Stacey Boundary conditions - SPECFEM2D GPU

[etotten93]

Hi There,

I have recently started using SPECFEM2D GPU (fortran) to great effect (thanks to the developers and maintainers) - I notice that PML ABCs are not implemented when running on GPU, so I have had to fall back to Stacey boundary conditions. At present, they are not proving very effective (see attached snapshots on a test run).

So my questions are as follows:

1. Is there a way to increase/up the strength of the Stacey ABCs in the Par_file/similar
2. If not, is the next best practice to pad the left, bottom and right (top is free surface) boundary with a thick layer of uniform velocity such that any artificial/numerical reflections never hit any seismometer receiver (this is my priority)?

Many thanks in advance for your time and help!

Eoghan

[danielpeter]

hi Eoghan,

yes, for GPU simulations only Stacey is provided for now. thus, the work-around is as you suggested, i.e., enlarge the domain to make sure that the boundary reflections hit your stations after the time of interest.