[BLOWUP] NaNs after switching from ERA5 to CMEMS wind forcing

run / config issues
1

Hello CROCO users,

I’m running an interannual simulation using CROCO with OpenMP, forced by ERA5 and GLORYS12. For surface forcing, I’ve recently replaced the original ERA5 winds with CMEMS reanalyzed winds (uwnd, vwnd, and wspd) in the bulk files.

Here’s the situation:

  • The simulation with the original ERA5 winds ran successfully over the entire period using the same boundary and initial conditions.
  • After replacing only the wind variables with CMEMS data, the simulation runs fine from June 2020 to August 2020.
  • In September 2020, the model suddenly crashes with NaNs in energy diagnostics:

MAIN: started time-stepping

STEP time[DAYS] KINETIC_ENRG POTEN_ENRG TOTAL_ENRG NET_VOLUME trd
26496 7549.00000 1.948888391E-03 4.6616186E+01 4.6618135E+01 8.4612796E+14 33
26497 7549.00116 NaN NaN NaN NaN 31

MAIN - number of records written into history file(s): 1
number of records written into restart file(s): 1
number of records written into averages file(s): 31

MAIN: Abnormal termination: BLOWUP

I initially suspected higher resolution of CMEMS winds might be doing something, so I reduced dt from 300 to 100 and the to 50, but the blowup still occurs at the same point. I verified the temporal consistency of the bulk files (timestamps seem fine), so I’m a bit stuck.

Thanks for any ideas on what could be my mistake.

2

Hi, Based on the situation you describe, the problem may be in the wind values that resulted from the CMEMS interpolation to your grid in September. You can check the horizontal field during those months and look for any atypical values. What parameters are you using for your surface forcing? Can you share your cppdefs? Best regards,

3

Hello! It was actually a problem with the data from the blk file, that it contained corrupted values but for some reason it was just that month (I created all of them in a for cyle). I assume it was an issue with the writing, or just bad luck. But now its working. Thanks for the reply!