[MITgcm-devel] another bug in growth.F ?

[Martin Losch]

Dimitris,

thanks for your explanation, I'll get a chance to look at the code  
more carefully tomorrow (maybe).

About the advection: I have not throroughly tested the advection  
scheme 1 (1st order upwind), currently the only one that we (Jean- 
Michel and I) believe in. It will most likely lead to AREA > 1 in  
some cases. I have implemented advection of snow with the default  
scheme (turn on with SEAICEadvSnow=.true.), but have not tested that,  
either. It should work as long as you use LAD=2 (default, backward  
euler scheme) and should NOT work with LAD=1 (Leap-frog).

I believe that the advection of snow is probably not so important and  
recommend not using it for now.

Martin

On 10 Nov 2006, at 17:27, Dimitris Menemenlis wrote:

> Martin,
>
>> Thanks for sorting these things out (and checking them in). I am  
>> looking
>> forward to trying them out.
>
> Not sure if things are fully sorted out yet.  Jinlun's HSNOW fix  
> still blows up
> tangent-linear sensitivity
>
>> As a first impression from my 2x2 degree simulation, your changes  
>> to growth.F
>> dramatically increases seaice thickness in the southern ocean (and  
>> also snow
>> thickness), further, snow is only where there is ice (good), and  
>> my Weddell
>> Gyre is much better represented now.
>
> Technically what that bit of code is trying to do is to use any  
> residual heat in the surface level temperature, i.e., if surface  
> level temperature is still above sea-ice freezing point after  
> melting all the sea ice, to melt any residual
> snow.  So overall I expect surface water to be a bit colder and  
> fresher than without this fix, which is consistent with thicker sea- 
> ice and thicker snow over that sea ice, if does not completely melt  
> in the summer.
>
> But most Antarctic sea ice does melt during southern hemisphere  
> summer, so there should not be more than a year's worth of  
> accumulation in most places.
>
> But why is the tangent linear still so sensitive?
>
> In the hi-res cube-sphere tests should I turn on HSNOW advection?   
> Is that code fairly robust and debugged now?
>
> Cheers, Dimitris
>
> _______________________________________________
> MITgcm-devel mailing list
> MITgcm-devel at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-devel