Re: [ferret_users] Regird tripolar grid

["William S. Kessler" <william.s.kessler@xxxxxxxx>]

This is very good advice (thanks, Russ).

As the link says "In general, regridding is a difficult task to  
perform accurately and without producing noise or spurious results."  
This discussion is useful and thoughtful, and applies not just to  
model output. One cannot assume that regridding done by simple  
interpolation (as in the first message below) will result in fields  
that conserve properties, especially products of properties (like  
fluxes), derivatives or integrals. Computation should be done as much  
as possible on the model's native grid, then regridding done to make a  
plot.

Ferret's CURV_TO_RECT function implements the GFDL algorithm, and is  
better than interpolation, but better still is working on the native  
grid until it is time to make a plot. And checking is always necessary.

Billy K

On Mar 6, 2013, at 15:57 , Russ Fiedler wrote:

> Hi,
>
> The MOM distribution includes tools for the postprocessing of model  
> data
> on tripolar grids onto regular lat/lon grids. You need the underlying
> grid_spec.nc file and need to create an exchange grid for your 2  
> degree
> resolution grid so that the interpolations are done correctly.
>
> See the file
>
> src/postprocessing/regrid/regrid.F90
>
> in your MOM distribution. See also
>
> http://data1.gfdl.noaa.gov/~arl/pubrel/r/mom4p1/src/mom4p1/doc/mom4_manual.html#Postprocessing%20regrid%20tool
>
>
>
> If you are using the newer MOSAIC style of grid file I think you want
> the fregrid tool in
>
> src/tools/fregrid
>
> Russ
>
>
>
>
> On Thu, 2013-03-07 at 09:27 +1100, Ansley Manke wrote:
> > Hi,
> > Ferret needs special handling for these grids. Grids like the  
> > tripolar
> > grid are called curvilinear grids in the Ferret documentation. So  
> > check
> > under "curvilinear" or "regridding, curvilienar" in the documentation
> > index for the CURV_TO_RECT and CURV_TO_RECT_MAP function.
> >
> > These functions are correct but slow. We have some newer ones on the
> > horizon, using methods and code from the Earth System Modeling  
> > Framework.
> >
> > Ansley
> >
> > On 3/6/2013 1:34 PM, Boyin Huang - NOAA Federal wrote:
> > > Hi Ferreters,
> > > Can we regrid tripolar grid data from GFDL MOM model outputs (NetCDF
> > > format) to regular 2x2 grid using ferret?
> > > I used the following script that failed in the Arctic region
> > > use tos_day_GFDL-ESM2G_historical_r1i1p1_18610101-18651231.nc
> > > SET MEMORY/SIZE=320
> > > define axis/x=0e: 358e:2/unit=degree xlon
> > > define axis/y=88S:88N: 2/unit=degree ylat
> > > define grid/x=xlon/y=ylat mygrid
> > > list/file="tos.dat"/format=stream/L=1:1825/clobber tos[g=mygrid]
> > >
> > > Any suggestions are welcomed.
> > > Thanks
> > >
> > > --
> > > Boyin Huang
> > > Ingest and Analysis Branch
> > > Global Climate Application Division
> > > NCDC/NOAA, Room-514
> > > Phone 828-271-4851
> > > Fax 828-271-4022, 828-271-4246