Greetings!
I am working on some code to convert BUFR (the NCEP flavor) observation
files to netCDF. The NCEP BUFR files that we archive contain multiple
subsets. For example, the "adpsfc" file that we ingest can contain any
of the following (from bufrtab.000):
|------------------------------------------------------------------------------|
| MNEMONIC | NUMBER | DESCRIPTION |
|----------|--------|----------------------------------------------------------|
| | | |
| NC000000 | A51001 | MSG TYPE 000-000 SYNOPTIC-LAND, RESTRICTED (WMO RES 40) |
| NC000001 | A63200 | MSG TYPE 000-001 SYNOPTIC - FIXED LAND |
| NC000002 | A61192 | MSG TYPE 000-002 SYNOPTIC - MOBIL LAND |
| NC000007 | A63206 | MSG TYPE 000-007 AVIATION - METAR / SPECI |
| NC000008 | A63247 | MSG TYPE 000-008 NPN AND MAP PROFILER SURFACE |
| NC000010 | A56100 | MSG TYPE 000-010 PRODUCTS (SHEF) NOT IN ANY OTHER TANK |
| NC000011 | A63214 | MSG TYPE 000-011 AFOS PRODUCTS (PRECIP) (SHEF) |
| NC000012 | A63215 | MSG TYPE 000-012 SUPPLEMENTARY CLIMATOLOGICAL DATA |
And usually multiple records ("subsets") are available for any one of
the mnemonics listed above. While many of the subsets in a particular
group share common parameters, there are also parameters that are unique
to a mnemonic.
I believe I can handle mimicking the BUFR format (multiple subsets per
file) using netCDF "groups". Am I correct that groups are only
available with netCDF-4? I'm concerned that this would limit user
access, if this is the case. We have many uses world wide and I have no
idea what the distribution of users are that can handle netCDF-4.
Is there a way to push multiple subsets into a single netCDF-3 file? If
not, that is fine and I can accept creating one netCDF file per subset
if need be. I'm just asking about the technical possibility.
Many thanks!
-kevin.
--
Kevin Manross
NCAR/CISL/Data Support Section
Phone: (303)-497-1218
Email:manross@xxxxxxxx <mailto:manross@xxxxxxxx>
Web:http://rda.ucar.edu
