch,
R.F., 1984. New K-Ar
Rifts-Principal
V.R.,
of the upper mantle
rift: results from Kilboume
States. J. Geophys.
R.E.
Publ., pp. 127-143.
Murthy,
and Moses, T.H., Jr., 1971. Heat flow in the western
Sclater,
K.H..
A.J. and composition
Mexico. Geochim.
Schmucker,
M., 1988. Geothermal
United
rift.
6437-6450. Olsen,
Irving,
a young continental Sass, J.H.,
12,375. Morgan,
M.F.,
and trace element
J.W., 1986. Ultramafic
late
Roden,
thermobarometry.
70: l-37.
a review. Rev. Geophys.,
Mineral.
Pyroxene Contrib.
1988.
Experimental
melting
relations
Petrol.,
98: 257-276.
thermometry Mineral.
dein the
Petrol.,
in simple 62: 129-139.
and