Possible inconsistency in stormIdentifier assignment in operational tropical cyclone BUFR products (IFS and AIFS-ENS)

AIFS, IFS and OpenIFS
,
1

Hello,

I have found what appears to be an inconsistency in the stormIdentifier field of the operational tropical cyclone BUFR products.

I examined both the IFS and AIFS-ENS TC BUFR files from the ECMWF Open Data S3 bucket and found the same behaviour in both products.

The files examined are:

20260626/12z/aifs-ens/0p25/enfo/20260626120000-360h-enfo-tf.bufr
20260626/12z/ifs/0p25/enfo/20260626120000-360h-enfo-tf.bufr

20260626/18z/aifs-ens/0p25/enfo/20260626180000-360h-enfo-tf.bufr
20260626/18z/ifs/0p25/enfo/20260626180000-144h-enfo-tf.bufr

20260627/00z/aifs-ens/0p25/enfo/20260627000000-360h-enfo-tf.bufr
20260627/00z/ifs/0p25/enfo/20260627000000-360h-enfo-tf.bufr

20260627/06z/aifs-ens/0p25/enfo/20260627060000-360h-enfo-tf.bufr
20260627/06z/ifs/0p25/enfo/20260627060000-144h-enfo-tf.bufr

20260627/12z/aifs-ens/0p25/enfo/20260627120000-360h-enfo-tf.bufr
20260627/12z/ifs/0p25/enfo/20260627120000-360h-enfo-tf.bufr

The stormIdentifier values changed as follows.

2026-06-26 12 UTC

HIGOS      -> stormIdentifier = 09W
MEKKHALA   -> stormIdentifier = 08W

2026-06-26 18 UTC

HIGOS      -> stormIdentifier = 09W
MEKKHALA   -> stormIdentifier = 08W

2026-06-27 00 UTC

HIGOS      -> stormIdentifier = 08W
MEKKHALA   -> stormIdentifier = 08W

At this cycle, two different tropical cyclones in the same BUFR file had the same stormIdentifier.

The two messages have different longStormName values and clearly different locations.

2026-06-27 06 UTC

HIGOS      -> stormIdentifier = 08W
MEKKHALA   -> stormIdentifier = 08W

The same duplication is still present.

2026-06-27 12 UTC

MEKKHALA   -> stormIdentifier = 07W

HIGOS is no longer present in the BUFR file at this cycle.

The same behaviour was reproduced in both the IFS and AIFS-ENS TC BUFR products.

According to the JTWC operational numbering, I expected:

MEKKHALA -> 07W
HIGOS    -> 08W

Therefore I would like to ask:

  1. Is this a known issue in the TC BUFR products?
  2. Was there any correction to the tropical cyclone metadata or BUFR encoding between the 2026-06-27 06 UTC and 12 UTC cycles?
  3. Is stormIdentifier intended to match the JTWC storm number for named tropical cyclones?
  4. For identifying the same tropical cyclone across forecast cycles, is longStormName currently the recommended identifier instead of stormIdentifier?

Thank you very much for your help.

2

Dear Kotaro Takahashi,

Thank you for this report.

This is a known behaviour and stems from a report from more than one meteorological centre.

  • ECMWF receives tropical cyclone reports from both the Met Office- SXXT50 bulletin, which compiles information from JTWC and other centres with tropical cyclone monitoring responsibilities- and from RSMCs. When both are available, the report issued by an RSMC supersedes the report issued by a non-RSMC to trigger the TC forecast products.
  • This dual-source approach helps mitigate the risk of missing data if one reporting source stops sending reports.
  • It also enables tropical cyclone forecast products to be triggered as soon as a report is issued by any of the reporting sources.
  • The main caveat is the inconsistency between sources in the assigned storm ID.

We will update our documentation to reflect this information.

Emma

3

Dear Emma,

Thank you very much for the explanation.

I understand that the storm ID may differ between reporting sources and may therefore change or temporarily be duplicated in the TC BUFR products when reports from different centres are used or superseded.

I will take this behaviour into account when processing the data.

Thank you also for updating the documentation.

Best regards,

Kotaro Takahashi