V5.10 introduces spinups for Surfex experiments, CREDAS, and default ouput hierarchy. It also fixes a few bugs

News in ECLIS V5.10

• CREDAS : a CREDAS is a CNRM-CM REference input DAta Set ; with ECLIS, this allows to greatly simplify the design of an experiment based on a reference CNRM-CM experiment ; this is described at length in two other articles, one for the concept of CREDAS, which is too a large xtent independent of ECLIS, and one for the use of a CREDAS in ECLIS
• Handling cycled spinups in Surfex offline runs : two types of spinups are handled. See relevant page.
• the name of BCOND files is checked at install against the GEOM parameter ; size is also checked for tl127
• script ECLIS/testing/test.sh allows for non-regression test on some ten varied configurations of ECLIS experiments on tori (coupled or not, with plugins or not...)
• the various conventions have been sorted out about the date used in naming the Arpege LBC or Nudging data file for the last day of the simulation (data of 1st day of following month, in ’GRED’ file). See the documentation at top of plugins/atm_lbc_nudging. In short : CAIAC and ASTER should use pattern ’DDDN’, and EAC and MEMO should use ’DDDD’
• parameter EXPARCHDIR stands for EXPeriment ARCHive root DIRectory ; if set, it will drive the organization of experiment ouputs on the archive machine, in an ’ECLIS-standard’ way, with the tree showing below ; however, you can set other values for any of the directories holding outputs (e.g. OCERESARCH, ICEOUTARCH, SFXOUTARCH, see page on parameters ...). The tree begins at directory EXPARCHDIR and is organized (if all components are used) as :
  • atm
    • output
    • restart
  • sfx
    • output
    • restart
  • oce
    • output
    • restart
  • ice
    • output
    • restart
  • riv
    • output
    • restart
  • cpl
    • output
    • restart
• Bug fixes
  • complementing Arpege restart with Surfex field (using parameter ASCEXE) was not OK for monthly restarts (but only for the 6-hourly ones)
  • in plugin atm_lbc_nudging : issue with date in GRE and GRED naming scheme
  • V5.7 and V5.8 were bugged regarding the fetching of atmospheric forcing files for Nemo stand-alone runs, when time-interpolation of forcing was invoked
  • in Nemo only when no SeaIce model is used
  • in cleaning the file transfer directory FTDIR
  • take IOSNAMREF into account even if no detached IO server is run
  • transfer of empty files to the archive was erroneously seen as a failed transfer
  • rebuild of Nemo outputs was suppressing its inputs even if rebuild failed (e.g. on tori front-end when IOServer was not used)
  • Nemo stand-alone Seaice outputs were not archived (since an undocumented version)