[MITgcm-support] CD scheme/OBCS

[Jean-Michel Campin]

Hi Holly,

Just a short comment: there are more than 6 experiments which do not use CD-scheme,
(among the coarse res: global_ocean.cs32x15, tutorial_baroclinic_gyre, 
tutorial_global_oce_in_p) since the default is useCDscheme=F, so that when 
it's not explicitly set, it's not used.
But I agree that probably too many tested exp use it.

Cheers,
Jean-Michel

On Tue, Jan 31, 2012 at 09:33:22AM -0500, Holly Dail wrote:
> Hi Martin -
> 
> Thanks for the advice.  I know the CD scheme is not favored, and there are many other options.  I've tried a wide variety of schemes, but I'm somewhat stuck with trying the CD scheme for legacy reasons (I'm trying to approximate the results of an older model run that used the CD scheme).  Jean-Michel gave me a patch to try, I'm currently testing it.
> 
> Its interesting that there are so many verification experiments that use the CD scheme, given that it is out of favor.  An extremely rough count (using grep useCD */*/data | grep TRUE | wc -l) in the  verification directory shows 36 examples with CD scheme on, 6 with it off.  Of the 'off' experiments, there are none that are standard forward model 3-D configs:
> 
> 1D_ocean_ice_column/input/data: useCDscheme=.FALSE.,
> 1D_ocean_ice_column/input_ad/data: useCDscheme=.FALSE.,
> OpenAD/input/data: useCDscheme=.FALSE.,
> OpenAD/input_ad_singlelayer/data: useCDscheme=.FALSE.,
> tutorial_cfc_offline/input/data: useCDscheme=.FALSE.,
> tutorial_dic_adjoffline/input_ad/data: useCDscheme=.FALSE.,
> 
> Thanks again for your advice -
> Holly
> 
> On Jan 31, 2012, at Jan 31 , 5:57 AM, Martin Losch wrote:
> 
> > Hi Holly,
> > 
> > the CD scheme may not be what you want for stability? Did you try using some bi-harmonic viscosity instead?
> > 
> > The CD scheme adds viscosity of unknown size (but depends on the coupling parameter) and is thus hard to control, therefore is it not very popular (even) with its creator.
> > 
> > Martin
> > 
> > On Jan 30, 2012, at 11:29 PM, Jean-Michel Campin wrote:
> > 
> >> Hi Holy,
> >> Will take a look (but I don't promise anything).
> >> Cheers,
> >> Jean-Michel
> >> 
> >> On Mon, Jan 30, 2012 at 03:36:31PM -0500, Holly Dail wrote:
> >>> Hello -
> >>> 
> >>> Has anyone implemented the CD scheme with OBCS?  I see comments on this list from Martin from 2006 saying it shouldn't be too hard and giving some hints on what needs to be done.  Has anyone tried it?  Have any advice?
> >>> 
> >>> As to where I'm at, I've commented out the stop in OBCS and built the code with both OBCS and cd_code compiled in.  It blows up after about 90 days in my setup, not surprisingly since I haven't made any fixes to the code yet.  My setup is Atlantic basin lat/lon grid at 1 degree with a single open boundary at the southern end.
> >>> 
> >>> Thanks in advance for any advice -
> >>> Holly
> >>> _______________________________________________
> >>> 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
> > 
> > 
> > _______________________________________________
> > 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