A 3D simultaneous source field test processed using alternating projections: a new active separation method

In 2008, a wide azimuth simultaneous source (SimSrc) 3D data set was acquired over an area coinciding with an existing conventional wide azimuth survey. Using two (nearly) simultaneous sources provided twice the number of shots, an increased shot density and twice the fold for essentially the same acquisition time. Although the improved sampling for the SimSrc data was predicted to yield improvements in processes such as 3D multiple attenuation and migration, cross-talk between the sources was a concern. Cross-talk can be suppressed by known techniques such as separation algorithms that estimate the individual shots and thereby separate SimSrc records into their constituent shots. However, separation algorithms available at the time suffered from aliasing and were not effective for severely shot-aliased data such as those encountered in the experimental data set, which had a 75-m in-line shot spacing, half that typical for wide azimuth geometries. In this paper we present a new separation technique called the Alternating Projection Method (APM), which is demonstrated to be robust in the presence of aliasing. Through comparisons with conventional wide azimuth data we show that 3D multiple removal and migration are both aided by the increase in shots. Moreover, the high quality of the APM separated data indicates that prestack products such as migration angle gathers will have better quality than was achieved a few years ago using a passive separation approach. We found that by carefully selecting and tailoring the method's projection operators, excellent separation results can be achieved efficiently.