[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

20020225: GRIB to netcdf (cont.)



>From: "Hoeth, Brian" <address@hidden>
>Organization:  JSFC
>Keywords:  200202211951.g1LJpex01567 gribtonc

Brian,

>No, we're wanting to convert GRIB files to netcdf for display in McIDAS.  We
>have been using GRIB decoder software that will create McIDAS GRID files,
>but our decoders don't work on some of the larger GRIB files that NCEP is
>pushing out now.  

I have been putting together a new gribdec.tar.Z packaging and an
addendum for Unidata McIDAS that supports the larger grids that are
floating around out there (up to 400000 points).  The new gribdec.k
routine that I am using will decode all of the CONDUIT grids (grids
that one could FTP from the NIC) except the ensembles.

The other thing I added to the my bundling of gribdec.tar.Z is a
routine that will decode grib messages directly into to images (AREA
files).  This is a fusion of gribdec.pgm and grdimg.pgm.  It works
nicely for everything I have used it for in the 4 days since I put it
together.

>So, we were looking into possibly grabbing your GRIB to netcdf converter
>instead of having to alter our decoders (no sense re-inventing the wheel).

I would be happy to let you use my new gribdec.tar.Z package.  The good
thing about them is they are almost 100% code from XCD (which I believe
you have; true?).

>Once in netcdf format, the data should be able to be displayed via ADDE in
>McIDAS right?

Yes, but you will still be faced with the maximum size of grids that
McIDAS can handle.  This is why I will be upgrading gridparm.inc in my
next McIDAS addendum (along with the XCD header file grib.h).

In addition, I have modified the XCD routine btab.c to do a better job
of creating navigation from grib messages with grid type 255.  The next
thing I have been thinking of doing is adding a true Lambert Conformal
nav module (nvxlcc.dlm) to McIDAS and upgrading grddef.for to support
that navigation.

- Aside - the Lambert Conformal nav modules in McIDAS support Lambert
Conformal Secant (nvxlamb.dlm) and Lambert Conformal Tangent
(nvxtanc.dlm), but only for grids where the measure in the X and Y
dimensions is equal.  This has caused me quite a bit of pain recently,
so I figure I will just bite the bullet and add a general, correct
Lambert Conformal nav.  What do you think?

Tom