Geological Society of America
Special Paper 388
2005
The deep mantle thermo-chemical boundary layer:
The putative mantle plume source
Thorne Lay*
Earth Sciences Department, University of California–Santa Cruz, Santa Cruz, California 95064, USA
ABSTRACT
Hypothetical narrow, cylindrical thermal plumes from great depths in the Earth,
which are commonly invoked to account for relatively stationary and persistent concentrations of volcanism at the surface, are intimately linked to the notion of instabilities and upwellings of a deep thermal boundary layer in the mantle. If the mantle is
undergoing whole-mantle convection with no internal stratification, the primary thermal boundary layer within the interior that may serve as a plume source is located
above the core-mantle boundary. If the mantle convection regime is globally or locally
stably stratified by compositional layering, thermal boundary layers should exist on
the margins of the compositional contrasts, and any such boundary layers could serve
as plume sources. Seismological, mineral physics, and geodynamic evidence favors the
existence of a complex thermo-chemical boundary layer in the lowermost few hundred
kilometers of the mantle, the so-called D region, involving extensive, if not global,
chemical stratification. Although chemical stratification of the mid-mantle has not
been ruled out, the lack of direct evidence for global existence of thermal boundary
layers in the mid-mantle means that the lowermost mantle region is generally invoked
as the source of thermal plumes that give rise to melting and chemical anomalies associated with hotspot volcanism. Observational constraints on this putative mantle
plume source are summarized here, along with consideration of the attendant implications for plume existence.
Keywords: D″ region, core-mantle boundary, superplumes, thermal plume, mantle heterogeneity.
INTRODUCTION
The mantle plume hypothesis (Morgan, 1971, 1972; Anderson and Natland, this volume) postulates the existence of
narrow cylindrical upwellings that drain hot material from an
internal thermal boundary layer in the mantle, giving rise to massive pressure-release melting within and below the lithosphere,
thereby accounting for hotspot volcanism. Although various as-
pects of the shallow melting and dynamic system predicted for
thermal plumes can be explained by alternate mechanisms, all
notions of thermal plumes share the need for a deep thermal
boundary layer from which the plumes originate. Characterization of any mantle thermal boundary layers at depth thus has
direct relevance to assessment of the plume hypothesis.
The existence of thermal boundary layers in a planet has two
main prerequisites: heat must flow across a persistent boundary,
*E-mail: [email protected].
T. Lay, 2005, The deep mantle thermo-chemical boundary layer: The putative mantle plume source, in Foulger, G.R., Natland, J.H., Presnall, D.C., and Anderson,
D.L., eds., Plates, plumes, and paradigms: Geological Society of America Special Paper 388, p. 193–205. For permission to copy, contact [email protected].
© 2005 Geological Society of America.
193
194
T. Lay
and there must be density or viscous stratification of the system
that permits superadiabatic thermal gradients to develop and
persist adjacent to the boundary (e.g., Stacey and Loper, 1983;
Christensen, 1984; Olson et al., 1987). These requirements
are certainly satisfied in the Earth at the surface and at the coremantle boundary. Dramatic compositional and viscosity contrasts across these two boundaries combine with the ongoing
cooling of the mantle and the core to give rise to thermal boundary layers at the top and bottom of the mantle. If the mantle
convection regime is further dynamically stratified as a consequence of chemical or rheological layering, thermal boundary
layers with superadiabatic thermal gradients will have developed across the stratified boundaries as heat has sought its way
to the surface. Advocates of the mantle plume hypothesis thus
invoke boundary layer instabilities either from the core-mantle
thermal boundary layer or from some mid-mantle thermal boundary layer associated with stratification of the mantle flow regime
(e.g., Kellogg et al., 1999).
The presence of a global seismic velocity discontinuity near
650 km depth, combined with the change in seismic velocity
gradients above and below the discontinuity, prompted early
consideration of chemical layering of the mantle at the base of
the transition zone (e.g., DePaolo and Wasserburg, 1976). Recognition that the 650 km discontinuity can be explained well by a
phase transition in (Mg,Fe)2SiO4 (with possible contribution
from a garnet to perovskite transition) removed the need for a
chemical contrast at this depth (e.g., Anderson, 1967). The endothermic phase change at 650 km probably impedes radial
transport of material and may be accompanied by a rheological
transition with important dynamic effects (e.g., Dziewonski,
this volume). No globally extensive seismic velocity discontinuity has yet been established at depths between the base of the
transition zone and the core-mantle boundary, and although it is
plausible that density stratification does exist and is simply difficult to resolve with seismic waves (Anderson, 2002), there is
as yet no strong observational basis for asserting that there are
global thermal boundary layers in the mantle anywhere other
than near the surface and the core-mantle boundary (e.g., Castle and van der Hilst, 2003). Thus advocates of thermal plumes
usually invoke deep mantle origins, although upwellings from
the transition zone or from a mid-mantle discontinuity have not
actually been ruled out.
Seismic wave analysis has revealed that there is significant
structural heterogeneity near the base of the mantle, with strong
radial seismic velocity gradients (or discontinuities) and strong
lateral velocity variations. The lowermost 200–300 km of the
mantle, designated by seismologists as the D″ region, has the
most pronounced variations, and these are more strongly manifested in shear wave velocities (Vs) than in compressional wave
velocities (Vp). These seismic wave velocity variations provide
evidence that a major thermo-chemical boundary layer exists on
the mantle side of the core-mantle boundary, extending at least
several hundred kilometers upward into the lower mantle (e.g.,
Lay et al., 2004). Of particular importance are two large-scale
Should these be expressed as Vs and Vp (with “V” italic and “s” and “p” subscripted)? If so, please indicate the change wherever these terms appear in the page
proofs of your chapter
regions of low shear velocity in the deep mantle below the central and southern Pacific Ocean and below the southern Atlantic
Ocean, Africa, and the southern Indian Ocean (Dziewonski et al.,
1993). These regions have horizontal dimensions of 2000–3000
km and shear velocity anomalies (dVs) averaging –3% to –5%. Should
The common assumption that low shear velocity in the mantle any part
of this be
is a consequence of relatively hot, buoyant material has led to in italics
designation of these features as “superplumes,” an unfortunate or be subterm that has dynamic implications that are actually at odds with scripted?
some of what is known about these regions and with the conventional fluid dynamic definition of a thermal plume. I will call
these regions large low-velocity provinces, a dynamically neutral description appropriate for the uncertain nature of these
structures.
The two large low-velocity provinces in the lower mantle
have strong seismic velocity expressions in the D″ region, but it
appears that some portions of them extend upward (or are overlain by separate low-velocity material) 800–1000 km above the
core-mantle boundary, significantly exceeding the conventional
thickness associated with D″. Weaker velocity anomalies may
extend to even shallower depths. Evidence for chemical contrast
between large low-velocity provinces and surrounding mantle,
summarized later, suggests that large low-velocity provinces are
massive chemical heterogeneities in the lower mantle, and as
such they may be enshrouded in thermal boundary layers. Thus
even though large low-velocity provinces may not be buoyantly
ascending or may be unable to penetrate up to the surface, they
may still give rise to thermal plumes shed from their boundary
layers. If they are long-lived they may significantly affect the
overall thermal transport in the deep mantle.
