Gubbins D., Herrero-Bervera E. Encyclopedia of Geomagnetism and Paleomagnetism

Подождите немного. Документ загружается.

as a thermal boundary layer

Originally the D

layer was identified as a distinct region immediately

above the Core-mantle boundary (CMB) by its seismic properties (see

, seismic properties), and these remain our principal source of infor-

mation on the structure of this region of Earth’s interior (Loper and

Lay, 1995; see Earth structure, major divisions). Its character as a ther-

mal boundary layer follows from the requirement that the core be cool-

ing at a sufficient rate (see Core-mantle boundary, heat flow across)

so that the outer core is in a state of convective motions sufficiently

vigorous to drive the geodynamo. In the bulk of the mantle, the

upward transfer of heat associated with core cooling is accomplished

by convection: upward movement of hot material and downward

movement of cold. Since vertical motions are inhibited near the base of

the mantle by the strong density contrast between the mantle and the core,

vertical transport of heat from the core must be accomplished primarily by

thermal conduction in the lowermost part of the mantle. Large temperature

gradients are required to conduct the requisite heat (see Mantle, thermal

conductivity); this is the essence of a thermal boundary layer.

The average temperature contrast across the D

layer can be estim-

ated by two complementary methods. The first involves extrapolation

of the temperatures across the mantle and the core. These extrapola-

tions start from known fixed points and assume that adiabatic tempera-

ture gradients exist in the bulk of the mantle and the core. The fixed

point in the mantle is the phase transition at 410 km depth, while that

in the core is the inner core boundary where the outer core material is

at its freezing point. The greatest uncertainty in this calculation is the

amount that the freezing temperature is depressed due to impurities

in the outer core. Estimates of the temperature contrast obtained from

extrapolation range from 200 to 1600



The second method for estimating the temperature contrast across

the D

layer is to use a crude approximation of the equation of heat

conduction:

q ¼ kAðDTÞ=ðDzÞ

where q is the total heat transferred across the CMB, k is the thermal

conductivity, A is the area of the CMB, DT is the temperature contrast

across the D

layer, and Dz is its thickness. Knowing the other

factors, this may be solved for DT. The most uncertain factor in this

formula is Dz.

The heat that is transferred from the core has two effects on the

material comprising the D

layer: it makes the material less dense,

leading to convective motions which transfer the heat upward through

the mantle, and it reduces the viscosity. The thermal and dynamical

structures of the D

layer are governed by the equations of conser-

vation of mass, momentum, and energy, with the latter two being

strongly coupled through the dependence of viscosity on temperature.

A simple and plausible pattern of convective motion consists of a very

slow descent of cold mantle material toward the CMB nearly every-

where and horizontal motion very close to the CMB toward a small

number of discrete sites, from which warm mobile material rises into

the mantle, forming thermal plumes (Stacey and Loper, 1983). Thermal

diffusive thickening of the D

layer is balanced by the downward

motion, leading to a layer of (nearly) steady thickness. Due to the strong

variation of viscosity with temperature, the D

layer has a double struc-

ture. Most of the temperature contrast occurs across the thicker portion,

in which motions are predominantly downward. Lateral motion is con-

fined to a thinner layer close to the CMB, which is hottest and hence

has lowest viscosity.

Lateral (geographic) variations in the temperature of the descending

material causes lateral variations in the thermal structure of the D

layer and consequently lateral variations in the rate of heat transfer

from core to mantle (see Inhomogeneous boundary conditions and

the dynamo). By means of these variations the mantle provides some

control on the structure of convection in the core, and of the Earth’s

magnetic field. In particular, variations in the thermal structure of the

layer on the timescale typical of mantle convection (10

years)

are the most plausible cause of the observed changes in the frequency

of reversals on that timescale.

David Loper

Bibliography

Loper, D.E., and Lay, T., 1995. The core-mantle boundary region.

Journal of Geophysical Research, 100: 6397–6420.

Stacey, F.D., and Loper, D.E., 1983. The thermal-boundary-layer inter-

pretation of D

and its role as a plume source. Physics of the Earth

and Planetary Interiors, 33:45–55.

Cross-references

Core-Mantle Boundary, Heat Flow Across

, Composition

, Seismic Properties

Earth Structure, Major Divisions

Inhomogeneous Boundary Conditions and the Dynamo

Mantle, Thermal Conductivity

, ANISOTROPY

Dynamic processes within the Earth can align minerals and inclusions.

The resulting rock fabric leads to seismic anisotropy, which means

that seismic velocity at a position varies as a function of the direction

of wave propagation. In contrast, isotropy refers to the case where

velocity is not directionally dependent. For reference, seismic hetero-

geneity (or inhomogeneity) refers to variation in velocity with posi-

tion. Conventional seismic imaging techniques (e.g., travel-time

tomography) invert data for the isotropic velocity heterogeneity of

the mantle, and as such offer insights into thermal and chemical struc-

tures. There are though a growing number of methods for estimating

seismic anisotropy. As anisotropy results from deformation processes,

such analyses offer added insight into the dynamical nature of the

Earth. Although we are still in the early stages of mapping anisotropy,

it appears that it is most pronounced in the boundary layers of

the Earth, regions where deformation is greatest. One such boundary

layer is the D

region (see D

as a boundary layer; Core-mantle

boundary).

The lowermost mantle boundary layer, referred to as the D

region

or layer, marks the boundary region between the solid silicate (rocky)

mantle and the liquid iron core. This region, which extends to a few

100 km above the Core-mantle boundary (CMB), is the site of the

most dramatic variations in velocity structure in the lower mantle

(see D

, seismic properties). Proposed explanations for the anomalous

properties of D

include the effects of a phase change in the region,

iron infiltration from the core, primordial material from early mantle

differentiation, and subducted material pooling at the CMB. Numerous

arguments suggest that mantle plumes may originate in D

. The ther-

mochemical nature of this region will influence thermal, gravitational,

topographic, and electromagnetic core-mantle coupling, which in turn

influence the dynamo, changes in the length of the Earth’s day, and

inner core development (see Core-mantle coupling; Core-mantle

boundary, heat flow across). Hence, studies of D

anisotropy offer

insights into the region’s influence on large-scale Earth processes.

There are three primary difficulties in studying D

anisotropy. The

first concerns the complexity of seismic wave propagation in anisotro-

pic media. The second involves removing the competing effects of

anisotropy from other regions of the mantle. The third lies in the inter-

pretation of anisotropy mechanisms.

146 D

, ANISOTROPY

Wave propagation in anisotropic media

P- and S-wavefronts emanating from a point source in an isotropic

homogeneous medium are spherical. The ray direction (direction of

energy transport) is normal to the wavefront, the particle motion of

the P-wave is also normal to the wavefront, and the particle motion

of the S-wave is tangential to the wavefront and normal to the ray

(transverse). The elasticity of an isotropic medium is described by

two independent parameters (e.g., the Lamé parameters).

