Extended Diffraction Tomography

TL;DR

Extended diffraction tomography is a new seismic inversion method that decomposes complex 2D/3D problems into independent 1D problems, allowing for parallel processing, robustness to noise, and improved accuracy with realistic reference models.

Contribution

This paper introduces extended diffraction tomography, enabling flexible reference models and parallelizable inversion of seismic data, enhancing robustness and computational efficiency.

Findings

Effective decomposition into 1D problems

Robustness to noise demonstrated

Benefits of realistic reference media shown

Abstract

We present the development of extended diffraction tomography, a new approach to the solution of the linear seismic waveform inversion problem. This method has several appealing features, such as the use of arbitrary depth-dependent reference models and the decomposition of the full 2D or 3D inverse problem into a large number of independent 1D problems. This decomposition makes the method naturally highly parallelizable. Careful implementation yields significant robustness with respect to noise. Several synthetic examples are shown which characterize the benefits of our method and demonstrate the usefulness of choosing realistic 1D reference media.