| Popis: |
Summary We apply visco-elastic full waveform inversion (FWI) to a 50-km long controlled-source seismic survey at the San Andreas Fault (SAF) to obtain high resolution P-wave and S-wave velocity models. The profile consists of 63 explosive sources and 912 3-component receivers. At the current state of the study, we derived individual source wavelets for the different shots at low frequencies (3–6 Hz). Several preprocessing steps (time / offset windowing, trace specific weights, 3D-2D-compensation) are evaluated considering the similarity of the derived source signatures. To estimate the seismic data from 2D wave propagation, a 3D-to-2D-compensation function is derived to account for different divergence of P-, S-and their converted phases. Homogeneous velocity models and coinciding wave paths are assumed. Traveltime models from Ryberg et al. (2012) and Hole et al. (2006) are used to derive velocity starting models. Attenuation is estimated from Qp and Qs t*-tomography models after Bennington et al. (2008). Density is estimated using Gardner’s (1974) relation. The forward modeling is based on a time-domain visco-elastic FD-algorithm of Robertsson et al. (1996). The inversion is performed in the frequency-domain. |
