[MITgcm-support] Questions on curvilinear grid

Martin Losch Martin.Losch at awi.de
Mon May 26 03:12:38 EDT 2008


Luca and Gianmaria,

the angle is the angle between geographical north and "grid  
north" (that is the y/j-direction of the grid). In "model/src/ 
rotate_spherical_polar_grid.F" there is a routine that computes these  
angles and the corresponding parameters cos(angle)/sin(angle) on C- 
points (angleCosC/angleSinC). I assume that you could use this  
routine to compute the angles "on the fly" rather than reading them  
in, but I never tried this in a non-spherical grid yet. Let me know  
if it works and we could consider calling this routine from  
ini_curvilinear_grid, if they are not provided "by the user".

Unfortunately, there is no verification experiment for the  
curvilinear grid (that is not a cubed-sphere grid), but have a look  
at this example (that does not use netCDF for input):
<http://dev.mitgcm.org/cgi-bin/viewcvs.cgi/MITgcm_contrib/arctic40km/? 
only_with_tag=MAIN>
That should give you an example of how to do it (with the CPP-Flag   
OLD_GRID_IO)

Martin

On 16 Mar 2008, at 18:32, Luca Liberti wrote:

> Hi,
> we are trying to make a simulation using a curvilinear orthogonal  
> grid in spherical coordinates.
> Starting from the grid nodes coordinates computed by my meshing  
> program we calculate
> all the relevant grid parametrs  such as  Ac,Ag,As,Aw, dxc,dyc  
> etc.and put them in a
> NetCDF file.  However it is not clear to us what the grid  
> orientation parameters
> angleCosC and angleSinC mean. What is the angle they are refferring  
> to?
> Our best guess would be the one between the positive x direction on  
> the grid (i to i+1) and the
> geographic north. Are we right?
> Also, having nx x ny nodes in the grid, most of the grid parameters  
> contain less elements.
> For instance Ac is defined in (nx-1)x(ny-1) positions and Aw in  
> (nx-2)x(ny-1).
> Since all the arrays have the same dimensions shall we put the  
> resulting values starting from
> element I=1 and J=1 (meaning that all the missing values will be  
> located at the I and J upper
> limits)or not ?
>
> Thanks in advance for your help
> Luca Liberti and Gianmaria  
> Sannino<luca.liberti.vcf>_____________________________________________ 
> __
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-support




More information about the MITgcm-support mailing list