Re: confusion with record dimension using gribtonc

• To: Rorik Peterson <ffrap1@xxxxxxxxxxxxxx>
• Subject: Re: confusion with record dimension using gribtonc
• From: Robb Kambic <rkambic@xxxxxxxxxxxxxxxx>
• Date: Thu, 22 Jan 2004 16:53:05 -0700 (MST)

On Thu, 22 Jan 2004, Rorik Peterson wrote:

I have been having some problems with the record dimension when
decoding  Alaska-region GRIB data obtained via ftp from NCEP at:

ftp://ftpprd.ncep.noaa.gov/pub/emc/mmb/mmbpll/alaska10.t00z/

The GRIB files available here are updated daily, each one corresponding
to a forecast time (0-48 hours).  I sequentially process them with
'gribtonc' into a single netCDF file

[/user2/ldm/AWIPS]$ perl -e 'foreach(glob "A*"){ `gribtonc wrf249.cdl
out.nc < $_`;}

However, I always have a complete record of missing data somewhere in
the middle.  Using today's data, I narrowed it down to when I added the
15-hour forecast file; at which point two records were added instead of
one, and the last one was completely empty.

Here I decode just one GRIB file:
[/user2/ldm/AWIPS]> rm out.nc;gribtonc wrf249.cdl out.nc < AWPREG15.tm00

Then dump the pertient information of what is in there
[/user2/ldm/AWIPS]$ ncdump -v valtime_offset out.nc | egrep "record
UNLIMITED|valtime_offset ="
        record = UNLIMITED ; // (2 currently)
 valtime_offset = 15, 15 ;

and then look at the data
[/user2/ldm/AWIPS]$ ncks -H -v u -d x,100,100 -d y,100,100 -d level,4,4
out.nc
level[4]=500

record[0] level[4]=500 y[100]=986.789 x[100]=986.789 u[540324]=2.17085
record[1] level[4]=500 y[100]=986.789 x[100]=986.789 u[1673253]=-9999

x[100]=986.789

y[100]=986.789

Rorik,

You already narrowed the problem down to either a bad data product or an
error in gribtonc.  First test would be to run gribtonc with the -v and
-l - flags, ie gribtonc -v -l -  .....    Hopefully this should shed some
light on the error. Next test would be to use gribdump to check if there
is an unusual header in the grib products.  In my recent work, I've found
a wrong header on only one product in a group of 2000 files. The
projection was wrong on the product.  ie gribdump -h <gribfile>

Can you send me the wrf249.cdl file you are using so I can use it.

I downloaded the AWPREG15.tm00 file and will let you know the results.

Robb...

You can trust me that the entire second record is full of -9999, which
is missing_value.  Is this a problem with the GRIB data, or gribtonc?
All the GRIB files are the same size, more or less, so I don't know
where the second set of (missing) data is coming from.  Any insight in
what to look at next is appreciated.

rorik

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