3D Multisource Full-Waveform Inversion using Dynamic Random Phase Encoding

We have developed a multisource full-waveform inversion algorithm using a dynamic phase encoding strategy with dual-randomization—both the position and polarity of simultaneous sources are randomized and changed every iteration. The dynamic dual-randomization is used to promote the destructive interference of crosstalk noise resulting from blending a large number of common shot gathers into a supergather. We compare our multisource algorithm with various algorithms in a numerical experiment using the 3D SEG/EAGE overthrust model and show that our algorithm provides a higher-quality velocity tomogram than the other methods that use only monorandomization. This suggests that increasing the degree of randomness in phase encoding should improve the quality of the inversion result.