Problems with PML

[InesAdd]

Hello everyone,

I am using specfem3d to perform simulations of acoustic wave propagation. I started with very simple simulations: a monopole with a frequency of f = 20Hz in an open homogeneous domain (containing water c=1500 m/s). I compared the numerical results with the analytical solution of monopole propagation and found some differences which I believe are related to the presence of spurious reflections caused by the not optimal performance of the C-PML conditions.

I noticed that the code works better when the domain is perfectly symmetrical (a cube), while the results significantly worst when the domain is asymmetrical. I am attaching some plots showing the results of the simulations I conducted (the source is located in the middle of the y and z and at x=100m, so it is not perfectly in the center of the cube). Could someone explain to me how to make the PML work better? Or if there are any validations of specfem3d for acoustic simulations?

Thanks in advance.