In addition to large low-velocity provinces, there are more
localized regions of relatively low velocity in the lower mantle,
some of which underlie surface hotspots and some of which
do not. With as yet poorly resolved lateral dimensions of 200–
500 km, those low-velocity regions with significant near-vertical
continuity are commonly interpreted as plumes rising through
the mantle (e.g., Goes et al., 1999; Bijwaard and Spakman,
1999; Nataf, 2000; Montelli et al., 2004; Zhao, 2004), with the
explicit assumption that low shear velocity indicates hot, buoyant material.
Combining the general expectation of thermal boundary
layers above the core-mantle boundary and on the margins of
major chemical boundaries, the deep sources for plumes are
commonly envisioned, of late, as shown in Figure 1. “Classic”
plumes are viewed as narrow conduits with diameters of 50–
200 km rising directly from the thermal boundary layer at the
core-mantle boundary, some with sufficient plume flux to ascend
2900 km through the mantle conveying excess temperatures that
induce large-scale pressure-release melting at locales such as
Hawaii and Iceland (e.g., Stacey and Loper, 1983; Davies, 1988;
Sleep, 1990). If D″ is a distinct chemical layer, such plumes may
rise from the top of that layer rather than directly from the coremantle boundary (e.g., Tackley, 1998). Hotspot regions of the
south Pacific and Africa are, perforce, attributed to plumes ris-
The deep mantle thermo-chemical boundary layer
Andes
S. America
Pacific Ocean
Hawaii
195
Atlantic Ocean
410
660
+
-
-
+
East Pacific
Fise
Africa
+
TBL
-
++
-
CMB
Figure 1. Schematic mantle cross-section showing key structural elements of the near-surface thermo-chemical boundary layer involved in plate
tectonics and notions of how deep mantle thermal plumes may rise from either a thermal boundary layer (TBL) near the core-mantle boundary
(CMB) or from the thermal or mechanical boundary layers around large low-velocity provinces of distinct composition found below the Pacific
and southern Atlantic Oceans. Globally extensive phase transitions exist at depths near 410 and 660 km.
ing from the large low-velocity provinces in the deep mantle beneath them (e.g., Jellinek and Manga, 2002; Romanowicz and
Gung, 2002). Yet other hotspots are attributed to shallow plate
tectonic processes (e.g., Courtillot et al., 2003; Anderson, this
volume). Entrainment of material from the core, D″, or large
low-velocity provinces in ascending plumes from the deep mantle is commonly advocated as an explanation for geochemical
anomalies in hotspot magmas (e.g., Hofmann and White, 1982;
Hart et al., 1992; Hofmann, 1997; Albarede and van der Hilst,
2002), although the level of confidence with which these can be
tied to any deep boundary layer is not high.
Given the array of possibilities for a deep mantle source of
thermal plumes, it is worthwhile to consider what is actually
known about the lowermost mantle structure to provide a basis
for assessing conceptual notions such as those in Figure 1.
Fortunately several recent review articles summarize the multidisciplinary contributions to our current understanding of lowermost mantle structure (e.g., Lay et al., 1998a; Garnero, 2000;
Lay and Garnero, 2004; Lay et al., 2004). As those references
provide comprehensive reading lists and citations, the goal here
will be to cast summary observations and constraints on the structure and processes near the core-mantle boundary in the context
of the debate (e.g., Anderson, this volume) about the existence
and nature of thermal plumes in the mantle.
ATTRIBUTES OF THE CORE-MANTLE
THERMO-CHEMICAL BOUNDARY LAYER
Thermal Structure
The temperature drop across the surface thermal boundary
layer is well established as being ~1200–1300° (e.g., Green and
Falloon, this volume), but there is great uncertainty about the
temperature contrast across the core-mantle boundary layer. The
latter contrast must be inferred based on extrapolations from the
few relatively robust temperature tie-points in the deep interior
constrained by experiments, the temperatures of upper mantle
phase transitions in olivine, and the estimated freezing point
of core alloy at the inner core boundary. One can estimate the
temperature contrast across the core-mantle boundary when
these tie-point estimates, themselves subject to substantial uncertainty due to imprecisely known compositional effects, are
extrapolated down or up along mantle and core adiabats, respectively, under the assumption that there are no regions of
superadiabatic gradients anywhere other than near the coremantle boundary. This procedure yields a temperature contrast
across the core-mantle boundary on the order of 1000–2000°
(e.g., Williams, 1998; Boehler, 2000), comparable to the lithospheric temperature drop. The core-mantle boundary itself is
nearly isothermal as a result of the very low viscosity of core
material, with a temperature somewhere between 3300 K and
4800 K, and any instability in the overlying mantle thermal
boundary layer could thus bring very hot material upward into
the mantle. The observation that excess temperatures associated
with hotspot magmas are usually found to be less than 200° (see
Anderson and Natland, this volume; Green and Falloon, this volume; Presnall and Gudfinnsson, this volume) raises immediate
questions about the viability of thermal plumes rising directly
from the core-mantle boundary due to the large excess temperature estimates for the boundary layer. Given secular cooling of
the core, the temperature contrast of ancient plumes would be
even higher, and Archaean komatiites are often interpreted in
this light; however, questions are now arising about the excess
temperatures actually associated with these melts (e.g., Parman
and Grove, this volume). Albers and Christensen (1996) and
Farnetani (1997) have explored the excess temperature issue,
196
T. Lay
with the latter establishing that unless the core-mantle boundary temperature estimates are in gross error, it is far easier to
account for the relatively modest temperature excess of hotspot
melts if a much weaker boundary layer serves as the source of a
thermal plume.
If the lower mantle is chemical layered—for example, if D″
is a compositionally distinct layer—the estimated temperature
contrasts across the upper boundary of D″ and the core-mantle
boundary can be reduced by a factor of two, helping to resolve
the excess temperature problem (Farnetani, 1997). Dynamic
instabilities of a thermal boundary layer above a dense, compositionally distinct layer tend to draw in material from the upper
portion of the thermal boundary layer, not sensing the full
temperature increase. This same argument also holds for thermal boundary layers at shallower depths than D″. A key issue
that then emerges is the longevity of a chemical contrast in the
face of steady entrainment (e.g., Sleep, 1988; Tackley, 2000).
Relatively small density increases of 1%–3% allow chemical
layers to persist for a long time for entrainment models that
account for strong pressure effects on material properties such
as thermal expansion (e.g., Zhong and Hager, 2003). Pressure
effects also serve to increase the temporal and spatial scales of
lower mantle features so that very large long-lived features are
favored at the base of the mantle.
Absolute temperature issues aside, the best argument for a
thermal boundary layer at the base of the mantle is that the heat
flux through the core-mantle boundary needed to sustain the
geodynamo is estimated to be from 1 to 15 terrawatts (TW),
which is a very uncertain number but almost always a positive
value (except in a few exotic models), implying heatflow from
the core into the mantle (e.g., Buffett, 2002; Lay et al., 2004).
Given estimates of a global surface heat flux on the order of
42–44 TW (e.g., Kellogg et al., 1999), there is a correspondingly
large uncertainty in the degree to which the mantle is heated
from below versus heated from within, with the former essential
for genesis of thermal plumes by boundary-layer instabilities.
Lower values of core-mantle boundary heat flux are easier to
reconcile with an old inner core and with reasonable initial Earth
temperatures. If chemical layering in the mantle produces any
mid-mantle thermal boundary layers one can obtain lower coremantle boundary heat flux estimates in some of the calculations.
Reassessment of the estimates of global surface heat flux (e.g.,
Hoffmeister and Criss, this volume) may lower the heat flux
values for the surface and the core-mantle boundary as well.
