Multi-objective Full Waveform Inversion
Our team focuses on developing signal models for seismic imaging that incorporate all the physical principles of wave-propagation. Our algorithms are related to full waveform inversion (FWI), which is a popular imaging tool in both exploration and global seismology. The advantage of FWI is that it could potentially result in an accurate and reliable characterization of complex geological structures with high resolution. However, as these algorithms aim to explain all details in the measured seismograms, they are not simple - advanced mathematical techniques and high-performance algorithms are required to process large volumes of seismic data that are subject to full-waveform models. In other words, gradient-based optimizations are prone to local-minima convergence, resulting in imperfect data fitting. Our research focuses on developing auxiliary signal models that assist FWI to produce accurate reconstruction of subsurface properties.
