[MITgcm-devel] problems with llc grid?

[Ed Hill]

On Wed, 23 May 2007 17:15:00 +0200 Martin Losch wrote:
> On 23 May 2007, at 16:54, Ed Hill wrote:
> >
> > The dyC values match at the edges (as they should!) but these edge
> > values are higher than they "ought to be" based on a smoothness
> > argument.  That is, the adjacent dyC lengths within each tile are
> > appreciably smaller than the dyC values right at the edges.
> >
> > That's the problem, right?

> Yes, but
> 1. I don't know if it is really a "problem"
> 2. it also affects rAs, therefore I do not understand what you mean  
> by "balance between un-even grid lengths and un-even grid cell  
> areas". To me it looks like dy[C,F,U,G] is quite large along the
> edge of face3 and AS A CONSEQUENCE rAs is also quite large along the
> same edge. Where is the balance? According to your description I
> would expect that as dyC gets smoother, the cell areas get less
> smooth. Don't misunderstand me: I am not saying this has got to be a
> problem. I am just wondering if this is what you expect. (please have
> a look at eddy.csail.mit.edu:~mlosch/llc/grid.t009.nc and the other
> files to see what I mean).
> If you think that this is OK, I'll shut up and live with this grid.
> 
> I am not sure that this feature of the grid is actually causing the  
> problems I am having. There are so many other possibilities.


Hi Martin,

You're absolutely right that making the lengths of adjacent cells more
similar tends to make their areas more similar.  This is something that
one can certainly do by (gently) adjusting the edge spacing within
the grid generation scripts.

The "trade-off" that I was thinking about is what happens at the
corners.  For global cube grids, there is a trade-off between making
the corner grid cells larger and the smoothness at the tile edges.
Since the llc has its corners effectively removed from the domain (over
land), then it probably does not matter how small the corner cells
become.  So, you have a good point!

I think the llc grid you are using can certainly be improved from a
smoothness standpoint.  How much?  I don't know.  But, with some 
effort, one can indeed tune the spacing a bit to make the areas and the
lengths smoother at the edges and at transition between grid types
(lat-lon to cap).

I'm very sorry but I won't have free time to work on grid generation
for the next ~2 months.  If someone would like to dig further into
these scripts to make some adjustments for smoothing then I could meet
with them one weekend and try to give them a "running start"...

Ed

-- 
Edward H. Hill III, PhD  |  ed at eh3.com  |  http://eh3.com/
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
URL: <http://mitgcm.org/pipermail/mitgcm-devel/attachments/20070523/7c4dcaf8/attachment.sig>