In contrast the elasticity of an anisotropic medium is parameter-

ized by 21 independent elastic parameters. Three independent waves

propagate and in a homogeneous anisotropic medium; their wavefronts

are not spherical. The ray direction is in general not perpendicular to

the wavefront and the particle motion of the P-wave and two S-waves

are no longer normal and tangential to the wavefront, respectively. The

most unambiguous indicator of anisotropy is the observation of two

independent and orthogonally polarized shear waves. As a shear wave

in an isotropic medium impinges on an anisotropic region it will split

into two shear waves. The two shear waves will continue to propagate

separately upon leaving the anisotropic region, retaining a “memory”

of the anisotropy. Such splitting is described by the time separation

between the two shear waves, dt, and the polarization of the fast shear

wave, f (Silver, 1996; Savage, 1999). Such analysis has been used

extensively to map anisotropy in the uppermost mantle. Should these

two waves encounter another anisotropic medium, they will again

split, spawning four independent shear waves. Consequently wave

propagation in heterogeneous anisotropic media becomes complicated

and it can be difficult to isolate regions of anisotropy.

The number of independent elastic parameters required to describe

wave propagation in anisotropic media is further reduced by the sym-

metry of the medium. For example, a single crystal of olivine is orthor-

hombic in symmetry and its elasticity is described by nine elastic

parameters. A commonly assumed style of anisotropy is transverse iso-

tropy, which is a form of hexagonal symmetry and is described by five

elastic parameters. Such a medium with a vertical axis of symmetry is

commonly referred to as having vertical transverse isotropy (VTI)

symmetry or radial anisotropy. A characteristic of wave propagation

in such media is that the fast and slow shear waves are polarized in

either the horizontal direction (SH) or the vertical plane (SV). Mechan-

isms for such anisotropy include the periodic horizontal layering of

contrasting materials or the alignment of crystals in a horizontal glide

plane with no fixed slip direction.

Evidence for anisotropy in D

A range of indicators suggest that the lowermost mantle is anisotropic.

The decay of phases diffracted along the core-mantle boundary show

dramatic regional variations and Doornbos et al. (1986) first sought

to explain this in terms of VTI anisotropy in the D

region (see

reviews in Lay et al., 1998 and Kendall and Silver, 1998).

anisotropy has been more clearly documented through obser-

vations of splitting in shear wave phases that travel hundreds of

kilometers through this region (S, ScS and Sdiff; Figure D1 ) (see Seis-

mic phases). The magnitude of the splitting, dt, varies considerably

(0 to >7 s) with both region and path length through D

(Kendall

and Silver, 1998; Lay et al. 1998). Although this is the site of the lar-

gest shear wave splitting magnitudes in the Earth, the splitting is

accrued over long path lengths and estimates of anisotropy in the

region are 0–3% ((v

fast

v

slow

)/v

ave

, where v refers to shear wave velo-

city). To date, analysis has primarily concentrated on measuring the

separation between shear wave arrivals on the transverse and radial

components of a three-component seismometer. In other words, the

polarization of the fast shear wave, f, is either in the vertical plane

or the horizontal direction (SV vs. SH). Such analysis implicitly

assumes VTI anisotropy as more general forms of anisotropy would

yield fast and slow shear wave energy on both the radial and transverse

components. Recently, Panning and Romanowicz (2004) have inverted

shear wave data to produce a global map of D

VTI anisotropy

(Figure D2/Plate 11c). Future directions involve developing techniques

to invert more general forms of anisotropy (i.e., nonVTI).

A difficultly in measuring D

anisotropy is ruling out the effects of

anisotropy in shallower parts of the mantle. For example, estimating

splitting in SKS phases is a now standard technique for studying upper

mantle anisotropy (see Silver, 1996; Savage, 1999). Any interpretation

of D

anisotropy must be correct for near-source and near-receiver

anisotropy (see Kendall and Silver, 1998). Analysis of SKS splitting at

permanent stations provides a correction for near-receiver anisotropy,

although care must be taken as this may be directionally dependent.

Using very deep earthquakes can mitigate near-source anisotropy, as

the transition zone and middle mantle are thought to be largely isotro-

pic (Mainprice et al., 2000). Alternatively, relative changes in shear

wave splitting in phases that turn above and below D

can be used

to isolate the anisotropy to the lowermost mantle.

The style of anisotropy in high velocity and hence probably colder

regions of D

is quite consistent. Here shear wave energy on the

transverse component leads that on the radial component. Regional

studies of D

beneath the Caribbean, Siberia, Alaska, Australia and

the Indian Ocean suggest a roughly VTI anisotropy. Recently, detailed

analysis has shown that the anisotropy beneath the Caribbean is not

purely VTI: the orientation of the symmetry axis varies up to 20



from

the vertical (Garnero et al., 2004). In warmer and more ambient

regions the style of D

anisotropy appears to be more complex and

the magnitude more variable. Specifically, the form of anisotropy

Figure D1 Seismic phases used to analyze D

anisotropy. UMA refers to upper mantle anisotropy. The SdS phase reflects off

a discontinuity that lies at the top of the D

region. At distances beyond 100



the phase S and ScS merge and diffract along the

CMB (S

diff

, ANISOTROPY 147

beneath the Pacific appears to vary in style and magnitude over rela-

tively short distances. Furthermore, D

anisotropy is not ubiquitous;

regions beneath parts of the Pacific and Atlantic Ocean appear to be

isotropic.

Cumulatively, these results suggest that anisotropy correlates with

regional variations in other properties of D

(see D

, seismic proper-

ties). Although we are still in the early stages of mapping out the mag-

nitude and orientation of such anisotropy, we can loosely group the

observations into categories based on regional characteristics. One is

associated with regions where slabs are predicted to subduct into the

lower mantle. Such regions are characterised by high D

shear veloci-

ties and a more VTI style of anisotropy. Another category involves

sites of inferred mantle upwelling, for example, beneath Hawaii. Such

regions are characterised by lower than average shear velocities and a

more complex style of anisotropy. Finally, initial results suggest that

more ambient regions of D

appear to be isotropic.

Causes of D

anisotropy

A range of mechanisms can cause anisotropy, including the preferred

alignment of crystals, grains, or inclusions, and the fine-scale layering

of contrasting materials. As the mantle convects, rock deformation will

be accommodated by mechanisms such as dislocation, diffusion, grain

boundary migration, and dynamic recrystallization (Karato and Wu,

1993). Dislocation glide and creep mechanisms are effective in aligning

crystals and producing a lattice preferred orientation (LPO). In order to

evaluate such effects in the lower mantle we need to know the minerals

present, their single-crystal elastic properties, and how they deform. We

are still a long way from knowing all of this information, but recent

advancements in laboratory experiments and theoretical calculations

are offering new insights. Candidate minerals for LPO anisotropy

include perovskite, magnesiowustite, columbite (Stixrude, 1998), and

the recently discovered post-perovskite phase (Murakami et al., 2004)

(see D

, composition). In high-velocity regions, post-perovskite and

magnesiowustite are likely candidates for D

anisotropy and perovskite

is a likely candidate for D

anisotropy in low-velocity regions.

Alternatively, the anisotropy may be associated with well-ordered

subseismic scale heterogeneity. Oriented inclusions or layering can

effectively generate anisotropy if there is sufficient contrast in material

properties. For example, aligned melt-filled tabular inclusions generate

anisotropy very effectively (Kendall and Silver, 1998).

