A Review of Geophysical Constraints on the Deep Structure of the Tibetan Plateau, the Himalaya and the Karakoram, and their Tectonic Implications

The Tibetan Plateau, the Himalaya and the Karakoram are the most spectacular consequences of the collision of the Indian subcontinent with the rest of Eurasia in Cainozoic time. Accordingly, the deep structures beneath them provide constraints on both the tectonic history of the region and on the dynamic processes that have created these structures. The dispersion of seismic surface waves requires that the crust beneath Tibet be thick: nowhere less than 50 km, at least 65 km, in most areas, but less than 80 km in all areas that have been studied. Wide-angle reflections of P-waves from explosive sources in southern Tibet corroborate the existence of a thick crust but also imply the existence of marked lateral variations in that thickness, or in the velocity structure of the crust. Thus isostatic compensation occurs largely by an Airy-type mechanism, unlike that, for instance, of the Basin and Range Province of western North America where a hot upper mantle buoys up a thin crust. The P-wave and S-wave velocities in the uppermost mantle of most of Tibet are relatively high and typical of those of Precambrian shields and stable platforms: V_p = 8.1 km s^-1 or higher, and V_s≈ 4.7 km s^-1. Travel times and waveforms of S-waves passing through the uppermost mantle of much of Tibet, however, require a much lower average velocity in the uppermost mantle than that of the Indian, or other, shields. They indicate a thick low-velocity zone in the upper mantle beneath Tibet, reminiscent of tectonically active regions. These data rule out a shield structure beneath northern Tibet and suggest that if such a structure does underlie part of the plateau, it does so only beneath the southern part. Lateral variations in the upper-mantle structure of Tibet are apparent from differences in travel times of S-waves from earthquakes in different parts of Tibet, in the attenuation of short-period phases, Pn and Sn, that propagate through the uppermost mantle of Tibet, and in surface-wave dispersion for different paths. The notably lower velocities and the greater attenuation in the mantle of north--central Tibet than elsewhere imply higher temperatures there and are consistent with the occurrence of active and young volcanism in roughly the same area. Surface-wave dispersion across north--central Tibet also requires a thinner crust in that area than in most of the plateau. Consequently the relatively uniform height of the plateau implies that isostatic compensation in the north--central part of Tibet occurs partly because the density of the relatively hot material in the upper mantle is lower than that elsewhere beneath Tibet, the mechanism envisioned by Pratt. Several seismological studies provide evidence consistent with a continuity of the Indian Shield, and its cold thick lithosphere, beneath the Himalaya. Fault-plane solutions and focal depths of the majority of moderate earthquakes in the Himalaya are consistent with their occurring on the top surface of the gently flexed, intact Indian plate that has underthrust the Lesser Himalaya roughly 80-100 km or more. P-waves from explosions in southern Tibet and recorded in Nepal can be interpreted as wide-angle reflections from this fault zone. P-wave delays across the Tarbela network in Pakistan from distant earthquakes indicate a gentle dip of the Moho beneath the array without pronounced later variations in upper-mantle structure. High Pn and Sn velocities beneath the Himalaya and normal to early S-wave arrival times from Himalayan earthquakes recorded at teleseismic distances are consistent with Himalaya being underlain by the same structure that underlies India. Results from explosion seismology indicate an increase in crustal thickness from the Indo--Gangetic Plain across the Himalaya to southern Tibet, but Hirn, Lepine, Sapin and their co-workers inferred that the depth of the Moho does not increase smoothly northward, as it would if the Indian Shield had been underthrust coherently beneath the Himalaya. They interpreted wide-angle reflections as evidence for steps in the Moho displaced from one another on southward-dipping faults. Although I cannot disprove this interpretation, I think that one can recognize a sequence of signals on their wide-angle reflection profiles that could be wide-angle reflections from a northward-dipping Moho. Gravity anomalies across the Himalaya show that both the Indo--Gangetic Plain and the Himalaya are not in local isostatic equilibrium. A mass deficit beneath the plain is apparently caused by the flexure of the Indian Shield and by