Conditional abort within the barotropic loop

Dear all,

I have a question. Usually when you get a crash with Nemo, it tells you that the velocity is crazy or the salinity below zero somewhere. But since the test is done at ever time step, which is also every baroclinic time step if you use time splitting, then it is not done at every barotropic time step. So sometimes your model crashes because the final values of the barotropic mode are NaNs, which makes everything is just NaN in the abort files, and the location of the crash does not make any sense. So there is no clue of why it crashed, and where.
Before I start coding something to modify this, I was just wondering if someone had written a fix for this.

Thanks in advance,


Salut Robinson, I did, a long time ago… basically adding a debug option to check on the minimum value of ht, hv and hu on all tiles at each barotropic step and stopping if one of them goes negative…

Bonjour Frédéric,

Thanks, I will try…

Bonne journée,