It seems that the physical process responsible for the anisotropy

beneath paleoslab regions is different from that beneath regions of man-

tle upwelling. The infiltration of core material into the D

region could

lead to inclusions of low-velocity material. Should these inclusions be

aligned, they may provide an explanation for the anisotropy beneath

the central Pacific. This is a region for which there is evidence for an

ultralow velocity zone at the base of the mantle and it has been argued

that these may be due to the presence of partial melt (see ULVZ, ultralow

velocity zone). Alternatively, the preferred orientation of perovskite

could explain the anisotropy. With either mechanism the magnitude

and orientation of the anisotropy must vary over short length scales.

There are a number of arguments that the anisotropy in high-velocity

regions is associated with the accumulation of paleoslab material in

(Kendall and Silver, 1998). Regions of anisotropy beneath the

Caribbean, Alaska, northern Asia, Australia, and the Indian Ocean cor-

relate with predicted CMB locations of the slabs. The high isotropic

shear velocities of this region can be explained by the retained slab ther-

mal anomaly and the high strains associated with the slab material col-

liding with the rigid CMB can align constituent crystals or inclusions.

In summary, there are clear regional variations in the style of

anisotropy that are very likely associated with different physical

processes. In a sense, these variations are analogous to those observed

between continental and oceanic regions in the upper mantle boundary

layer. As our knowledge of D

mineral physics grows and new seismic

arrays provide improved coverage, studies of D

anisotropy will offer

new insights into this important but somewhat enigmatic region.

Michael Kendall

Figure D2/Plate 11c Map of seismic S-wave anisotropy (x ¼ V

)inD

based on a global waveform tomography inversion

for a 3D model of radial anisotropy (VTI) (from Panning and Romanowicz, 2004). Also shown are areas of detailed regional analysis

of D

anisotropy: white indicates regions wherein D

horizontally polarized S-waves (SH) are faster than a vertically polarized S-waves

(SV); red shows regions where D

appears to be isotropic; green shows a broad region of D

beneath the Pacific where the style of

anisotropy appears to be complex and quite variable.

148 D

, ANISOTROPY

Bibliography

Doornbos, D.J., Spiliopoulos, S., and Stacey, F.D., 1986. Seismologi-

cal properties of D

and the structure of a thermal boundary layer.

Physics of the Earth and Planetary Interiors, 57: 225–239.

Garnero, E.J., Maupin, V., and Fouch, M.J., 2004. Variable azimuthal

anisotropy in the Earth’s lowermost mantle. Science, 306(5694):

259–261.

Karato, S. and Wu. P., 1993. Rheology of the upper mantle: A synth-

esis. Science, 260: 771–778.

Kendall J-M. and Silver, P.G., 1998. Investigating causes of D

aniso-

tropy. In Gurnis, M., Wysession, M.E., Knittle, E., and Buffett, B.

A. (eds.), The Core-Mantle Boundary Region. Geodynamic Series,

28, Washington, D.C.: American Geophysical Union, pp. 97–118.

Lay, T., Williams, Q., Garnero, E.J., Kellogg, L., and Wysession, M.E.,

1998. Seismic wave anisotropy in the D

region and its implications.

In Gurnis, M., Wysession, M.E., Knittle, E., and Buffett, B.A. (eds.),

The Core-Mantle Boundary Region. Geodynamic Series, 28,

Washington, D.C.: American Geophysical Union, pp. 299–318.

Mainprice, D., Barruol, G., and Ben Ismail, W., 2000. The seismic ani-

sotropy of the Earth’s mantle: From single crystal to polycrystal.

In Karato, S., Forte, A.M., Liebermann, R.C., Masters, T.G., and

Stixrude, L. (eds.), Earth’s Deep Interior: Mineral Physics and Tomo-

graphy from the Atomic to the Global Scale. Geophysical Monograph,

117, Washington, D.C.: American Geophysical Union, pp. 237–264.

Murakami, M., Hirose, K., Kawamura, K., Sata, N., and Ohishi, Y.,

2004. Post-perovskite phase transition in MgSiO

. Science, 304:

855–858.

Panning, M., and Romanowicz, B., 2004. Inferences of flow at the

base of the Earth’s mantle based on seismic anisotropy. Science,

303: 351–353.

Savage, M., 1999. Seismic anisotropy and mantle deformation: What

have we learned from shear-wave splitting? Reviews in Geophy-

sics, 37:65–106.

Silver, P.G., 1996. Seismic anisotropy beneath continents: Probing the

depths of geology. Annual Review Earth Space Sciences, 24:

385–432.

Sitxrude, L., 1998. Elastic constants and anisotropy of MgSiO

perovskite, periclase, and SiO

at high pressures. In Gurnis, M.,

Wysession, M.E., Knittle, E., and Buffett, B.A., (eds.), The Core-

Mantle Boundary Region. Geodynamic Series, 28, Washington

D.C.: American Geophysical Union, pp. 83–96.

Cross-references

Core-Mantle Boundary

Core-Mantle Boundary, Heat Flow Across

Core-Mantle Coupling, Electromagnetic

Core-Mantle Coupling, Thermal

Core-Mantle Coupling, Topographic

as a Boundary Layer

, Composition

, Seismic Properties

Seismic Phases

ULVZ, Ultralow Velocity Zone

, COMPOSITION

Both the mantle and the core of the Earth are fundamentally affected

by the properties of the material at the base of the mantle. This region

controls the heat flow between the core and the mantle; it may be the

source region for mantle plumes; and it is the lowermost window

through which the magnetic field of the Earth must propagate to reach

the surface. That the base of the mantle differs seismically from the

overlying mantle material was recognized nearly a century ago by

Gutenberg (1913). The D

zone was originally defined as a distinct

zone by Bullen (1949) as the layer with anomalously low increases

in seismic velocity with depth in the lowermost several hundred of

kilometers of the mantle. Since the importance of mantle convection

was recognized, the chemical composition of this lowermost 300 km

of Earth’s 2900 km-thick mantle has widely been viewed as approxi-

mately equivalent to that of the overlying lower mantle, with its seismic

properties perturbed by a large increase in temperature distributed over

the lowermost few hundred kilometers of the mantle. The superadia-

batic gradient associated with the basal boundary layer of the Earth’s

convecting mantle likely involves a gradual temperature increase of

1000 K to perhaps as much as 2000 K (if the layer is internally stra-

tified) across the D

zone. Such shifts in temperature can produce

decreased gradients of seismic velocities across this zone, a result that

generally agrees with laterally averaged seismic models of Earth’s

mantle, such as the Preliminary Reference Earth Model (PREM:

Dziewonski and Anderson, 1981).

However, improvements in seismic observations have driven a reas-

sessment of the possible roles of chemical differentiation and chemical

heterogeneity within D

. The specific seismic observations include:

a sporadic seismic discontinuity involving an increase of 1–3%

in shear and perhaps in compressional velocity at heights usually

200–300 km above the core-mantle boundary (CMB); a laterally het-

erogeneous zone 5–40 km above the CMB with decreases in compres-

sional and shear velocity of the order of 10% and 30%, respectively

(the ultralow velocity zone, or ULVZ); two large-scale slow structures

beneath Africa and the western Pacific spanning lateral dimensions of

2000 km and extending to heights of 1000 km above the CMB,

with shear decrements of 3% or larger; and pervasive and laterally

varying anisotropy within the D

layer (see Gurnis et al., 1998; Lay

et al., 1998; and D

, seismic properties, and D

, anisotropy). The che-

mical characteristics of D

have thus emerged as a multidisciplinary

problem, incorporating the anomalous seismic characteristics of this

layer, the somewhat ill-constrained properties of Earth materials at

the extreme conditions of the CMB (pressure of 135 GPa and tempera-

ture of 3500–5000 K), and the effects of changes in chemistry on the

electromagnetic properties of the deep mantle, and thus on the geo-

magnetic field.

