Re: [gembud] problems with NAM 104/255 CONDUIT files

David,

The missing value points are being scaled by the 10**-2 from Pascal to
mb for plotting. The unpacking routine routine used by the plotting
programs in gemlib.a needs a check for missing value before the scaling
is applied. 

I have provided the updated source code at:
http://www.unidata.ucar.edu/downloads/gempak/nawips-5.10.4/source/gb2ugem.c

You can update a source distribution by placing this file in
$GEMPAK/source/gemlib/gb

The updated gemlib.a with:
cd $GEMPAK/source/gemlib/gb
make clean all
ar rv $GEMLIB/gemlib.a gb2ugem.o         <-- needed for linux/gmake only
make clean

The relink your plotting programs:
cd $GEMPAK/source/programs/gd
make clean all install programs_nc programs_gf
make clean

cd $GEMPAK/source/programs/gui/nmap2
make clean all install
make clean

cd $NAWIPS/comet/garp
make all install
make clean

cd $NAWIPS/unidata/programs
make clean all install programs_nc programs_gf
make clean

I'll repack the binary distributions on the download
site as the builds are complete here.

Steve Chiswell
Unidata User Support

On Tue, 2007-11-06 at 13:18 -0800, David Ovens wrote:
> CONDUIT users and Gembuds,
> 
> As you know, the CONDUIT data feed recently dropped the NAM 104 grids
> in GRIB1 format.  We are receiving the 90-km NAM grids in GRIB2 format
> under the following pqact entry:
> 
>  CONDUIT        /nam\.t(..)z\.grbgrd.*grib2
> 
> Most everything looks fine; however, when I convert these files to
> GEMPAK using dcgrib2 from 5.9.4, 5.10.3, or 5.10.4, the PMSL field is
> all fouled up.  The following URL shows what I get for a 'CINT=0' setting: 
>   http://www.atmos.washington.edu/~ovens/2007110600_badpmsl.gif
> 
> Here's how I decoded the file:
> 
>  $GEMEXE/dcgrib2 -d - 2007110600_nam104.gem < \
>  /home/disk/data/conduit/nam.20071106/nam.t00z.grbgrd00.tm00.grib2
> 
> I suspect this is a problem with the GRIB2 file (my first guess) or
> the way GEMPAK is decoding the sea-level pressure field.
> 
> Does anyone have a fix for this yet?
> 
> Thanks,
> 
> David
-- 
Steve Chiswell <address@hidden>
Unidata