What was the problem ?
The daily od glb3x2 processing at ECMWF produced zero downdraft entrainments/detrainments during the period:
2004-05-09 to 2005-07-12As a result, no tracer mass is transported downwards by the downdraft .
Who should be worried ?
Everybody who has used 'od-fc-ml60-glb3x2-sub' files for the period May 2004 to July 2005 during the same period.
Affected files have been storred in the following archives:
ecmwf : ec:/nlh/TM/meteo/od/fc/ sara : /shared/kelder/TM/meteo/od/fc/ knmi : mos:/fa/ks/TM/meteo/od/fc/The affected monthly tar files are smaller than the proper ones:
177367040 2004/01/ml60/glb3x2/sub_200401_21p06.tar 167362560 2004/02/ml60/glb3x2/sub_200402_21p06.tar 178503680 2004/03/ml60/glb3x2/sub_200403_21p06.tar : 180592640 2004/04/ml60/glb3x2/sub_200404_21p06.tar <-- ok 150845440 2004/05/ml60/glb3x2/sub_200405_21p06.tar <-- zeros! 133058560 2004/06/ml60/glb3x2/sub_200406_21p06.tar : 137461760 2004/07/ml60/glb3x2/sub_200407_21p06.tar 136611840 2004/08/ml60/glb3x2/sub_200408_21p06.tar 129546240 2004/09/ml60/glb3x2/sub_200409_21p06.tar 128829440 2004/10/ml60/glb3x2/sub_200410_21p06.tar 120709120 2004/11/ml60/glb3x2/sub_200411_21p06.tar 127232000 2004/12/ml60/glb3x2/sub_200412_21p06.tar
The affected files are reprocessed in July 2005 .
A check on zero downdraft entrainments/detrainments has been added to the meteo routine of TM5. Please update your base version to 2005-07-15 or later.
This was a pure processing error, not a programming error. Might be caused by linking a mixture of older and newer object files.
To avoid this kind of failures, the source of the daily processing will be recompiled every day from now on.