Large Low-Velocity Provinces
Seismological constraints on the temperature field in D″ are
even more indirect than those from geodynamic and geomagnetic
arguments, with lack of constraint on the compositional effects
and on the high-pressure thermal expansion leading to large
uncertainties (e.g., Trampert et al., 2001; Julian, this volume).
Perhaps the most robust statement is that seismology reveals the
existence of large-scale lateral variations in the seismic velocity
structure near the base of the mantle (Fig. 2; see also Ritsema,
this volume), with a large degree 2 component in the shear wave
structure. Recent global seismic tomography models for D″ are
discussed in detail by Lay et al. (2004), but for our immediate
purpose there are two key attributes of the D″ boundary layer
relevant to the plume question.
The first is that two large low-velocity provinces, as mentioned previously, are evident in all of the shear wave models,
separated by a circum-Pacific ring of relatively high-velocity
material. The primary wave (P-wave) velocity expression of these
two large low-velocity provinces is much more muted, but there
is less agreement among Vp models, and Vp fluctuations are
only ~20%–50% of those in the Vs models. As the resolution of
seismic tomography improves, internal coherence of the large
low-velocity provinces may diminish, but the large-scale pattern
will persist (see Dziewonski, this volume). The predominance
of long-wavelength structure runs counter to expectations for a
very hot, inviscid thermal boundary layer beset by many instabilities giving rise to plumes: some process imposes large-scale
structure on the boundary apart from the intrinsic boundary layer
temperature structure. Possibilities include disruption of the hot
thermal boundary layer by cooled downwelling material in the
lower-mantle circum-Pacific high-velocity ring or the existence
of large-scale chemical heterogeneities embedded in the boundary layer. If there is a thermal contribution to the large-scale
heterogeneities, attendant variations in radial temperature gradients will impose a spatially varying heat flux boundary condition
on the core flow regime (e.g., Glatzmaier et al., 1999), which
may play an important role in reversals of the magnetic field.
The second important attribute of the D″ boundary layer to
be gleaned from these low-resolution tomography models is that
in some well-sampled regions, such as under the central Pacific,
there are strong decorrelations between Vp and Vs variations, or
at least much stronger relative perturbations in Vs versus Vp.
This indicates that thermal variations are not the sole explanation for the large-scale patterns. Inversions for bulk sound velocity variations (which depend on bulk modulus and density,
but not on rigidity) emphasize the differences in behavior of Vp
and Vs and suggest some anticorrelation with Vs patterns on
large scales (Fig. 3). This behavior cannot be readily accounted
for by temperature alone, and although there are many technical
challenges to bulk sound velocity estimation, most seismologists
agree that compositional heterogeneity is probably important on
large scales in D″ (e.g., Masters et al., 2000). The differences in
global sampling for P-waves and shear waves (S-waves) in the
deep mantle is significant, but even localized determinations
of velocity structure, as for the central Pacific region illustrated
in Figure 3, support a difference in the behavior of Vp and Vs
that can be accounted for only by bulk modulus and shear modulus perturbations of opposite sign. This appears to be the case
for both the Pacific and African large low-velocity provinces,
although complex relationships between Vp and Vs are documented in other regions as well (e.g., Wysession et al., 1999;
Saltzer et al., 2001).
The deep mantle thermo-chemical boundary layer
A
dVs: Grand
B
197
dVs: Mengin and Romanowicz
4
0
-2
dVs (%)
2
-4
C dVp: Zhao
D
dVp: Boschi and Dziewonski
0
dVp (%)
2
-2
Figure 2. Maps of S velocity (top row) and P velocity (bottom row) variations relative to the global average values in the D″ region at a depth
~200 km above the core-mantle boundary in representative global mantle seismic tomography models. S velocity variations are from Grand
(2002) and Mégnin and Romanowicz (2000), and P velocity variations are from Zhao (2001) and Boschi and Dziewonski (2000). Positive values
(blue) indicate higher-than-average velocities, and negative values (red) indicate lower-than-average velocities. The scale bars differ: S velocity
variations are two to three times stronger than P velocity variations. Large-scale patterns dominate, suggesting the presence of coherent thermal
and chemical structures on continental-type scales.
Attributes of the large low-velocity provinces are summarized in Figure 4, and the many seismic observations associated
with these structures are discussed at length by Lay and Garnero
(2004). The lateral gradients on the edges of these structures are
very abrupt, resolved by detailed waveform and traveltime
analyses (not by global tomography) to be on the scale of tens
of kilometers (Ni et al., 2002). For some portions of the large
low-velocity provinces, sharp lateral gradients persist into the
mid-mantle, 800–1000 km above the core-mantle boundary.
There are ongoing debates about the internal velocity structure
of the large low-velocity provinces, but most researchers agree
that the Vs reductions are much stronger than Vp reductions.
The strong lateral gradients and anomalous Vp/Vs observations
strongly suggest a distinct chemistry of large low-velocity
provinces. At the high pressures near the base of the mantle, the
sensitivity of seismic wave velocity to temperature variations is
significantly suppressed relative to that in the shallow mantle
(e.g., Chopelas, 1996; Julian, this volume), so very large temperature increases would be required to account for the Vs reductions, yielding very hot large low-velocity provinces. However,
the apparent need for a compensating bulk modulus increase (to
explain why Vp is not proportionally low) requires a chemical
change, which alone may suffice to reduce the rigidity with no
thermal anomaly (candidate materials are not yet constrained).
While pushing the limits of normal mode analysis, there is evidence that large low-velocity provinces have higher than normal
density (e.g., Ishii and Tromp, 1999, 2004).
Taking the information presently available at face value, one
could argue either that large low-velocity provinces are massive,
hot upwellings with distinct chemistry that gives rise to their
density and Vp/Vs characteristics (“superplumes”) or that they
are chemically distinct piles of dense material that may or may
not have any thermal anomaly, but do have steep lateral gradients
sustained by the shear flow of the surrounding high-viscosity
198
A
T. Lay
Locations of large low S velocity provinces in D"
Bulk sound velocity, model Sb10l18
1.5
0.0
δVb (%)
B
-1.5
Shear and compressional velocity profiles
PREM-H
PREM-H
SPAC
PPAC
2000
300 km
-3% dVs
300 km
Depth (km)
2200
2400
Sharp
Boundaries
-1- -3% dVs Decrease
2600
δVs 1.7%
δVp 0.75%
-3-12% dVs
velocities
2800
6.3
6.8
ULVZ
7.0
7.2
Vs (km/s)
Should
this be
expressed
as Vb
(with “V”
ital and
“b” subscripted?
7.4
12.8
13.2
13.6
CMB
ULVZ
14.0
Vp (km/s)
(Vp2–4/3Vs2)0.5)
Figure 3. (A) Map of bulk sound velocity (Vb =
variations in the D″ region from the tomographic model Sb10l18 of
Masters et al. (2000). Comparison with the shear and compressional
velocity models in Figure 2 shows anticorrelation of Vb and Vs in D″
in the central Pacific, where relative reductions in Vs are more than 2.7
times those in Vp. The small box indicates the region of a detailed analysis of S- and P-wave triplications by Russell et al. (2001). (B) Models
SPAC and PPAC obtained by Russell et al. (2001) for the central Pacific, along with a reference structure (PREM-H), a modification of the
PREM model of Dziewonski and Anderson (1981), with D″ having
the same velocity gradients as in the overlying lower mantle rather than
reduced gradients as in PREM. This is viewed as an empirical homogeneous, adiabatic reference model relative to which effects of thermal
and chemical heterogeneity in the boundary layer should be assessed.
