[MITgcm-devel] TUV.mat

[Jean-Michel Campin]

Hello Ed,

I don't know how those TUV.mat files have been generated.
(see my e-mail to the devel list on March 31, 2005).
I am now using a kind of "average" rotating angle
(base on one of those TUV.mat) with only 2 matrix
(instead of 4) that correspond exactly to the cos & sin
of a unique rotation angle.
you can have a look at MITgcm/utils/cs_grid/README

See you,

Jean-Michel

On Thu, Jul 21, 2005 at 11:35:17AM -0400, Ed Hill wrote:
> 
> Hi Alistair, Jean-Michel, & others,
> 
> Do you folks know where to find the script(s) used to generate the
> "TUV.mat" file for the uvcube2latlon() function?  I know J-M was looking
> into this recently but I didn't see a "final" answer.
> 
> I did locate three different copies of the file in CVS:
> 
>   MITgcm/verification/global_ocean.cubed32x32x30/matlab/TUV.mat
>   MITgcm/verification/solid-body.cs-32x32x1/input/TUV.mat
>   MITgcm_contrib/enderton/Diagnostics/DiagUtility/TUV.mat
> 
> where the 1st and 3rd are identical and have a MatLAB header that
> reports:
> 
>   MATLAB 5.0 MAT-file, Platform: GLNX86, 
>     Created on: Wed May  1 07:40:10 2002
> 
> and the 2nd contains four arrays used for vector rotations that are up
> to 7% different than the previous two and has a header reporting:
> 
>   MATLAB 5.0 MAT-file, Platform: LNX86, 
>     Created on: Fri May 11 18:52:01 2001
> 
> Is one "better" than the other?  The up-to-7% difference seems like
> something that shouldn't be ignored.  So it would be really nice to know
> how it was/is/ought-to-be generated.
> 
> Ed
>