NSERC 2011 DNOISE progress report

The main thrust of the DNOISE project is focused on the following researth themes: [1] seismic acquisition design and recovery from incomplete data with the goal to reduce acquisition costs while increasing the spatial bandwidth and aperture of seismic data; [2] Removal of the 'surface nonlinearity' by simultaneous estimation of the source signature and the surface-free Green's function by inverting the surface-related multiple prediction operator; [3] Reduction of the computational complexity of full-waveform inversion (FWI) by randomized dimensionality reduction; [4] "Convexification" of FWI to remove or at least diminish the adverse effects of non-uniqueness that has plagued FWI since its inception; The first three themes are directed towards removing major impediments faced by FWI related to the costs of acquiring data, the computational costs of processing and inverting data, and to issues with source calibration and surface-related multiples. The final theme is more 'blue sky' and tries to incorporate ideas from migration-velocity analysis into the formulation of full-waveform inversion. Aside from these themes, we will continue to work on seismic data acquisition schemes that favor sparsity-promoting recovery and on the development of large-scale solvers using recent developments in convex and stochastic optimization.