The sizes of the velocity discontinuities at 2661 km depth are indicated,
along with the presence of a 10 km–thick ultra-low-velocity zone for
models SPAC and PPAC. From Lay et al. (2004).
deep mantle. Numerous investigations of the dynamics of chemical plumes have been spurred by these observations (e.g., Tackley, 2000; McNamara and Zhong, 2004), but a key question
places all interpretations in doubt: what is the relative contribution of thermal versus chemical heterogeneity to the seismic velocity anomalies and to any positive or negative buoyancy? This
Figure 4. Top: Two massive provinces with thick regions of low S velocity exist in the lower mantle: under the southern Pacific and under
the southern Atlantic, Africa, and the southern Indian Ocean. Bottom:
A south-to-north (left-to-right) cross-section sketch of the latter feature
is shown, with a 200–300 km–thick layer in D″ having an abrupt decrease in Vs of 1–3% overlying a layer with average Vs decreases of
3%–5%. Under Africa this low-velocity body appears to extend as much
as 800 km above the core-mantle boundary (CMB), and it may have
an ultra-low-velocity zone (ULVZ) just above the CMB. It has sharp
lateral boundaries in D″ and in the mid-mantle. The low-S-velocity
province under the central Pacific is not yet as fully imaged, but appears
comparable in size and in the strength of the velocity reductions.
issue is wide open, but in the context of considering scenarios
such as that in Figure 1, we can accept with confidence that there
are large chemical anomalies in the deep mantle concentrated near
the core-mantle boundary, possibly with localized extensions
upward beyond the conventional limits of the D″ region. Whether
or not there are strong thermal boundary layers surrounding
these regions that are capable of detaching and rising as plumes
depends on whether large low-velocity provinces have relatively
high, normal, or low temperatures. If the structures are piles of
chemically distinct material with no major heat flux through
their boundaries, shear flows in the surrounding mantle could
still transport hot material from the core-mantle boundary layer
The deep mantle thermo-chemical boundary layer
up along the margins of the large low-velocity provinces, with
instabilities arising from the complex flow geometry (e.g.,
Jellinek and Manga, 2002). Thorne et al. (2004) report a statistical correlation between the location of surface hotspots and
strong lateral gradients near the margins of low-velocity regions
in the lowermost mantle (not only for the two large low-velocity
provinces), which may provide some support for the latter notion. While the statistical significance of spatial correlations of
this type is open to question, it is noteworthy that hotspots tend
to not be centered over the core regions of large low-velocity
provinces, which may bear upon their dynamic status.
199
Detected ULVZ presence (light) and absence (dark)
Ultra-Low-Velocity Zones
2200
2400
Depth, km
It is important to recognize that the seismically characterized
large low-velocity provinces are not equivalent to the “ultralow-velocity zones” detected just above the core-mantle boundary (e.g., Garnero et al., 1998; Thorne and Garnero, 2004). The
latter features occur rather extensively (Fig. 5) and involve very
strong reductions in both Vp (up to 10%) and Vs (up to 30%),
and there is evidence that they have strong (10%) density increases. The ultra-low-velocity zones are typically only a few
kilometers to a few tens of kilometers thick, juxtaposed against
the core-mantle boundary, and have sharp upper boundaries that
reflect short-period seismic waves. The velocity anomalies in Vs
and Vp have a ratio of 2.5 to 3, which is compatible with the
presence of partial melting (e.g., Williams and Garnero, 1996;
Lay et al., 2004), with likely iron partitioning into partially segregated melts possibly accounting for excess density. As seen in
Figure 5, the properties of ultra-low-velocity zones are remarkable in terms of their velocity perturbations, and their location
next to the core-mantle boundary, inarguably the hottest place
in the mantle, strongly suggests a thermal influence associated
with the thermal boundary layer there. However, there are interesting issues, such as, why, if there is dense melt, it has not
completely drained to the core, and whether these regions represent core material that has penetrated into the mantle. The
dynamics and thermal transport of ultra-low-velocity zones are
just beginning to be explored numerically.
Figure 5 indicates that ultra-low-velocity zones occur in
many regions of the core-mantle boundary layer. Williams et al.
(1998) present a correlation analysis suggesting that the distribution of surface hotspots is better correlated with this distribution of ultra-low-velocity zones than with that of the large-scale
low-velocity regions of D″. The significance of such correlations
can be questioned because some ultra-low-velocity zones appear
to actually have very localized three-dimensional configurations,
involving mounds rather than thin lenses (e.g., Helmberger et al.,
1998; Wen and Helmberger, 1998), so their horizontal dimensions may be much smaller than indicated in Figure 5, which
is based on Fresnel zones for 1D structures. Plumes rising from
ultra-low-velocity zones should have large temperature excesses
and should quickly drain out the localized structures, so it seems
difficult to invoke ultra-low-velocity zones as sources for major
PREM
PREM
Pacific
Pacific
2600
2800
ULVZ
3000
5
7
8
6
S Velocity, km/s
ULVZ
13
14
12
P Velocity, km/s
Figure 5. A map (top) showing regions where features of an ultra-low
velocity zone (ULVZ) have either been observed (light gray) or are not
observed (dark gray), and (bottom) seismic velocity models for PREM
and the central Pacific region, where a ULVZ is detected immediately
above the core-mantle boundary (bottom). The ULVZ structures in various regions have P velocity drops of 4%–10% and S velocity drops
of 12%–30%. In general, regions with ULVZ in the D″ region tend
to occur in regions of low S velocities, but this is not universally true
(cf. Fig. 2). Some ULVZ regions may involve much more localized
domelike heterogeneities of smaller lateral extent than indicated by the
1D velocity profiles.
plumes. If dense melts separate out from upwelling boundary
layer instabilities, the ultra-low-velocity zones could be consequences rather than causes of thermal plumes.
The likelihood that partial melting or a melt phase is involved in ultra-low-velocity zones raises the possibility that the
lower mantle is near the eutectic (the lower mantle assemblage
is very likely to have a eutectic) of the bulk mantle composition
(e.g., Akins et al., 2004). The effects of partial melt on seismic
velocities are much more pronounced than those of subsolidus
temperature variations, and this fact motivates consideration
of large low-velocity provinces as regions with very small melt
fractions, perhaps as a consequence of their chemical anomaly
200
T. Lay
(e.g., Lay et al., 2004). Ultra-low-velocity zones might then be
viewed as relatively high-melt-fraction portions of D″, while
large low-velocity provinces have a low melt fraction, with both
differing chemically from overlying mantle. As it is unclear
whether melts in D″ will be buoyant or will sink, the importance
for the plume issue of any partial melting in ultra-low-velocity
zones or large low-velocity provinces remains unclear. Ultralow-velocity zones may represent partial melting of chemically
distinct material and need not be related to the bulk lowermantle eutectic at all. Not all regions of large low-velocity provinces are underlain by ultra-low-velocity zones (the northern
and western portion of the Pacific large low-velocity province
is, and so is the northernmost region of the African large lowvelocity province; see Figures 5 and 4), and some ultra-lowvelocity zones may underlie regions of high velocity in D″, so
no generally valid interpretation of the connection between
these low-velocity features has emerged.
