Re: gribtonc

• To: laps@xxxxxxxxxxxxxxxxxxxxx
• Subject: Re: gribtonc
• From: Robb Kambic <rkambic@xxxxxxxxxxxxxxxx>
• Date: Thu, 6 Aug 1998 10:22:08 -0600 (MDT)

On Wed, 5 Aug 1998 laps@xxxxxxxxxxxxxxxxxxxxx wrote:

robb,

i tried   % ncgen -o output.nc lga.cdl . this just created a
netcdf version of he cdl file.  i added in the missing info
to my cdl file. i put kg/kg for the rh units and datetime
id and model id. when i added in the model_id(nmodels) :

        long    model_id(nmodels) ;
                model_id:long_name = "generating process ID number" ;

i get a core dump but no errors. if i use the eta.cdl file, it writes

a huge netcdf file with missing values instead of numbers.

Dave,

The problem is that the cdl file doesn't match the data, it's not anything
to do with udunits.  I would start again with the eta.cdl file as a
template for ncgen.  I would add the needed vars for the laps cdl.  The
above core dump could be cause by nmodels not being defined.  After I got
the ncgen working, then use gribdump to compare the input to the cdl file.
It's an interation type of process.   BTW, did you try to contact FSL to
obtain the laps cdl file?  They should probably have one that should work
or be a better starting point than from scratch. I don't know who to
contact, but maybe their web page could help.

Robb...

i have a eta.cdl file that came with the decoders that i used to fix
the problems in my lga.cdl file.  my goal is to create a netcdf file
to use with laps (fsl). i have examples of what those files are supposed
to look like. am i missing something?

dave

==============================================================================
Robb Kambic                                Unidata Program Center
Software Engineer III                      Univ. Corp for Atmospheric Research
rkambic@xxxxxxxxxxxxxxxx                   WWW: http://www.unidata.ucar.edu/
==============================================================================