Full Waveform Inversion Based on Wavefield Correlation

Low frequencies and initial model play a crucial role in full waveform inversion (FWI). However, in traditional seismic exploration, the frequency range of signals recorded by receivers is limited, and a good initial model for FWI is not available easily, leading to FWI algorithms easily fall into local minima. We propose a wavefield correlation method to improve cycle skipping issue of FWI. Test of 2D frequency domain FWI based on Marmousi model shows that the result of inversion has a better result than conventional method.