In blended sources processing, many conventionally acquired shot gathers are phase-encoded and blended together to form supergathers to reduce the computational cost and I/O burden of migration. Romero et al. (2000) first explored this idea with the wave-equation migration of synthetic data associated with the Marmousi model. They produced acceptable images with less cost than the conventional method. The limitation of their approach was that the blended sources images were always no better in quality than the corresponding conventional images, because the blended sources introduced unacceptable crosstalk noise into the final migration section. Krebs et al. (2009) presented their full waveform inversion result with blended sources encoded by random encoding functions. Their computational efficiency was increased by a factor of 50 compared to standard full waveform inversion and their method has been mostly tested for a fixed-spread acquisition geometry. The extension of blended sources processing to marine acquisition is a topic of current research.