ECHAM crashes with "FATAL ERROR in mo_time_base:Set_JulianDay: Fatal error!"

[ackerlar]

Hi,

I am trying to resume an AWI-ESM2 run from an existing PI control run. The parent EXP ends with the year 2249 and I want want the new simulation start with the year 1850. However, I get this error:

28/29 day in months invalid.
FATAL ERROR in mo_time_base:Set_JulianDay: Fatal error!

I also tried to resume from years 2248, 2250, 2251 but get always the same error. Does anyone know where this comes from? Seems like an issue with a leap year but the ECHAM restart files actually don't have any time information, do they?

[ackerlar]

ECHAM output is this

  0:  Read date (initial/restart):  2. Jan      350 00:00:00
  0:  Basic initialization of time manager <echam time manager> done.
  0:  Experiment resumed at:  1. Jan     2250 00:00:00
  0:  Set manager to resumed time step + 1.
  0:  Namelist file namelist.echam reopened
  0:  Reset resume date from namelist:  1. Jan     1850 00:00:00
  0:  Time step and start date evaluation done.
  0:  Initial state of time manager <echam time manager> is locked now!
  0:  State of manager >>echam time manager<<
  0:  Step counter :    133240320
  0:  Time step [s]:    450.000000000000
  0:  Initial date:  2. Jan      -50 00:00:00
  0:  Current date:  1. Jan     1850 00:00:00
  0:
  0:  Using DT_STOP for model stop.
  0:  Stop experiment at:  1. Jan     1851 00:00:00
  0:  Previous date: 31. Dec     1849 23:52:30
  0:  Next date    :  1. Jan     1850 00:07:30

To be honest I don't have a clue what tat means... Why is the initial date 2. Jan 350?

[ackerlar]

Okay, the information is in the fdate and nstep attributes in the ECHAM restart files... I changed them manually, so that I get:

  0:  Read date (initial/restart):  1. Jan     2249 00:00:00
  0:  Basic initialization of time manager <echam time manager> done.
  0:  Experiment resumed at:  1. Jan     2250 00:00:00
  0:  Set manager to resumed time step + 1.
  0:  Namelist file namelist.echam reopened
  0:  Reset resume date from namelist:  1. Jan     1850 00:00:00
  0:  Time step and start date evaluation done.
  0:  Initial state of time manager <echam time manager> is locked now!
  0:  State of manager >>echam time manager<<
  0:  Step counter :       280512
  0:  Time step [s]:    450.000000000000
  0:  Initial date:  1. Jan     1846 00:00:00
  0:  Current date:  1. Jan     1850 00:00:00
  0:
  0:  Using DT_STOP for model stop.
  0:  Stop experiment at:  1. Jan     1851 00:00:00
  0:  Previous date: 31. Dec     1849 23:52:30
  0:  Next date    :  1. Jan     1850 00:07:30

And the model runs for me. Are there any pitfalls, one needs to be aware of when changing these attributes?