[MITgcm-support] Finale: north-south or south-north gridding?

jschwarz at awi-bremerhaven.de jschwarz at awi-bremerhaven.de
Thu Sep 21 16:55:23 EDT 2006


I've repressed the urge to rebel against convention and regridded everything from south to north.. 
Thanks for a very enjoyable discussion (:
 jill 

ps. In case any of you decides to take up remote sensing, be warned that we plot things from north to south with 0,0 in the top left corner!

----- Original Message -----
From: Baylor Fox-Kemper <baylor at mit.edu>
Date: Friday, September 22, 2006 3:37 am
Subject: Re: [MITgcm-support] north-south or south-north gridding?

> Dear MITgcmers (especially Jill, Martin and Dimitris)
> 
> Beware the pseudo-vector!
> 
> While true vector operations, e.g., grad, div, addition, scalar  
> multiplication, survive a mirror reflection unaffected x,y,z->-x,-
> y,- 
> z (and most physics does, too, except for elementary particle, weak 
> 
> force interactions, see http://lhcb-public.web.cern.ch/lhcb-public/ 
> html/cpviolationtoc.htm), a pseudo-vector changes sign under mirror 
> 
> reflection!  How do you get a pseudo-vector? you take a cross- 
> product, or curl.  So, VORTICITY is a PSEUDOVECTOR, and so is f.  
> The  
> problem is, the momentum and tracer equations are rotation and  
> reflection invariant, but this is only because there are doubled 
> sign  
> reversals, i.e., the coriolis parameter should change sign but the  
> coriolis acceleration shouldn't!
> 
> So, there are very tricky reconstructions to make when you start  
> messing with mirror reflections, as some things should change sign  
> and others don't.  It is very easy to get messed up!  I would  
> recommend against running a model in such a way that you have to 
> flip  
> the sign of some things and not others unless you are very 
> confident  
> that you know the sign of everything that's supposed to come out.   
> You can really get messed up with potential vorticity and  
> streamfunctions, for example, where you will have to flip the sign 
> of  
> not only f, but also curl v, etc.  You have to remember to use a  
> right-hand rule or a left-hand rule depending on whether you've  
> reflected or not.
> 
> Cheers, and good luck if you try!
>    -Baylor
> 
> 
> On Sep 21, 2006, at 11:10 AM, Martin Losch wrote:
> 
> >
> > On Sep 21, 2006, at 4:13 PM, Dimitris Menemenlis wrote:
> >
> >> Hi Jill, any MITgcm "feature" that has not been tried before (or 
> 
> >> recently) is unlikely work OK the first time around.  There will 
> 
> >> possibly/probably be one or two places in the code where a  
> >> particular MITgcm contributor was not thinking of NZ at time of  
> >> addition.  What about running on a +50 to +80 grid but then  
> >> looking at "that" solution through a mirror or, equivalently I  
> >> think, plotting it on a -80 to -50 grid.  Would that work?  
> Dimitris>
> > I think that's perfect! That way, even the direction of the  
> > coriolis force would "look" right in the mirror (o:
> >
> > M.
> >
> >>
> >>
> >> _______________________________________________
> >> 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
> 



More information about the MITgcm-support mailing list