[MITgcm-devel] bug in seaice thermodynamics

[Jean-Michel Campin]

Hi,

We found an annoying bug in pkg/seaice, when trying to use the 
thermodynamics part only. Will check-in a fix soon.

The annoying aspect is that it might go unnoticed/unexplained. 
Will give some details of when it shows up and what it implies,
in case someone fall into the same trap.

The problem is that with SEAICEadvArea,SEAICEadvHeff = F
(default is SEAICEadvArea,SEAICEadvHeff = T, but one might be tempted
to switch them off to just use the seaice thermodynamics)
areaNm1 & hEffNm1 are not filled, so that, e.g., seaice_solve4temp 
does not update TICE which remains warm and ice can grow with almost 
no upper limit (Ross was getting 60.m of ice during the 1rst winter 
in some spots).

when: SEAICE_GROWTH_LEGACY defined,
 trying to use just the thermodynamics (SEAICEuseDYNAMICS=F),
 then one would expect that with no diffusion and no ice-velocity, 
 and defaults advection scheme (should not matter but the problem 
 is that it does), switching off SEAICEadvArea or SEAICEadvHeff
 would not change the results. Note that it works for the snow, 
 I can turn on/off SEAICEadvSnow, results don't change.

I think this bug is quiet old (probably as old as the 2 parameters
SEAICEadvArea,SEAICEadvHeff, 2007-04-30; I checked with a version
of the code from late Sept 2010, and was there already), but the 
side effects got worst in v.1.131 of seaice_growth.F (TICE no longer 
updated) or in v.1.135 (now getting a Floating exception if
I just set SEAICEadvArea=F but keep SEAICEadvHeff=T )

I might also swith those 2 flags off in one of the 
thermodynamics only seaice verification experiment.

Cheers,
Jean-Michel