[MITgcm-support] OBCS for sea ice!

[Martin Losch]

Hi Liqun,

I think the boundary conditions along solid walls are appropriate: no flow (normal velocity = 0) and you can select no-slip (free-slip seems to be the default for some reason), and no further BC (for ice strength) are required.
To be honest, I have not thought seriously about open boundary conditions for the stress terms yet, because, as I said earlier, the model was not really intended to be run in this kind of configuration. Currently you can specify thickness, concentration, and velocities along the open boundary. Then there are a couple of CPP-flags that control the code in obcs_adjust_uvice.F, I guess you can try those. But maybe you need to use completely different BCs. I have seen papers were people use boundary conditions for the ice strengh P (variable press, or press0), e.g. P=0 on the boundary. Please have a look around and try things out. If you find a good solution we can think of incorporating it into the distributed code.

What instabilities in the interior are you talking about? What do they look like? Again, I would need more details to provide better support. In general the default solver does a good job of smooth solution, but at high resolution, sharp linear kinematic features appear (as in your previous plot). These are actually interesting and sought for and not a problem of the code. I do not recommend to use the EVP solver, it's always a little noisy, and the free-drift code is really not proper sea-ice dynamics solver. We uses it for special cases and for debugging.

Martin

On Jun 14, 2013, at 2:22 AM, <liqun_hu818 at yahoo.com.cn> wrote:

> Hi, Martin:
>       I have tested different solver for the sea ice dynamics, and one questions is that what is  the boundary conditions for the sea ice internal stress calculation under default? I think unrealistic BCs for the calculation of ice stress could be one potential problem for the instability characteristics, which happened in the interior region, but not started from the boundary. When I use the freedrift configuration, it seems that the boundary condition of sea ice has been correctly used, the modeled variables were not perfect, but more resonable.
> Best wishes,
> Liqun
>