Significant progress has been made in recent years in mapping lateral
heterogeneities in the seismic velocity structure of the mantle [
Woodward and Masters, 1991; Pulliam et al., 1993; Forte
et al. 1993; Su et al. 1994; Vasco et al., 1994].
The velocity variations may arise from a combination of temperature,
composition, and phase, with temperature believed to play the dominant
role. Determining the temperature variations associated with the velocity
heterogeneities places constraints on the temperature differences driving
mantle convection and enables more direct comparison with dynamic flow
models and geophysical observables such as the geoid, heat flow, and plate
motions. At ambient pressure, the scaling coefficients between between
seismic velocity and temperature have now been measured in a variety of
mantle minerals [ Anderson et al., 1992; Isaak, 1992]. Sound
velocity measurements at high P and T under shock compression have
shown that the compressional velocity-temperature scaling relation at
deep mantle pressures (
100 GPa) is 
5 times smaller than
ambient pressure values [ Duffy and Ahrens, 1992, 1994]. From
this work, it was estimated that long-wavelength velocity heterogeneities
in the deep lower mantle correspond to root-mean-square thermal anomalies
of 
150 K. The potentially important role of anelasticity in
velocity-temperature scaling relations has been discussed by
Karato [1993].
In the most detailed studies relating thermal anomalies to seismic
tomography results to date, Yuen et al. [1993] and Cadek
et al. [1994] estimated that the large-scale anomalies found in the lower
mantle have outer temperatures of 400 K above the surrounding mantle and
thermal anomalies in excess of 1000 K near the plume center. Extremely
cold anomalies, which could be related to avalanche events from the
upper mantle, are also found in the lower mantle. The possible role
of radiative heat transfer in the formation of large-scale lower mantle
anomalies has been examined [ Matyska et al., 1994].
In studies of diffracted and reflected waves in the D
layer near the base of the mantle, Wysession et al. [1993, 1994]
argue that thermal anomalies as large as 400-1000 K might explain seismic
velocity perturbations near the CMB.
Thus, in the deep lower mantle, it is becoming increasingly apparent that
large thermal anomalies may exist, despite comparatively low levels of
seismic heterogeneity in much of the region.