Page 54 - Global Tectonics

THE INTERIOR OF THE EARTH 41

the central Andes and Kamchatka InSAR measurements using glacial rebound is the role of transient creep,
have been used to evaluate volcanic hazards and the where the strain rate varies with time under constant
movement of magma in volcanic arcs (Pritchard & stress. Because the total strains associated with
−3
Simons, 2004). In southeast Iran, InSAR data have been rebound are quite small (≤10 ) compared to the large
used to determine the deformation field and source strains associated with mantle convection, transient

parameters of a magnitude M w = 6.5 earthquake that creep may be important during post-glacial isostatic
affected the city of Bam in 2003 (Wang et al., 2004). The rebound (Ranalli, 2001).
combined use of GPS and InSAR data have revealed the In contrast to the upper mantle, much of the
vertical displacements associated with a part of the San lower mantle is seismically isotropic, suggesting that
Andreas Fault system near San Francisco (Fig. 8.7b). diffusion creep is the dominant mechanism associated

with mantle flow at great depths (Karato et al., 1995).
Unlike dislocation creep, diffusion creep (and also
2.10.6 Deformation superplastic creep) result in an isotropic crystal struc-
ture in lower mantle minerals, such as perovskite and
in the mantle magnesiowüstite. Large uncertainties about lower
mantle rheology exist because lower mantle materials
Measurements of seismic anisotropy (Section 2.1.8) and are difficult to reproduce in the laboratory. Neverthe-

the results of mineral physics experiments have been less, advances in high-pressure experimentation have

used to infer creep mechanisms and flow patterns in the allowed investigators to measure some of the physi-
mantle (Karato, 1998; Park & Levin, 2002; Bystricky, cal properties of lower mantle minerals. Some
2003). The deformation of mantle minerals, including measurements suggest that lower mantle rheology
olivine, by dislocation creep results in either a preferred strongly depends on the occurrence and geometry of
orientation of crystal lattices or a preferred orientation minor, very weak phases, such as magnesium oxide
of mineral shapes. This alignment affects how fast (Yamazaki & Karato, 2001). Murakami et al. (2004)
seismic waves propagate in different directions. Mea- demonstrated that at pressure and temperature con-
surements of this directionality and other properties ditions corresponding to those near the core–mantle
potentially allow investigators to image areas of the boundary, MgSiO 3 perovskite transforms to a high-
mantle that are deforming by dislocation creep (Section pressure form that may influence the seismic charac-

2.10.3) and to determine whether the flow is mostly teristics of the mantle below the D″ discontinuity

vertical or mostly horizontal. However, these interpre- (Section 12.8.4).
tations are complicated by factors such as temperature, Unlike most of the lower mantle, observations at the
grain size, the presence of water and partial melt, and base of the mesosphere, in the D″ layer (Section 2.8.5),
the amount of strain (Hirth & Kohlstedt, 2003; Faul indicate the presence of seismic anisotropy (Panning &
et al., 2004). Romanowicz, 2004). The dominance of V SH polariza-
Most authors view power-law (or dislocation) tion over V SV in shear waves implies large-scale horizon-

creep as the dominant deformation mechanism in the tal flow, possibly analogous to that found in the upper
upper mantle. Experiments on olivine, structural evi- 200 km of the mantle. The origin of the anisotropy,
dence in mantle-derived nodules, and the presence of whether it is due to the alignment of crystal lattices or
seismic anisotropy suggest that power-law creep to the preferred orientation of mineral shapes, is uncer-
occurs to a depth of at least 200 km. These results tain. However, these observations suggest that D″ is a
contrast with many studies of post-glacial isostatic mechanical boundary layer for mantle convection.
rebound (Section 2.11.5), which tend to favor a diffu- Exceptions to the pattern of horizontal flow at the base

sion creep mechanism for flow in the upper mantle. of the lower mantle are equally interesting. Two excep-

Karato & Wu (1993) resolved this apparent discrep- tions occur at the bottom of extensive low velocity
ancy by suggesting that a transition from power-law regions in the lower mantle beneath the central Pacifi c
creep to diffusion creep occurs with depth in the and southern Africa (Section 12.8.2) where anisotropy
upper mantle. Diffusion creep may become increas- measurements indicate the onset of vertical upwelling
ingly prominent with depth as pressure and tempera- (Panning & Romanowicz, 2004).
ture increase and stress differences decrease. A source Another zone of seismic anisotropy and horizontal
of potential uncertainty in studies of mantle rheology flow similar to that in the D″ layer also may occur at