The seismic observations of structural complexity within D

have

driven a broad range of proposals of possible chemical changes that

might be present within this region. These possibilities include: descent

of melts from the overlying mantle, enriching this zone in incompatible

and volatile elements; interaction of material with the underlying core,

producing iron enrichment, and contributions from subducted slab

material. Additionally, a different chemistry of this region might have

been created early in the history of the planet, through early differentia-

tion processes or perhaps even during planetary accretion. All such pro-

posed changes in chemistry must not dramatically alter the seismic

velocity or density in this region from that within the lowermost

mantle: seismic travel-time and normal mode constraints show that

the bulk of D

can only differ in velocity and density by a few percent

from that within the deep mantle (see Earth structure, major divisions).

This constraint is, however, relaxed in the lowermost 5–40 km of the

mantle, where dramatic changes of seismic wave velocities have been

observed.

From a geomagnetic perspective, the importance of the chemistry of

the lowermost mantle lies primarily in its effect on the propagation

of Earth’s magnetic field through the basal layer of the mantle. If the

electrical conductivity of the lowermost mantle is markedly elevated

from that of the overlying material, through enrichment in iron, the

presence of partial melt, or both, then temporal variations of the mag-

netic field could be attenuated within such a layer (see Mantle, electri-

cal conductivity, mineralogy). Moreover, any electromagnetic coupling

between the core and mantle would generate a torque on the overlying

mantle: such a torque has been associated with shifts in the nutation of

the Earth and changes in the length of day (e.g., Buffett, 1992). If high

conductivity zones were heterogeneously distributed near the base of

, COMPOSITION 149

Earth’s mantle, then the mantle could exercise a structural control on

changes in virtual geomagnetic pole positions during reversals of the

polarity of Earth’s magnetic field, potentially forcing the reversing

pole to follow geographically preferred paths (Aurnou et al., 1996).

Core-mantle chemical interactions

The most straightforward way to determine the composition of D

would be to analyze samples derived from this region. However, the

ability to geochemically sample D

directly via surface volcanism is

controversial. For such sampling to occur, an upwelling would have

to transit nearly 3000 km of Earth’s mantle and retain a chemical sig-

nature that could be unambiguously associated with its basal origin. It

is possible that a portion of Earth’s hot spots may originate from D

and as such might be expected to carry a geochemical signature of this

region or perhaps even of the underlying core. Some evidence exists

that some hot spots and flood basalts carry anomalously high

187

Os/

183

Os ratios (e.g., Walker et al., 1995). The radiogenic parent

187

Os is

187

Re: Re is a strongly siderophilic (iron-loving) element

and is thus anticipated to be strongly enriched in Earth’s core.

Re enrichment is similarly observed in iron meteorites. In contrast,

Os is generally lithophilic, partitioning into silicates. Thus, elevated

187

Os/

183

Os ratios are associated with the sampling of Re-rich regions.

The most parsimonious explanation for such ratios is that some hot

spots and flood basalts may record a geochemical signature of interac-

tion with Earth’s core. The amount of chemical interaction necessary

to generate the observed ratios is not large, but it nevertheless may

provide prima facie evidence for chemical interaction between Earth’s

core and mantle.

How might such core-mantle chemical interactions occur? A direct

means for producing such interactions is through chemical reactions

of the silicate of the lowermost mantle with the liquid iron-rich alloy of

the outer core. Such reactions, which involve oxidation of core-derived

iron and reduction of mantle silicon, can be described using chemical

formulae such as

ðMg

1x

ÞSiO

ðmantle materialÞþ3½ð1  xÞsFe

¼ xMgSiO

þ sSiO

þ½ð1  xÞsFeSi

þ½3ð1  xÞ2sFeO:

Here, s is the amount of silica produced in the reaction and (1x)is

the amount of iron in the reacting mantle material. Such reactions have

been both experimentally verified to occur and are thermodynami-

cally driven to the right by a negative volume change of the reaction

(e.g., Knittle and Jeanloz, 1991). The intriguing aspect of these reac-

tions is that the products involve both iron-free silicates, which are

both likely to be of low density and electrical conductivity, and rela-

tively high density, high electrical conductivity iron alloys. Accord-

ingly, core-mantle reactions can lead to a suite of reaction products

with highly heterogeneous physical properties. The key questions

associated with such core-mantle reactions are mechanical in nature:

over what length-scale such reactions occur at the CMB, how effi-

ciently reaction products are entrained within flow of the overlying

mantle or underlying core (exposing fresh mantle material to reaction),

and whether the low-density products are able to separate from the

high-density products. In short, whether the mantle develops a thin

reacted rind at its base, or whether reacted zones can be dynamically

entrained into the overlying mantle, producing an enrichment of iron

alloys and replenishment of the reacting zone in the lowermost mantle

is ill-constrained. The possibility that topographic effects might be

important in enhancing such chemical reactions has been suggested

(e.g., Jeanloz and Williams, 1998): any topography on the CMB

involves either silicate material intruding into iron or vice versa (see

Core-mantle boundary topography). In such cases, the gravitational

effects that drive the separation of the core and the mantle material

are locally relaxed, with core fluid prospectively being able to infiltrate

the mantle material without traveling against gravity.

Subduction and D

The role of slab (and particularly oceanic crust) enrichment in D

has

frequently been invoked as a possible cause of geochemical heteroge-

neity in this region. If ancient oceanic crust is sequestered in D

, then

this zone would be enriched in Ca, Al, and Si relative to the overlying

mantle. Moreover, volatile elements such as hydrogen and carbon that

were retained within the slab during its transit through the mantle

might also be more abundant in this zone. A compelling case for the

existence of such slab-related enrichment would require several sepa-

rate lines of evidence: first, seismic evidence that slabs penetrate to

CMB depths; second, a sufficiently elevated density of subducted

material that it would stably reside for long times within D

; and third,

a possible mechanism by which oceanic crust can efficiently separate

from its underlying depleted mantle. Each of these lines of evidence

is not fully developed: seismic tomography shows that slab-related

seismic anomalies thicken substantially as they enter into the lower

mantle, but these anomalies do not appear to be easily tracked into

the lowermost 500–1000 km of the mantle. As the full slab is com-

prised of oceanic crust overlying depleted mantle material, the chemis-

try of the slab as a whole lies close to that of the bulk mantle.

Therefore, a means by which oceanic crust can delaminate from its

depleted complement and accumulate within D

is required if the

dregs of ancient crustal subduction are a major chemical component

of the lowermost mantle. Accordingly, while slabs may generate geo-

chemically distinct regions within D

, the available evidence neither

requires nor precludes the presence of ancient slabs within D

Partial mel ting

Partial melting and associated melt segregation are well known means

of geochemical differentiation in the shallow regions of the planet.

