Anti-Aliased Non-Periodic Source Separation

In wavefield signal apparition, acquisition with periodic variations in source activation parameters shifts all or part of a signal cone out to, e.g., the Nyquist wavenumber, enabling perfect separation of simultaneous-source (sim-source) data in a diamond-shaped region below a certain temporal frequency. Cyclic convolution in the spatial frequency domain can be exploited to separate data with known non-periodic encoding functions. Such a separation comprises a least-squares reconstruction in the frequency-wavenumber (fk) domain. By applying the reconstruction in the S-transform domain and using an exact expression for NMO stretch, it is possible to perform anti-aliased source separation in conjunction with an invertible NMO or when pre-processing renders the encoding time and space-variant.