Co-Authors: Xin Cheng, Kun Jiao, and Denes Vigh , all from Schlumberger

The conventional least-squares based full waveform inversion (FWI) is known to suffer from cycle skipping for typical field data that lacks very low-frequency energy. Various strategies have been proposed to mitigate this fundamental obstacle of FWI in order to enable a robust FWI based model building flow. These strategies consist of both data-domain modifications of the misfit functional and image-domain migration velocity analysis variants. The studies of various strategies have yielded encouraging perspectives regarding the use of FWI in model building process.

In this talk, we will focus on alternative FWI strategies in order to achieve robust inversion based model building flows. More specifically, we will review and compare some of the promising data-domain objective function modification strategies (, e.g., travel-time or  optimal transport based objective functions), and we will investigate their capabilities in model building with various geometric settings for both diving and reflected energy. With both synthetic and real data examples, we will demonstrate the robustness of a FWI based model building process, which employs various FWI strategies at different model building stages.

Speaker Biography: Dong Sun, Schlumberger
Dong Sun has worked as research geophysicist in the Full Waveform Inversion & RTM team at Schlumberger since 2012. Before joining Schlumberger, he received his PhD degree in Computational and Applied Mathematics from Rice University at Houston where he worked on differential semblance based waveform inversion under the guidance of Dr. Bill Symes. His research interests include: Seismic Imaging and Inversion, Signal Processing, Numerical Simulation and Optimization.