To: Michelle Mainelli, NOAA/NCEP
The Unidata Users Committee recognizes the importance of the GRIB and
BUFR data formats to
the meteorological research and education communities of which we are a
part. These formats
have traditionally been difficult for non-expert users to read, and
continued growth in their use and in their complexity makes it urgent
that this difficulty be mitigated wherever possible. In particular,
ambiguity in which external tables were used by the encoding software
threatens the ability for decoding software to correctly interpret
archived data.
The Unidata Users Committee requests that Michelle Mainelli (NOAA/NCEP),
in her role as representative to the Unidata Users Committee, forward
the following recommendations to the IPET-DRC (Inter-Programme Expert
Team on Data Representation and Codes). The Users Committee understands
that Jeff Ator (also of NOAA/NCEP) is a representative on the IPET-DRC,
and will deliver the recommendations for consideration at the Team's
next meeting, scheduled for the week of May 21, 2012.
The recommendations are:
1) Improve handling of GRIB and BUFR tables by creating a web registry
of machine readable tables and by requiring the inclusion of a reference
to a registered table in GRIB and BUFR records. The reference would be
an unambiguous
signature of the table, such as an MD5 checksum. Both WMO standard
tables and
local tables should be required to be registered.
2) Create reference software and/or a web service that can be used to
validate that a BUFR/GRIB message is well-formed.
It should read the message metadata and apply the registered tables,
and also decode the data itself, returning a representation of all the
metadata
and data in a well understood encoding (e.g. ASCII or XML) so that other
BUFR/GRIB
reading software can be validated against this "reference implementation".
3) Establish an international GRIB and BUFR users group to share
knowledge in
an informal way, where users can ask questions and get advice.
4) Encourage WMO members to take part in a community effort to create a
standard mapping of GRIB parameter tables to the CF (Climate and
Forecast Convention) standard names.
More details can be found in this short paper:
https://www.unidata.ucar.edu/staff/caron/papers/GRIBarchivals.pdf
Thank you in advance for your consideration.
