Re: [ferret_users] interpolation

[Ansley Manke <ansley.b.manke@xxxxxxxx>]

Hi David,
By default Ferret assumes that cell edges are midway between grid 
points, in the absence of other information. Using the concept of cell 
boundaries/edges one can define the axis in whichever way you like, 
however in this case it seems that Ferret's definition of the cells on 
the axis is the one you already want:

yes? use levitus_climatology

yes? let zz = z[gz=temp]
yes? let zlo = zboxlo[gz=temp]
yes? let zhi = zboxhi[gz=temp]

yes? list zz, zlo, zhi, zz-zlo, zhi-zz
            DATA SET: 
/home/porter/tmap/fer_dsets/data/levitus_climatology.cdf
            DEPTH (m): 0 to 5000
Column  1: ZZ is Z[GZ=TEMP]
Column  2: ZLO is ZBOXLO[GZ=TEMP]
Column  3: ZHI is ZBOXHI[GZ=TEMP]
Column  4: EX#4 is ZZ-ZLO
Column  5: EX#5 is ZHI-ZZ

                ZZ    ZLO    ZHI   EX#4   EX#5
0       /  1:     0.     0.     5.    0.0    5.0
10      /  2:    10.     5.    15.    5.0    5.0
20      /  3:    20.    15.    25.    5.0    5.0
30      /  4:    30.    25.    40.    5.0   10.0
50      /  5:    50.    40.    62.   10.0   12.5
75      /  6:    75.    62.    88.   12.5   12.5
100     /  7:   100.    88.   125.   12.5   25.0
150     /  8:   150.   125.   175.   25.0   25.0
200     /  9:   200.   175.   250.   25.0   50.0
300     / 10:   300.   250.   350.   50.0   50.0
400     / 11:   400.   350.   500.   50.0  100.0
600     / 12:   600.   500.   700.  100.0  100.0
800     / 13:   800.   700.   900.  100.0  100.0
1000    / 14:  1000.   900.  1100.  100.0  100.0
1200    / 15:  1200.  1100.  1350.  100.0  150.0
1500    / 16:  1500.  1350.  1750.  150.0  250.0
2000    / 17:  2000.  1750.  2500.  250.0  500.0
3000    / 18:  3000.  2500.  3500.  500.0  500.0
4000    / 19:  4000.  3500.  4500.  500.0  500.0
5000    / 20:  5000.  4500.  5000.  500.0    0.0

David Wang wrote:

> Hey all,
>
> Here I have a question that may not be very relevant, but...
>
> I used Levitus climatology as the initial hydrographic fields for a 
> z-coordinates model. My preprocessings include interpolating Levitus 
> onto the model grid which generally is denser than Levitus' in the 
> upper ocean and intact in depths. This was done in Ferret using simple 
> implicit linear interpolation associated with regridding, something like
>
> let tmp = theta[GX=newlon,GY=newlat];
> let newtheta = tmp[GZ=newdepth];
>
> I recently found that in Levitus dataset cell edges reside at the 
> middle points between two adjacent grid points on z coordinates, not 
> the opposite (i.e., grid points sit at the midpoints between two 
> edges). I don't know if Ferret treats the two coordinate arrangements 
> differently concerning regridding interpolation, or does it matter or 
> not. Furthermore is there better (or more accurate) way to prepare 
> these initial data?
>
> Thanks, David