Re: gribtocdl (fwd)

Arild,

I don't know if you are keeping your own version of gribtocdl but this bug
was sent to me. The fix is to replace km with m in gdes.c for Polar and
Lambert projections.

robb...

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

---------- Forwarded message ----------
Date: Tue, 31 May 2005 15:39:57 -0600
From: John Caron <address@hidden>
To: Robb Kambic <address@hidden>
Subject: Re: gribtocdl

Robb Kambic wrote:

>On Tue, 31 May 2005, John Caron wrote:
>
>
>
>>the file in /upc/share/testdata/grid/netcdf/nuwg/charlie.nc
>>
>>has
>>
>> float Dx(nav);
>>     :long_name = "x-direction grid length";
>>     :units = "km";
>>   float Dy(nav);
>>     :long_name = "y-direction grid length";
>>     :units = "km";
>>
>>has units of km, but the value is in fact in meters.
>>
>>
>>
>
>
>
im ignoring the "km" at the moment, assuming its meters (which it is!)
but id like to use whatever units are there.

>
>
>
>>I assume this  is a bug in gribtocdl, since here are the global attributes:
>>
>> :record = "reftime, valtime";
>> :lbls = "lbls_bot, lbls_top";
>> :lfhg = "lfhg_bot, lfhg_top";
>> :liso = "liso_bot, liso_top";
>> :lpdg = "lpdg_bot, lpdg_top";
>> :history = "2004-12-01 15:43:34 - created by gribtocdl";
>> :title = "Enter model definition here";
>> :Conventions = "NUWG";
>> :GRIB_reference = "Office Note 388 GRIB";
>> :GRIB_URL = "http://www.nco.ncep.noaa.gov/pmb/docs/on388/";;
>> :version = 0.0; // double
>>
>>
>>
>>The best fix for now is to change the units to meters, as that is
>>hardcoded in my reader. However, I would like to "do the right thing"
>>and believe whatever units are specified.
>>
>>Im wondering how widespread this bug is ? Who uses grib2cdl and grib2nc ?
>>
>>We should perhaps consider replacing  them with an nj22 based version,
>>which produces CF instead of NUWG.
>>
>>
>>
>
>i agree but nj22 can't do thin or tiled grids yet, gribtonc handles these
>fine.  also, don't know if display programs can handle the CF conventions.
>
>
good point, we'll probably need both for now.