Hi,
what is the sea surface elevation you start with? I guess you 
experience the following:

If you set up some salinity (density) distribution with large 
horizontal gradients it appears, that the barotropic mode 
rapidly adjusts to the density field in such a way, that the 
barotropic (sea level) pressure gradient balances the vertical 
integral of the baroclinic pressure gradients. (A very old 
adjustment problem which has been discussed first by Marsigli in 
the 16th or 17th century .. I have forgotten. I remember there 
is some remark on this in the textbook of Gill.) The point is, 
that the currents, which give a rough geostrophic balance to 
these gradients are missing in your initial field. Since the 
geostropic balance is the leading balance in most cases, the 
result may be very large currents which damage completely your 
initial field and may violate CFL criteria. 

A way out is not simple. To initialize a new model I am doing 
the following:
- edit the tracer subroutine and set the tracer time tendency to 
zero. (In some cases the vertical mixing must be switched on to 
get a reasonable surface boundary layer. This depends on your 
problem.) This means, that you have tracers constant.
- Run your model (with a sufficiently) small timestep. For 
simplicity you could switch off wind forcing. The result is a  
set of fields, tracers (baroclinic pressure gradient), sea 
level, baroclinic and barotropic velocities, where the (fixed in 
time) vertically integrated sea level gradient approximately 
balances the barotropic pressure gradient and currents are in 
geostrophic balance with the pressure fields. 
- I do not know, how long you must run the model to reach a 
stable state.
- Alternatively, you could diagnose the geostropic baroclinic 
velocity field from the initial data and could initialize the 
model with these fields. I propose to use FERRET for this 
purpose. 

- If you have found a sufficiently stable model state, replace 
the simplified executable with the original one and use the last 
restart file to initalize the new model.

- An alternative is to use the sourceterms in tracer to relax 
the tracer fields to the initial configuration.

It is possible, that I am completely wrong with my idea. Anyway, 
the problem described above is typical for free surface code. 
You need some spinup for every model configuration.

Greetings,
Martin




-------------------------------------------------------
This SF.Net email is sponsored by: INetU
Attention Web Developers & Consultants: Become An INetU Hosting Partner.
Refer Dedicated Servers. We Manage Them. You Get 10% Monthly Commission!
INetU Dedicated Managed Hosting http://www.inetu.net/partner/index.php
_______________________________________________
FMS-mom4 mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/fms-mom4

Reply via email to