Mantle dynamics and seismic tomography

Abstract

Three-dimensional imaging of the Earth's interior, called seismic tomography, has achieved breakthrough advances in the last two decades, revealing fundamental geodynamical processes throughout the Earth's mantle and core. Convective circulation of the entire mantle is taking place, with subducted oceanic lithosphere sinking into the lower mantle, overcoming the resistance to penetration provided by the phase boundary near 650-km depth that separates the upper and lower mantle. The boundary layer at the base of the mantle has been revealed to have complex structure, involving local stratification, extensive structural anisotropy, and massive regions of partial melt. The Earth's high Rayleigh number convective regime now is recognized to be much more interesting and complex than suggested by textbook cartoons, and continued advances in seismic tomography, geodynamical modeling, and high-pressure–high-temperature mineral physics will be needed to fully quantify the complex dynamics of our planet's interior.

The advent of the theory of plate tectonics about 30 years ago established that most near-surface geological phenomena such as earthquakes, volcanoes, and mountain belts can be understood in the context of a unifying model of interacting surface plates. However, our understanding of this system has largely been limited to detailed kinematics of plate motions, leaving the nature of the driving motions in the interior as a great puzzle. Questions such as what is the configuration of convection, and how are surface tectonics controlled by internal processes, have long been raised, but there was a lack of tools and evidence to evaluate various hypotheses. Most views regarding mantle dynamics remained highly speculative until recently.

Seismic tomography emerged in the early 1980s, providing a major probe of the dynamical system of which plates are just the surface veneer. This technique is similar to noninvasive medical techniques used to image human interiors, although seismic tomography uses elastic waves rather than x-rays. Tomographic images are extracted from many crisscrossing paths of P and S waves through the planet, revealing regions of higher or lower than average seismic velocity at a given depth. The velocity variations are caused by both chemical and thermal variations, which can be related to the density fluctuations that drive convective flow. Earthquake sources generate the elastic waves that are recorded at seismic stations around the world, and it has taken substantial time to accumulate a sufficient number of data to image the interior with high resolution. The first generation of low-resolution models of internal structure now has advanced to the point that major dynamical questions are being resolved.

The first-order question in mantle dynamics is whether mantle convection takes place in mantlewide convective cells or whether it involves a layered system, with separate flow regimes in the upper mantle (above 650 km) and lower mantle. One of the most exciting results from the last 5 years is the verification of deep penetration of former oceanic lithosphere into the lower mantle (1–3). Tomography shows thickened tabular extensions of subducted material to depths as great as 2,000 km, directly below deep subduction zones where earthquakes take place in oceanic slabs down to about 650 km depth. Thus, strictly layered mantle convection can now be ruled out with good confidence. But many puzzling questions remain; for example, there are some down-going plates that seem to deflect and flatten near 650 km deep (3), with their descent apparently being resisted. Current thinking is that this flattening is caused by the nature of the 650-km deep boundary, which involves a solid-solid endothermic phase transition in major mantle minerals that causes colder material to undergo the phase transition (to a denser phase) at higher pressures. The dynamical consequence of this global phase boundary is to inhibit penetration of the dense lower mantle by cold, untransformed downwellings. Despite this resistance, oceanic plate in the upper mantle continues to sink, accumulating near the phase boundary. After some period, the accumulated materials may slowly sink further, transforming to the denser phase, resulting in a dynamic instability. Numerical simulations indicate that penetration occurs rapidly once it starts (4, 5), akin to an avalanche. This modulation of the descending flow by the phase boundary may account for the variety of geometries of currently subducting oceanic lithosphere and the thickening of the lower mantle slab material.

Some tomographic revelations are less obviously connected to plate motions at the surface. Most dramatic are two massive, antipodal regions of anomalously low seismic velocity in the lower mantle (6, 7): one under the South Pacific and the other under the Southern Atlantic and Africa. Low seismic velocities usually are inferred to correspond to hotter than average structure, which should be less dense and therefore upwelling. These features are thus thought to be either massive thermal plumes (superplumes) or suites of unresolved smaller upwellings. These low-velocity regions underlie shallow active hotspot centers in the South Pacific as well as the vigorous tectonic rifting of the African continent. However, the exact relationship between these two large-scale lower mantle anomalies and near-surface tectonics is not yet clear.

Seismic tomography also has revealed complex structure in the boundary layer at the base of the mantle, where heat from the core produces a hot thermal boundary layer. The dynamics of this hot boundary layer are distinctive from the cold stiff plates of the surface boundary layer, but a comparable role in mantle dynamics is likely. The deep boundary layer has now been characterized as having extensive, laterally varying layering (analogous to the continent/ocean variations of the surface boundary layer), extensive seismic anisotropy presumably caused by mineral alignments or structural fabrics induced by shear flow (analogous to the anisotropy of oceanic plates), and localized regions of partial melting (analogous to mid-ocean ridge magma chambers). The complex structures now being revealed by high-resolution tomography indicate that the deep boundary layer is a major site of highly active chemical and thermal processes affecting mantle dynamics and evolution (8, 9).

In summary, recent advances in seismic tomography have brought the realization that mantle dynamics has more multiscale complexity than had been recognized. This is reflected in complex patterns of mantle seismic velocity anomalies that could not be anticipated solely on the basis of surface plate motions. Given the high Rayleigh number of the mantle, such complexity is not unexpected, but it is clear that improved resolution of the structure of the interior on small-scale lengths will be critical to advancing interpretations of the chemical, thermal, and dynamical processes. This endeavor requires concerted efforts in seismology, in high-resolution geodynamical modeling with complex processes such as melting being accounted for, and in experimental constraints on high-pressure, high-temperature mineral properties.