Open aidanheerdegen opened 5 years ago
@nichannah I am looking into this
I would like to solve #193 using this approach, but I am confused about what timing information ACCESS-OM2 is currently using from cice_in.nml
. I am using 1deg_jra55_ryf
as a test run, with the runtime set to 1 month, so the first time it runs there is no difference, and then each subsequent run istep0
is incremented by 35040, which is what npt
is set to in cice_in.nml
.
But if we look in the ice_diag.d
files, it seems this is all being ignored
$ grep istep0 archive/output00?/ice/ice_diag.d
archive/output000/ice/ice_diag.d: istep0 = 0
archive/output001/ice/ice_diag.d: istep0 = 496
archive/output002/ice/ice_diag.d: istep0 = 944
archive/output003/ice/ice_diag.d: istep0 = 1440
archive/output004/ice/ice_diag.d: istep0 = 1920
$ grep npt archive/output00?/ice/ice_diag.d
archive/output000/ice/ice_diag.d: npt = 496
archive/output001/ice/ice_diag.d: npt = 448
archive/output002/ice/ice_diag.d: npt = 496
archive/output003/ice/ice_diag.d: npt = 480
archive/output004/ice/ice_diag.d: npt = 496
So is any of this stuff necessary? Seems like libaccessom2 is taking care of all of this, and we can do away with the reading from the previous cice_in.nml
completely.
Currently the cice5 driver copies the
cice_in.nml
from thework
directory to therestartXXX
directory as it contains model timing information that is needed to correctly restart CICE.cice_in.nml
is also a configuration file, so it is confusing that a file named the same is in two places.The relevant information required to correctly restart CICE5 should be extracted from
cice_in.nml
and saved to a new namelist file in the restart directory. Saycice_restart.nml
. This is clearer from a user point of view. Especially when the error pops up thatpayu
can't findcice_in.nml
when trying to restart an old run, wherecice_in.nml
was assumed to be in the correspondingoutputXXX
directory, but that directory no longer exists.