Full waveform inversion of DAS field data from the 2018 CaMI VSP survey

Matthew Eaid, Scott Keating, Kristopher A. Innanen

Carbon capture and storage has become a key research area for diverting carbon dioxide gas away from the atmosphere by storing it in deep subsurface reservoirs. Seismic data are a key technology for monitoring the injected carbon dioxide to ensure that it remains in the target formation and does not migrate into regions where it may pose a risk. Distributed acoustic sensing permits permanent installation of receivers in borehole geometries, allowing for highly repeatable sampling of transmission wavefield modes which is crucial for seismic monitoring surveys. To fully leverage the data supplied by DAS fibers, the inverse methods developed by Eaid et al. (2020) must transfer to DAS data acquired in the field both in isolation and in combination with accelerometer data. In 2018, the Consortium for Research in Elastic Wave Exploration Seismology acquired a 3D walkaway-walkaround VSP survey into both three-component accelerometers and DAS fiber. In this report the DAS fiber data are inverted using isotropic-elastic full waveform inversion. A method for the source wavefield estimation and a log-derived model parameterization are found to be crucial for convergence of the inverted models in FWI. Inverting the DAS data independently provides robust parameter estimates of the subsurface P-wave velocity, S-wave velocity, and density structure with a good fit between modeled and field data. However, inverting both datasets together using a newly formulated objective function that provides a means of controlling the relative emphasis on DAS and accelerometer data is shown to have a stabilizing effect on the inverted models when compared to using either dataset alone.