Detected D" S velocity discontinuity
2300
Depth, km
Seismic Velocity Discontinuities
The strong lateral gradients at the margins of large lowvelocity provinces appear to extend right down to the core-mantle
boundary, either as vertical or as steeply dipping boundaries
(e.g., Wen et al. 2001; Ni and Helmberger, 2003). While the
possibility exists that the boundaries are caused by lateral transitions in partial melt fraction, it is more likely that they involve
gradients in composition. In regions of D″ well removed from
the large low-velocity provinces, there is evidence for regionally
extensive seismic velocity stratification, with seismic velocity
discontinuties involving increases in Vp and Vs ~200–300 km
above the core-mantle boundary (e.g., Wysession et al., 1998;
Lay and Garnero, 2004). Vs increases of 2%–3% have been
widely observed (Fig. 6), with Vp increases tending to be weaker
and more variable. The strongest discontinuities are observed in
circum-Pacific regions where the large-scale structure of D″ has
high Vs (Fig. 2); however, there is evidence for both Vp and Vs
increases beneath the central Pacific, near the northern margin
of the Pacific large low-velocity province (e.g., Russell et al.,
2001). The latter observations complicate efforts to account for
the D″ velocity discontinuities as thermal or chemical anomalies
associated with subducted oceanic slab (e.g., Grand, 2002), and
they have prompted notions either of a phase change or of chemical layering of D″.
The possibility of a phase change near the top of D″ has recently been boosted by the discovery of a postperovskite phase
transition in MgSiO3 perovskite (e.g., Murakami et al., 2004),
which appears to have potential to account for the D″ seismic
velocity discontinuities (Tsuchiya et al., 2004a). Theoretical calculations suggest a large positive Clapeyron slope for this phase
transition (Tsuchiya et al., 2004b), which would result in large
topography on the transition boundary and also in increased
potential for dynamic instability of the boundary layer in regions
where the postperovskite phase is present (Matyska and Yuen,
2004). There is a weak correlation between shallower depths of
PREM
SLHA
SLHE
SYL1
SGHP
SWDK
SYLO
SGLE
2500
2700
2900
7.0
7.2
7.4
S Velocity, km/s
Figure 6. A map (top) of regions with large-scale (>500 km) coherent
structures having S velocity increases of 2.5%–3% at depths 130–300 km
above the core-mantle boundary, and (bottom) representative S velocity models obtained from waveform modeling of data from localized
regions of the lowermost mantle. The S velocity models are 1D approximations of the local structure, and in most regions it appears that
there is substantial variability in the depth and strength of the velocity
increase on 100–300 km lateral scale lengths. Most regions with clear
S-wave velocity discontinuities have high S-wave velocities in D″ (cf.
Fig. 2) and underlie regions of extensive subduction of oceanic lithosphere in the past 200 million years, but the central Pacific is an important exception.
the D″ discontinuities and higher-velocity regions in large-scale
tomographic models. To the extent that the volumetric velocity
variations are an indicator of thermal variations, this is consis-
The deep mantle thermo-chemical boundary layer
Detected strong (dark) or weak (light) D" anisotropy
2300
Depth, km
tent with a large positive Clapeyron slope for a phase change
(e.g., Sidorin et al., 1999). However, the patchy occurrence of
D″ discontinuities favors the existence of the phase change primarily in relatively cool regions of high seismic velocity. Thus
one might not expect any phase change to exist or for it to be
deeper in the mantle in hot areas (large low-velocity provinces?),
which is at odds with the observations of discontinuities under
the central Pacific several hundred kilometers above the coremantle boundary (e.g., Russell et al., 2001). It is also curious that
the margins of large low-velocity provinces do not yet clearly
indicate any transition to regions with a positive velocity increase at the top of D″. What little evidence exists for structure
at the tops of large low-velocity provinces tends to favor small Vs
decreases at about the same depth as Vs increases in other regions
(e.g., Wen 2001; Ni and Helmberger, 2003). Some combination
of chemical heterogeneity and a phase change must be invoked
to account for the full suite of observations, and at present there
are substantial uncertainties in doing so (see Lay et al., 2004).
The other likely explanation for D″ discontinuities involves
chemical heterogeneities, either of primordial nature or possibly
replenishing accumulations of oceanic crustal components delaminated from slabs in the deep mantle. The occurrence of
strong D″ discontinuities beneath regions of subduction over the
past 100 million years and beneath regions of relatively high Vs
in the mid-mantle is often invoked in favor of an oceanic slab
connection. There are only a few regions where tomographic
images show high velocities extending continuously across the
mantle, with the region below the Gulf of Mexico perhaps the
best example (e.g., Grand, 2002), but raypath sampling in the
tomography studies is quite limited and may accentuate apparent continuity in finely parameterized tomographic models (see
Dziewonski, this volume), so connections between D″ structure
and shallow subduction remain tenuous. The correlation between
regions of historical subduction and mantle heterogeneity at
depths of 800–1000 km are relatively strong (e.g., Scrivner and
Anderson, 1992; Wen and Anderson, 1995, 1997), but this correlation weakens significantly with depth.
Because the interpretation of deep mantle seismic velocity
structures alone is ambiguous, additional observations are
needed. Seismologists have detected seismic anisotropy in many
regions of D″ (Fig. 7) and are seeking to establish a probe of the
flow and strain system in the boundary layer and its relationship
to flow in the mid-mantle (e.g., Lay et al., 1998b; Moore et al.,
2004). While progress is being made in mapping the existence
and properties of anisotropy in D″, the most general current observation is that regions of relatively high shear velocity, where
strong D″ discontinuities are observed, tend to have the most
pronounced anisotropy, and the orientation is close to being vertical transverse isotropy. This observation can be accounted for
by dynamic models in which low temperatures and high stresses
associated with mid-mantle downwellings drive underlying
regions of D″ into a dislocation deformation regime in which
lattice preferred orientation in (Mg,Fe)O can develop over large
scales (e.g., McNamara et al., 2002). It remains to be determined
201
PREM
Vsh
Vsv
PREM
Vsh
Vsv
2500
2700
2900
7.0 7.2 7.4
7.0 7.2 7.4
S velocity, km/s
S velocity, km/s
Figure 7. A map (top) showing locations where relatively strong (>1%)
vertical transverse isotropy has been detected in region D″ (dark shading) or where weak or nonexistent anisotropy has been observed (light
shading), and (bottom) examples of shear velocity models proposed for
various regions of D″ where the velocity structure has been found to be
anisotropic. The velocity models, which are compared to the PREM
reference structure, have velocities a few percent higher for horizontally polarized S-waves (SH) than for vertically polarized S-waves (SV),
which is consistent with vertical transverse isotropy. S-wave anisotropy
in D″ has been detected in regions with high S velocities (Fig. 2) and
strong S velocity discontinuities (Fig. 6) and in regions with low S velocities and weak or nonexistent S velocity discontinuities.
whether a postperovskite phase transition in the deep mantle plays
a significant role in the anisotropy in the boundary layer, but
there does appear to be some correspondence between the D″
velocity discontinuities and the onset of detectable anisotropy.
