[MITgcm-support] convective adjustment

[Baylor Fox-Kemper]

Hi,
   I have been fiddling with the convection schemes recently, and would 
like clarification as well.  I will tell you how I think it works, and 
perhaps Jean-Michel et al. can correct me?
   Cheers,
     -Baylor

There are currently 3 types of parameterizations of convective mixing 
available in addition to the regular vertical diffusivity and 
viscosity.  They all differ from regular vertical diffusivities in that 
their behavior depends on the density field.

1)  Instantaneous mixing by rearrangement of density pairs.  This is 
the simplest convective adjustment scheme, although it may cause odd 
numerical instabilities (Cessi, P., and W.R. Young, 1996. Some 
unexpected consequences of the inter-action between convective 
adjustment and horizontal diffusion. Physica D., 98, 287-300.)  It can 
be used by setting cadjfreq in your data file section PARM03.  If you 
set it greater than zero, then this will be the interval between checks 
to see if convective adjustment is necessary.  If you set it less than 
zero, then it checks every time step.

2)  Mixing by increased diffusivity when there is unstable density 
profile.  This is very similar to 1), except instead of instantaneously 
swapping densities between adjacent levels, the model just cranks up 
the vertical diffusivity to a value which is set by the parameter 
ivdc_kappa which you specify in data file section PARM01.  All you need 
to do to use this instead of 1) is to set this parameter.  cadjfreq 
must also be set, and I believe in this case it specifies the frequency 
which the model checks to see if it needs to turn up or turn down the 
diffusivity.

3) K-Profile Parameterization (KPP).  This is a paramerization scheme 
for 3-d turbulence.  It can be used in place of the other forms of 
convective adjustment, as it will also serve to mix away unstable 
density profiles.  However, it also includes variable viscosity and a 
non-local flux parameterization and a number of hidden parameters.  
Basically, to use KPP, you need to put useKPP=.TRUE. in data.pkg in 
your execution directory and put "kpp" in packages.conf (which can be 
in your build directory or your code directory).  You can then edit 
KPP_OPTIONS.h if you'd like to change what KPP is doing.

Useful KPP References:

Large, W.G., McWilliams, J.C., Doney, S.C., 1994. Oceanic vertical 
mixing: A review and a model with a nonlocal boundary layer 
parameterization. Reviews of Geophysics 32, 363–403.

Large, W.G., 1998. Modeling and parameterizing the oceanic planetary 
boundary layer. In: Chassignet, E.P., Verron, J. (Eds.), Ocean Modeling 
and Parameterization. Kluwer Academic Publishers, Netherlands, pp. 
81–120.


4)  Another type of convection is resolved non-hydrostatic convection.  
This will require very fine resolution and small time steps...and it is 
a whole other issue...


Cheers,
    -Baylor



On Jan 25, 2005, at 6:37 AM, Dan Conipo wrote:

> Hello.
> I am use the code to simulate the circulation in a box
> ocean. I started
> using the data file of the experiment 2 in the
> tutorial.
> I do not understand how the convective mixing works
> there. It seems
> that
> the mixing is linked just to the vertical diffusivity.
> Regards,
> Daniele
>
>
> 		
> __________________________________
> Do you Yahoo!?
> Take Yahoo! Mail with you! Get it on your mobile phone.
> http://mobile.yahoo.com/maildemo
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://dev.mitgcm.org/mailman/listinfo/mitgcm-support
>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: text/enriched
Size: 3580 bytes
Desc: not available
URL: <http://mitgcm.org/pipermail/mitgcm-support/attachments/20050125/77d60b20/attachment.bin>