Global tomography from ambient seismic noise cross-correlation

Ambient seismic noise is generated in the ocean and recorded worldwide. In order to extract Green function between stations from seismic noise records, we consider the analytical signal and compute the phase correlation to to extract phase coherent signals which after stacking build up the empirical Green function. The ambient noise phase cross-correlations are stacked using the time-frequency domain phase weighted stack (Schimmel et al. 2011). This method is applied to noise data in the period band 30 to 250sec. A bootstrap approach is used to measure group velocity between pairs of stations and to estimate the corresponding error. We show that it is necessary to stack 2 years of data in order to determine reliable group velocity measurements. It is observed by analyzing the convergence that less data are required to extract reliable group velocities at short period than at long periods. This data processing is applied to 150 stations of the global networks GEOSCOPE and GSN. Global maps of group velocities and the corresponding errors are then estimated and inverted to obtain the 3D S-wave model. CRUST2.0 model is used and the S-wave model below the crust is determined using a simulated annealing method in which the number of splines that describes the model is adapted within the inversion. This model is the first global S-wave velocity model derived from seismic noise recordings in the wide period band 30-250sec.