A pilot-scale experiment for carbon dioxide (CO2) sequestration was undertaken at the Nagaoka test field in Japan. Time-lapse crosswell seismic tomography was conducted to detect and monitor the movement of CO2 injected into an aquifer. We applied difference analysis with data normalization (DADN) to the time-lapse data to eliminate false images that were apparent in a conventionally processed difference section. Conventional difference analysis calculates travel-time delays after inversion, whereas the DADN method calculates them from raw travel-time records before inversion. Thus, fewer errors are generated with the DADN method compared to a conventional inversion analysis. We applied the DADN method to time-lapse tomography data recorded before and after the injection of CO2 and computed the velocity variation in a subsurface section, which clearly showed the distribution of CO2 flooding within a high permeability zone in the aquifer and showed no CO2 leakage into the caprock. Our results also show the maximum velocity decrease as a result of CO2 injection was about 9%, which is close to the results obtained in laboratory experiments. Finally, numerical simulations were inverted to test the effectiveness of the conventional and DADN methods in dealing with noise. These tests showed that the DADN method effectively reduces unique coherent noise for particular receiver and source combinations. We concluded that the DADN method provides useful data for monitoring the flow of CO2 sequestered in underground aquifers.
Copyright(C) Research Institute of Innovative Technology for the Earth (RITE). All rights reserved.