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

Re: Gribtonc changes

On Tue, 27 May 2003, Jim Cowie wrote:

>
> Hi Robb,
>
> Here's my $0.02:
>
> 1) GRIB2 support - you didn't mention this but I really hope that support
>     for GRIB2 is included. Last I heard, the NWS is going to start using
>     this before too long.

Hiya Jim,

Yes there will definitely be grib2 support.  I'm actually working on this
now.


>
> 2) The biggest problem we have with gribtonc is that all the product
>     tables are hard-coded. We often get grids from FSL or elsewhere
>     which don't use the NCEP grib tables, so we have to modify the code.
>     Externalizing this to a set of (optional) tables would solve this
>     problem.
>


Yea, hard coding tables is bad.  Actually I've been working on making XML
files of all the tables. So far, I've got more than 25 tables. The tables
are located at:

http://www.unidata.ucar.edu/projects/THREDDS/tech/index.html#STD

I would like to hear comments about the tables.


Robb...




> -jim
>
>
>
> Robb Kambic wrote:
> > Hiya,
> >
> > I'm writing to inform you about some proposed changes to the output
> > netCDF files produced by the gribtonc (GRIB-to-netCDF) decoder.  There
> > are about 35 sites that have installed the decoders package as far as I
> > know.  I'm currently in the process of updating the package, including
> > the CDLs to make them more useful and to add attributes required by
> > new conventions.
> >
> > The original CDLs followed the NUWG conventions:
> >
> >   http://www.unidata.ucar.edu/packages/netcdf/NUWG/
> >
> > which haven't been modified in several years,
> >
> > Lately the CF conventions
> >
> >   http://www.cgd.ucar.edu/cms/eaton/cf-metadata/index.html
> >
> > are becoming established as a more complete set of conventions for
> > model output.  Also, we are considering including spatial and temporal type
> > variables to make the NetCDF file more GIS compatible.  Another idea is to
> > enhance the metadata about the date for THREDDS servers and other NetCDF
> > clients.
> >
> > The CDL variable names will continue to follow the NUWG conventions
> > for backward capability. The long_name attribute has been standardized
> > by concatenation of the NUWG variable name plus the NUWG level name
> > using the word "at" as a connector. So, the T_trop long_name is
> > "Temperature at tropopause" in all the CDLs.  This supports
> > consistency in applications such as Unidata's IDV (Integrated Data
> > Viewer). For similar reasons, we are adding attributes required by
> > other conventions such as CF, such as the standard_name attribute.  A
> > sample variable will look like:
> >
> >         float  T_trop(record,y,x) ;
> >                T_trop:long_name = "Temperature at tropopause" ;
> >                T_trop:standard_name = "air_temperature" ;
> >                T_trop:units = "degK" ;
> >                T_trop:GRIB_parameter_number = 11 ;
> >                T_trop:GRIB_level_flag = 7 ;
> >                T_trop:_FillValue = -9999.f ;
> >                T_trop:navigation = "nav" ;
> >
> >
> > Some other additions to the variable attributes are the 
> > GRIB_parameter_number,
> > and the GRIB_level_flag attributes. These are being included so other
> > conventions may refer to the NCEP GRIB tables as a cross reference.  This 
> > will
> > permit more flexible usage for the NetCDF files.
> >
> > Another variable has been added called forecast_time, a human readable
> > string representing the valid forecast times. To make the  data
> > monotonically increasing, the code sets records for all the valid times in 
> > the
> > netCDF file on initialization.
> >
> > When netCDF GIS applications become available, the new NetCDF files
> > will already have the needed information for correct display. We are also
> > considering temporal variables so time bounding boxes can be created for the
> > datasets.
> >
> > This is a brief summary of the kind of modifications we are making to
> > the decoders package.  If you have other ideas about what should be
> > included or omitted, please send in your comments.
> >
> >
> > Robb...
> > ===============================================================================
> > Robb Kambic                            Unidata Program Center
> > Software Engineer III                          Univ. Corp for Atmospheric 
> > Research
> > address@hidden                 WWW: http://www.unidata.ucar.edu/
> > ===============================================================================
>
>
> --
> Jim Cowie
> NCAR/RAP
> address@hidden
> 303-497-2831
>
>

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