Feasibility of coupled hydrogeophysical inversion for characterization and monitoring of subsurface CO2 injection

We apply a coupled hydrogeophysical inverse modeling approach for the characterization and monitoring of multiphase flow processes associated with subsurface CO₂ injection. Specifically, our approach incorporates (1) a hydrological forward model that simulates CO₂ injection and corresponding hydrological measurements, (2) petrophysical models that relate the simulated state variables, such as CO₂ saturation, temperature and pressure, to geophysical properties, such as the elastic moduli and electrical conductivity, (3) forward models that simulate time-lapse geophysical measurements, and (4) an optimization algorithm that allows for the estimation of unknown parameters given the available data. Building on previous work, we demonstrate the approach at first using a simplified synthetic example. We evaluate the sensitivity of different hydrological and geophysical (crosswell seismic and electrical resistivity) data types to the parameters of interest, and the sensitivity of the results to different modeling assumptions, for example, regarding the petrophysical models and hysteresis in relative permeability and capillary pressure. In addition, we apply the approach to a small-scale pilot experiment in which super-critical CO₂ was injected into a brine formation while hydrological and seismic measurements were performed. More specifically, continuous active-source seismic monitoring (CASSM) data were collected in a crosswell configuration using a fixed source and sensors, allowing for the continuous monitoring of seismic waveforms at a high temporal sampling rate. In addition to the CASSM data, the arrival time of CO₂ in a monitoring well was measured, as was the time-varying pressure response to a fall-off test conducted prior to CO₂ injection. Compared to a previous study in which the experiment and measurements were modeled in a loosely coupled fashion, we find that an improved match between the measured and simulated observations can be obtained, resulting in a refined understanding of heterogeneity and CO₂ migration at the site. Overall, application of the proposed approach using crosswell data appears promising for the relatively small scale considered (10s of meters), and refining the petrophysical models through laboratory studies will likely help decrease uncertainty in parameter estimates. The approach is being extended to a larger scale (100s or 1000s of meters) by incorporating surface-surface and surface-borehole geophysical data, surface deformation data, and pressure data in distant monitoring wells.