The work presented herein shows that the RWE forward modeling approach has significant potential, which motivates additional research on a number of topics. While the next development stage would be applying the technique to a 2D dataset, it is important to speculate that this approach should upscale to 3D seismic data volumes with relative ease. This statement is support by the observation that a single iteration of waveform inversion should theoretically be equivalent to a wave-equation migration of three frequencies in a 3D data volume. Importantly, migrations of this size are routine in industry, which suggests that 3D RWE waveform inversion is computationally feasible.
The results also motivate additional work towards improving the amplitude accuracy of the one-way RWE operators. An initial improvement would be including the amplitude factor described by Zhang et al. (2003), which is missing from conventional wavefield extrapolation Claerbout (1985). Use of this term should generate data closer to the amplitudes predicted by ray theory. Future work could also implement the one-way operators of Thomson (2005) that derived from a "true-amplitude" one-way wave-equation that incorporates reflection and multiple-scattering losses.