Full waveform inversion algorithm based on a time-shift nonlinear operator

An accurate initial model is crucial for full waveform inversion (FWI). In this paper, we propose a new way to build an initial model for conventional FWI by developing a new FWI algorithm based on a new time-shift nonlinear operator. We apply the new time-shift nonlinear operator to the waveform, define a new misfit function and derive the corresponding gradient operator. Numerical results using synthetic data from the Overthrust model demonstrate that compared with traditional FWI, the new algorithm is less sensitive to the traveltimes error. Using the inverted model of the new algorithm as the initial model, conventional FWI can obtain much better results.