DISCUSSION AND CONCLUSIONS
Given the many degrees of freedom that exist for the properties and configuration of hypothesized chemical heterogeneities,
phase changes, partial melting, and thermal structure in D″, it is
possible to account for the diverse seismic observations with
202
T. Lay
various scenarios, but difficult to formulate definitive tests of
each hypothesis. Most discussions of the properties in the lowermost mantle invoke a hybrid scenario involving a partially
mixed boundary layer disrupted by downwelling slab material,
with a phase change and large piles of chemical heterogeneities
(Fig. 8). This scenario is motivated by and can be reconciled
with all of the seismic observations discussed in this paper (see
Lay and Garnero, 2004), and one can then superimpose thermal
plumes originating either within the thermal boundary layer at
the core-mantle boundary or above large low-velocity provinces,
resulting in the “vision” depicted in Figure 1. Direct demonstration of the existence of small-plume structures is at the frontier
of deep Earth seismic imaging efforts, and caution is warranted
with respect to all published interpretations of small-scale heterogeneities. This scenario requires some mechanism for reducing
the excess temperature of plumes rising directly from the core-
A
mantle boundary, and it suggests that chemical differences may
exist between plumes rising from large low-velocity provinces
and those rising from direct contact with the core (as may be
argued in the case of the Dupal anomaly, e.g., Wen et al., 2001).
An alternate scenario for the core-mantle boundary layer,
considered by Lay et al. (2004), involves reconciling the full
suite of D″ observations with a globally chemically stratified
lowermost mantle (Fig. 8B). This approach, motivated by the
first-order chemical stratification of the planet, suggests that
lateral variations of partial melting within the chemically distinct layer account for the diversity of seismic observations. This
concept would involve moderate thermal boundary layers at the
core-mantle boundary and at the top of the compositional layer.
Instabilities arising from the upper boundary layer could then
ascend to the surface as in Figure 1, presumably with all plumes
involving entrainment of similar compositional heterogeneities
Hybrid Slab/CBL/PC Model
1500
2891
B
Thermo-Chemical Boundary Layer Model
1500
MID-MANTLE FLOW REGIME
COLD
HIGH
HEAT FLUX
2891
COOLED
WARM
LOW
HEAT FLUX
HOT
HOT
VERY LOW
HEAT FLUX
HOTTEST
Figure 8. Scenarios for the thermo-chemical boundary layer in the D″ region based on seismic observations. (A) The hybrid thermo-chemical boundary layer scenario for the deep mantle, in which subducting slabs penetrate to the core-mantle boundary, providing thermal and chemical anomalies that
will eventually rise back up in the mantle while hot dense chemical anomalies of either primordial or
slab-related origin are swept into large piles under upwellings. Either a phase change or a radial gradient in structural fabric exists as well, giving rise to reflections from the top of D″. (B) The global
thermo-chemical boundary layer scenario, in which the lowermost mantle is a dense chemically distinct layer, possibly of primoridal nature, which remains unmixed but thermally coupled to overlying
flow. Variable heatflow out of the chemical layer occurs in response to the configuration of mantle
flow, leading to lateral variation in thermal structure across the boundary layer. Topography is induced
on the chemical layer by the mid-mantle flow as well. Either of these scenarios may be overlain by
thermal boundary layer instabilities yielding plumes as in Figure 1, but at present the direct observational basis for them is weak. Modified from Lay et al. (2004).
The deep mantle thermo-chemical boundary layer
from the D″ layer and having smaller excess temperatures than
would plumes arising directly from the core-mantle boundary.
However, such plumes are not a necessary consequence of this
scenario, and the majority of seismic observations can be accounted for without instabilities in the upper boundary layer.
Overall, the lowermost mantle appears to have substantial
chemical and thermal heterogeneity, and this complex thermochemical boundary layer may plausibly serve as a source of thermal instabilities that rise through the mantle. However, a sober
assessment of our current understanding and observations is that
there are many unknowns with regard to the nature of this deep
boundary layer, and it is quite plausible that there are, in fact, no
thermal plumes rising from the region. Thus much more work
will be required before we can attach much confidence to cartoon notions such as that of Figure 1.
ACKNOWLEDGMENTS
This paper was motivated in part by discussions at the 2004
CIDER (Cooperative Institute for Deep Earth Research) workshop, which was supported by the National Science Foundation
(NSF) under Grant PHY99-07949. This research was supported
by NSF grant EAR-0125595. Ed Garnero collaborated with the
author on several studies from which graphics and ideas have
been drawn. Don Anderson, Gillian Foulger, Bruce Julian, and
Lianxing Wen provided helpful comments. Contribution no. 479,
Center for the Study of Imaging and Dynamics of the Earth,
Institute of Geophysics and Planetary Physics, University of
California–Santa Cruz.
REFERENCES CITED
Akins, J.A., Luo, S.-N., Asimow, P.D., and Ahrens, T.J., 2004, Shock-induced
melting of MgSiO3 perovskite and implications for melts in Earth’s lowermost mantle: Geophysical Research Letters, v. 31, L14612, doi: 10.1029/
2004GL020237.
Albarede, F., and van der Hilst, R.D., 2002, Zoned mantle convection: Philosophical Transactions of the Royal Society of London (Ser. A), v. 360,
p. 2569–2592.
Albers, M., and Christensen, U.R., 1996, The excess temperature of plumes rising from the core-mantle boundary: Geophysical Research Letters, v. 23,
p. 3567–3570, doi: 10.1029/96GL03311.
Anderson, D.L., 1967, Phase changes in the upper mantle: Science, v. 157,
p. 1165–1173.
Anderson, D.L., 2002, The case for irreversible chemical stratification of the
mantle: International Geology Review, v. 44, p. 97–116.
Bijwaard, H., and Spakman, W., 1999, Tomographic evidence for a narrow
whole mantle plume below Iceland: Earth and Planetary Science Letters,
v. 166, p. 121–126, doi: 10.1016/S0012-821X(99)00004-7.
Boehler, R., 2000, High-pressure experiments and the phase diagram of lower
mantle and core materials: Reviews of Geophysics, v. 38, p. 221–245, doi:
10.1029/1998RG000053.
Boschi, L., and Dziewonski, A.M., 2000, Whole Earth tomography from delay
times of P, PcP, and PKP phases: Lateral heterogeneities in the outer core
or radial anisotropy in the mantle?: Journal of Geophysical Research,
v. 105, p. 13,675–13,696, doi: 10.1029/2000JB900059.
Buffett, B.A., 2002, Estimates of heat flow in the deep mantle based on the
power requirements for the geodynamo: Geophysical Research Letters,
v. 29, no. 12, doi: 10.1029/2001GL014649.
203
Castle, J.C., and van der Hilst, R.D., 2003, Searching for seismic scattering off
mantle interfaces between 800 and 2000 km depth: Journal of Geophysical
Research, v. 108, 2095, doi: 10.1029/2001JB000286.
Chopelas, A., 1996, Thermal expansivity of lower mantle phases MgO and
MgSiO3 perovskite at high pressure derived from vibrational spectroscopy:
Physics of the Earth and Planetary Interiors, v. 98, p. 3–15, doi: 10.1016/
S0031-9201(96)03175-5.
Christensen, U., 1984, Instability of a hot boundary layer and initiation of thermochemical plumes: Annales Geophysicae, v. 2, p. 311–319.
Courtillot, V., Davaille, A., Besse, J., and Stock, J., 2003, Three distinct types of
hotspots in the Earth’s mantle: Earth and Planetary Science Letters, v. 205,
p. 295–308, doi: 10.1016/S0012-821X(02)01048-8.
Davies, G.F., 1988, Ocean bathymetry and mantle convection, A: Large-scale
flow and hotspots: Journal of Geophysical Research, v. 93, p. 10,467–10,480.
DePaolo, D., and Wasserburg, G.J., 1976, Nd isotope variations and petrogenic
models: Geophysical Research Letters, v. 3, p. 249–252.
