Time-Lapse Topography Inversion from Gas–Plume Monitoring

Summary Carbon dioxide sequestration projects require geophysical monitoring and verification to meet community and stakeholder expectations. Mapping of plume movement is a key component of this obligation. Time lapse seismic provides excellent data on plume evolution, but this imposes the need for subsurface models that consistently predict the observed movement. Seal topography is a primary driver of plume migration, together with fault geometry and permeability. Traditional 3D multiphase flow models are too heavy for use as forward models in inversions for topography, and standard fixed-geometry geocellular models are inconvenient for topographical inversion. Using simplified flow models for plume evolution using topography and plume-thickness variables, we have developed an efficient framework for topography inversion where seismic information on plume thicknesses is available. The framework relies on a reduced physics model that describes a bouyancy-driven plume via a simple PDE-controlled thickness field conserving volume. Fast and efficient inversion is possible via either adjoint-state methods or full-Jacobian methods based on reduced basis representations. The method is illustrated using field data from CO2CRC's Otway Project in Australia, using plume evolution over a 4 month injection period with pre and post-facto seismic monitoring. The resulting inversion predicts surface topography trends conformable with other sources of data.