Utilisation des corrélations de bruit micro-sismique pour l'analyse des propriétés du champ d'onde et l'imagerie crustale

摘要

Since 10 years, crustal tomography based on noise cross-correlations have been used in many regions. In most studies, the noise to signal ratio is used to select the correlations for the tomography. According to the theory, the cross-correlations of two noise records converge to the Green function in positive and negative parts if noise sources are homogeneously distributed around the stations. Therefor, the time symmetry of correlations is related to the quality of its convergence to the Green function. The symetry properties are related to the spatial distribution of noise sources and synchronisation between stations. We use this property to control the quality of SIMBAAD dataset and some data have been corrected. We use a particle motion analysis and a beamforming on all correlations to identify some sources of P waves in deep oceans. In the Aegean-Anatolian, noise sources are homogeneously distributed at 15 s but this distribution becomes heterogeneous at longer periods. For each pair of stations, we compare different components of the correlation tensor and we check the time symmetry to obtain the closest dispersion curve to the Green function. For each frequency, we inverse all travel times to reconstruct the group velocity maps for Rayleigh and Love waves. We inverse local dispersion curves to construct a 3D model of S wave velocity in the crust in the Aegean Anatolian domain. We observe some lateral changes for S wave velocity which are coherent to the geological structures (Sedimentary basin, continental crust, ...). We also prove there is some lateral changes of the depth of the Moho discontinuity. Our observations are coherent with previous geophysical studies. The Moho depth seems to be related with the recent deformation (younger than 15 My).