the low density of the sedimentary rock in the basin formed by the flexure. The mass excess in the Himalaya seems to be partly supported by the strength of the Indian plate, for which the flexural rigidity is particularly large. An increase in the Bouguer gravity gradient from about 1 mGal km^-1 (1 mGal = 10^-3 cm s^-2) over the Indo--Gangetic Plain to 2 mGal km^-1 over the Himalaya implies a marked steepening of the Moho, and therefore a greater flexure of the Indian plate, beneath the Himalaya. This implies a northward decrease in the flexural rigidity of the part of the Indian plate underlying the range. Nevertheless, calculations of deflections of elastic plates with different flexural rigidities and flexed by the weight of the Himalaya show larger deflections and yield more negative gravity anomalies than are observed. Thus, some other force, besides the flexural strength of the plate, must contribute to the support of the range. A bending moment applied to the end of the Indian plate could flex the plate up beneath the range and provide the needed support. The source of this moment might be gravity acting on the mantle portion of the subducting Indian continental lithosphere with much or all of the crust detached from it. Seismological studies of the Karakoram are consistent with its being underlain by particularly cold material in the upper mantle. Intermediate-depth earthquakes occur between depths of 70 and 100 km but apparently do not define a zone of subducted oceanic lithosphere. Rayleigh-wave phase velocities are particularly high for paths across this area and imply high shear wave velocities in the upper mantle. Isostatic gravity anomalies indicate a marked low of 70 mGal over the Karakoram, which could result from a slightly thickened crust pulled down by the sinking of cold material beneath it. Geophysical constraints on the structure of Tibet, the Himalaya and the Karakoram are consistent with a dynamic uppermost mantle that includes first, the plunging of cold material into the asthenosphere beneath southern Tibet and the Karakoram, as the Indian plate slides beneath the Himalaya, and second, an upwelling of hot material beneath north--central Tibet. The structure is too poorly resolved to require such dynamic flow, but the existence for both a hot uppermost mantle beneath north--central Tibet and a relatively cold uppermost mantle beneath southern Tibet and the Karakoram seem to be required. Both group and phase velocities of Rayleigh waves and Love waves are delayed along paths crossing Tibet. The low velocities require a crustal thickness in excess of 50 km, and for most regions in excess of 60 km. Crustal thicknesses in excess of 80 km can be ruled out for all paths studied, and for most of Tibet, a crustal thickness of 65-70 km seems required. Clear evidence for lateral heterogeneity beneath Tibet is provided not only by body waves (discussed below) but also by surface waves (Brandon & Romanowicz 1986), which show an area of lower uppermost shear-wave velocity and thinner crust in north--central Tibet than elsewhere in the plateau. These variations might explain the differences in group velocities measured by different workers, and the different structures that they deduced, but if so, they also render the regionalization of surface-wave dispersion into arbitrary tectonic provinces risky. Although Rayleigh-wave phase velocities can resolve large differences in upper-mantle velocities for regions the size of Tibet, constraints on these velocities are best derived from body waves. Thus, with the exceptions of Brandon & Romanowicz's (1986) detailed investigation of north--central Tibet, the study of southernmost Tibet by Jobert et al. (1985) and that of Romanowicz (1982) for the northeasternmost part of the plateau, I do not think that surface waves have placed an important bound on the velocity in the upper mantle beneath Tibet. The seismic data are broadly consistent with partial melting of the uppermost mantle of north--central Tibet, where recent volcanism has been observed. Correspondingly, there is no suggestion of such low velocities, and such high temperatures, in the mantle elsewhere beneath Tibet, for which late-Cainozoic volcanism has not been reported. The results are also consistent with a slightly thinner crust in north--central Tibet than farther south, suggesting that both Airy and Pratt isostasy share compensation for north--central Tibet's great height. Finally, the average shear-wave velocity in the upper mantle of southern Tibet seems to be higher than that in northern Tibet, but neither is the degree of difference well determined, nor is the location of the transition from one to the other well mapped.