High-Resolution Migration With Stable SVD-Based Inverse Hessian Compensation

Abstract

Conventional Kirchhoff migration is commonly affected by limited resolution, reflector broadening, and sidelobe contamination. Here, we propose DSVD-IHCM, a migration-imaging method that uses singular-value decomposition-based damped inverse Hessian compensation. The method first constructs local point-spread functions (PSFs) from the migration responses of scattering points and then explicitly forms an image-domain Hessian by spatial interpolation. A stable inverse Hessian operator is then constructed by applying damping to the singular-value inverse and is applied to the conventional migration image to reduce blurring and improve resolution. To avoid the memory and computational burden associated with explicit storage and global inversion of the full Hessian, we use blockwise construction, blockwise inversion, and weighted stitching. Experiments with a horizontally layered model, the Marmousi model, and field seismic data show that the constructed Hessian captures the Kirchhoff migration response. Compared with conventional migration and gradient-based image-domain least-squares methods, DSVD-IHCM more effectively compresses the main lobe, suppresses sidelobes and artifacts, and improves reflector focusing and spatial resolution. The method also remains stable and robust for noisy data, providing a practical route to high-resolution image-domain migration compensation.