Implicit elastic full waveform inversion for the VSP data: snowflake data from Carbon Management Canada Newell County Facility

Tianze Zhang, Xiaohui Cai, Kristopher A. Innanen

Carbon capture and storage (CCS) has risen as a critical research domain, aiming to reduce carbon dioxide emissions by confining the gas deep within subsurface reservoirs.Seismic data are indispensable in monitoring the sequestered carbon dioxide to ensure its confinement to intended zones and prevent encroachment into areas of potential risk. In 2018, the Consortium for Research in Elastic Wave Exploration Seismology executed a 3D walkaway-walkaround VSP survey, integrating both three-component accelerometers and DAS fibres. This investigation employs implicit full waveform inversion to determine the baseline model based on 2018’s accelerometer data. This implicit elastic full waveform in-version harnesses neural networks to produce elastic models. The neural network’s weights are optimized to generate refined elastic models that minimize data misfit, obviating the need for precise initial models. A comparison of inversion outcomes with well-log data is encouraging, and the alignment between synthetic and observed data further underscores its promise