[MITgcm-devel] problem with SEAICEadvSnow

[Martin Losch]

Dimitris,

hice is only used as input to seaice_budget_ice. It control the  
penetration of short wave radiation and the effective conductivity of  
heat through the ice. There it appears in the denominator, and so  
does hsnow (in seaice_budget_ice):
>          effConduct=
>      &        XKS/(hSnwLoc(I,J)/hice(I,J)+XKS/XKI)/hice(I,J)
> C     These two formulations give different results than
> C     original formulation. Bummer!
> C         effConduct= 1. _d 0 / ( hSnwLoc(I,J)/XKS + hice(I,J)/XKI )
> C         effConduct= XKS*XKI / ( hSnwLoc(I,J)*XKI + hice(I,J)*XKS )

Maybe you can try
A: one of the two other formulations that may be more stable the the  
original one (but do not give the same results in testreport)
B: cap hsnwloc the same way as hice in seaice_growth.


>> SEAICEadvScheme   = 1, ! this is not the default. The default is 2  
>> (2nd order
>> central differences with modified euler time step, the original code)
>
> Is "SEAICEadvScheme = 1" supposed to be more stable than "2"?

Yes, 1st order upwind ("SEAICEadvScheme = 1" ) is more stable and  
very smooth. Actually too smooth. In the end we would like to go with  
33, but not sure if the works properly (I think it does).

Martin

>
> D.
> _______________________________________________
> MITgcm-devel mailing list
> MITgcm-devel at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-devel