Search

Summary A combination of complex geology and limitations imposed by the surface seismic acquisition geometry results in seismic images contaminated by variabilities in wave propagation effects such as illumination. Consequently, imprints of variable illumination compromise the reliability of the amplitude and phase within the seismic image. Additionally, today's conventional methods of amplitude inversion assume that the seismic amplitudes are representative of the earth's acoustic and elastic properties and do not compensate for variable illumination effects. We present a subsalt case study from the Gulf of Mexico demonstrating least-squares migration in the image domain with point-spread functions (PSF) ability to simultaneously correct for illumination effects and produce a higher-resolution image of thin sand beds located in close proximity to base of salt. We discuss our approach for mitigating scattering effects present in the PSFs arising from high-contrast impedance boundaries within the earth model. Finally, we show the results from a poststack depth-domain inversion, providing true amplitude images and acoustic impedance volumes and allowing reliable assessment of the viability of previously identified drilling targets and enabling corresponding augmentation of prospect interpretation.

Summary Successful time-lapse studies require special care when it comes to the removal of undesirable artefacts caused by the differences in acquisition geometries. By attempting to repeat the source and receiver geometries between surveys as precisely as possible, any subsequent 4D noise is minimized. However, in some cases it is not possible to repeat the survey geometries between vintages. This is the case when a towed streamer survey is compared with an OBS acquisition. The image domain approach for correcting illumination differences between 4D datasets builds on wave equation reflectivity inversion using Point Spread Functions (PSFs). In this two-step least-squares imaging method, the reflectivity is recovered by explicitly computing multi-dimensional PSFs using wave-equation modeling and de-convolving these PSFs with the final migrated image. We define a 4D formulation which is not dependent on geological and/or reservoir production constraints by introducing the concept of cross-survey PSFs (XPSFs). As shown using synthetic data examples, the joint reflectivity inversion process delivers superior results when compared to separate inversions as it ensures a more robust recovery of the 4D effects. The presented new methodology using cross-survey Point Spread Functions (XPSFs) ensures consistency of the wavefields for recovering the 4D signal.

Summary Full waveform inversion (FWI) is a class of inversion strategies that aim to invert the full characteristics of the seismic wavefield, amplitude, and phase of all arrivals, using the most realistic (or most appropriate) physics to describe the behaviour of the earth. One particular challenge is the incorporation of non-seismic data, such as well-log information, to provide an accurate measurement of the velocity at particular points in the subsurface. Well-log sonic velocity measurements do not match directly with seismic velocity information, because the instrumentation is different and measurements are made for waves at a different frequency regime. We propose to match a well-log derived seismic image to the migrated the seismic data as an image space constraint in FWI. We show that the sensitivity of image domain well-log constraints extends throughout the over/side-burden of wells. This provides a significant advantage over conventional model space constraints that only have sensitivity in the immediate vicinity of the wells. We demonstrate and justify our method using synthetic experiments in a complicated, realistic subsurface setting, using a model from the SEAM time-lapse pilot.