Thus, the likely existence of partial melt in the lowermost regions

of Earth’s mantle may imply that enrichment in incompatible

elements—those elements that partition strongly into silicate liquids

relative to solids—occurs at these depths. The idea that partial melt

exists in the lowermost mantle is driven by the observation that decre-

ments of 10% in compressional wave velocity and 30% in shear

velocity exist in heterogeneously distributed zones in the lowermost

5–40 km of the Earth’s mantle—the ultralow velocity zones (ULVZ,

q.v.). Such decrements likely require 6–30% partial melt (with the

amount depending on how the melt is distributed) entrained in parts

of the lowermost mantle (Williams and Garnero, 1996). The stable

retention of melt at the base of Earth’s mantle (rather than it buoyantly

rising to the top of Earth’s mantle) is probably produced by two

tandem, complementary effects. First, the dramatic increase in tem-

perature within D

relative to the overlying mantle (see D

as a

boundary layer) may produce sufficiently high temperatures that the

base of the mantle lies above its solidus—the temperature at which

melting initiates in multicomponent systems. Second, melts at pres-

sures corresponding to those at the base of the mantle may be denser

than their coexisting solids, due to both structural effects associated

with highly densified amorphous structures and the preferential parti-

tioning of iron into silicate liquids. Indeed, the available constraints

on the ULVZ indicate that this region has an increase in density rela-

tive to the overlying mantle material of about 10% (Rost et al.,

2005): a result fully consistent with a marked enrichment in iron, with

the additional iron most likely being oxidized. Thus, melts likely sta-

bly stratify near the base of the mantle, and any partial melt that occurs

in the deep lower mantle may descend into D

, enriching this zone in

both volatile elements (H and C), and likely incompatible major ele-

ments such as Fe, Ca, and Al. Additionally, radiogenic heat-producing

elements such as K and U might be enriched within the CMB region.

150 D

, COMPOSITION

However, the melting behavior of the mantle material at the extreme

pressures and temperatures of the CMB is not yet well-constrained,

so the geochemical consequences of partial melting and melt segrega-

tion are largely inferred based on experiments at substantially less

extreme conditions. The key point about the geochemical effects of

partial melting in D

is that they are not exclusive of either core-mantle

reactions or the presence of slab-derived components within D

. Core-

mantle reactions may lower the melting point of mantle material and

partial melt could facilitate the transfer of reacted material away from

the CMB. Similarly, if the melting temperature of ancient oceanic crust

(or, indeed, any portion of the slab) is lower than that of normal mantle

material, then slab-derived melts could descend into D

. As such,

enrichment of D

in slab-related geochemical components could be

driven by partial melting processes. Lastly, the importance of partial

melting in D

is likely to have been far greater earlier in Earth history.

If secular cooling of Earth’s mantle has occurred over the age of the

Earth, then partial melt is anticipated to have been more abundant in

the deep mantle early in Earth history, and today’sD

could (at least par-

tially) represent the solidified residue of partial melt.

Primordial chemical layering

A rather different genesis for a chemically distinct D

was proposed by

Ruff and Anderson (1980). They proposed that the composition of the

lowermost mantle might differ markedly from that of the overlying

material due to effects associated with Earth’s accretion. Such primor-

dial layering might have arisen through early accretion of material

enriched in Ca, Al, and Ti. These elements are abundant in the initial

condensates of the protoplanetary nebula, and the idea was simply that

these elements might have heterogeneously accreted to the center of

the protoEarth during the earliest stages of planetary accretion. Subse-

quently, they might have been displaced to the region surrounding the

core due to the gravitational instability of subsequently accreted dense,

iron-rich material. The dilemma is that this layer must not only persist

through impacts of planetary-sized objects on the early Earth, but also

endure without mixing throughout Earth history: both the lack of a

clearly global discontinuity at the top of D

and the generally similar

seismic properties of the bulk of D

to the overlying mantle make such

broadscale, profoundly chemically distinct, and primordial layering

unlikely. However, if primordial layering does persist within D

,it

may be less likely to be associated with accretionary effects, and more

likely to be produced by differentiation processes within a primordial

magma ocean. Such a magma ocean is widely postulated to have existed

early in Earth history due to accretional heating associated with giant

impacts on the early Earth. In this sense, “primordial” chemical differen-

tiation of D

could operationally differ little from the effects of partial

melting of Earth’s deep mantle over the course of Earth history.

However, such early chemical differentiation may also be affected

by the presence within the deep mantle of a recently discovered

post-perovskite phase [Murakami et al., 2004]. This ultrahigh pressure

phase of (Mg,Fe)(Si,Al)O

may be the dominant mineral within D

and its pressure of onset appears to correspond with the depth of the

seismic discontinuity often viewed as the top of D

(see D

, seismic

properties). Little is known about the chemical behavior or affinities

of this CaIrO

-structured phase, beyond the likelihood that the transi-

tion to this phase has a positive Clapeyron slope. However, depending

on the degree to which D

has equilibrated with the overlying mantle

material, D

could be enriched in elements that are more compatible

within the post-perovskite structure than within the silicate perovskites

of Earth’s lower mantle. Speculatively, if iron preferentially partitions

into this phase, then a net iron enrichment of D

could arise through

progressive equilibration between D

and the overlying mantle. Such

equilibration could be generated through selective transport of material

across the phase boundary, with progressive fractionation of compatible

(and perhaps heavy) elements into the lowermost mantle.

Quentin Williams

Bibliography

Aurnou, J.M., Buttles, J.L., Neumann, G.A., and Olson, P.L., 1996.

Electromagnetic core-mantle coupling and paleomagnetic reversal

paths. Geophysical Research Letters , 23: 2705–2708.

Buffett, B.A., 1992. Constraints on magnetic energy and mantle con-

ductivity from the forced nutations of the Earth. Journal of Geo-

physical Research, 97: 19581–19597.

Bullen, K.E., 1949. Compressibility-pressure hypothesis and the

Earth’s interior. Monthly Notices of the Royal Astronomical

Society, Advances in Geophysics - Supplement, 5: 355–368.

Dziewonski, A.M., and Anderson, D.L., 1981. Preliminary reference

Earth model. Physics of the Earth and Planetary Interiors, 25:

297–356.

Gurnis, M., Wysession, M.E., Knittle, E., and Buffett, B.A., (eds.)

1998. The core-mantle boundary region, Washington, D.C.: Amer-

ican Geophysical Union.

Gutenberg, B., 1913. Uber die Konstitution des Erdinnern, erschlos-

sen aus Erdbebenbeobachtungen. Physikalische Zeitschrift, 14:

1217–1218.

Jeanloz, R., and Williams, Q., 1998. The core-mantle boundary region.

Reviews in Mineralogy, 37: 241–259.

Knittle, E., and Jeanloz, R., 1991. Earth’s core-mantle boundary:

Results of experiments at high pressures and temperatures. Science,

251: 1438–1443.

Lay, T., Williams, Q., and Garnero, E.J., 1998. The core-mantle

boundary layer and deep Earth dynamics. Nature, 392: 461–468.

Murakami, M., Hirose, K., Kawamura, K., Sato, N., and Ohishi, Y.,

2004. Post-perovskite phase transition in MgSiO

. Science, 304:

855–858.

