Re: [ferret_users] Converting from one polar stereographic grid (curvilinear) to another

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

Hi Rob,
Ferret has just the one set of tools, curv_to_rect_map, curv_to_rect, 
and rect_to_grid for regridding to or from curvilinear grids. If you 
have a way of doing the computation you want in Fortran or C, we could 
help you turn that into a Ferret external function.

Note there are also the functions SAMPLEXY_CURV, which you could 
possibly use to just sample data from the fine grid at the desired 
locations on the coarse grid.  Maybe give that some thought.

Ansley

Rob Briggs wrote:
> Good Morning,
>
> I'm pre-processing some data and wish to regrid from a high resolution
> polar stereo grid (curvilinear) to another lower resolution grid (my
> model is for Antarctica so all my data is in polar stereo). 
>
> My target grid, from fort.92.40.nc, is in km
> My source grid, source.nc, is in m 
>
> yes? use fort.92.40.nc
> yes? show grid H[d=1]
>     GRID GOZ3
>  name       axis              # pts   start                end
>  X1        X (km)             141 r   -2800                2800
>  Y1        Y (km)             141 r   -2800                2800
>  normal    Z
>  TIME      T (years)            1 r   -29999               -29999
>
> yes? use source.nc
> yes? show grid TEMP[d=2]
>     GRID GPX1
>  name       axis              # pts   start                end
>  X11       X (meter)         1160 r   -2797500             2997500
>  Y11       Y (meter)         1120 r   -2797500             2797500
>  normal    Z
>  normal    T
>
> And then I want to do something like this, 
> yes? let var=TEMP[g=H[d=1]]
>
> Then I would have my regridded variable. If I had a way to convert grid
> GOZ3 in km to a grid in m I think I would be there. Is there away to do
> this (without using curv_to_rect, I believe I could it with
> curv_to_rect, but I'm hoping there might be a way to do it without,
> curv_to_rect is very slow and I've not had sucess with it yet on
> continental scale data). 
>
> regards,
>
> Rob Briggs
>
>