[MITgcm-devel] TUV.mat

[Jean-Michel Campin]

Hi Alistair,

I have a question about the TUV.mat file that is used in uvcube2latlon.m
(there is one copy of this TUV.mat file in 
verification/global_ocean.cubed32x32x30/matlab/)

It contains 4 fields: TUu, TUv,  TVu, & TVv
and to rotate a vector from the CS grid onto a lat-lon referential,
I would expect the folowing:
 TUu= cos(alpha)
 TUv=-sin(alpha)
 TVu= sin(alpha)
 TVv= cos(alpha)
where alpha is the angle between the E-W direction 
and the CS X-direction at the center of the tracer-cell (equal to the
angle between the S-N direction and the CS Y-direction).
But when I conpute TUu-TVv or TUv+TVu,
I don't get exactly zero, with some differences ~ 2.24.e-3 near
some corners, and not the same patern for all the 4 equatorial faces.

Is there something I am missing ?

Thanks,

Jean-Michel