Deep Learning-Driven Subsurface Characterization for Affordable CO₂ and Reservoir Monitoring (opens in new tab)

Time-lapse seismic is one of the most effective tools for monitoring subsurface processes or property changes caused by hydrocarbon production, CO2 sequestration, geothermal development, and many other activities. To substantially reduce the cost and turnaround time of time-lapse seismic data processing projects, we developed a deep-learning network for rapid characterization of subsurface property changes by establishing direct nonlinear mapping from premigration seismic data to subsurface p...