FULL WAVEFORM INVERSION WITHOUT LOW FREQUENCY USING WAVEFIELD PHASE CORRELATION SHIFTING METHOD

Full Waveform Inversion (FWI) of seismic data is a high resolution subsurface imaging tool and there is a lot of effort to fully industrialize it. The method, which uses a gradient based data fitting approach to minimize the misfit between observed and simulated waveforms, strongly requires either a good initial model or low frequency data for the convergence to the global solution. In this paper, we address the cycle skipping phenomena when none of the two above mentioned requirements are met. Then we propose the wavefield phase correlation shifting (WPCS) method to help reduce the influence of local minimum. Compared to the conventional FWI methods, a better inversion result could be achieved by applying the WPCS method without the low frequency components. We test the WPCS method on the Marmousi model and the result is improved compared with conventional method.