Sparse inversion based deblending in CMP domain using Radon operators
Kai Zhuang, Daniel O. Trad, Amr Ibrahim
We implemented a deblending framework in the CMP domain to test the efficacy of deblending outside the commonly used receiver domain. By operating in the CMP domain instead of the common receiver domain, dipping reflectors are centered as opposed to apex shifted, this allows us to implement a simple hyperbolic Radon operator to decrease processing time taken to invert for a deblended data set versus an apex shifted operator. The Radon operator is posed as an inversion problem using a L1 model norm to support focusing in the Radon domain allowing better mapping of data back to their focused gathers. Implementation of deblending in an inversion-based framework is a relatively newer route to exploring deblending, with previous source separation implementations being denoising the pseudo-deblended data. Inversion based deblending allows us to explain all the data by refitting back to the blended dataset using the blending operator.