Robust source signature deconvolution and the estimation of primaries by sparse inversion

The past few years had seen some concentrated interest on a particular wavefield-inversion approach to the popular SRME multiple removal technique called Estimation of Primaries by Sparse Inversion (EPSI). EPSI promises greatly improved tolerance to noise, missing data, edge effect, and other physical phenomenon generally not described by the SRME relation (van Groenestijn and Verschuur, 2009a,b). It is based on the premise that it is possible to stably invert for both the primary impulse response and the source signature despite beforehand having no (or very limited) explicit knowledge of latter. The key to successful applications of EPSI, as shown in very recent works (Savels et al., 2010), is a robust way to reconstruct very sparse primary impulse response events as part of the inversion process. Based on the various successful demonstrations in literature, there is a very strong sense that EPSI will also play an important role in future developments of source signature deconvolution and the general recovering of wavefield spectrum.