Accommodating Changes in the Data Stream at NRL/FNMOC: The TAC to BUFR Migration

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Tuesday, 6 January 2015: 9:30 AM
131AB (Phoenix Convention Center - West and North Buildings)
Patricia M. Pauley, NRL, Monterey, CA; and D. Tyndall, C. McGregor, L. Troutman, and G. Hoisington

Numerical Weather Prediction (NWP) centers such as NCEP and FNMOC obtain world-wide observations in real time via the World Meteorological Organization's (WMO) Global Telecommunications System (GTS). Although the GTS has undergone infrastructure updates over the years, the codes used for radiosonde and surface land data have not changed since the days of 300 baud teletype machines. However, the WMO has mandated the migration from Traditional Alphanumeric Codes (TAC) (such as the TEMP code used for radiosonde data and the SYNOP code used for surface land data) to Table Driven Code Forms (TDCF), which primarily means Binary Universal Form for the Representation of meteorological data (BUFR). BUFR was first approved for operational use by the WMO in 1988. The characteristics of BUFR that give it an advantage over TAC are that it is 1) universal across data types, 2) self-describing, 3) table-driven, and 4) binary (Ator, 2010). In 2003, the WMO approved the migration of TAC to BUFR, with data to be disseminated in both TAC and BUFR formats for a limited while data users make the transition. The official WMO timetable has the parallel distribution of TAC and TDCF messages for radiosonde and surface land data ending in November 2014. Since many countries are not yet generating BUFR messages for radiosonde and surface land data, TAC messages for at least some countries will likely continue past the deadline.

This paper details work done at the Naval Research Laboratory Monterey (NRL) and at Fleet Numerical Meteorology and Oceanography Center (FNMOC) to use BUFR messages in addition to TAC messages, in preparation for the discontinuation of the TAC messages. The work at FNMOC includes identifying GTS bulletin identifiers for the BUFR messages, adapting the BUFR decoder for each variant of BUFR, and writing the decoded data in text file for internal use. The work at NRL includes adapting the code in the data assimilation system to read the decoded BUFR data in addition to the decoded TAC data, performing a duplicate check between the TAC and BUFR data, adapting the quality control code to use the blend of TAC and BUFR data, and accommodating differences in the BUFR data, which include high resolution (2-second) radiosonde soundings and balloon drift information, both of which are only available in BUFR. Difficulties encountered in all of these aspects will be presented, as well as errors found in the BUFR data and efforts to report the errors to the originating data providers.