Target-oriented diversity stacking for virtual-source imaging and monitoring

Abstract

Interferometric virtual-source (VS) redatuming crosscorrelates downgoing waves with the corresponding upgoing waves to convert records from surface-source gathers to virtual gathers at the buried receiver locations. It can be viewed as the cancellation of common parts of the raypaths from surface sources to different buried receivers. As part of this process, a stacking operator — a uniform or simple offset function — is applied to weight and sum the surface-source array to form the VS. The stacking operator should preserve sources associated with effective cancellations of common raypaths and suppress ineffectively cancelling sources. The VS records should show reduced effects of overburden complexity, therefore providing improved image quality as well as improved repeatability in time-lapse monitoring. However, complex near-surface effects such as intricate shallow structures and variable weathering layers can severely distort the raypaths. As a result, sources associated with ineffective-raypath cancellation can produce substantial artifacts, instead of being spatially suppressed by the conventional stacking operator. To address these issues, we propose a data-driven VS method with a diversity-stacking theme in which each individual source contribution is weighted by certain quality measures. Specifically, we predict the upgoing wavefields using the conventional VS response and use the quality of these predictions compared with the original upgoing wavefields to approximate the weight of each source for the diversity stacking. Compared with previous methods, the new VS approach provides improved image quality and repeatability based on a pilot field of 13 time-lapse surveys, which reduced a significant repeatability problem across a 17-month survey gap.