Approximate constant density acoustic inverse scattering using dip‐dependent scaling

Abstract

This abstract presents a computationally efficient method to approximate the inverse of the Hessian or normal operator arising in a linearized inverse problem for constant density acoustics model of reflection seismology. Solution of the linearized inverse problem problem involves construction of an image via prestack depth migration, then correction of the image amplitudes via application of the inverse of the normal operator. The normal operator acts by dip‐dependent scaling of the amplitudes of its input vector. This property permits us to efficiently approximate the normal operator, and its inverse, from the result of its application to a single input vector, for example the image, and thereby approximately solve the linearized inverse scattering problem. We validate the method on a 2D section of the Marmousi model to correct the amplitudes of the migrated image.