A three-dimensional mechanical model of the Tibetan crust explains both the first-order features of GPS surface velocities and the contrast in the types of earthquake between northern and southern Tibet. See Letter p.79
The collision of India with Eurasia has built the Tibetan Plateau1, by far the largest area of high topography on the planet, as the Indian plate has thrust beneath Tibet. The crust of the plateau is approximately twice the typical thickness, and great debates have raged over its development and mechanical properties. Temperatures are high within the thickened crust2,3, which implies significant weakness of the middle and lower crust4,5. Several groups have proposed that the middle crust of Tibet is fluid enough to decouple the upper crust from the underthrusting Indian lithosphere, such that Tibetan middle crust might be extruded through a low-viscosity channel flow from beneath the high plateau either southward4 or eastward5. But others have argued6 that deformation at all depths in the Tibetan crust is coherent. On page 79, Copley et al.7 use a three-dimensional (3D) mechanical model to explain the pattern of faulting and surface deformation in Tibet, and conclude that mechanical coupling between underthrust India and Tibet must be strong in southern Tibet. This is inconsistent with channel-flow models for southern Tibet4.
The Tibetan Plateau deforms pervasively, as shown by Global Positioning System (GPS) measurements8,9, and earthquakes in the brittle upper crust occur over the entire plateau7. Thrust faulting, indicative of shortening and crustal thickening, is found only on the margins of the plateau. In the southern interior, the earthquakes occur almost entirely on normal faults, indicating extensional stress, whereas in the northern interior strike–slip faulting dominates.
This contrast in the type of faulting between the northern and southern interior of Tibet has long been noted, but cannot be explained by changes in stresses induced by topography or plate motions alone. Despite the change in style of faulting, the surface velocities obtained from GPS measurements at sites across the entire plateau show substantial and nearly uniform strain between the major strike–slip faults, specifically contraction in the direction of plate convergence and extension orthogonal to it10. A successful mechanical model needs to explain both the difference in the style of faulting between the northern and southern regions and the relatively uniform surface strain field shown by the GPS data.
Copley et al.7 approximate the Tibetan crust using a 3D viscous model, with long-term (geological timescale) deformation driven by imposed velocity boundary conditions and topographically induced stresses. The model predicts the stress field (internal forces) within the crust, which determines the style of faulting, and the motions of points on the surface, which can be compared to the GPS data. The authors tested different boundary conditions, reflecting vertical mechanical coupling or decoupling between the Tibetan crust and the underlying Indian lithosphere, which is assumed to be rigid.
The model with strong mechanical coupling best fits both the difference in the style of faulting and the GPS observations. In northern Tibet, which is not underlain by rigid Indian lithosphere, the model predicts a stress state that promotes strike–slip faulting, with surface deformation that reflects shortening in the direction of plate convergence and extension orthogonal to it. This deformation results from a combination of compression from plate motions and topographically induced stresses. In southern Tibet, where the crust is coupled to underthrust India, the vertical mechanical coupling induces an additional component of shear stress. This significantly changes the stress field so that normal faulting and east–west extension are favoured instead of strike–slip faulting (Fig. 1).
The authors7 have succeeded in capturing the contrast between the observed faulting patterns in the northern and southern regions with a simple mechanical model, while also matching the first-order features of the observed GPS velocity field in northern Tibet. In southern Tibet, however, the GPS velocities contain a significant component due to the recoverable, elastic response to the build-up of stress on the Main Himalayan Thrust (MHT; Fig. 1) during the earthquake cycle. The shallow part of the MHT is frictionally locked, slipping mainly in large earthquakes. The presence of this locked region causes contraction normal to the Himalayan arc that extends a considerable distance into southern Tibet. With the next large earthquake, this region will spring back southwards, releasing centuries of stored strain energy in seconds to minutes. Together with an existing model for this elastic strain component11, Copley et al.'s long-term deformation model can explain the first-order features of the GPS velocities across Tibet.
Their model does not match some details of the GPS velocity field. It underestimates the rate of east–west extension across the southern part of the plateau. In addition, it approximates Tibet as a continuous medium, and cannot include localized slip on the major strike–slip fault systems. In addition, the elastic model used for the strain from the MHT11 does not work well for the region east of 90° E. This area near the eastern end of the Himalayan arc is complicated, and features a southward jump of the convergent front to the Shillong plateau in India. Further work will be needed to determine whether any of these factors simply reflect local variations or potential problems with the model.
Copley and colleagues7 have placed important new constraints on the mechanical properties of Earth's continental lithosphere in its most extreme environment, and forced a critical evaluation of the channel-flow models for Tibet. Their model makes testable predictions of the average viscosity of the middle and lower crust in Tibet. The great debate is not finished, but it may have been channelled in a new direction.
Royden, L. H., Burchfiel, B. C. & van der Hilst, R. D. Science 321, 1054–1058 (2008).
Francheteau, J. et al. Nature 307, 32–36 (1984).
Nelson, D. et al. Science 274, 1684–1687 (1996).
Beaumont, C., Jamieson, R. A., Nguyen, M. H. & Lee, B. Nature 414, 738–742 (2001).
Clark, M. K. & Royden, L. H. Geology 28, 703–706 (2000).
England, P. & Molnar, P. Science 278, 647–650 (1997).
Copley, A., Avouac, J.-F. & Wernicke, B. P. Nature 472, 79–81 (2011).
Wang, Q. et al. Science 294, 574–577 (2001).
Gan, W. et al. J. Geophys. Res. 112, B08416 (2007).
Chen, Q. et al. J. Geophys. Res. 109, B01403 (2004).
Bettinelli, P. et al. J.Geodesy 80, 567–589 (2006).