-
Notifications
You must be signed in to change notification settings - Fork 7
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Fixes #210. Added adjustable tolerance to the rasterization routine. … #663
Fixes #210. Added adjustable tolerance to the rasterization routine. … #663
Conversation
…Added some heuristic logic in mkMOMAquaRaster.F90 to define a different tolerance (default is 1.0e-12, and under some conditions we change it to 1.0e-5). This fixes the failures with 5 degrees MOM6 rasters. Aslo updated the code and the build for MITgcm rasters
|
||
! Get source grid directory and destination raster file names | ||
!------------------------------------------------------------ | ||
|
||
i = command_argument_count() | ||
i = iargc() |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@atrayano Can you bring back the Fortran standard command_argument_count
and get_command_argument
rather than iargc
and getarg
? I'd like to keep those old extensions from coming back
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@mathomp4 Good catch! Done
@atrayano, would you please give me and @biljanaorescanin the exact set of commands you use to generate the tile/raster files? I recall you do it all manually i.e., without using We are unable to reproduce your output. |
@sanAkel I just have sent you a long chat message in Teams. In essence, what I did when I manually generated the tile grid for you was to generate 3 rasters (cube, land (i.e. Pfafstetter) and tripolar, but using slightly modified source code. I had hardcoded a different tolerance 1.0e-5 instead of 1.0e-12 in the rasterize.H, the recompiled GEOS, and the 3 rasters were generated with the new binaries. I am nearly 100% sure that the tripolar and the land rasters were generated this way, and I am less certain what I did for the cube - it could have been generated with the 1.0e-12 tolerance. If it is critical, I could check. Then the rasters were combined pretty much the same ways the make_bcs script combines them. In this PR, I decided it is overkill to downgrade the tolerance for the land and the cube, and effectively it set to 1.0e-5 ONLY for the MOM raster and under certain conditions, depending on the MOM resolution and the raster size. Otherwise, the code sets the tolerance to 1.0e-12. I believe this approach is better, but obviously I cannot expect 0 diff with the old result. I could work with you or @biljanaorescanin to reproduce the old result |
Thank you @atrayano. This info should help us figure it out. If we can't next week, will reach out. |
@biljanaorescanin when we tag up, best to pull this recent commit from @atrayano and build before we test again. Thanks! |
All, I have converting this to draft until all issues have been fully resolved. When it is ready for testing, please undraft |
Following is my review and reasons for marking it ready to merge. All the steps that I took to test are also provided. ✅ It builds with the ✅ For NLv3, c12 and T1MOM6 (72x36) options,
✅ Using the new boundary conditions, I remapped
i.e., the set of BCs are now pointing to: mepo status
|
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
It works! That is, I could obtain a:
- Set of restarts,
- A run (1-day) that finished without any problem(s).
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Cmake seems good!
…Added some heuristic logic in mkMOMAquaRaster.F90 to define a different tolerance (default is 1.0e-12, and under some conditions we change it to 1.0e-5). This fixes the failures with 5 degrees MOM6 rasters. Aslo updated the code and the build for MITgcm rasters