[MITgcm-devel] Re: sea-ice

[Jean-Michel Campin]

Hi Dimitri,

> Jean Michel, I notice you decided not to change temperature directly
> within sea-ice model.  Where do you do the change so that I can do the
> same with pkg/seaice.
> 
> Cheers, D.

I did significant changes in the Therm.Sea-Ice pkg 
that explain why I changed also the name.
Since Stephanie was not really using the previous version
(I mean, not what Curtis checked-in a year ago), this new 
name should not be a problem.

Basically, I carefully tested that this version conserve
energy, fresh water and salt, but only did short test
(20 years) with NCEP forcing.
I choose not to change temperature but to return all the
sea-ice effects as fluxes (in Qnet, saltFlux and EmPmR).
The reason is that part a the freezing (that contributes to
the increase of ice-thickness but not ice extend) was already
returned as a heat flux. Therefore it seems logical to me 
to treat all the freezing components in the same way. 
I always use forcing_In_AB=.FALSE., and it seems to work like
this, but needs more and longer tests.
I don't use surfaceTendencyTice (since regarding the time-stepping, 
it's not an instantaneous adjustment) but it's not a problem (I mean,
the SeaIce pkg can perfectly use surfaceTendencyTice if necessary).

Completely different:
I noticed that you changed 2 data files:
> removed redundant monitorFreq in data files
Does it cause problems on some platforms ?
I did it on purpose, to show that 1 parameter can be specified 2 times,
and that in this case, only the last one matters. Also because  
monitorFreq =1.,
is quiet useful in making sure that the monitor is written 
at each time step, but might confuse some users who will think
that monitorFreq is set with iteration number units.

Cheers,

Jean-Michel