[MITgcm-support] Re: single-layer configuration problem update

Patrick Heimbach heimbach at MIT.EDU
Tue Aug 30 10:59:15 EDT 2005


Sergey,

several things:
1. I thought you were using the SINGLE_LAYER_MODE flag(?)
2. The std. output has NAN's in cg2d after a few timesteps,
   but you're saying the model doesn't explode.
   Can you clarify?
3. The alternative, simplere forcing would be via
   S/R external_fields_load.F
   and not using exf (i.e. remove 'exf' from packages.conf)
   An example is in verification/global_ocean.90x40x15/code/

-Patrick



  > -----Original Message-----
  > From: mitgcm-support-bounces at mitgcm.org 
  > [mailto:mitgcm-support-bounces at mitgcm.org] On Behalf Of 
  > Sergey Vinogradov
  > Sent: Tuesday, August 30, 2005 10:50 AM
  > To: mitgcm-support at mitgcm.org
  > Cc: Rui Ponte
  > Subject: Re: [MITgcm-support] Re: single-layer 
  > configuration problem update
  > 
  > 
  > Martin:
  > I don't know of any other way than EXF to introduce the 
  > variable forcing 
  > into the mitgcm. The resulting field is not exploding 
  > though, it stays 
  > equal to i.c.
  > Anyway, SINGLE_LAYER_MODE is probably what I was looking for in a 
  > context of my experiments.
  > Thank you,
  > Sergey
  >  
  > Martin Losch wrote:
  > 
  > > Sergey,
  > > that's really puzzling. Have you tried the same thing without the 
  > > EXF-package? (I have had unexpected behavior with this 
  > package before).
  > >
  > > However, if --- as you suspect --- no forcing fields 
  > introduced by EXF 
  > > is not "felt" by the model, then it wouldn't explode, but 
  > would just 
  > > keep quiet without moving at all. So I think, that on the 
  > contrary, 
  > > the model feels too much of the forcing; or something completely 
  > > different is going on.
  > >
  > > I can't remember your configuration, and obviously, I am 
  > too stupid to 
  > > search the MITgcm-support email list effectively so I 
  > didn't find your 
  > > previous emails, but were you using the CPP-flag 
  > SINGLE_LAYER_MODE 
  > > (kicks out most of thermodynamics)? However, even if you 
  > were using 
  > > it, I don't see where it should interfere with the EXF package.
  > >
  > > Martin
  > >
  > > On Aug 30, 2005, at 3:49 PM, Sergey Vinogradov wrote:
  > >
  > >>  Hi all,
  > >>
  > >>  I just want to update you on my quest in solving the 
  > problem with 
  > >> single-layer run under dynamic forcing (and maybe get 
  > some useful 
  > >> feedback from you).
  > >>
  > >>  The issue was erroneous computation of ocean dynamics 
  > (sea-level, 
  > >> bottom pressure) in a single layer configuration with a 
  > use of EXF 
  > >> package interface for the time-varying atm. pressure loading.
  > >>
  > >>  Appearently, it is not a problem of just pressure 
  > loading. It is far 
  > >> more general problem, as I tried similar experiment with 
  > wind stress. 
  > >> It looks like any forcing that is introduced through EXF 
  > in one-layer 
  > >> run, is not "felt" by the model.
  > >>
  > >>  Below is the exerpt from the model output with one 
  > layer under wind 
  > >> stress at the beginning of forward integration where you 
  > can see the 
  > >> error accumulating in cg solver with each step. Similar 
  > run with 23 
  > >> layers computes reasonable fields.
  > >>
  > >>  I'd appreciate any suggestions/comments/tips very much!
  > >>  Regards,
  > >>  Sergey
  > >>
  > >> (PID.TID 0000.0001) // 
  > >> =======================================================
  > >>  (PID.TID 0000.0001) // End MONITOR dynamic field statistics
  > >>  (PID.TID 0000.0001) // 
  > >> =======================================================
  > >>  (PID.TID 0000.0001)  MDSREADFIELD: opening global file: 
  > >> /anvil/scratch/day07/svv/forcing/NCEP_R6_ustr_1999
  > >>  (PID.TID 0000.0001)  MDSREADFIELD: opening global file: 
  > >> /anvil/scratch/day07/svv/forcing/NCEP_R6_ustr_1999
  > >>   cg2d: Sum(rhs),rhsMax =   5.83105612406154E-15  
  > 2.52789606828280E+01
  > >>   cg2d: Sum(rhs),rhsMax =   4.66381008031730E-09  
  > 2.03846096766287E+01
  > >>   cg2d: Sum(rhs),rhsMax =  -7.72491057666330E-11  
  > 1.63318316358843E+02
  > >>   cg2d: Sum(rhs),rhsMax =   7.82384650122037E-10  
  > 3.94671184216402E+02
  > >>   cg2d: Sum(rhs),rhsMax =   1.84482164570718E-10  
  > 1.39786242903642E+03
  > >>   cg2d: Sum(rhs),rhsMax =  -3.86795120889058E-10  
  > 5.51987500042437E+03
  > >>   cg2d: Sum(rhs),rhsMax =   5.60930842031203E-10  
  > 2.96468380272229E+04
  > >>   cg2d: Sum(rhs),rhsMax =  -4.60258786649532E-11  
  > 1.03144921478204E+06
  > >>   cg2d: Sum(rhs),rhsMax =   1.93486835808804E-12  
  > 3.16417950241384E+08
  > >>   cg2d: Sum(rhs),rhsMax =  -2.47236726757771E-12  
  > 2.35353664301718E+13
  > >>   cg2d: Sum(rhs),rhsMax =  -3.50420466366502E-16  
  > 2.44341352738558E+23
  > >>   cg2d: Sum(rhs),rhsMax =  -2.42051059455915E-16  
  > 1.68257839407874E+43
  > >>   cg2d: Sum(rhs),rhsMax =  -2.02970152146811E-16  
  > 1.42078931888867E+83
  > >>   cg2d: Sum(rhs),rhsMax =  -4.29685153894816E-16  
  > 9.27740314802081+162
  > >>   cg2d: Sum(rhs),rhsMax =                    NAN         
  >           INF
  > >>   cg2d: Sum(rhs),rhsMax =                    NAN  
  > 0.00000000000000E+00
  > >>   cg2d: Sum(rhs),rhsMax =                    NAN  
  > 0.00000000000000E+00
  > >>   cg2d: Sum(rhs),rhsMax =                    NAN  
  > 0.00000000000000E+00
  > >>  (PID.TID 0000.0001)  MDSREADFIELD: opening global file: 
  > >> /anvil/scratch/day07/svv/forcing/NCEP_R6_ustr_1999
  > >>   cg2d: Sum(rhs),rhsMax =                    NAN  
  > 0.00000000000000E+00
  > >>   cg2d: Sum(rhs),rhsMax =                    NAN  
  > 0.00000000000000E+00
  > >>
  > >
  > >
  > > _______________________________________________
  > > MITgcm-support mailing list
  > > MITgcm-support at mitgcm.org
  > > http://mitgcm.org/mailman/listinfo/mitgcm-support
  > 
  > 
  > _______________________________________________
  > MITgcm-support mailing list
  > MITgcm-support at mitgcm.org
  > http://mitgcm.org/mailman/listinfo/mitgcm-support
  > 




More information about the MITgcm-support mailing list