The construction of a space-variant point spread function for enhancing deep reflector imaging

Abstract Least-squares migration (LSM) is a commonly used method for high-precision subsurface imaging. In recent years, image-domain LSM based on the approximated Hessian matrix has attracted widespread attention. However, the wavelet in the Hessian matrix is typically assumed to be fixed, which limits the imaging resolution. In this study, we analyze the imaging resolution affected by wavelet at different depths, and develop an image-domain LSM that constructs a space-variant point spread function (PSF) based on time-varying wavelets. First, a polynomial spectral simulation method is introduced to extract time-varying wavelets from each segmented seismic record in the frequency domain. Then, the space-variant PSF is constructed using the extracted time-varying wavelets at different depths, thereby effectively approximating the Hessian matrix. Finally, a non-iterative image-domain LSM is applied based on the constructed space-variant PSF to enhance the imaging resolution of deep reflectors. Numerical tests on both synthetic and field data demonstrate the effectiveness and practical potential of the proposed method for high-resolution imaging of complex geological structures.