I am trying to investigate some aspects of extreme precipitation in MPAS. To do so I am running idealized simulations on a flat limited area domain. This means hexagons.
I have found the MPAS-Tools/mesh_tools GitHub repository to be quite helpful in creating custom planar hexagonal meshes for this purpose. The program planar_hex, in particular, is great for generating meshes. However, meshes created by that tool seem to be strictly periodic, as the cell/edge/vertex neighbors at cells in the outermost row & column of the mesh connect across to the other side. Removing those connections in the file seems to cause MPAS to be very unhappy.
I then found the periodic_hex tool in MPAS-Tools, which also appears to generate a periodic grid based on a namelist file. But that tool also includes mark_periodic_boundaries_for_culling.py, a Python script that says it can help perform what I want - to remove the outermost cells to create a non-periodic planar mesh. Running this tool requires then running a final script in the mesh_conversion_tools directory, called MpasCellCuller.x.
Okay, so I did this with a test domain and then ran a simulation. However, pretty much all variables are NaNs. So did I miss something in this process? Is there some other script in MPAS-Tools that I need to run on the output from MpasCellCuller.x?
Alternately, I tried "fooling" MPAS into using LBCs for a limited area idealized run. I used an existing real-data setup to determine the necessary structure for LBC files and used that to create "dummy" LBC files appropriate for my simulation. I noted from trial and error that the IC file created by init_atmosphere requires a few additional arrays that mark specified and relaxation zone cells for the LBCs, which I did according to section 7.1 of the attached MPAS-atmosphere tech note (hard to find online). I then set apply_lbcs = true in the namelist and modified the streams.atmosphere file - the immutable "lbc_in" stream in particular - to use the dummy LBC files I created. Those simulations also produce mostly NaNs, except I see non-NaN values seemingly in areas where the LBCs are "specified".
This is where I am stuck.
I have found the MPAS-Tools/mesh_tools GitHub repository to be quite helpful in creating custom planar hexagonal meshes for this purpose. The program planar_hex, in particular, is great for generating meshes. However, meshes created by that tool seem to be strictly periodic, as the cell/edge/vertex neighbors at cells in the outermost row & column of the mesh connect across to the other side. Removing those connections in the file seems to cause MPAS to be very unhappy.
I then found the periodic_hex tool in MPAS-Tools, which also appears to generate a periodic grid based on a namelist file. But that tool also includes mark_periodic_boundaries_for_culling.py, a Python script that says it can help perform what I want - to remove the outermost cells to create a non-periodic planar mesh. Running this tool requires then running a final script in the mesh_conversion_tools directory, called MpasCellCuller.x.
Okay, so I did this with a test domain and then ran a simulation. However, pretty much all variables are NaNs. So did I miss something in this process? Is there some other script in MPAS-Tools that I need to run on the output from MpasCellCuller.x?
Alternately, I tried "fooling" MPAS into using LBCs for a limited area idealized run. I used an existing real-data setup to determine the necessary structure for LBC files and used that to create "dummy" LBC files appropriate for my simulation. I noted from trial and error that the IC file created by init_atmosphere requires a few additional arrays that mark specified and relaxation zone cells for the LBCs, which I did according to section 7.1 of the attached MPAS-atmosphere tech note (hard to find online). I then set apply_lbcs = true in the namelist and modified the streams.atmosphere file - the immutable "lbc_in" stream in particular - to use the dummy LBC files I created. Those simulations also produce mostly NaNs, except I see non-NaN values seemingly in areas where the LBCs are "specified".
This is where I am stuck.