Dziewonski, A.M., and Anderson, D.L., 1981, Preliminary Reference Earth
Model: Physics of the Earth and Planetary Interiors, v. 25, p. 297–356, doi:
10.1016/0031-9201(81)90046-7.
Dziewonski, A.M., Foret, A.M., Su, W.-J., and Woodward, R.L., 1993, Seismic
tomography and geodynamics, in Aki, K., and Dmoska, R., eds., Relating
geophysical structures and processes: The Jeffereys volume: Washington,
D.C., American Geophysical Union Geophysical Monograph 76, p. 67–105.
Farnetani, C.G., 1997, Excess temperature of mantle plumes: The role of chemical stratification across D: Geophysical Research Letters, v. 24, p. 1583–
1586, doi: 10.1029/97GL01548.
Garnero, E.J., 2000, Lower mantle heterogeneity: Annual Reviews of Earth and
Planetary Science, v. 28, p. 509–537, doi: 10.1146/annurev.earth.28.1.509.
Garnero, E.J., Revenaugh, J.S., Williams, Q., Lay, T., and Kellogg, L.H., 1998,
Ultralow velocity zone at the core-mantle boundary, in Gurnis, M., et al.,
eds., The core-mantle boundary region: Washington, D.C., American Geophysical Union, p. 319–334.
Goes, S., Spakman, W., and Bijwaard, H., 1999, A lower mantle source for
central European volcanism: Science, v. 286, p. 1928–1931, doi: 10.1126/
science.286.5446.1928.
Glatzmaier, G.A., Coe, R.S., Hongre, L., and Roberts, P.H., 1999, The role of
the Earth’s mantle in controlling the frequency of geomagnetic reversals:
Nature, v. 401, p. 885–890, doi: 10.1038/44776.
Grand, S.P., 2002, Mantle shear-wave tomography and the fate of subducted
slabs: Philosophical Transactions of the Royal Society of London (Ser. A),
v. 360, p. 2475–2491.
Hart, S.R., Hauri, E.H., Oschmann, L.A., and Whitehead, J.A., 1992, Mantle
plumes and entrainment: Isotopic evidence: Science, v. 256, p. 517–520.
Helmberger, D.V., Wen, L., and Ding, X., 1998, Seismic evidence that the source
of the Iceland hotspot lies at the core-mantle boundary: Nature, v. 396,
p. 251–255, doi: 10.1038/24357.
Hofmann, A.W., 1997, Mantle geochemistry: The message from oceanic volcanism: Nature, v. 385, p. 219–229, doi: 10.1038/385219a0.
Hofmann, A.W., and White, W.M., 1982, Mantle plumes from ancient oceanic
crust: Earth and Planetary Science Letters, v. 57, p. 421–436, doi: 10.1016/
0012-821X(82)90161-3.
Ishii, M., and Tromp, J., 1999, Normal-mode and free-air gravity constraints on
lateral variations in velocity and density: Science, v. 285, p. 1231–1236,
doi: 10.1126/science.285.5431.1231.
Ishii, M., and Tromp, J., 2004, Constraining large-scale mantle heterogeneity
using mantle and inner-core sensitive normal modes: Physics of the Earth
and Planetary Interiors, v. 146, p. 113–124, doi: 10.1016/j.pepi.2003.06.012.
Jellinek, A.M., and Manga, M., 2002, The influence of a chemical boundary
layer on the fixity, spacing and lifetime of mantle plumes: Nature, v. 418,
p. 760–763, doi: 10.1038/nature00979.
Kellogg, L.H., Hager, B.H., and van der Hilst, R.D., 1999, Compositional stratification in the deep mantle: Science, v. 283, p. 1881–1884, doi: 10.1126/
science.283.5409.1881.
Lay, T., and Garnero, E.J., 2004, Core-mantle boundary structures and processes,
in Hawkesworth, C., and Sparks, S., eds., State of the planet, Washington,
204
T. Lay
D.C., American Geophysical Union, Geophysical Monograph 150,
p. 25–41.
Lay, T., Garnero, E.J., and Williams, Q., 2004, Partial melting in a thermochemical boundary layer at the base of the mantle: Physics of the Earth and
Planetary Interiors, v. 146, p. 441–467, doi: 10.1016/j.pepi.2004.04.004.
Lay, T., Williams, Q., and Garnero, E.J., 1998a, The core-mantle boundary layer
and deep Earth dynamics: Nature, v. 392, p. 461–468, doi: 10.1038/33083.
Lay, T., Williams, Q., Garnero, E.J., Kellogg, L., and Wysession, M.E., 1998b,
Seismic wave anisotropy in the D″ region and its implications, in Gurnis,
M., et al., eds., The core-mantle boundary region: Washington, D.C., American Geophysical Union, p. 229–318.
Masters, G., Laske, G., Bolton, H., and Dziewonski, A.M., 2000, The relative
behavior of shear velocity, bulk sound speed, and compressional velocity
in the mantle: Implications for chemical and thermal structure in Earth’s
deep interior, in Karato, S., et al., eds., Mineral physics and tomography
from the atomic to the global scale: Washington, D.C., American Geophysical Union, p. 63–87.
Matyska, C., and Yuen, D.A., 2004, The importance of radiative heat transfer
for superplumes with a deep mantle phase transition: Earth and Planetary
Science Letters, v. 125. p. 255–266.
McNamara, A.K., and Zhong, S., 2004, Thermochemical structures within a
spherical mantle: Superplumes or piles?: Journal of Geophysical Research,
v. 109, B07402, doi: 10.1029/2003JB002847.
McNamara, A.K., van Keken, P.E., and Karato, S.-I., 2002, Development of
anisotropic structure in the Earth’s lower mantle by solid-state convection:
Nature, v. 416, p. 310–314, doi: 10.1038/416310a.
Mégnin, C., and Romanowicz, B., 2000, The three-dimensional shear velocity
structure of the mantle from the inversion of body, surface, and highermode waveforms: Geophysical Journal International, v. 143, p. 709–728,
doi: 10.1046/j.1365-246X.2000.00298.x.
Montelli, R., Nolet, G., Dahlen, F.A., Master, G., Engdahl, E.R., and Hung, S.H.,
2004, Finite-frequency tomography reveals a variety of plumes in the mantle: Science, v. 303, p. 338–343, doi: 10.1126/science.1092485.
Moore, M.M., Garnero, E.J., Lay, T., and Williams, Q., 2004, Shear wave splitting and waveform complexity for lowermost mantle structures with lowvelocity lamellae and transverse isotropy: Journal of Geophysical Research,
v. 109, p. B02319, doi: 10.1029/2003JB002546.
Morgan, W.J., 1971, Convective plumes in the lower mantle: Nature, v. 230,
p. 42–43.
Morgan, W.J., 1972, Deep mantle convection plumes and plate motions:
Bulletin of the American Association of Petroleum Geologists, v. 56,
p. 203–213.
Murakami, M., Hirose, K., Kawamura, K., Sata, N., and Ohishi, Y., 2004, Postperovskite phase transition in MgSiO3: Science, v. 304, p. 855–858, doi:
10.1126/science.1095932.
Nataf, H.C., 2000, Seismic imaging of mantle plumes: Annual Reviews of
Earth and Planetary Science, v. 28, p. 391–417, doi: 10.1146/annurev.
earth.28.1.391.
Ni, S., and Helmberger, D.V., 2003, Ridge-like lower mantle structure beneath
South Africa: Journal of Geophysical Research, v. 108, B2, 2094, doi:
10.1029/2001JB001545.
