The thermal structure of Tibetan crust and upper mantle (Invited)

Almost 30 years ago I was a co-author of a paper that suggested the mantle lithosphere beneath Tibet had been removed by a convective instability caused by shortening. Surface wave tomography carried out by a number of groups now clearly shows that this proposal is wrong, and that cold lithosphere is still present everywhere beneath the plateau. It has not been removed by delamination, and, at least in the northern part, the mantle beneath the Moho is hotter than that at greater depths. Furthermore the thickness of the lithosphere, of ~300 km, is greater than that beneath any other continental region and strongly suggests that the lithosphere has been shortened. The Iranian plateau shows similar features on a smaller scale. This unexpected behaviour can be understtod if the density of the lithosphere is a function of both its temperature and its composition. Mantle nodules brought up by melts show that much of the continental lithosphere has been depleted by melt removal, leaving a harzburgite whose density is substantially less than that of the fertile mantle. The trace element composition of basaltic rocks from northern Tibet shows that their source rocks were harzburgites that had been enriched by a few percent of metasomatic melt. The melting is probably occurring in the shallow mantle beneath the Moho, where the shear wave velocity is low and where the temperature is increasing because of downward conduction of heat generated by radioactive decay within the crust. This behaviour will occur wherever the thickness of the crust and lithosphere have been increased by shortening, and is likely to have been even more important than it is now at earlier times in the Earth's history.