Search found 4 matches

by PenguinDrum
Wed Dec 03, 2025 8:21 am
Forum: ROMS Discussion
Topic: ROMS model blow up in new version while work fine in older version
Replies: 7
Views: 753

Re: ROMS model blow up in new version while work fine in older version


Did you really use the syntax

!#define ...

with an exclamation mark before the #define in your CPP definitions file?

I would have expected that to throw a syntax error in the C-Preprocessor.

I am concerned that the options you think you have selected may not be what you actually got ...
by PenguinDrum
Fri Nov 28, 2025 4:50 am
Forum: ROMS Discussion
Topic: ROMS model blow up in new version while work fine in older version
Replies: 7
Views: 753

Re: ROMS model blow up in new version while work fine in older version


Hi, PenguinDrum

You may check if you are using COARE3.5 in your newest ROMS.

If yes, this may be the same issue I encountered several weeks ago. And I found this was relating to the COARE3.5.

Here are several options you can try and see if your problem still exist:

- recompile the ROMS with ...
by PenguinDrum
Fri Nov 28, 2025 4:48 am
Forum: ROMS Discussion
Topic: ROMS model blow up in new version while work fine in older version
Replies: 7
Views: 753

Re: ROMS model blow up in new version while work fine in older version


Hi

How do you compile ROMS? I had a similar problem when trying the sea ice model, but when I compiled with -mcmodel=large, the program worked fine.


Hi talebu,
Thanks for your reply and recommdation. I compile ROMS with the default build_roms.sh with intel mpi 2023. The problem remians after I ...
by PenguinDrum
Thu Nov 27, 2025 3:14 am
Forum: ROMS Discussion
Topic: ROMS model blow up in new version while work fine in older version
Replies: 7
Views: 753

ROMS model blow up in new version while work fine in older version

Hello everyone,

I have encontoured a problem when migrating my roms version from V4.1 to the newest develop branch on github.
The same configuration works fine on older version, but blow up in the newest version after 20-30 steps with KEchar = NaN, PEchar = NaN.
I wonder if there is significant ...