The World Meteorological Organization (WMO) has been working for TDCF migration, which is transition from observation reports and forecast messages in TAC (traditional alphanumeric codes) into TDCF (table-driven code forms) namely BUFR (binary universal representation) and CREX. For most of message types, the parallel distributions were scheduled up to November 2014.
Some countries actually stopped their TAC distributions, and
it turned out that the meteorology community needs more work to get fully ready to the transition, especially for upper-air sounding. The situation was complicated since WMO tried to change not only code but reporting practice, as discussed in following forum thread:
it turned out that the meteorology community needs more work to get fully ready to the transition, especially for upper-air sounding. The situation was complicated since WMO tried to change not only code but reporting practice, as discussed in following forum thread:
Traditionally, the alphanumeric messages for a single ascent of radiosonde are supposed to be disseminated in four parts:
The standard isobaric levels are 1000, 925, 850, 700, 500, 400, 300, 250, 150, and 100 for Part A; and 70, 50, 30 and 10 hPa for Part C. Weather charts are supposed to be drawn at those levels (JMA's charts are available at http://www.jma.go.jp/jp/metcht/kosou.html).
The significant levels (in Parts B and D) supplements standard levels so that linear interpolation of all reported data (both standard and significant levels) gives enough accuracy. Roughly they are chosen at kinks of plots of temperature, humidity, and winds (a good illustration of temperature siglevs is given at ECMWF website). Please see the Manual on Codes Volume I.1 for the complex rules for selecting significant levels.
Then, why do we split messages at 100 hPa? The answer is time of delivery. The 100 hPa level lies 16 km above sea level. Typical radiosonde has 6 m/s speed. It takes 45 minutes to reach that level. The time may double if we wait for the end of ascent: that's why WMO member states agreed to split the report.
In the new paradigm of BUFR, there is a rule called B/C 20 and 25 to determine how to migrate from TAC to BUFR:
https://www.wmo.int/pages/prog/www/WMOCodes/BC_Regulations/BC20-PILOT.pdf
https://www.wmo.int/pages/prog/www/WMOCodes/BC_Regulations/BC25-TEMP.pdf
It describes two messages should be sent for a single ascent:
That was correct, but the meaning or value of the reporting style was not stressed. So implementers simply used the template only and ignored the text regulations. Most implementations simply converts existing TAC to BUFR in parts.
It was supposed that BUFR would get the mainstream, i.e. observation systems directly generated BUFR and TAC were made as a compatibility service. By doing so we could eliminate various problems of TAC reports. But that mainstreaming did not take place in many WMO members. I suspect the change of the rule is even not known for many people working for observation systems. So the part-for-part converter had to be chosen.
There are many benefits of BUFR-mainstreamed data processing, which is basically straightforward handling of numbers. For example 1000's digit of geopotential height in lower troposphere is omitted in FM 35 TEMP. If 700 hPa height is 3451 metres, only digits "451" is shown in the report, and the recipient has to guess the lost digit using "common sense". It is usually 3451 m in most of the world, but it can be 2451 m in very cold places like Antarctica, so it is a headache for automated system. A simple-minded converter always generates BUFR with 3451 m value for TAC input "451" which may be incorrect. But the benefit has not so stressed for these decades, and I don't think many understand it.
Right now there are many proposals to change the part-by-part converted BUFRs to distinguish Parts A-D. But I'd really remind the situation filled by a whole bunch of unimplemented regulations and ununderstood good intentions. I think we must be really careful for changing rules, since current one takes many years to be understood partly.
- Part A: standard isobaric levels up to 100 hPa
- Part B: significant levels up to 100 hPa
- Part C: standard isobaric levels above 100 hPa
- Part D: significant levels above 100 hPa
The standard isobaric levels are 1000, 925, 850, 700, 500, 400, 300, 250, 150, and 100 for Part A; and 70, 50, 30 and 10 hPa for Part C. Weather charts are supposed to be drawn at those levels (JMA's charts are available at http://www.jma.go.jp/jp/metcht/kosou.html).
The significant levels (in Parts B and D) supplements standard levels so that linear interpolation of all reported data (both standard and significant levels) gives enough accuracy. Roughly they are chosen at kinks of plots of temperature, humidity, and winds (a good illustration of temperature siglevs is given at ECMWF website). Please see the Manual on Codes Volume I.1 for the complex rules for selecting significant levels.
Then, why do we split messages at 100 hPa? The answer is time of delivery. The 100 hPa level lies 16 km above sea level. Typical radiosonde has 6 m/s speed. It takes 45 minutes to reach that level. The time may double if we wait for the end of ascent: that's why WMO member states agreed to split the report.
In the new paradigm of BUFR, there is a rule called B/C 20 and 25 to determine how to migrate from TAC to BUFR:
https://www.wmo.int/pages/prog/www/WMOCodes/BC_Regulations/BC20-PILOT.pdf
https://www.wmo.int/pages/prog/www/WMOCodes/BC_Regulations/BC25-TEMP.pdf
It describes two messages should be sent for a single ascent:
- Standard and significant levels up to 100 hPa
- Standard and significant levels above 100 hPa
That was correct, but the meaning or value of the reporting style was not stressed. So implementers simply used the template only and ignored the text regulations. Most implementations simply converts existing TAC to BUFR in parts.
It was supposed that BUFR would get the mainstream, i.e. observation systems directly generated BUFR and TAC were made as a compatibility service. By doing so we could eliminate various problems of TAC reports. But that mainstreaming did not take place in many WMO members. I suspect the change of the rule is even not known for many people working for observation systems. So the part-for-part converter had to be chosen.
There are many benefits of BUFR-mainstreamed data processing, which is basically straightforward handling of numbers. For example 1000's digit of geopotential height in lower troposphere is omitted in FM 35 TEMP. If 700 hPa height is 3451 metres, only digits "451" is shown in the report, and the recipient has to guess the lost digit using "common sense". It is usually 3451 m in most of the world, but it can be 2451 m in very cold places like Antarctica, so it is a headache for automated system. A simple-minded converter always generates BUFR with 3451 m value for TAC input "451" which may be incorrect. But the benefit has not so stressed for these decades, and I don't think many understand it.
Right now there are many proposals to change the part-by-part converted BUFRs to distinguish Parts A-D. But I'd really remind the situation filled by a whole bunch of unimplemented regulations and ununderstood good intentions. I think we must be really careful for changing rules, since current one takes many years to be understood partly.
