Iterative Dip-Steering Median Filter: A New Approach to Separation of Simultaneous Sources

Simultaneous source acquisition technology, also referred to as ‘blended acquisition’, involves recording two or more shots simultaneously. Despite the fact that conventional processing procedures can still produce acceptable images from the blended data, it is still desirable to separate the blended data into single shot gathers, which allows to reduce the crosstalk noise and to allow application of standard pre-stack processing, such as statics, velocity analysis, etc. In this paper we introduce a new and simple Iterative Dip-Steering Median Filter (IDS_MF) in order to separate the blended seismic shot gathers. This method extends the well-known conventional median filter which works well with horizontallyaligned coherent events but cannot handle steep events, especially events with conflicting dips. It first identifies the dominant dips inside a processing window by a Fourier-Radial transform in the frequency-wavenumber domain. A median filtering is then applied along the dominant dip in order to separate the signal. The data are processed along the remaining dominant dips in an iterative manner until all signals have been retained. The IDS_MF method can be used for random or spike noise attenuation. It does not require that reflections or events are hyperbolic and can be applied on any gather such as Common-shot, CMP, Common-receiver and Common-offset. IDS_MF does not require special acquisition geometries and works well for both 2D and 3D data as well as land and marine seismic surveys. The main limitation of the method is that it relies on local coherency. It produces very good deblending results when the residual statics are properly accounted for. Field data is used to demonstrate the performance of the proposed de-blending technology.