On the suitability of BUFR and GRIB for archiving data

A major weakness of the BUFR and GRIB formats is their dependence on external tables. For GRIB, one must have the correct tables in order to understand the meaning of the data. For BUFR, the correct tables are needed to understand the data but also to even parse the binary file. An important design flaw in these formats is that there is no way for a reader to know for certain which tables the writer used. Compounding this are implementation practices which make this problem occur more often than might be expected. This makes BUFR and GRIB not suitable as long-term storage formats. In order to correct this problem, 1) there must be a foolproof way for software to know which tables were used when the data was written, and 2) there must be an authoritative registry of tables.