Rost, S., Garnero, E.J., Williams, Q., and Manga, M., 2005. Seismic

constraints on a possible plume root at the core-mantle boundary.

Nature, 435: 666–669.

Ruff, L., and Anderson, D.L., 1980. Core formation, evolution and

convection: A geophysical model. Physics of the Earth and Plane-

tary Interiors, 21: 181–201.

Walker, R.J., Morgan, J.W., and Horan, M.F., 1995. Osmium-187

enrichment in some plumes—Evidence for core-mantle interaction.

Science, 269: 819–822.

Williams, Q., and Garnero, E.J., 1996. Seismic evidence for partial

melt at the base Earth’s mantle. Science, 273: 1528–1530.

Cross-references

Core-Mantle Boundary Topography

as a Boundary Layer

, Anisotropy

, Seismic Properties

Earth Structure, Major Divisions

Mantle, Electrical Conductivity, Mineralogy

, SEISMIC PROPERTIES

Across most of Earth’s lower mantle, P- and S-wave seismic velocities

increase with depth in a fashion consistent with adiabatic self-

compression of uniform composition material, and there is relatively

little lateral variation in seismological structure (see Earth structure,

major divisions). However, the lowermost few hundred kilometers of

the mantle, called the D

region, has been recognized since the

1940s as having anomalous seismic properties, with reduced seismic

velocity gradients with depth and strong lateral variations in structure.

This prompted designation of D

as a distinct layer in the early label-

ing of Earth’s shells. Seismological characterization of inhomogeneity

of the D

region has advanced dramatically in recent decades, and the

unusual seismic properties of D

are now commonly interpreted as

, SEISMIC PROPERTIES 151

manifestations of complex thermal and chemical boundary layers at

the base of the mantle (see D

as a boundary layer). This region is

of importance to the Mantle and the core dynamical systems.

At a depth of 2891 km below the surface, the core-mantle boundary

(CMB) separates the two principal chemical layers of the Earth, the

molten metallic-alloy outer core and the highly viscous silicate- and

oxide-mineral mantle (see Core-mantle boundary). The contrast in

density between the mantle rock and the core alloy exceeds the density

change from air to rock at Earth’s surface, and the contrast in viscosity

is comparable to that between rock and ocean water. The superadia-

batic temperature contrast from the core into the mantle is estimated

to be as high as 2000



C, possibly exceeding that across the litho-

spheric thermal boundary layer, and there is heat flow of 2–10 TW

across the boundary (44 TW flows through the Earth’s surface) (see

Core-mantle boundary, heat flow across). Over time, dense materials

in the mantle are likely to have concentrated in D

, while light dross,

expelled from the core may have accumulated at the CMB (see D

composition). The D

boundary layer is relatively accessible to seismic

imaging, and its seismic properties are discussed here (also see D

anisotropy; D

as a boundary layer; D

, composition; ULVZ, ultralow

velocity zone). Garnero (2000), Lay et al. (1998a), and many papers in

Gurnis et al. (1998) provide reviews of ongoing research topics and

conceptual models for processes occurring in D

One-dimensional seismic velocity models for D

One-dimensional global seismic velocity models, such as the Prelimin-

ary Reference Earth Model (PREM) of Dziewonski and Anderson

(1981), usually have reduced velocity gradients in the deepest 150 km

of the mantle, reflecting the global departure of D

velocity structure

from that of the overlying lower mantle. For several decades, efforts

have been made to interpret the low (or even negative) velocity gradients

in such average models of D

structure as the result of a superadiabatic

temperature increase across a thermal boundary layer; large temperature

increases above the CMB can be reconciled with the one-dimensional

models (see D

as a boundary layer; Core-mantle boundary, heat

flow across). However, studies over the past half century have estab-

lished that there is a very substantial heterogeneity in D

on a wide vari-

ety of scale lengths; thus, there is no meaningful “average” structure for

the region useful for a robust interpretation of the boundary layer.

It appears more useful to discuss the seismic properties of D

by empha-

sizing the lateral variations in structure, as is the case for the surface

boundary layer.

Large-scale seismic velocity patterns

Seismic tomography has been applied to develop three-dimensional

seismic velocity models for the entire mantle for over 15 years. This

method extracts three-dimensional P-wave and S-wave velocity fluc-

tuations relative to one-dimensional reference models by inversion of

massive data sets with crossing raypaths for which seismic wave arrival

times are measured, exploiting the localized sensitivity of seismic waves

to velocity structure around the path on which they propagate through

the Earth. These three-dimensional velocity inversions indicate that het-

erogeneity in the mid-lower mantle is relatively weak, but that in the

lowermost 300–500 km of the mantle heterogeneity increases with

depth, becoming stronger in D

than anywhere else in the Earth except

for the uppermost mantle where major thermal, chemical, and partially

molten heterogeneities are known to exist. The most surprising aspect

of the D

seismic velocity heterogeneity is that it is dominated by

large-scale patterns, with large continuous volumes of high- or low-

velocity material having scale lengths of hundreds to thousands of kilo-

meters. This predominance of large-scale variations is observed in both

S-wave and P-wave velocities, as indicated by the representative models

for lateral variations in D

structure shown in Figure D3a/Plate 11a.

S-wave velocity variations (dVs) relative to the global average

velocity in D

(Figure D3a/Plate 11a.) are positive (relatively high

velocity) beneath the circum-Pacific, with relatively low-velocity

regions found under the central Pacific and south Atlantic/Africa.

High-velocity regions are likely to be at lower temperature than low-

velocity regions, although chemical differences may contribute to the

variations. Recent tomographic models have 4% S velocity varia-

tions in D

, two or three times stronger than those found in the mid-

lower mantle, and comparable to the 5–8% variations at depths near

150 km in the upper mantle. While mid-mantle variations tend to have

more spatially concentrated heterogeneities, there is a significant

degree of radial continuity of velocity anomaly patterns throughout

the lower mantle, possibly linking subduction zones at the surface to

high velocity regions in D

and hot spot distributions at the surface

to low velocity regions in D

P-wave velocity variations, dVp, in current tomographic models

(Figure D3b/Plate 11b) involve 1.0–1.5% fluctuations, with low-

velocity areas beneath the southern Pacific and south Atlantic/Africa

and high velocities under eastern Eurasia. There is not a strong increase

in P velocity heterogeneity in D

relative to that in the overlying mantle

as there is for S-wave velocity, which can be attributed to the expected

greater sensitivity of S velocity to strong thermal variations in a hot

thermal boundary layer.

As apparent in Figure D3/Plate 11a,b, there is a general spatial

correlation between S- and P-wave velocity patterns, as expected for

temperature-induced variations, but in some regions, such as beneath

the northern Pacific, this correlation breaks down. In other regions,

such as beneath the central Pacific, the S-wave variations are much

stronger than the P-wave variations even though both have the same

sign. The decorrelation of dVp and dVs and the variation in their

relative strengths provide strong evidence that thermal variations alone

cannot explain the large-scale patterns of seismic velocity heterogene-

ity, so there is likely to be chemical heterogeneity present in D

Figure D3/Plate 11a,b Maps of S velocity (a) and P velocity

(b) variations relative to the global average values in D

at a depth

near 200 km above the CMB in representative global mantle

seismic tomography models. S velocity variations are from Me

gnin

and Romanowicz (2000) and P velocity variations are from 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 2–3 times stronger than P velocity variations.

