Curvelet-based migration amplitude recovery

Migration can accurately locate reflectors in the earth but in most cases fails to correctly resolve their amplitude. This might lead to mis-interpretation of the nature of reflector. In this thesis, I introduced a method to accurately recover the amplitude of the seismic reflector. This method relies on a new transform-based recovery that exploits the expression of seismic images by the recently developed curvelet transform. The elements of this transform, called curvelets, are multi-dimensional, multi-scale, and multi-directional. They also remain approximately invariant under the imaging operator. I exploit these properties of the curvelets to introduce a method called Curvelet Match Filtering (CMF) for recovering the seismic amplitude in presence of noise in both migrated image and data. I detail the method and illustrate its performance on synthetic dataset. I also extend CMF formulation to other geophysical applications and present results on multiple removal. In addition of that, I investigate preconditioning of the migration which results to rapid convergence rate of the iterative method using migration.