Multi-Parameter Full Waveform Inversion

Full Waveform Inversion is presently the most advanced seismic imaging technology. By reconstructing the full seismic wavefield, we can recover all of the components of the elastic or viscoelastic tensor. Inverting for multiple classes of parameters is required to solve meaningful near-surface problems in geotechnical studies or in hydrogeology. It is, however, challenging, as parameters have different sensitivities and become coupled. My research group develop new ways to simultaneously reconstruct the earth properties, in particular shear and pressure velocities and attenuation. One promising avenue is the use of deep learning to engineer new and effective regularization functions.

Automated data processing using machine learning

Machine learning promises to automate and replace many traditional jobs. Geophysical data processing is no exception. My research group investigate the automation of the traditional seismic data processing workflow. Seismic processing is labour intensive and require extensive expertise, which prevents this powerful technique from being used in many fields of applications. By using deep learning, we can replace and simplify many of the complex processing steps required to perform imaging (for example velocity analysis). We can also envision automating the whole chain of operations with reinforcement learning. Along the way, important questions must be answered: can we train machines to understand the underlying physical laws behind geophysical imaging ? What are the best features to feed our neural nets, and how can we bridge the gap between simulation and real data applications ?

Characterizing permafrost using seismic attenuation

Extensive offshore areas of the Beaufort Sea are underlain by terrestrial permafrost that has been transgressed as a result of post-glacial sea level rise.  As a consequence, the offshore permafrost is degrading, causing substantive changes in its physical properties and potentially liberating free gas or dissociating permafrost gas hydrates. Assessing the state of permafrost degradation is thus critical to understanding the impacts of these phenomena, especially for climate change. Conventional processing of marine seismic surveys is not adapted to the physical properties of permafrost, which exhibits high attenuation and highly variable velocities. A promising approach is to use and measure the attenuation of seismic waves caused by the presence of ice. My research group studies the fundamental seismic properties of frozen porous media, with the aim of assessing the potential of seismic attenuation to characterize the permafrost integrity. This would pave the way to use powerful new tools like viscoelastic full waveform inversion to quantitatively assess the integrity of permafrost on a large scale.

See the project posting for more information.