Re: 20020610: Unidata Decoders and modifications (fwd)

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

---------- Forwarded message ----------
Date: Fri, 30 Aug 2002 13:16:36 +0000
From: Arild Burud <address@hidden>
To: Robb Kambic <address@hidden>
Subject: Re: 20020610: Unidata Decoders and modifications (fwd)

Robb Kambic wrote:
> 
> On Wed, 12 Jun 2002, Russ Rew wrote:
> 
> Arild,
> 
> Yes, we are interested in incorporating the changes into the decoders
> package. Of course we will give you the credit for developing the
> gribtocdl program.  At this time, I would appreciate a response back
> giving Unidata Program Center the right to incorporate the code into the
> Decoders packages. It was a clever idea that  works very well. At

We have discussed it in our office, Unidata Program Center can of course
incorporate the updated code into the Decoders package.
Just make sure that Terje Reite is also credited for his work on
gribtocdl, we have made the program together.
The work we have done with decoding was intended to be available for any
researchers, if they find it useful.

> this time, I've only done minor testing on gribtocdl but I'm very limited
> on time.  Do you have any updates since I got the package in June?  Also,
> once include the gribtocdl code will be under support by Unidata

Yes, I have made a minor change to the programs: In the user-supplied
conversion table it is now allowed to specify units as
"none"/"None"/"NONE" or simply a blank string, without errormessages
during decoding. The updated version is again available as
http://noserc.met.no/formkonv/decoders-2.3.8n.tar.gz
No other changes have been done, the manpages and other information was
not updated.

> Again, you have done a great job in creating gribtocdl program and I'm
> sure it will be used by our community.
> 
> Thanks
> 
> I like the idea of the using the  user-supplied conversion table for
> parameter names, this is a nice feature.

Nice it is, but still there are some features from the original gribtonc
that have the potential of creating problems during decoding to
netcdf/cdl. As an example, take the special handling of parameters based
on their number. This is done several places in the package, such as 
mslparam() and lisoparam(). These may cause problems when the
userdefined conversion table is used - and I have no good solution to
such problems either.

There may be more problems that I just haven't come across yet.

We have seen other problems when we want to convert ECMWF GRIB-files to
netCDF. Problems that are caused by limitations in the existing decoders
package. For instance, the model number used in the GRIB data has been
seen to change from tie to time in the original GRIB data. It should
have been possible to allow the model number to be a variable number
instead of a constant number, so that datasets can be added with the
same CDL even if model number is updated. (I may have missed something
in my understanding of CDL...)

Yours
Arild Burud
-- 
---------------------------------------------------------------------
address@hidden       System Developer          Tel.: +47 22963035
http://www.met.no                                  Fax.: +47 22963050
Norwegian Service Centre for Climate Modelling,  http://noserc.met.no
Norwegian Meteorological Institute,    PB 43 Blindern,    N-0313 Oslo
---------------------------------------------------------------------