This thesis presents a 3-D least-squares wave-equation migration that yields regularized common image gathers for amplitude versus angle (AVA) analysis. It is an extension of 2-D least-squares migration to the 3-D case using the common-azimuth approximation.; The thesis generalizes least-squares wave-equation migration with various types of regularization. In particular, an efficient preconditioning strategy is adopted to decrease the cost of the iterative inversion. The thesis also proposes a new scheme that combines a smoothness regularization in the ray parameter direction with a sparseness regularization in the depth direction to further improve the resolution of seismic images. The problem is solved by an iterative re-weighted least-squares conjugate gradients algorithm (IRLS).; Extensive tests on synthetic and real data show that regularized iterative migration/inversion enhances event continuity in common image gathers. More importantly, the method is validated by a careful comparison of the inverted common image gathers and synthetics obtained from well log data. In addition, the vertical resolution of the inverted image is also improved as a consequence of increased coherence in common image gathers, sparse regularization in the depth direction, and by implicitly introducing migration deconvolution in the inversion. |