Approximate 3D full-waveform inversion for rapid assessments of multiazimuth VSP data: Synthetic feasibility study for a carbon sequestration project

ABSTRACT Within carbon sequestration projects, seismic full-waveform inversion (FWI) is a promising tool to monitor reservoir changes at the injection site. Rock heterogeneity can produce complex 3D plume structures that demand 3D FWI technology, which generally requires considerable computational efforts and run times and therefore has limited applicability for low-cost, rapid assessment of monitoring data. To mitigate the limitations of expensive 3D FWI, a pseudo-3D FWI procedure was proposed that achieves a reasonable trade-off between the quality of timelapse analysis and computational efficiency when applied to a synthetic imaging scenario modeled after the Carbon Management Canada Newell County Field Research Station (FRS) near Brooks, Alberta. Inversion of simulated multioffset, multiazimuth vertical seismic profiling data using the proposed method outperformed independent 2D FWI of separate source lines in terms of model consistency along the intersection, due to the underlying coupling of the model spaces. This justified an interpolation into a 3D volume that showed good agreement with 3D FWI results and true models, as containment within the reservoir layer and directionality of the plume were inferred accurately. The good performance of the pseudo-3D FWI approach can be traced back to minimal out-of-plane wave effects in this synthetic data set, which originated from the horizontally layered subsurface encountered at the FRS and modeled in this study. Furthermore, the results indicated improved robustness toward noise and source sparsity when compared to benchmark 3D FWI solutions, which showed elevated background variations with increasing source sparsity. However, because the pseudo-3D FWI procedure provides major computational benefits in comparison to full 3D FWI while providing satisfactory plume imaging in situations where out-of-plane wave effects in the data are minimal, there is value in the method within the frame of near real-time monitoring of carbon dioxide plumes at the FRS and potentially other carbon sequestration projects hosted in benign geologies.