Pseudo-3D elastic FWI: Structurally coupled inversions of intersecting 2D planes

Anton Ziegon, Kristopher A. Innanen

Elastic full-waveform inversion (FWI) is a key technology to provide more complete imaging and monitoring of subsurface operations, as it is able to invert multi-component data into three independent elastic parameters that allow for a more complete reservoir description. In the case of geologic carbon dioxide (CO₂) sequestration, it has a big potential for reliable monitoring, measurement, and validation (MMV) of the CO₂ plume. Usually, the plume develops a complex 3D structure within the reservoir due to the heterogeneity of the rock properties, such that 2D imaging alone is not sufficient for the monitoring task and 3D elastic FWI is needed. The problem is that 3D elastic FWI is computationally very expensive and challenging, and therefore it is not feasible for realistic applications. In this report, we present a pseudo-3D elastic FWI concept to get more meaningful 3D information from intersecting 2D planes while avoiding the computational burdens of the 3D procedure. This is achieved by introducing structural coupling along the intersection of the different 2D sections of a walkaway-walkaround VSP study such that the resulting models are near-identical along the intersection, which ultimately improves the consistency and interpretation of the results. Synthetic testing on a simple 3D block anomaly model confirms that the coupling is improving the model consistency along the intersection and therefore providing more meaningful pseudo-3D information about the subsurface structure. The tests also revealed challenges associated with this form of coupling, mitigation strategies to overcome these issues, and possible limitations of the pseudo-3D procedure. The results of this report are promising enough to motivate further studies, such that this report is concluded with a list of possible research objectives for future testing and developing. The application of this procedure to the Snowflake field data, a walkaway-walkaround VSP data set, is identified as the ultimate goal of this study and therefore future testing should push this procedure in that direction.