Image-domain versus data-domain velocity analysis based on true-amplitude subsurface extended migration

We compare the traditional image-domain migration velocity analysis technique with a new approach developed in the data domain. From a reflectivity model obtained after true-amplitude migration and multiplied by an annihilator, new data are modelled under the Born approximation in the extended domain. The new objective function is simply the l2-norm of the data to be minimised. We discuss the advantages and limitations of the new approach versus the image-domain approach on two synthetic 2d models. The two approaches differ by the nature of the oscillations observed in the velocity gradients and by the relative weights contributing to the velocity update.