SEISMIC FULL-WAVEFORM INVERSION USING DECOMPOSED P-WAVEFIELD

Here we describe the development of a seismic full-waveform inversion method which employs P-wavefield decomposition to obtain accurate velocity information. Briefly, P-wavefield decomposition for multi-component data was Performed with Helmholtz decomposition in elastic media and an objective function. To achieve efficient inversion, application of a back-propagation technique is essential. Therefore, a stress tensor was used for P-wavefield decomposition to allow application of a back-propagation technique. Our Drowsed inversion algorithm was validated with synthetic data obtained from the Marmousi2 velocity model which simulated an ocean bottom, multi-component survey. The subsurface information obtained with our inversion method was more accurate in regard to velocity and structure compared with a conventional elastic inversion method. In addition, the application of our inversion method to synthetic data simulating an ocean bottom seismometer survey which uses a small number of receivers also obtained better results in a numerical test.