[MITgcm-support] OBCS/EXF in Checkpoint68d

[Jean-Michel Campin]

Hi Roberta,

Thanks for submitting a clear explanation of the problem.
The fix is not yet in master branch (for now, pull-request #562 has not been merged yet)
but will probably get there later today.

Cheers,
Jean-Michel

On Tue, Nov 16, 2021 at 12:20:31PM +0100, Roberta Sciascia wrote:
> Hi Oliver, 
> 
> Great to hear that it has been fixed!
> It was an unlucky timing to download a new checkpoint but the perfect model setup to find the bug. 
> 
> Hope is all well with you too. 
> 
> Cheer, Roberta 
> 
> 
> 
> 
> > On 15 Nov 2021, at 21:17, Roberta Sciascia <roberta.sciascia at sp.ismar.cnr.it> wrote:
> > 
> > Hi everyone, 
> > 
> > I???ve recently dowloaded the MITgcm checkpoint68d and stumbled upon an error in the OBCS/EXF handling that I suppose might be related to the commit 
> > ???Add interannual monthly forcing to exf ???. 
> > I???ve run this same configuration successfully on an older version of the MITgcm (checkpoint67b) but wanted to use the option OBCSbalanceSurf so I???ve upgraded to a newer version. 
> > 
> > I run a configuration with OBCS read from files and time levels controlled from the EXF_NML_OBCS. 
> > In my specific case, I have only one open boundary, the southern one and here is how my EXF_NML_OBCS looks like 
> > 
> >  &EXF_NML_OBCS
> > 
> > #
> > 
> >  obcsSstartdate1   = 20160101,
> > 
> >  obcsSstartdate2   = 000000,
> > 
> >  obcsSperiod       = 21600.,
> > 
> > #
> > 
> > The configuation was compiling correctly but failing to run with the following error
> > 
> > (PID.TID 0000.0001) *** ERROR *** CAL_CHECKDATE: Invalid month in date(1)=         0
> > 
> > 
> > (PID.TID 0000.0001) *** ERROR *** CAL_FULLDATE: fatal error from cal_CheckDate
> > 
> > This error is caused by the field OBCSN that in my case is not defined and doesn???t have a Startdate set. However, I suppose that with the modification to the file exf_getffield_start.F <https://github.com/MITgcm/MITgcm/compare/checkpoint68d...master#diff-5c86463f942c7c23fd7c4b443144629c9036aa58922d7e67cc6ccb3db613665f> and specifically the condition in the new lines 80-81  
> > 
> > 
> > IF ( useCAL .AND. .NOT.(fld_period.EQ.-12. .OR. & (fld_period.EQ.-1 .AND. useYearlyFields)) ) THEN
> > 
> > is true even for non-active OBC hence the model is unable to run because in the case of a non-active boundary the date is equal to zero. 
> > 
> > Reverting to the old configuration IF ( useCAL .AND. fld_period.GT.0. ) THEN solved the problem for me but I don???t think my OBCS configuration is a corner case and thought this might be useful to everyone. 
> > 
> > Hope this is somehow clear
> > 
> > Cheers, 
> > Roberta 
> > -- 
> > Roberta Sciascia
> > 
> > Consiglio Nazionale delle Ricerche
> > ISMAR - Istituto di Scienze Marine
> > Sede Secondaria di Lerici 
> > Forte Santa Teresa ??? 19032 Lerici (SP), Italy
> > tel: +39 0187 1788903
> > 
> > 
> > 
> > 
> > 
> 

> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mailman.mitgcm.org/mailman/listinfo/mitgcm-support