[MITgcm-support] EXCH2 grid parameter question clarification

[Jean-Michel Campin]

Hi,

I think I understand  the problem: without blank tiles,
my previous answer might work with exch2. However, with blank
tiles, the coordinates (and the default coriolis set-up) will 
not be right.
This needs to be fixed (can't promise something right now, 
but it's on the list), and in the mean time,
as a temporary fix, and if you are not using exf-interpolation,
could try to specify coriolis from files:
setting (in file "data", namelist PARM01):
 selectCoriMap=3,
and providing files fCoriC.bin, fCoriG.bin & fCorCs.bin
with Coriolis parameter at grid-cell center, at grid-cell corner
and f'=2.Omega.cos(Phi) for the 3rd one.

Cheers,
Jean-Michel

On Wed, Nov 02, 2011 at 06:29:29PM -0400, Jean-Michel Campin wrote:
> Hi Michael,
> 
> I did not look into details (will have to check this),
> but you don't need to put all the tile in X dir
> (it has been like this for most of your CS-grid example 
> which use pkg/exch2, but it's not stricktly necessary).
> There is an example which use pkg/exch2 (but not useCubedSphereExchange):
> in verification/MLAdjust, see code/SIZE.h and code/SIZE.h_mpi
> This might well fix your problem.
> 
> Cheers,
> Jean-Michel
> 
> On Wed, Nov 02, 2011 at 10:51:43PM +0200, Michael Siccha wrote:
> > Dear all,
> > first of all I am sorry for having formulated my question so awkwardly.
> > I will be more precise this time.
> > I am running a regional model of the Red Sea using MPI and (among
> > others) the EXCH2 package in a single facet configuration. My grid is
> > subdivided into 32 tiles, 8 in the horizontal and 4 in the vertical,
> > comprising the area between ~ 10°N,32°E and ~ 30°N,47°E. Due to the
> > geometry of the Red Sea I have 17 blank tiles. In order to conform with
> > the EXCH2 package my SIZE.h file declares the 15 wet tiles as 15
> > processes in x direction, as nPx=15. The model runs nicely, the output
> > files show a geometrically correct Red Sea and the data obtained make
> > sense on the first glance (not much data yet), but when I read the
> > automatically generated grid parameter files ('XC', 'YC',...) the
> > latitudinal coordinates of ALL tiles lie in the range of 10°N and 15°N,
> > that is the lowermost row of my grid (as declared in SIZE.h), while they
> > should of course span the full range between 10°N and 30°. Similarly are
> > the grid cell face areas all computed with these coordinates and only
> > correct in the lowermost row of my grid. The listing of grid parameters
> > in the output text file of the executable is the same as in the
> > described files. No type of output method (useSingleCpuIo, GlobalFiles,
> > or the standard setting of individual files for each tile) changes the
> > situation. 
> > So my question is whether there is way to obtain the correct global grid
> > cell parameters as output while using EXCH2 ?
> > I am using the 'rA' variable in some code modifications and am (just
> > slightly) worried whether the model uses the correct values...
> > Thanks in advance
> > Michael Siccha
> > 
> > 
> > _______________________________________________
> > 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