Research

Summary

My interest of research focus on seismic imaging from the Earth's structure to the source.

Seismic structural imaging is benefiting from full-waveform constrain over the years. I am using full-waveform tomography methods combined with various types of misfit function measurements in order to build a better seismic velocity model. Anisotropy and physical attenuation (Q) are also considered. A new full-waveform tomograpic model with transverse isotropy of the contiguous U.S and surrounding region is coming out.

Earthquake source imaging is extremely important for source parameter studies, earthquake hazard assessment and also structural tomography. I am using time-reversal, adjoint tomography and back-projection to perform earthquake source and tsunami source imaging. I am also evaluating different source imaging methods aiming at the relationship between energy radiators derived from different methods (utilizes different frequency components) and different source kinematic parameters.

Attenuation is critical for accurate seismic imaging in complex media especially gas chimney and half-saturated sediments. Strong attenuation produces severe phase rotation and dispersion along with energy dissipation. I developed a set of Q-compensated reverse time migration (QRTM) algorithms with finite-difference operators. The method supports fine-grid parallelization and GPU acceleration, and works with both frequency-independent and frequency-dependent Q with tilted transverse isotropy velocity. It improves the performance of industrial-sized QRTM significantly and has potential to be applied in global seismology.

Q-compensated Reverse Time Migration

Machine learning