Large-scale patterns dominate, suggesting the presence of coherent

thermal and chemical structures.

152 D

, SEISMIC PROPERTIES

(Masters et al., 2000). In fact, if bulk sound velocity variations, dVb, are

estimated from P and S arrival times, they are found to have anticorrela-

tion with dVs on very large scale, which requires bulk modulus and

shear modulus to have opposite sign perturbations. It is important to

recognize that the patterns in Figure D3/Plate 11a,b are relative to the

global average velocities at D

depths, and that these averages do not

necessarily define “ normal” mantle structure. This results in uncertainty

in interpretations of the velocity fluctuations (for example, high-velocity

regions may be anomalously low temperature, perhaps due to cool

down-wellings that have disrupted an overall hot boundary layer, or

low-velocity regions may be anomalously hot regions, perhaps partially

melted, within a chemically distinct, high-velocity layer). Resolution of

finer scale seismic properties in D

is pursued in order to overcome this

baseline uncertainty of seismic tomography imaging.

Seismi c velocity discont inuities in D

The presence of several percent lateral variations in S and P velocities

concentrated within the lowermost mantle suggests the possibility of

rapid velocity increases and/or decreases at the top of the D

region. Var-

ious seismic observations suggest that rapid increases in velocity with

depth are indeed present at depths from 130–300þ km above the CMB

in many regions. Rapid shear velocity increases of 1.5–3% are found over

intermediate-scale (500–1000 km) regions, with velocity structures and

spatial sampling as indicated in Figure D4.While thePREMmodel

shows a smooth velocity increase with depth in the lower mantle, with

being suggested only by a drop in rate of velocity increase with depth

150 km above the CMB, other shear velocity models obtained for loca-

lized regions have an abrupt increase in velocity near the top of the D

region. The velocity discontinuities are required to account for observed

S-wave reflections that precede the reflections from the CMB. Similar

models are obtained for localized regions for P-waves, usually with a

smaller velocity increase of 0.5–1%. The increase in velocity may be

distributed over a few tens of kilometers or it may be very sharp, and in

some regions it can vary in depth by as much as 100 km over lateral scales

of just 200 km. Wysesssion et al. ( 19 98 ) r ev ie w m an y o bs er va ti on s a n d

models for this D

seismic velocity discontinuity. While localized one-

dimensional models are undoubtedly oversimplified, they do strongly

suggest that some regions of D

appear to effectively be layered, with

t he r ma ll y a nd c o mp os it io na ll y d is ti nc t m at er ia l.

Comparison of Figure D3a/Plate 11a and Figure D4 suggests that

the S velocity discontinuity is commonly observed in areas of high

S-wave velocity in D

. An abrupt increase in velocity with depth is

not expected for a thermal boundary layer structure, so most interpre-

tations of this correlation invoke the notion of localized ponding of

cool subducting lithospheric slabs that have sunk to the lowermost

mantle, retaining enough thermal and chemical anomalies to account

for the high seismic velocity and strong radial gradients. Alternatively,

chemically distinct high-velocity material may be present in these

regions, perhaps involving ancient accumulated material concentrated

during core formation or segregated oceanic crustal materials that have

accumulated over Earth history. The observation of a weak and vari-

able S velocity increase beneath the central Pacific, a relatively low-

velocity region far from any historical subduction zone, complicates

any attempt to interpret the velocity discontinuity as the result of

recent slab thermal anomalies.

A possible explanation for the D

seismic discontinuities has been

provided by the discovery in 2004 of a phase change that occurs in

magnesium silicate perovskite, MgSiO

, the dominant mineral struc-

ture in the lower mantle (Murakami et al., 2004) . At pressure and tem-

perature conditions close to those at the top of D

, a slightly denser

post-perovskite mineral is expected to form, with a relatively strong

increase in shear modulus and a small or even negative change in bulk

modulus. This is expected to occur preferentially in lower temperature

regions of D

, which may account for the intermittence of the D

reflections. The effects of chemistry on this phase transition and its

implications for boundary layer dynamics are active areas of research.

Lo w-veloc ity provinc es in D

The central portions of the largest regions with very low shear velocity

seen in Figure D3a/Plate 11a appear to have very strong decreases in

velocity from above and laterally. For example, the low-velocity region

beneath the southern Atlantic and Africa has been modeled as having

an abrupt 1 to 3% decrease 250–300 km above the CMB, with average

velocities in D

that are 3–5% lower than for PREM (Figure D5). It

appears that these are distinctive low-velocity provinces, comparable in

thickness and lateral extent to the high-velocity regions, and having

abrupt lateral boundaries (e.g., Wen et al., 2001). The sub-African pro-

vince appears to thicken to 800 km or so beneath the continent, with a thin

Figure D4 Representative S velocity models obtained from

waveform modeling of data from localized regions of the

lowermost mantle (top), and a map of regions with large-scale

(>500 km) coherent structures having S velocity increases of

2.5–3%, 130–300 km above the CMB (bottom). The S velocity

models are one-dimensional 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

(Figure D3/Plate 11a,b) and underlie regions of extensive

subduction of oceanic lithosphere in the past 200 Ma, but the

central Pacific is an important exception.

, SEISMIC PROPERTIES 153

region of ultralow velocity right above the CMB (see below). Lateral gra-

dients on the margins of the low-velocity province are very abrupt, on the

scale of tens of kilometers, both in D

and in the upward extension into

the lower mantle.

The large low-velocity provinces are commonly assumed to be very

hot regions, and have been called Superplume provinces, under the

assumption that they are sources of upwelling flow rising from a hot

thermal boundary layer at the CMB. Indeed, they do underlie large

regions of low velocity in the lower mantle as well as concentrations

of hot spot volcanoes at the surface, some having distinctive chemical

anomalies. However, the strong lateral gradients and apparent endur-

ance of these structures suggest that there is distinctive chemical com-

position associated with the low-velocity provinces, and the extent to

which thermal and chemical buoyancy trade-off is not yet established.

P velocity anomalies of the low S velocity provinces are much smaller,

suggesting the possibility of partial melting of these regions, which

raises the possibility that these are simply the hottest (and partially

molten) regions within a global chemically distinct layer. The strong

temperature increase expected across the D

region may approach

or exceed the eutectic solidus of lower mantle mineral assem-

blages, allowing partial melting to take place in the boundary layer

(Lay et al., 1998a).

Ultralow velocity zones in D

The most dramatic seismic velocity reductions that have been detected

in D

are associated with very thin layers or lumps of material just

above the CMB that are called ultralow velocity zones (ULVZs) (see

ULVZ, ultralow velocity zone). Figure D6 shows P and S velocity

models characteristic of a region with a ULVZ (the central Pacific).

