search

CESR-lab / ucla-roms

GNU General Public License v3.0
3 stars 8 forks source link

BGC diagnostics are zero in some tiles inconsistently #12

Open ubbu36 opened 3 months ago

ubbu36 commented 3 months ago

I'm adding this as an issue for visibility.

I'm having problems where the bgc_avg_dia output is zero in some tiles. This is in turn affecting the Alk and DIC fields.

This seems to be happening in different tiles during different runs, even with the same restart file and same executable. I've attached a few movies here that illustrate the issue in two different runs.

https://github.com/CESR-lab/ucla-roms/assets/69470952/3d693812-963f-40c4-9a9b-0c14642b9379

https://github.com/CESR-lab/ucla-roms/assets/69470952/2918030f-b280-4a22-8d94-3140f291d0cd

matt-long commented 3 months ago

@ubbu36, I thought you were also seeing the block structure in differences between tracers in different runs. Is that the case?

matt-long commented 3 months ago

All the BGC diagnostics you point to are contingent on the pH solver converging. Is it possible that that the pH solve is failing silently? Why would infect the whole block?

ubbu36 commented 3 months ago

@matt-long yes, I'm seeing the differences in tracers between runs, but they tracers are not 0, they are just different because, presumably, the BGC is failing. You can only see the difference in tracers if two runs are subtracted like in the video attached here:

https://github.com/CESR-lab/ucla-roms/assets/69470952/5ecc69e6-eb91-44f6-bf6d-1578d596b069