M=2*N is the criteria of triangular truncation Gaussian grids, such as NCAR cam AGCM T31(48*96), T42(64*128) , T85(128*256). For same rhomboidal truncation number of Gaussian grids, latitude resolution is higher than trianguler, such as GFDL AGCM R15, R30
Tn: grid_lon >= 3n + 1, grid_lat > (3n+1) / 2
Rn: grid_lon >= 3n + 1, grid_lat > (5n+1) / 2 (n is truncation number)
Model: Typical Grid:
T42 128 x 64
T63 192 x 96
R30 96 x 80
R42 128 x 108
our model is rhomboidal truncation 42 (R42) resolation AGCM, I want to convert spherical harmonic coefficients to grid space by NCL, but I'm puzzled about 'shsgC' input array form. My spectral coefficients array is (2,43*43), I don't know how to turn it to (2,nlat,N), then get grid space (nlat,nlon) by NCL function 'shsgC'. (nlat=108,nlon=128).
======= 2005-05-28 13:32:00 您在来信中写道：=======
>shagC is for Gaussian grids. All Gaussian grids we deal with have
>the characteristic that the number of longitudes [M] is twice the size
>of the number of latitudes [N]. ie M=2*N
>Your example grid does not satisfy that criteria.
>Is 108x128 a real world gaussian grid?
>I wrote a simple script to test romb_trunc and tri_trunc.
>Run the script. It seems to work ok.
>Note: non-gaussian grids are assumed to have pole points.
>So you would use shaeC and a grid might be shaped like
>N=73 and M=144.
ncl-talk mailing list
This archive was generated by hypermail 2b29 : Tue May 31 2005 - 08:13:59 MDT