Introduction

Wave-equation migration velocity analysis Biondi and Sava (1999) has recently emerged as a promising new technology with the potential to overcome the difficulties encountered in complex structures by traveltime-based velocity analysis methods.

Briefly, in wave-equation migration velocity analysis (WEMVA), we iteratively update the slowness model with perturbations in slowness ($\Delta S$) obtained by inversion from perturbations in image ($\Delta R$), which, by definition, is the difference between the current image and a better-focused image (Figure 1). The key ingredient of WEMVA is the better-focused image with which we compare the original image. Here, residual migration plays a very important role, because it has remarkable properties of image enhancement.

Several residual migration methods are capable of improving the migrated images. A good choice is Stolt prestack residual migration, which is not only fast and robust, but can also be formulated as a velocity-independent procedure Sava (1999).

Another important element of WEMVA is the ability to convert images to angle-domain common-image gathers Prucha et al. (1999) to assess the quality of the velocity used in imaging. When the velocity is incorrect, different events in CIGs are not flat, but rather point up or down, and therefore are a very clear guide to where and how the velocity map needs improvement.

 

flow
Figure 1 WEMVA flow chart. We start with the recorded data and an initial guess about the slowness model (background slowness). We compute the background wavefield by recursive downward continuation from the surface and image (background image). We then apply an image enhancement procedure (residual migration) to get a better-focused image. From the two images we compute the perturbation in image, which we can invert for the perturbation in slowness. Finally, we update the background slowness and repeat the loop until convergence is achieved.

In this paper, I present an example of how residual migration can be used to improve the quality of images. I use a synthetic model with features relevant to real data in complex structures: dipping beds, reverse faults, and zones of severe distorsion. Throughout the project, I make extensive use of angle-domain CIGs.