This region, which is lower velocity than PREM over a 200–300 km

thick D

region, has a thin layer 10 to 40 km thick and laterally quite

extensive (500–1000 km across), with P velocity reductions of 4 to

10% and S velocity reductions of 12 to 30%. In some regions, such

as under Iceland, the ULVZ appears to be in a localized mound a

few hundred kilometers across, but with comparably strong velocity

reductions to the more extensive areas of ULVZ under the Pacific

(Garnero et al., 1998). Comparing Figure D6 and Figure D3a/Plate

11a indicates that ULVZs are commonly found within the margins of

large-scale low-velocity regions, including at least portions of the

major low S velocity provinces. Existence of a melt component in

the ULVZs is the likely explanation for their strong velocity effects.

The melt may involve either core-material that has infiltrated into the

mantle, or partial melting of mantle material. In either case ULVZs

are likely to involve thermochemical anomalies, probably with iron

enrichment that causes the melt component to be dense, concentrated

near the CMB (see: D

, composition; Core-Mantle boundary; ULVZ,

ultralow velocity zone).

Figure D6 Seismic velocity models for PREM and the central

Pacific region, where an ULVZ is detected right above the CMB

(top). The ULVZ structures in various regions have P velocity

drops of 4 to 10% and S velocity drops of 12 to 30%. The map

(bottom) shows regions where ULVZ features have either been

observed (light gray) or are not observed (dark gray). In general,

regions with ULVZ in D

tend to be found in regions of low S

velocities (Figure D3a/Plate 11a).

Figure D5 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/southern Indian Ocean (bottom).

A cartoon sketch of the latter feature is shown (top), with a

200–300 km thick layer in D

having an abrupt decrease in

velocity of 1 to 3% overlying a layer with average velocity

decreases of 3 to 5%. Under Africa this low velocity body appears

to extend upward to as much as 800 km above the CMB, and it

may have an ultralow 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 well imaged,

but appears comparable in size and strength of velocity reductions.

154 D

, SEISMIC PROPERTIES

Seismi c velocity anisotrop y in D

As is the case for upper mantle structure, the velocity structure in D

anisotropic, with seismic velocities being dependent on the direction of

propagation and the polarization of ground shaking (see D

, aniso-

tropy ). This results in shear wave splitting, where an S-wave will sepa-

rate into two components with orthogonal polarization, one traveling

slightly faster than the other. By measuring the polarizations and

travel-time difference between the fast and slow S-waves, it is possible

to determine the anisotropic characteristics of the medium. The results

of such analyses of S-wave splitting for phases traversing D

are sum-

marized in Figure D7. In most cases, the data indicate S velocity mod-

els in which horizontally polarized (SH) vibrations travel with 1 to 3%

higher velocities than vertically polarized (SV) vibrations for raypaths

grazing horizontally through the D

region. This behavior is consistent

with vertical transverse isotropy (VTI), which can result from hexa-

gonally symmetric minerals with vertically oriented symmetry axes

or from stacks of thin horizontal layers with periodic velocity fluctua-

tions (Lay et al., 1998b). The regions with the best documented cases

for strong VTI in D

tend to have higher than average S velocities

(Figure D7 and Figure D3a/Plate 11a). There are observations favor-

ing slightly tilted (nonvertical) transverse isotropy, which results in

weak coupling of the SH and SV signals (the fast wave is close to

the SH polarization), as well as limited regions where SV signals are

found to propagate faster than SH signals (such as in the low S velo-

city region of the central Pacific). Obtaining mineralogical and dyna-

mical explanations for anisotropy in the lowermost mantle are areas

of active research (see: D

, anisotropy). There is great interest in

improving the characterization of D

anisotropy because it has the

potential to reveal actual deformation processes occurring in the

boundary layer.

Discussion and conclusions

The diverse seismic properties of D

appear to reflect the presence of

thermal and chemical boundary layers at the base of the mantle. There

is a predominance of large-scale structures and the strength of hetero-

geneities at intermediate and large-scales appears to be significantly

greater than in the overlying lower mantle. Small-scale (<10 km) het-

erogeneities also appear to be present, but are not clearly more pro-

nounced than in rest of the lower mantle. Interpretation of the seismic

properties of D

involves many uncertainties, and a wide spectrum of

scenarios and processes have been advanced to account for the observa-

tions. Thermal variations in D

play an important role in determining the

heat flux boundary condition on the core dynamic regime, along with

the possible role of D

as a source of upwelling thermal plumes, so per-

haps the greatest current challenge is to identify the relative contribu-

tions to the seismic properties of thermal versus chemical heterogeneity.

Thorne Lay

Bibliography

Boschi, L., and Dziewonski, A.M., 2000. Whole Earth tomography

from delay times of P, PcP, and PKP phases: Lateral heterogene-

ities in the outer core or radial anisotropy in the mantle? Journal

of Geophysical Research [solid Earth], 105: 13675–13696.

Dziewonski, A.M., and Anderson, D.L., 1981. Preliminary reference

earth model. Physics of the Earth and Planetary Interiors, 25:297–356.

Garnero, E.J., 2000. Lower mantle heterogeneity. Annual Reviews of

the Earth and Planetary Sciences, 28: 509–537.

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., Wysession, M.E., Knittle, E., and Buffett, B.A.

(eds.), The Core-Mantle Boundary Region. Washington, D.C.:

American Geophysical Union, pp. 319–334.

Gurnis, M., Wysession, M.E., Knittle, E., and Buffett, B.A. (eds.),

1998. The Core-Mantle Boundary Region. Washington, D.C.:

American Geophysical Union. 334 pp.

Lay, T., Williams, Q., and Garnero, E.J., 1998a. The core-mantle

boundary layer and deep Earth dynamics. Nature, 392: 461–468.

Lay, T., Williams, Q., Garnero, E.J., Kellogg, L., and Wysession, M.E.,

1998b. Seismic wave anisotropy in the D

region and its implica-

tions. In Gurnis, M., Wysession, M.E., Knittle, E., and Buffett, B.A.

(eds.), The Core-Mantle Boundary Region. Washington, D.C.: Amer-

ican Geophysical Union, pp. 299–318.

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 Karato, S., Forte, A.M., Liebermann,

R.C., Masters, G., and Stixrude, L., (eds.), Earth's Deep Interior:

Mineral Physics and Tomography from the Atomic to the Global

Scale, Washington, D.C.: American Geophysical Union, pp. 63–87.

Mégnin, C., and Romanowicz, B., 2000. The three-dimensional shear

velocity structure of the mantle from the inversion of body, surface,

and higher-mode waveforms, Geophysical Journal International,

143: 709–728.

Murakami, M., Hirose, K., Kawamura, K., Sata, N., and Ohishi, Y.,

2004. Post-perovskite phase transition in MgSiO

, Science, 304:

855–858.

Figure D7 Examples of shear velocity models (top) proposed for

various regions of D

where the velocity structure has been found

to be anisotropic, and a map (bottom) showing locations where

relatively strong (>1%) vertical transverse isotropy has been

detected in D

(dark shading) or where weak or nonexistent

anisotropy has been observed (light shading). The velocity

models, which are compared to the PREM reference structure,

have a few percent higher velocities for horizontally polarized

S-waves (SH) than for vertically polarized S-waves (SV), which is

consistent with vertical transverse isotropy. S-wave anisotropy in

has been detected in regions with high S velocities (Figure

D3a/Plate 11a) and strong D

S velocity discontinuities

(Figure D4), and in regions with low S velocities and weak, or

nonexistent S velocity discontinuities.

, SEISMIC PROPERTIES 155