[MITgcm-support] Freshwater OBCS problem

[Dustin Carroll]

Thanks Jody and Dimitris for all your helpful suggestions,

Bringing the Tstep way down to 3s fixed the problem with cg2d diverging.

Cheers,
Dustin

> 
> I also tried turning off the forcing and just setting certain locations of the initial T and S to (1 psu and 1 deg C). This causes the model to get negative salinities as well.
> 
> Well, this is a good clue.  My guess is that you are violating the CFL criterion, perhaps because your initial density profile is unstable and convecting (did you check it? I can never remember if we fill from the bottom up or top down, and often set the model up upside down, which is always embarrassing).  30 s time steps for 100x2.5 m resolution is not small enough, unless your model isn't doing anything.  I'd drop it to something more like 6 s, or even 3 s if you are running non-hydrostatic.  
> 
> Get the unforced case running first, and then you can worry about the forcing. 
> 
> Again, to get things going, I set dumpFreq=deltaT if I am having problems, and then plot the data to see where the output fails.  Often helps me figure out the problem.  Strongly suggest you make some plots of your data.
> 
> Cheers,   Jody