ULethbridge-PIMS Distinguished Lecture: Wenyuan Liao

Full waveform inversion (FWI) is a novel seismic wave equation-based data-driven approach characterized by exploiting full waveform information from all types of seismic data. It can image the subsurface with high resolution, up to half the propagated wavelength. In particular, the shallow part of the earth model can be greatly enhanced, and the depth image can be significantly improved. However, FWI is a computationally intensive iterative process that requires many iterations. Therefore, it is essential to use an efficient local optimization gradient-based method, which requires accurate computation of the derivatives of the misfit function with respect to the model parameters. In this talk, I will briefly review the gradient-based method for constrained optimization problems and the adjoint analysis tools used in efficient gradient computation. Based on that, I will formulate the FWI problem as a PDE-constrained optimization problem and then solve it through the framework of the adjoint state method. Several numerical applications will be presented to demonstrate the effectiveness and robustness of the proposed computational framework.