Characterization of a complex fault system by 2D acoustic Random Objective Waveform Inversion

Summary One key ingredient of a successful seismic full waveform inversion (FWI) is the choice of the objective function. Robust FWI approaches combine objective functions with different weights of the phase- and amplitude-mismatch between modelled and field data. Typically, the phase-mismatch is corrected first and the amplitude-mismatch later. However, the weighting of one waveform information over another can result in convergence issues. In contrast, the recently proposed Random Objective Waveform Inversion (ROWI) randomly applies different objective functions at each stage of the FWI. In this study, we created a ROWI based on the combination of the robust Global Correlation Norm (GCN) and an AGC-weighted l2-norm. While the GCN is able to fit waveforms of the direct wave and dominant reflections, the AGC-norm enhances weak reflections with the drawback of destroying the amplitude information. Therefore, both objective functions can complement each other. The proposed ROWI approach is applied to land-seismic data acquired above the Glucksstadt Graben in Northern Germany. The ROWI result reveals a complex, heterogeneous system of thrust faults and disturbed layers within the Graben, while a much simpler layering is present outside of the Graben. The ROWI result is consistent with a time-migrated reflection seismic image and borehole logs.