[MITgcm-support] SEAICE pkg, unstable HEFF

[Martin Losch]

Hi Mari,

there are many things that can go wrong with such a configuration, so that the information that you provide is not quite sufficient. If you want more help, I’d ask you to send all namelist files (data*) and all *_OPTIONS.h files that you have modified (ie., the *_OPTIONS.h files in the “code” directory), in particular the SEAICE_OPTIONS.h and CPP_OPTIONS.h file.

based on what you sent I can say this:
> On 29 Mar 2016, at 14:10, Mari Fjalstad Jensen <Mari.F.Jensen at uib.no> wrote:
> 
> Hi,
> I am having some problems with the seaice pkg.
> I am doing some simulations with a run for a 1000*2250 km domain, 5km horz resolution and 30 depth levels, SEAICE pkg on.
> 
> Freshwater is added through EmP in a roughly 25000 km^2 area.
> The run crashes after about 10 years when adding 0.08 Sv of precipitation. The problem seems to be in two grid cells where the sea-ice thickness (HEFF) increases with 50 m in one time step (400s). The grid cells are located just at the edge of where I add freshwater. I do not have this problem when adding less or no freshwater.
How do you add this? Isn’t 0.08 Sv *1e06 (m^3/Sv) / (25000e6m^2) = 3.2e-06 m/s of water a pretty high precipitation rate? How much less do you need to add to avoid the instability?
> 
> Any help with which SEAICE parameters to change would be highly appreciated.
> 
> # SEAICE parameters
> &SEAICE_PARM01
>      SEAICEwriteState   = .TRUE.,
>      SEAICEadvScheme    = 33,
>      SEAICE_initialHEFF = 0.0,
>      SEAICE_deltaTtherm = 400.,
>      SEAICE_deltaTdyn   = 400.,
>      SEAICE_deltaTevp   = 20.,
If you want to use EVP, I recommend using the new schemes of Madlen Kimmritz, which is implemented in the most recent MITgcm, see Kimmritz etal 2015. On the convergence of the modified elastic-viscous-plastic method of solving for sea-ice dynamics. Journal of Computational Physics, 296, 90-100, doi:10.1016/j.jcp.2015.04.051, 2015, and "The adaptive EVP method for solving the sea ice momentum equation”, just accepted in Ocean Modelling: <http://mitgcm.org/~mlosch/adaptiveEVP_accepted.pdf>
Some if this is already in the online-documentation. I recommend using parameters SEAICE_evpAlpha=SEAICE_evpBeta=500 and SEAICEnEVPstarSteps = 200 or so. That may not lead to fully converged solutions but they will be more or less smooth.
But I think your solution will be just as sane with the default VP parameters (i.e. just comment out deltaTevp), and much better (if smoothness and stability count) with order 10 non-linear iteration steps (NSPEUDOTIMESTEPS = 10 or if you have recent code SEAICEnonLinIterMax=10).
>      SEAICE_multDim     = 1,
>      ICE2WATR           = 0.92D0,
> SEAICEadvSnow = .FALSE.,
> SEAICEuseFlooding  = .FALSE.,
These two parameters only make sense if there is no snow in the simulations.

> 
> ----------
> I have tried varying diffusion/viscosity parameters and domain where I add freshwater without success..

Martin

> 
> Best Regards,
> Mari
> 
> 
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-support