Fix discrepancy between EUMETSAT and ROM SAF processing to BUFR

[Ian Culverwell]

Harald Anlauf (DWD) writes:

Dear all,

I've restricted the list of addressees to EUMETSAT and ROMSAF 
because this is only a minor technical issue.

There is inconsistent rounding in the last decimal places between 
the encoding at EUMETSAT and at the ROMSAF in different places.  
This applies to current data (as of today, 2016-07-06, 08:36 UTC).

It would be nice if this could be made consistent, and if wrong 
rounding and/or undesired truncation be identified and resolved.

I've attached what our "in-house" BUFR decoding software actually 
"sees" for a Metop-B sounding.

See the attachment for details, and sorry for the special language 
and spelling used in the description of the keys.  ;-)

Harald

The contents of harald's tar file are attached.

[Ian Culverwell]

Somehow this got copied to #473, so close this ticket as 'duplicated'.