Ni, S., Tan, E., Gurnis, M., and Helmberger, D., 2002, Sharp sides to the African
superplume: Science, v. 296, p. 1850–1852, doi: 10.1126/science.1070698.
Olson, P., Schubert, G., and Anderson, C., 1987, Plume formation in the
D″-layer and the roughness of the core-mantle boundary: Nature, v. 327,
p. 409–413, doi: 10.1038/327409a0.
Romanowicz, B., and Gung, Y.C., 2002, Superplumes from the core-mantle
boundary to the lithosphere: Implications for heat flux: Science, v. 296,
p. 513–516, doi: 10.1126/science.1069404.
Russell, S.A., Reasoner, C., Lay, T., and Revenaugh, J., 2001, Coexisting shearand compressional-wave seismic velocity discontinuities beneath the
central Pacific: Geophysical Research Letters, v. 28, p. 2281–2284, doi:
10.1029/2000GL012553.
Saltzer, R.L., van der Hilst, R.D., and Kárason, H., 2001, Comparing P and S
wave heterogeneity in the mantle: Geophysical Research Letters, v. 28,
p. 1335–1338, doi: 10.1029/2000GL012339.
Scrivner, C., and Anderson, D.L., 1992, The effect of post Pangea subduction
on global mantle tomography and convection: Geophysical Research Letters, v. 19, p. 1053–1056.
Sidorin, I., Gurnis, M., and Helmberger, D.V., 1999, Dynamics of a phase
change at the base of the mantle consistent with seismological observations: Journal of Geophysical Research, v. 104, p. 15,005–15,023, doi:
10.1029/1999JB900065.
Sleep, N.H., 1988, Gradual entrainment of a chemical layer at the base of the
mantle by overlying convection: Geophysical Journal, v. 95, p. 437–447.
Sleep, N.H., 1990, Hotspots and mantle plumes: Some phenomenology: Journal of Geophysical Research, v. 95, p. 6715–6736.
Stacey, F.D., and Loper, D.E., 1983, The thermal boundary-layer interpretation
of D″ and its role as a plume source: Physics of the Earth and Planetary
Interiors, v. 33, p. 45–55, doi: 10.1016/0031-9201(83)90006-7.
Tackley, P.J., 1998, Three-dimensional simulations of mantle convection with a
thermo-chemical basal boundary layer: D″, in Gurnis, M., et al., eds., The
core-mantle boundary region: Washington, D.C., American Geophysical
Union, p. 231–253.
Tackley, P., 2000, Mantle convection and plate tectonics: Toward an integrated
physical and chemical theory: Science, v. 288, p. 2002–2007, doi: 10.1126/
science.288.5473.2002.
Thorne, M.S., and Garnero, E.J., 2004, Inferences on ultralow-velocity zone
structure from a global analysis of SPdKS waves: Journal of Geophysical
Research, v. 109, B08301, doi: 10.1029/2004JB003010.
Thorne, M.S., Garnero, E.J., and Grand, S.P., 2004, Geographic correlation between hot spots and deep mantle lateral shear-wave velocity gradients:
Physics of the Earth and Planetary Interiors, v. 146, p. 47–63, doi: 10.1016/
j.pepi.2003.09.026.
Trampert, J., Vacher, P., and Vlaar, N., 2001, Sensitivities of seismic velocities
to temperature, pressure and composition in the lower mantle: Physics of
the Earth and Planetary Interiors, v. 124, p. 255–267, doi: 10.1016/S00319201(01)00201-1.
Tsuchiya, T., Tsuchiya, J., Umemoto, K., and Wentzcovitch, R.M., 2004a,
Elasticity of post-perovskite MgSiO3: Geophysical Research Letters, v. 31,
p. L14603, doi: 10.1029/2004GL020278.
Tsuchiya, T., Tsuchiya, J., Umemoto, K., and Wentcovitch, R.M., 2004b, Phase
transition in MgSiO3 perovskite in the Earth’s lower mantle: Earth and
Planetary Science Letters, v. 224, p. 241–248.
Wen, L., 2001, Seismic evidence for a rapidly-varying compositional anomaly at the base of the Earth’s mantle beneath Indian ocean: Earth and
Planetary Science Letters, v. 194, p. 83–95, doi: 10.1016/S0012-821X
(01)00550-7.
Wen, L., and Anderson, D.L., 1995, The fate of slabs inferred from seismic
tomography and 130 million years of subduction: Earth and Planetary
Sciences Letters, v. 133, p. 185–198, doi: 10.1016/0012-821X(95)00064-J.
Wen, L., and Anderson, D.L., 1997, Slabs, hotspots, cratons and mantle convection revealed from residual seismic tomography in the upper mantle:
Physics of the Earth and Planetary Interiors, v. 99, p. 131–143, doi:
10.1016/S0031-9201(96)03162-7.
Wen, L., and Helmberger, D.V., 1998, Ultra-low velocity zones near the coremantle boundary from broadband PKP precursors: Science, v. 279, p. 1701–
1703, doi: 10.1126/science.279.5357.1701.
Wen, L., Silver, P., James, D., and Kuehnel, R., 2001, Seismic evidence for a
thermo-chemical boundary at the base of the Earth’s mantle: Earth and
Planetary Science Letters, v. 189, p. 141–153, doi: 10.1016/S0012-821X
(01)00365-X.
Williams, Q., 1998, The temperature contrast across D″, in Gurnis, M., et al.,
eds., The core-mantle boundary region: Washington, D.C., American Geophysical Union, p. 73–81.
Williams, Q., and Garnero, E.J., 1996, Seismic evidence for partial melt at the
base of the Earth’s mantle: Science, v. 273, p. 1528–1530.
Williams, Q., Revenaugh, J., and Garnero, E.J., 1998, A correlation between
The deep mantle thermo-chemical boundary layer
ultra-low basal velocities in the mantle and hot spots: Science, v. 281,
p. 546–549, doi: 10.1126/science.281.5376.546.
Wysession, M., Lay, T., Revenaugh, J., Williams, Q., Garnero, E.J., Jeanloz, R.,
and Kellogg, L., 1998, The D″ discontinuity and its implications, in Gurnis, M., et al., eds., The core-mantle boundary region: Washington, D.C.,
American Geophysical Union, p. 273–298.
Wysession, M.E., Langenhorst, A., Fouch, M.J., Fischer, K.M., Al-Eqabi, G.I.,
Shore, P.J., and Clarke, T.J., 1999, Lateral variations in compressional/
shear velocities at the base of the mantle: Science, v. 284, p. 120–125, doi:
10.1126/science.284.5411.120.
Zhao, D., 2001, Seismic structure and origin of hotspots and mantle plumes:
205
Earth and Planetary Science Letters, v. 192, p. 251–265, doi: 10.1016/
S0012-821X(01)00465-4.
Zhao, D., 2004, Global tomographic images of mantle plumes and subducting
slabs: Insight into deep Earth dynamics: Physics of the Earth and Planetary
Interiors, v. 146, p. 3–34, doi: 10.1016/j.pepi.2003.07.032.
Zhong, S., and Hager, B.H., 2003, Entrainment of a dense layer by thermal
plumes: Geophysical Journal International, v. 154, p. 666–676, doi: 10.1046/
j.1365-246X.2003.01988.x.
MANUSCRIPT ACCEPTED BY THE SOCIETY JANUARY 3, 2005
Printed in the USA
Page 206 is blank