[MITgcm-support] Re: OBCS sponge and prescribe code checked in

Tue Sep 21 11:11:37 EDT 2004

Martin,

Quoting Martin Losch <mlosch at awi-bremerhaven.de>:

> Hi Patrick,
> thanks for the code, in particular for the sponge layer code!
> Here come more stupid questions:
> 1.Shouldn't the call ext_getobcs in obcs_calc.F come after the if  
> (useOrlanski???) then, else, endif? If I am not mistaken, then the  
> prescribed values of exf_getobcs get overwritten either by the  
> orlanski_???? or the alternative (if useorlanski????=.false.). I don't  
> know what to do with balance code then, should probably at the very  
> end, unless the prescribed values are never supposed to be balanced.

You're totally right, I was too quick in merging
3 different versions.
If the call exf_getobcs is used, then the IF (orlanski...) stuff
should not be used.
In that case, yes, the balancing needs to get done after everything
has been assigned. I guess I'll have to redo this.

> 2. need the read-in boundary fields be exchanged? (in  
> external_fields_load and all other exf_getforcing routines the forcing  
> fields are exchanged) I am aware that there are no exchange routines  
> for xz and yz slices, but is it save for more than 1 CPU the way it is  
> now? Just asking ...

Although the obcs fields ougth to be exchanged in principle, 
there are no xz-/yz exchanges, and no plan to do these.
The thing to do is to exchange the full fields
once the obcs have been added to these.
I forgot where this is done, will look.

-Patrick

> On Sep 21, 2004, at 1:38 AM, Patrick Heimbach wrote:
> 
> >
> > Hi Martin, Uli, and others,
> >
> > I've just checked in code in pkg/obcs/ that
> > * ALLOW_OBCS_PRESCRIBE allows prescibing time-dependent obcs values,
> > * ALLOW_OBCS_SPONGE sponge layers
> > * ALLOW_OBCS_BALANCE rudimentary balancing of normal barotropic flow
> >
> > tag is checkpoint55a_post
> >
> > The stuff I checked in at least doesn't break
> > the existing verif.'s that use obcs ;o)
> > The code has been / is used in a few open boundary setups,
> > it compiles ok, but no guarantee that the merging may have
> > broken some of the new stuff.
> > Any feedback on bugs, improvements etc are welcome.
> >
> > One feature/caveat is that the time-dependent prescibe
> > runs in conjunction with exf, cal., i.e.
> > S/R OBCS_CALC calls S/R EXF_GETOBCS
> > If someone wants to volunteer to call
> > external_fields_load alternatively, please.
> >
> > The files read are
> > xz files (northern, southern boundary)
> > yz files (eastern, western boundary)
> >
> > I've put a few "template" CPP options and runtime parameter files under
> > http://mitgcm.org/cgi-bin/viewcvs.cgi/MITgcm_contrib/heimbach/ 
> > obcs_prescribe/
> >
> > A verif. experiment will hopefully come soon.
> >
> > -Patrick
> >
> >
> >
> > Quoting Patrick Heimbach <heimbach at MIT.EDU>:
> >
> >>
> >> Hi Martin,
> >>
> >> there is code for sponge layers and code for reading
> >> 2D (xz, yz) boundart slices (was part of ECCO).
> >> I wanted to merge it with the main code last week,
> >> but didn't have time since I also had to catch up
> >> with the adjoint (-> c55).
> >> Will try it in the first half of this week
> >> (remind me ;o).
> >>
> >> -Patrick
> >>
> >> Quoting Martin Losch <mlosch at awi-bremerhaven.de>:
> >>
> >>> Hi,
> >>> now the obcs-package has got me. In the global_ocean.90x40x15
> >>> experiment, I would like to introduce open boundaries in the south
> >>> and/or north to emulate polar regions. Now the simplest and dirtiest
> >>> way to do it would be through restoring in external_forcing. But I  
> >>> want
> >>> to do it the hard and clean way and use the obcs package (o: A few
> >>> questions arise:
> >>> 1) as a general rule/how-to I start by modifying obcs_calc, so that
> >>> after calling obcs_calc, the boundary values at time myTime+deltaT  
> >>> are
> >>> known. Why myTime+deltaT, why not the current time=myTime?
> >>> 2) ALLOW_OBCS_SPONGE: Is there code for sponge layers, or am I
> >>> completely responsible for it myself (subroutine  
> >>> obcs_sponge_u/v/t/s)?
> >>> As far as I know, a sponge layer is not much more than a restoring
> >>> zone, correct? If I use a sponge layer (#define ALLOW_OBCS_SPONGE),
> >>> will I still have an open boundary? Or in other words: just having
> >>> restoring in external_forcing.F is not the same as ALLOW_OBCS_SPONGE
> >>> and do the restoring somewhere in obcs_sponge_u/v/t/s, etc? It seems
> >>> so.
> >>> 3) not quite OBCS, but related. I want to read my open boundary  
> >>> values
> >>> from a file (in calc_obcs), so I have to create code for that, fair
> >>> enough; but because I only want to read one xz-slice, for example,
> >>> (there is even an MDSIOREADFIELDXZ, impressive), I have a problem  
> >>> with
> >>> exchanging the xz-slice, or did I just not find the excange_xz_r?  
> >>> file?
> >>> Does anyone have experience with this?
> >>>
> >>> Martin
> >>>
> >>> _______________________________________________
> >>> MITgcm-support mailing list
> >>> MITgcm-support at mitgcm.org
> >>> http://dev.mitgcm.org/mailman/listinfo/mitgcm-support
> >>>
> >>
> >>
> >> --------------------------------------------------------
> >> Patrick Heimbach   Massachusetts Institute of Technology
> >> FON: +1/617/253-5259                  EAPS, Room 54-1518
> >> FAX: +1/617/253-4464             77 Massachusetts Avenue
> >> mailto:heimbach at mit.edu               Cambridge MA 02139
> >> http://www.mit.edu/~heimbach/                        USA
> >>
> >> _______________________________________________
> >> MITgcm-support mailing list
> >> MITgcm-support at mitgcm.org
> >> http://dev.mitgcm.org/mailman/listinfo/mitgcm-support
> >>
> >
> >
> > --------------------------------------------------------
> > Patrick Heimbach   Massachusetts Institute of Technology
> > FON: +1/617/253-5259                  EAPS, Room 54-1518
> > FAX: +1/617/253-4464             77 Massachusetts Avenue
> > mailto:heimbach at mit.edu               Cambridge MA 02139
> > http://www.mit.edu/~heimbach/                        USA
> >
> 
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://dev.mitgcm.org/mailman/listinfo/mitgcm-support
> 

--------------------------------------------------------
Patrick Heimbach   Massachusetts Institute of Technology
FON: +1/617/253-5259                  EAPS, Room 54-1518
FAX: +1/617/253-4464             77 Massachusetts Avenue
mailto:heimbach at mit.edu               Cambridge MA 02139
http://www.mit.edu/~heimbach/                        USA