Wave equation GSP prestack depth migration and illumination

Abstract

A heterogeneous medium with strong velocity contrast and complex subsurface structures presents a great challenge to seismic imaging. Prestack depth migrations based on the wave equation are accurate but very expensive, thereby limiting their use. However, recent advances in efficient 3D wavefield extrapolators and the continued increase in computing performance are making application of full 3D prestack wave equation depth imaging more affordable. In addition, since PreSDM was first employed for subsalt and carbonate imaging, there have been major advances in efficient and accurate 3D wavefield extrapolation technologies. The generalized screen propagator (GSP) is a one-way wave-equation-based wide-angle propagator that can provide high-resolution and high-fidelity subsurface images. GSP neglects up/down reverberations between heterogeneities and correctly handles forward multiple-scattering phenomena including focusing/defocusing, diffraction, and wave-interference effects. The algorithm alternates between the space-domain and wavenumber-domain via the fast Fourier transform (FFT). The operations within each domain are self-adaptive to the complexity of the medium, making this method robust in the presence of strong velocity contrasts. GSP can be directly applied for prestack shot-record depth migration. The forward downgoing wave from the source side and backward extrapolation of the upgoing wave from the receiver side independently use the same propagator. At each depth step, application of a correlation or deconvolution imaging condition determines a depth image. Figure 1 shows a prestack shot-record migration on the Sigsbee2 data set provided by the Smaart JV Project. Figure 1b shows that the top of salt is well imaged by the GSP migration even for the very steep-sided synclinal feature in the top salt surface. The amplitude and phase of the base salt are imaged well and have good continuity, and the diffractors beneath the salt are clearly reconstructed. The image quality in the finite-difference migration (Figure 1c) is generally comparable to the GSP result, but the …