Earth science

Roots of the matter

How far down does the ancient continental material that constitutes Earth's 'tectosphere' extend? Fresh interpretation of the behaviour of seismic waves helps in reconciling previous estimates.

Over the past three decades there has been vigorous debate over how thick the continents can be — that is, the depth to which the rigid crust and upper mantle reach before meeting convecting mantle that can flow and drive tectonic motion. On page 707 of this issue1, Gung and colleagues add new seismological interpretations that go some way to explaining the differing views.

The oldest continental rocks are more than 3.8 billion years old and there are extensive regions of continents, known as 'shields', that are older than 1 billion years. In contrast, the oldest oceanic material is only about 200 million years old, because of the cycle in which oceanic crust is created at mid-ocean ridges and subsequently destroyed as material is returned to depth in subduction zones (particularly around the Pacific). The preservation of old material as the continents move across the Earth's surface due to the relative motions of the tectonic plates is related to what lies beneath: samples brought to the surface through various eruptive processes indicate that there is a significant difference between the continental and oceanic environments.

Based on information from heat flow, geochemistry and the relative delay times of seismic waves in different settings, Jordan2 proposed the 'tectosphere' model, in which a zone moves with the motion of the plate lying beneath the old continental shields and would be expected to be about 400 km thick. More recent assessments of heat-flow data and geochemistry favour a zone no thicker than 250 km. A thickness of 200–250 km is also consistent with investigations of how the Earth has responded to the removal of the load caused by glaciers, and with regional seismological studies using seismic surface waves. But many seismological models of three-dimensional structure based on global observations would favour a zone extending to 400 km, based on the depths at which seismic shear waves travel at elevated speeds — shear waves have their particle motion perpendicular to the direction of propagation. Such higher seismic wave speeds indicate the presence of a region possessing distinct properties, undoubtedly cooler than its surroundings, but also likely to be distinct in composition.

Gung and colleagues1 help to reconcile these results. Their work is based on images of Earth structure derived from shear waves with different polarizations. The approach — seismic tomography — is akin to the medical tomography used to study the interior of the human body. In the seismic version, however, illumination of Earth's interior depends on the distribution of earthquakes generating seismic waves and of seismic stations producing high-quality data. Many different classes of wave-propagation phenomena are exploited to improve sampling deep into the Earth. The dominant energy from earthquakes is radiated as shear waves in which the deformation is transverse to their propagation path. Purely horizontally polarized shear waves have somewhat simpler characteristics than those with propagation in a vertical plane and so have been used in many studies. Gung et al. point out that the studies indicating a thick root to the continental shields make extensive use of data from the horizontally polarized waves.

When the authors segregate the information from vertically and horizontally polarized shear waves, they find that the resulting images show significant differences. The structure with fast wave speeds beneath most continental shields extends to at most 250 km for the vertically polarized waves, whereas the equivalent structures for horizontally polarized waves extend deeper, to nearly 400 km. Gung et al. suggest that this form of seismic anisotropy, with horizontally polarized shear waves travelling faster than those with vertical polarization, is due to the influence of mantle flow in the region between 250 and 400 km. Similar anisotropy has been found under the ocean basins3,4 in the depth range from 80 km to 250 km. The transition to anisotropy caused by mantle flow would then mark the seismic definition of the base of the 'tectosphere', with a depth of no more than 250 km beneath the ancient continental shields.

In most parts of the world it is difficult to provide any direct test of the results of Gung et al.1. But the distribution of earthquakes around the shield of northern Australia, a region that I have studied, is such that higher-frequency seismic waves can be used to probe structure beneath the shield. Here, similar anisotropy is found, with faster travel of horizontally polarized waves5 through a zone below 220 km that produces significant dissipation of wave energy and is hence likely to have lowered viscosity, promoting flow. A complicated structural transition to lowered seismic wave speeds beneath the ancient shield occurs at about 210 km and has itself been attributed to a different kind of anisotropy6. The patterns of seismic anisotropy are subtle and the strong effects associated with polarization are accompanied by variations with the direction of propagation7.

Overall, Gung et al. have provided a satisfying way forward in tackling a long-standing puzzle in Earth science. But on the evidence from northern Australia — and as might be expected — the picture will not be straightforward. The seismically defined base of the tectosphere is likely to be quite complex, depending on variations in the local tectonic environment.


  1. 1

    Gung, Y., Romanowicz, B. & Panning, M. Nature 422, 707–711 (2003).

    ADS  CAS  Article  Google Scholar 

  2. 2

    Jordan, T. H. Rev. Geophys. Space Phys. 13, 1–12 (1975).

    ADS  Article  Google Scholar 

  3. 3

    Ekström, G. & Dziewonski, A. M. Nature 394, 168–172 (1998).

    ADS  Article  Google Scholar 

  4. 4

    Montagner, J.-P Rev. Geophys. 32, 135–137 (1994).

    ADS  Article  Google Scholar 

  5. 5

    Tong, C., Gudmundsson, O. & Kennett, B. L. N. J. Geophys. Res. 99, 15783–15797 (1994).

    ADS  Article  Google Scholar 

  6. 6

    Leven, J., Jackson, I. J. & Ringwood, A. E. Nature 289, 234–236 (1981).

    ADS  Article  Google Scholar 

  7. 7

    Debayle, E. & Kennett, B. L. N. Earth Planet. Sci. Lett. 184, 339–351 (2000).

    ADS  CAS  Article  Google Scholar 

Download references

Author information



Corresponding author

Correspondence to B. L. N. Kennett.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Kennett, B. Roots of the matter. Nature 422, 674–675 (2003).

Download citation

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.