Closed aekiss closed 3 years ago
This bug has now been fixed in cice. access-om2 and the control submodules have been updated to use the fixed version of cice, and the control submodules also now specify chio=0.006, which is the value that was actually used before.
This bug has now been fixed in the latest cice executables
/g/data/ik11/inputs/access-om2/bin/cice_auscom_*_dfb833a_libaccessom2_44e8821.exe
so if you upgrade to these with an old configuration you must also set chio
= 0.006 in cice_in.nml
in order to reproduce the previous behaviour (since this is what it was actually using previously, irrespective of the chio
value set in cice_in.nml
).
The value of
chio
is only used once, to calculatecpchr
here: https://github.com/COSIMA/cice5/blob/16fcc6d5fdffb918e4076473562346e47d3df016/source/ice_therm_vertical.F90#L691 but thencpchr
is redefined here before it is used for anything https://github.com/COSIMA/cice5/blob/16fcc6d5fdffb918e4076473562346e47d3df016/source/ice_therm_vertical.F90#L737-L744fbot_xfer_type = 'constant'
is the default, which effectively hard-codeschio=0.006
, ignoring whatever was set in the namelist file. This bug seems to have been introduced in the upgrade to cice 5.1.2 in 2015. Thanks to Paul Sandery for pointing out the insensitivity to this parameter.