Deblending using hybrid Radon transform

Amr Ibrahim, Kai Zhuang, Daniel O. Trad

In this report, we add a hybrid Radon transform to the inversion-based deblending method. Deblending methods that use sparsity constraint relies on the similarity between the primary signal and the transform basis. This similarity results in the focusing of the coherent primary signal while simultaneously attenuating the incoherent blending noise. However, seismic data contains mixtures of signals with different travel time trajectories. Therefore, using hyperbolic functions to model seismic data that contains non-hyperbolic events will reduce sparsity (signal focusing) and thereby reduce the efficiency of the deblending algorithm. To overcome this issue, we expand the transform basis to account for more types of signals contained in the seismic data. In this report, we a hybrid of linear and hyperbolic Radon basis functions to match the moveout patterns of reflections and direct arrivals. Synthetic data examples show that this transform can efficiently deblend seismic reflections and direct arrivals.