Re: Comparing Two Curvilinear Variables

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

Hi -
As it so often happens, a topic comes up from many people at about
the same time.  The issue of data on curvilinear grids is one we are
actively working on in Ferret development, and some initial tools for
working with these grids will be available in a release at the end of this
summer. The first step will be a function to regrid data from a curvilinear
grid to a recangular grid of the user's choice and do comparisions there. 
We will also have a function to regrid from a rectangular to a 
curvilinear grid.

Ansley Manke

Daryl Herzmann wrote:

> Hi Ferret!
>
> I am in a dill of a pickle and have been trying to think like a Ferret 
> without much luck.  Perhaps a kind person could help me out with a 
> documentation reference I have probably missed :(
>
> I have MM5 model forecasted precip on one curvilinear grid.  The NetCDF 
> highlights are:
>
>    float rain_non(time, i_cross, j_cross) ;
>        rain_non:long_name = "ACCUMULATED NONCONVECTIVE PRECIPITATION" ;
>        rain_non:units = "cm" ;
>    float latitcrs(i_cross, j_cross) ;
>        latitcrs:long_name = "LATITUDE (SOUTH NEGATIVE)" ;
>        latitcrs:units = "degrees" ;
>    float longicrs(i_cross, j_cross) ;
>        longicrs:long_name = "LONGITUDE (WEST NEGATIVE)" ;
>        longicrs:units = "degrees" ;
>
> And I have NCEP stage4 precip data on a different curvilinear grid. Again, 
> the NetCDF highlights.
>
>    float APCP_SFC(YCells, XCells) ;
>        APCP_SFC:long_name = "Total precipitation" ;
>        APCP_SFC:units = "inch" ;
>    float longitude(YCells, XCells) ;
>        longitude:long_name = "longitude" ;
>        longitude:units = "degrees_east" ;
>    float latitude(YCells, XCells) ;
>        latitude:long_name = "latitude" ;
>        latitude:units = "degrees_north" ;
>
>
> So, thinking like a Ferret, I would think I need to regrid these two 
> curvilinear grids onto a common regular lat-lon grid and then proceed with 
> the comparison between the two.  Unfortunately, I can't figure out how to 
> do this :(  From the docs, I see this in Chp 8 Sec 8.2
>
> "To perform the analysis in a rectilinear coordinate system, the
> conversion of the curvilinear data is most easily done with SAMPLEXY_CURV
> (under development06/00 )."
>
> I can't seem to find that samplexy_curve was implemented. So I searched 
> the email archives and found this post which suggests to use scat2grid 
> functions.
>
> http://ferret.pmel.noaa.gov/Ferret/Mail_Archives/fu_2001/msg00084.html
>
> But those functions don't seem to work for me.  The resulting grid is 
> always "No valid data".  For example I did:
>
> yes? use APCP_06171200.nc  ! My NCEP data
> yes? DEFINE AXIS/X=100w:90w:0.5/UNITS=degrees x0
> yes? DEFINE AXIS/Y=40n:50n:0.5/UNITS=degrees y0
> yes? show data
>     currently SET data sets:
>    1> ./APCP_06171200.nc  (default)
> name     title                             I         J         K      L
> APCP_SFC Total precipitation              1:1121    1:881     ...    ...
> MAPPROJECTION
>          MapProjection                    ...       ...       ...    ...
> LONGITUDE
>          longitude                        1:1121    1:881     ...    ...
> LATITUDE latitude                         1:1121    1:881     ...    ...
>
> yes? LET outg = SCAT2GRIDGAUSS_XY(longitude, latitude, APCP_SFC, x[gx=x0], 
> y[gy=y0],0.5,0.5,2,2)
> yes? shade outg
>
>
> I am using Ferret 5.53 on Fedora Core 1 and am probably doing something 
> ignorant :(
>
> thanks,
>  daryl
>
>