Hi John,
As I mentioned before, the differences appear at the beginning of bufr
usage. Version 0 was experimental and probably all this is relevant only
for Ecmwf data. At the time no one was creating bufr data except us.
So in practice I would not be too much concerned about that. I would use
version 13 and make all necessary links. What happens in practice is
that data producers set in the section 1 in many cases wrong information
about local tables used and so on. Ecmwf software creates table names
out of information from section 1 to avoid any possible clashes.
The new links are needed only if software can not find table in the
BUFR_TABLES directory. The reason is usually false usage of local tables
and than table name contains origination centre, sub-centre master
version and local version. The tables to be loaded are printed if they
are missing and you can make links for those tables.
Best regards
Milan
Milan Dragosavac
ECMWF
Shinfield Park, Reading, Berkshire, RG2 9AX, UK
Tel: (+44 118) 949 9403
Fax: (+44 118) 986 9450
Telex: 847908 ECMWF G
E-mail: milan.dragosavac@xxxxxxxxx
John Caron wrote:
Hi Milan:
I see that in the BUFRDC tables, B0000000000098000000.TXT and
B0000000000098002001.TXT, the 013015 bit width changed from 10 to 12,
and apparently the scale also changed from 1 to 7:
version 0 (B0000000000098000000.TXT):
013015 SNOWFALL (AVERAGE RATE) M/S 1 0 10
and version 2 (B0000000000098002001.TXT):
013015 SNOWFALL (AVERAGE RATE) M/S 7 0 12
I see, as you mention, the change to SI units in version 2, which
affects the scales.
So this is a pretty clear example of changes to scale,reference and/or
bit width to operational descriptors. Without knowing these, any
messages that use those descriptors will be incorrectly decoded.
So how do we know when "links are needed" to other tables? Does the WMO
maintain canonical copies of previous table versions of the Table
documents that we can double check? Should we use BUFRDC tables as
canonical ?
Regards,
John
Milan Dragosavac wrote:
Dear John,
I think you can use the latest version 13 of the tables and make links
to any other tables with smaller version number. In my Bufr reference
manual I kept track on difference between version 0 and 2 and I believe
those are the only differences. However, I am not sure if those
differences refer only to Ecmwf's maintained tables or not. All
entries are for table B class 13 where units were changed and
consequently scales. The data width was changed only for 013015
snowfall (average rate) from 10 to 12.
To summarise, I think it is very safe to use version 13 and make links
to any other when needed.
Best regards
Milan
Milan Dragosavac
ECMWF
Shinfield Park, Reading, Berkshire, RG2 9AX, UK
Tel: (+44 118) 949 9403
Fax: (+44 118) 986 9450
Telex: 847908 ECMWF G
E-mail: milan.dragosavac@xxxxxxxxx
