Recently proposed structures and characteristics of the lowermost
mantle, along with schematic paths of the seismic waves used by (2) (PcP)
and (1) to map the thin region
of ultra-low seismic velocity.
From L. H. Kellogg, Mapping the core-mantle boundary, Science, 277, 646, August 1, 1997.
The boundary between Earth's core and mantle is a fundamental
discontinuity where silicate minerals come into
contact with molten iron alloys. Because it is so remote, nearly 2900 km below the surface, it can only be studied from afar. New
studies are allowing a glimpse of the lowermost part of the mantle and raising new questions about this region. A very thin layer in
which seismic velocities drop by 10% or more (1, 2) may indicate melting at the base of the mantle (3). The lower mantle may be
anisotropic, perhaps as a consequence of fine-scale compositional variations (see figure). Moreover, seismic studies have also
revealed several hundred meters of small-scale topography on the core-mantle boundary (4).
The wide-ranging recent work provides a clearer picture of fine-scale
structure in the region known as D", encompassing about 200 km of the lowermost
mantle (and ending at
the core-mantle boundary). In the 1980s seismic studies uncovered a discontinuous increase in the velocities of shear and compressional seismic waves in this layer.
More recently, the D" region has been shown to vary laterally in both thickness and in seismic velocity gradient [HN4] . For instance, a study of the southwestern Pacific (5)
found that the layer varied in thickness by about 100 km over a short distance, whereas in other locations the discontinuity has not been observed at all. A variety of mechanisms
have been proposed to account (alone or in combination) for the structure of D". As the core cools, a thermal boundary layer develops at the core-mantle boundary; this layer
likely becomes unstable from time to time, releasing upwelling plumes of hot material. Subducted slabs, or perhaps only the component that was once part of the oceanic crust,
may languish at the base of the mantle. A residue of core formation, too heavy to rise under the influence of the heat from the cooling core, may pile up under mantle upwellings.
A phase change is also a possibility. Or the boundary may be chemically active, with molten core iron altering the composition of nearby minerals. The variability of
velocity gradients and thicknesses suggests that more than one of these mechanisms may be at work.
A recent approach has been to use seismic waves that are diffracted
along the mantle side of the boundary to examine details of the structure.
The very small signal means that
multiple data from large arrays must be stacked. Within a thin (less than 20-km) layer at the base of the mantle, P wave velocities (nP) are reduced by about 10% (2).
This interpretation can vary, to some extent, with a tradeoff between layer thickness and the velocity reduction; thinner layers with a greater velocity reduction can also satisfy the
data in some locations.
What could cause such a severe drop in velocities over such a
small region? Williams and Garnero (3) proposed that the lowermost mantle
is partially molten. One
consequence would be a reduction in shear-wave velocities (nS); they predicted that nS should drop three times as much as nP within the layer (3). In their study of the
core-mantle boundary beneath Tonga, Japan, and South America, Revenaugh and Meyer (2) conclude that just such a drop is required, supporting the partial melt hypothesis.
They also note that the layer may vary in thickness from place to place, possibly because of interaction with mantle downwellings and upwellings. Whatever is happening at the
core-mantle boundary, structures there are likely heavily influenced by the pattern of convection in the lower mantle.
Another constraint on the nature of the core-mantle boundary
comes from studies of anisotropy of seismic velocities. In the upper mantle,
anisotropy depends on mineral
orientation and can indicate mantle flow direction. In the lowermost mantle, where the mineralogy is different, other mechanisms may be required, but anisotropy may still
provide information about local flow patterns. One possible mechanism for producing anisotropy is compositional layering; if the deep mantle contains material of varied
composition, convective flow may induce fine lamellae that would result in an orientation for the maximum and minimum nS. Some of the strongest evidence for anisotropy has
been found under Alaska and the Central Pacific (7), but to date only a few regions of the lower mantle have been examined.
These new approaches provide substantial insight, but they open
up more questions, such as whether the thin, low-velocity layer at the
base of the mantle is found around the
world, and whether the observations show partial melting. If so, melting in the lowermost mantle could have significant consequences for the compositional evolution of the D¢¢
layer and the thermal structure of upwelling plumes. Perhaps more complete maps of anisotropy along with the new estimates of small-scale topography (4) will indicate what
the flow patterns are and how any compositional variations interact with the dynamic lower mantle.
1.E. J. Garnero and D. V. Helmberger,
Geophys. Res. Lett. 23, 977 (1996).
2.J. Revenaugh and R. Meyer, Science 277, 670 (1997).
3.Q. Williams and E. J. Garnero, ibid. 273, 1528 (1996).
4.P. S. Earle and P. M. Shearer, ibid. 277, 667 (1997).
5.A. Yamada and I. Nakanishi, Geophys. Res. Lett. 23, 1553 (1996).
6.L. Vinnik et al., Geophys. Res. Lett. 22, 1657 (1995).
7.J. M. Kendall and P. G. Silver, Nature 381, 40 (1996).