search

SPECFEM / specfem3d_globe

SPECFEM3D_GLOBE simulates global and regional (continental-scale) seismic wave propagation.
GNU General Public License v3.0
91 stars 96 forks source link

add PML for the one-chunk case #277

Closed komatits closed 7 years ago

komatits commented 9 years ago

See https://github.com/geodynamics/specfem3d/issues/312

komatits commented 9 years ago

See also https://github.com/geodynamics/specfem3d_globe/issues/26 .

komatits commented 9 years ago

It will be easy to add PML in its current form once we have the new (very fast) DSM coupling; this comes from the fact that PML is currently implemented in SPECFEM3D only for external mesh faces that are parallel to X, Y or Z, but once we have the fast DSM calculation for any point we can easily switch (at least temporarily) from a mesh for a chunk of the Earth to a mesh for a cube (with the right Earth curvature and topography at its surface).

See the picture below.

This trick will of course not work if the chunk becomes big (in terms of aperture in degrees) because then the elements in the two upper corners could become too distorted due to Earth curvature).

Another (better) option would be to implement titled PMLs, i.e. PMLs for absorbing layers not parallel to X, Y or Z, but that is not done yet. (see also geodynamics/specfem3d#312 ). we_could_change_the_chunk_in_order_to_use_pml

komatits commented 9 years ago

Vadim Monteiller @vmont will do that in 2016 (if he joins our group), following a discussion with him at EGU'2015.

komatits commented 9 years ago

See also https://github.com/geodynamics/specfem3d/issues/498

komatits commented 7 years ago

Probably not needed any more, just use SPECFEM3D_Cartesian with a mesh designed for a chunk using Trelis, Gmsh or similar.