Large grain size in a random boundary condition is effective for low-frequency data. Randomly shaped grains work better than regularly shaped grains, and are also very effective in dealing with broadband sources. Of key importance is matching the grain size in the random boundary to the spectrum of the wavefield used in modeling. Applying such frequency-matched boundary conditions extends the utility of the random boundary condition beyond RTM, with possible applications including waveform inversion.
![[pdf]](icons/pdf.png){kind=icon}