Estimation of Water Within the Lithospheric Mantle of Central Tibet from Petrological-Geophysical Investigations

Investigations of the lithosphere and sub-lithospheric upper mantle by integrated petrological-geophysical modeling of magnetotelluric (MT) and seismic surface-wave data, which are differently sensitive to temperature and composition, allows us to reduce the uncertainties associated with modeling these two data sets independently, as commonly undertaken. We use selected INDEPTH MT data, which have appropriate dimensionality and large penetration depths, across central Tibet for 1D modeling. Our deep resistivity models from the data can be classified into two different and distinct groups: (i) the Lhasa Terrane and (ii) the Qiangtang Terrane. For the Lhasa Terrane group, the models show the existence of upper mantle conductive layer localized at depths of 200 km, whereas for the Qiangtang Terrane, this conductive layer is shallower at depths of 120 km. We perform the integrated geophysical-petrological modeling of the MT and surface-wave data using the software package LitMod. The program facilitates definition of realistic temperature and pressure distributions within the upper mantle for given thermal structure and oxide chemistry in the CFMAS system. This allows us to define a bulk geoelectric and seismic model of the upper mantle based on laboratory and xenolith data for the most relevant mantle minerals, and to compute synthetic geophysical observables. Our results suggest an 80-120 km-thick, dry lithosphere in the central part of the Qiangtang Terrane. In contrast, in the central Lhasa Terrane the predicted MT responses are too resistive for a dry lithosphere regardless its thickness; according to seismic and topography data the expected lithospheric thickness is about 200 km. The presence of small amounts of water significantly decreases the electrical resistivity of mantle rocks and is required to fit the MT responses. We test the hypothesis of small amounts of water (ppm scale) in the nominally anhydrous minerals of the lithospheric mantle. Such a small amount of water dramatically affects the resistivity but has no influence on the seismic velocities (and therefore, the calculated surface wave's dispersion curves are unaffected too). Three different proton conduction models for olivine conductivity (1 - Wang et al., 2006; 2 - Yoshino et al., 2009; 3 -Jones et al., 2012) and two water partition coefficients are tested. The presence of water in lithospheric mantle is decreased from 170 km to the LAB depth at 200 km. If we move this water-presentbottom boundary to shallower depth, the lithospheric mantle becomes too resistive. Our results favour a moderately wet (<0.01 wt%) mantle above the underthrusted Indian lithosphere, probably as a result of the dehydration processes. The presence of percolating water-rich fluids has the additional effect of lowering the solidus, and therefore facilitating partial melting in the warm lower crust of Lhasa.