[MITgcm-support] Format of curvilinear grid files

[Jean-Michel Campin]

Hi Christopher,

In fact, the additional row and column are only used to read in
the 2 "missing corners" for grid-variable located at the
grid cell corner (e.g., xG, yG, rAz ...) when using a cubed-sphere
topology.
What I refer by "missing corners" (at a face South-East corner i,j=Nx+1,1
and at a face North-West corner i,j=1,Ny+1) is related to the fact that 
on a cube, we have 2 more corners points than we have grid cell center.

Cheers,
Jean-Michel

On Mon, Jul 08, 2013 at 02:10:51PM +0200, Martin Losch wrote:
> To be honest: I don't know exactly, but from looking at model/src/ini_curvilinear_grid.F I assume that it is just discarded and replaced in the "call exch_*_rs" code.
> 
> An alternative way of specifying a curvilinear grid is with OLD_GRID_IO defined (see model/src/ini_curvilinear_grid.F for details), which will require only (Nx,Ny) fields, or, if you are bold (o:, do it the netcdf way.
> 
> Martin
> 
> On Jul 3, 2013, at 7:10 PM, Christopher L. P. Wolfe <clwolfe at ucsd.edu> wrote:
> 
> > Thanks Martin. Do you know what the model ends up doing with this extra information?
> > 
> > Christopher
> > 
> > On Jul 3, 2013, at 12:28 AM, Martin Losch <Martin.Losch at awi.de> wrote:
> > 
> >> Hi Christopher,
> >> 
> >> for the "G"-fields (XG,YG, etc). the values of the extra points should be pretting clear (i.e.  the N/E coordinates of the sNy/sNx'th cell). For the unclear values it's probably enough to insert some dummy values
> >> On Jul 3, 2013, at 6:28 AM, "Christopher L. Wolfe" <clwolfe at ucsd.edu> wrote:
> >> 
> >>> Hi all,
> >>> 
> >>> I've been trying to set up an experiment on a curvilinear grid with pre-computed grid spacings, areas, and angles. I thought I had written out all the grid descriptors in the proper format (18 descriptors, big endian, 64-bit), but I kept having the model try to read past the end of my grid file and crashing. Some poking around on the MITgcm website brought me to the Held-Suarez-on-a-cube-sphere tutorial (http://mitgcm.org/public/r2_manual/latest/online_documents/node137.html) where it mentions that the grid faces are 32x32 but the grid files contain 18 33x33 fields. This made me think that I totally don't understand what's supposed to be in the grid files.
> >>> 
> >>> Can anyone confirm that the grid files are indeed supposed to contain one extra point in each direction? If so, what defines the values of the grid descriptors on that extra point?
> >>> 
> >>> In any event, if anyone can point me to a matlab script or other such thing with an example of writing out curvilinear grid descriptors, I'd very much appreciate it.
> >>> 
> >>> Thanks!
> >>> Christopher
> >>> 
> >>> -----------------------------------------------------------
> >>> Dr. Christopher L. P. Wolfe
> >>> Research Oceanographer
> >>> Climate, Atmospheric Science, and Physical Oceanography
> >>> Scripps Institution of Oceanography, UCSD
> >>> clwolfe at ucsd.edu                           858-534-4560
> >>> -----------------------------------------------------------
> >>> 
> >>> 
> >>> 
> >>> 
> >>> 
> >>> _______________________________________________
> >>> 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