Full Wavefield Migration of Vertical Seismic Profiling data

Abstract

Until now, in most seismic imaging technologies, both surface and internal multiples are considered as noise. In today’s industrial practice, we see various methods for suppressing multiples before migration. This means that only a fraction of the recorded wavefield is used in imaging. In this thesis, we present a method termed full wavefield migration (FWM) that uses the multiple-reflections in the data to improve the illumination of the field in areas that cannot be reached by the primaries, to yield a better vertical resolution as well as to suppress migration artefacts caused by crosstalk of multiple-reflections. This thesis demonstrates the feasibility of full wavefield migration on a kind of borehole seismic known as vertical seismic profiling (VSP). We know that in today’s practice, images obtained using VSP data always suffer from poor illumination and small aperture effects. Therefore, we expect in VSP acquisition geometry, multiples can lead to significant improvement in illumination, both at the reservoir level as well as away from the well region. In this thesis, the advantage of using multiples in full wavefield migration has been demonstrated. We validated our algorithm on 2D synthetic and field VSP data. Full wavefield migration is posed as an inverse problem, where the parameters to be estimated are the subsurface reflectivities. We discuss an iterative forward modelling engine termed full wavefield modelling which is used in the inversion scheme. Full wavefield modelling allows us to compute the full wavefield (primaries and all multiples) in terms of estimated reflectivities. In the full wavefield modelling engine, we assume a scale-separation between the background migration velocity that governs only the one-way wavefield propagation and the reflectivity model that governs the two-way scattering. The modelling engine accounts for the non-linearity of the wavefield due to reflectivity, incorporating the transmission effects and multiple scattering at all depth levels. To solve the inverse problem, we have used iterative conjugate-gradient scheme, which is a local optimization method. We also presents a solution for imaging of blended source VSP data using FWM. The inversion-based imaging algorithm allows us to use any complex source wavefield without the need for a separate deblending (pre-processing) step. This thesis introduces the concepts of elastic full wavefield modelling and inversion. The elastic modelling of P and S waves is illustrated for a horizontally layered medium using a VSP geometry. The elastic imaging to estimate angle-dependent reflectivity parameters that incorporates mode-conversions in subsurface layers is an important area of future research. Nearly vertical structures such as salt-flanks pose a migration challenge for conventional FWM. We have also extended the FWM algorithm to incorporate turning-waves using horizontal one-way wavefield extrapolation. Using this extension, we illustrate that FWM can be used to image steep dips or near-vertical structures using the turning wavefield in VSP data. Alok Kumar Soni.