Time-lapse surface seismic inversion with thin bed resolution for monitoring CO2 sequestration: A case study from Cranfield, Mississippi

► Time-lapse seismic shows weak CO2 signature at Cranfield datasets. ► BPI derived acoustic impedances show improved resolution. ► CO2 caused Impedance decreases mostly occur at the top of the injection interval. ► The strong impedance decreases mostly occur around injection wells locations. The feasibility of carbon dioxide sequestration research at Cranfield, Mississippi is studied by injecting millions tonnes of CO2 into the lower Tuscaloosa sandstone Formation over a two year period. Time-lapse surface seismic surveys were recorded at pre-(2007) and post-(2010) injection stages to monitor the subsurface fluid plume. The injection interval, appearing as a thin layer in the well-log data, shows very weak signature of CO2 injection in the time-lapse seismic amplitude data. In order to improve the capability of tracking CO2 plume movement using seismic data, we have applied a basis pursuit inversion (BPI) method to the post-stack seismic datasets. This method of inversion incorporates a priori information as a wedge dictionary and employs a L1-norm optimization for obtaining solutions with improved resolution. The inverted time-lapse acoustic impedances show a strongly decreasing trend mostly at the top of the injection interval, which is in agreement with well-log measurements for CO2 saturation. Improved resolution time-lapse impedance mapping therefore is an effective tool for imaging the displacement of the CO2 plume.