A GLOBAL TOMOGRAPHIC MODEL OF SHEAR ATTENUATION IN THE UPPER-MANTLE

We present a global three-dimensional model of shear attenuation in the upper mantle, based on the measurement of amplitudes of low-frequency (100-300s) Rayleigh waves observed at stations of the Geoscope and Iris networks. Attenuation coefficients are measured on R1 and R2 paths using a method which minimizes the effects of focussing due to propagation in a three-dimensional elastic Earth. Through a series of tests which, in particular, involve the computation of synthetic models of attenuation and focussing, we demonstrate that long wavelength lateral variations in attenuation in the first 400-500 km of the mantle can indeed be resolved. The model is obtained in a two-step procedure. The first step consists in the computation of maps of Rayleigh wave attenuation at different periods, using an inversion method without a priori parametrisation, which involves the introduction of a correlation length, chosen here at 3000 km to optimize the trade-off between resolution and variance in the model. In the second step, after corrections for shallow structure, an inversion with depth is performed, assuming lateral heterogeneity is confined to depths between 80 and 650 km. The resulting model presents lateral variations in Q(beta) that are correlated with tectonic features, in particular ridges and shields in the first 250 km of the upper mantle. Below that depth the pattern shifts and becomes correlated with the hotspot distribution, particularly so if the buoyancy strength of hotspots is taken into account. Two major low-velocity zones appear to be located in the central pacific and beneath northern Africa, in the depth range 300-500 km. This pattern seems to continue at greater depth, but resolution becomes insufficient below 500 km to draw definitive conclusions. The smooth lateral variations retrieved are on the order of +/-50% down to 400 km. We propose an interpretation in terms of plume/lithosphere/ridge interaction in the upper mantle, arguing for de flection of the bulk of hot upwelling material from plumes towards ridges, which may be occurring between 200 and 300 km depth.