The Numerical Predicition Division (NPD) of Japan Meteorological Agency (JMA) has been working on TDCF migration of both surface and upper-air reports.
Surface
Use in NWP
- Assimilation of BUFR SYNOP reports was started on 16 October 2013.
- Currently coordinates (latitude, longitude, and elevation) in BUFR is ignored and replaced by coordinates from Volume A, WMO Pub. 9.
- Current implementation picks BUFR if TAC for the same station is absent; the weight of BUFRs is little now but will grow when TAC reports are stopped
- Figure below is the number of stations reporting Any SYNOP (black), Only TAC (red), TAC and BUFR (purple), and Only BUFR (blue), observed at NWP system in JMA. The number of BUFR (purple + blue) firstly exceeds 50 % of entire SYNOP reports in mid-October 2014. But there are still 40+% of stations which does not (probably cannot) send BUFR.
Trouble cases
Missing pressure for geopotential height
Starting at 11Z 23 January 2014, the SYNOP BUFR reports from the station 07497 BOURG ST-MAURICE contained:
- missing value for 0-07-004 PRESSURE, follewd by
- extremely large value for 0-10-009 GEOPOTENTIAL HEIGHT.
These values were decoded into 16 bit signed integers. The missing value was converted to minimum possible value (-32768e1 Pa). Unfortunately that was treated as valid value since geopotential was not missing. Then that caused floating point exception in computation of power of (now negative!) pressure. In the TAC (traditional alphanumeric) reports it was safely assumed geopotential is always missing when pressure is missing. We are not monitoring later reports regarding this issue, as the NWP system was fixed immediately. We also contacted the originating country, and they say it will be fixed.
Upper-air
Analysis of incoming data
description of algorithm to tell code form and parts
Use in NWP
- BUFR TEMP in U.S. was once used in assimilation from April 25 2013
- But it was stopped February 2014 (migration back to TAC)
- Starting the end of October 2014, the system uses BUFR if TAC is missing for the station/ship
Trouble cases
Geopotential height for US TEMP BUFR
Starting February 2014, most of U.S. TEMP BUFR reports contained unnatural geopotential height. This issue is explained detail in Errors in BUFR radiosonde data page by FNMOC.
Japan's TEMP BUFR
A few problems were found in TEMP BUFR reports sent from Tokyo. The TAC-BUFR conversion at RTH Tokyo has to be amended to get rid of the situation, and it is not possible to show specific schedule for now.
- When the geopotential height of 700 hPa goes below 2400 m, it is reported 1000 m higher than actual.
When there are significant levels below or equal to 1000 hPa level (i.e. pressure >= 1000), the pressure is reported 1000 hPa less than actual. This problem was fixed effective 2015-08-18T12Z.When the geopotential height of 1000 hPa goes below MSL, the reported value is same to coded digits in TAC TEMP (512 means 12 m). This problem was fixed effective 2014-02-24T12Z.
The issues are "missing top digit of data", lost in TAC encoding.