Custom MPAS Grid Generation

davidl2000

New member
I know it's probably been asked but I really think MPAS could benefit from having custom grid generation tools made by the MPAS developers, so there's integration and people can reliably and 100% make custom grids without being worried about if MPAS init_atmosphere will accept or reject it. I've tried so many attempts with MPAS-Tools and JIGSAWpy (which do work sometimes but are inconsistent) and fixes from this forum and even trying to rely on AI assistance and nothing seems to alleviate the issue. Custom grid generation is 100% required in my use cases to simulate severe weather outbreaks/hurricanes at high resolution in a refinement region with either a CONUS domain at 15km or global 60km with a 15km CONUS domain and refinement region that has to be customized for size and resolution. A lot of the stuff I've seen says it could be anything from obtuse triangles in the generated mesh to sphere radius inconsistencies to just init_atmosphere rejecting it with barely any indication as to why. Seems to be at least a semi-common occurrence here as well. Of course people are free to share their opinions or maybe even some fixes that's worked for them that I'm unaware of. I am not very skilled or adept with Fortran so I definitely can have missed something but I feel like my Python script for generating the meshes is not the issue.
 
I know it's probably been asked but I really think MPAS could benefit from having custom grid generation tools made by the MPAS developers, so there's integration and people can reliably and 100% make custom grids without being worried about if MPAS init_atmosphere will accept or reject it. I've tried so many attempts with MPAS-Tools and JIGSAWpy (which do work sometimes but are inconsistent) and fixes from this forum and even trying to rely on AI assistance and nothing seems to alleviate the issue. Custom grid generation is 100% required in my use cases to simulate severe weather outbreaks/hurricanes at high resolution in a refinement region with either a CONUS domain at 15km or global 60km with a 15km CONUS domain and refinement region that has to be customized for size and resolution. A lot of the stuff I've seen says it could be anything from obtuse triangles in the generated mesh to sphere radius inconsistencies to just init_atmosphere rejecting it with barely any indication as to why. Seems to be at least a semi-common occurrence here as well. Of course people are free to share their opinions or maybe even some fixes that's worked for them that I'm unaware of. I am not very skilled or adept with Fortran so I definitely can have missed something but I feel like my Python script for generating the meshes is not the issue.
i'm working on one
 
Back
Top