Seismic Depth Processing of a Real Dataset in the Gunsan Basin, Offshore Western Korea using Migration Velocity Analysis (MVA)

We applied the seismic depth processing to a real dataset using the migration velocity analysis (MVA). The survey area is the Gunsan Basin (Yellow Sea, offshore western Korea). The real seismic data were recorded on 480 receivers and the entire survey line is about 30 L-km. The source spacing and the receiver spacing are 50 m and 12.5 m, respectively. A seismic impulse source is generated by 4 air-gun arrays (4,578 cu. in.). The seismic depth processing using the MVA provides more reasonable velocity profile than the velocity from the conventional time processing procedure. The basic concept of the MVA is that the pre-stack depth migration (PSDM) image of a same reflection point is same for all different shots when the velocity is correct. In other words, all events on the common reflection point (CRP) gather should be flat if the velocity obtained by the MVA is correct. In this study, we used the tomography-based MVA technique which minimizes the depth residuals. Before running the MVA, a pre-processing in time domain is necessary to improve the S/N ratio and attenuate the multiples of the raw data. The pre-processing workflow is consists of low cut filtering, spherical divergence correction, swell noise attenuation, velocity analysis, predictive deconvolution, and radon filtering. We confirmed that the noise and the multiples are attenuated after pre-processing. The depth processing using the MVA is performed by several steps as follows. First, the time-domain stacking velocity as an initial velocity was used in the depth processing. The pre-processed data was sorted to the common offset data and the PSDM was performed. Next, the PSDM sections are sorted into the CRP gathers. We picked the residual moveout on every CRP to obtain the interval velocity. Finally, the entire workflow of the MVA is repeated until the CRP gathers are flat. From the MVA results, we could generate flatter CRP gathers and reasonable velocity model in depth domain after several iterations.