A methodology for 7D warping and deformation monitoring using time-lapse seismic data

Abstract

A methodology is presented for vector analysis of the image displacements (warping) between successive 3D seismic image volumes that provides 7D analysis (including lateral and vertical displacements) of in situ subsurface deformation around hydrocarbon reservoirs. The key challenges are (1) assessment of just vertical shifts is insufficient, and vector displacements should be determined; (2) robust vertical displacements can usually be derived, but lateral shifts are less well defined because of the generally smooth data character in a horizontal/horizon plane; (3) subvoxel resolution is necessary for correct matching and deformation analysis; (4) velocity and strain effects are intrinsically combined in time-lapse seismic images; (5) separation of accumulated and local effects is necessary; (6) apparently coherent and smooth displacement fields do not necessarily provide good strain analysis; (7) warping is easily degraded by noise, and good cross-matching is a prerequisite. To address these challenges, a full 3D, local warp vector derivation methodology is proposed, which involves (1) constraint using prior estimates, (2) local refinement with subvoxel resolution, and (3) 3D and vectorial conditioning using a deformable mesh with sensitivity to image-match quality. The warping approach is extended to separate accumulated from local effects and to analyze in situ deformation based on the displacement vector volume. This is achieved by a finite-element approach to determine an elemental pseudostrain tensor field and an iterative procedure to separate the pseudostrain into velocity and strain components. The approach up to the strain analysis, is demonstrated using a real data example, which indicates the potential of the methodology (accumulated overburden effects are separated to reveal a local compaction signature in the reservoir), but realistic, quantitative values of strain have not yet been realized.