Kui Xang

Non-linear inverse scattering in anisotropic elastic media using a modified Born iterative method

Kui Xang

Estimating an accurate subsurface model of the Earth is crucial for identifying the locations of reservoirs. With the development of long offsets and wide azimuth seismic data acquisition, we can extract multiple anisotropic parameters from the recorded data to describe the right physical features of the Earth’s subsurface. Recently, we present an iterative non-linear inverse scattering algorithm for the reconstruction of the elastic parameters and mass density. To solve the non-linear forward scattering problem, we transform the elastodynamic wave equation into an integral equation, and then solve that integral equation by a fast-Fourier-transform-accelerated Krylov space method. For the non-linear inverse scattering problem, we approximate it to a series of linear approximations by using the Born approximation. Then, we use a modified Born iterative method to invert the model update at each linear approximation. The modified Born iterative method is based on the explicit computation of the Fréchet derivative. In our algorithm, the Fréchet derivative and its adjoint operators are presented in matrix-free manners so that we can also use the fast Fourier transform to reduce the computational cost and memory demand. We formulate the multi-parameter non-linear inverse scattering problem using a modified matrix-free Born iterative method. The application of synthetic data examples demonstrates its capability to produce accurate inversion results.