[MITgcm-support] orlanski open boundary conditions

[Martin Losch]

Sorry,
I send the email prematurely. Originally, I had set ALL OB_*=100*-1, 
which then resulted in problems only at the north and east, where this 
parameter value is appropriate, at the south and west I got something 
like a solid(?) boundary and the effects I described did not appear.

But with the "correct" values
OB_Jnorth(:)=-1,OB_Ieast(:)=-1,OB_Jsouth(:)=1,OB_Iwest(:)=1
I get the problems at ALL boundaries. So there is some sort of symmetry.

I put two experiment configuration on (and the output obtained on 
sea.mit.edu)
sea:/data/mlosch/MITgcm/vortex
sea:/data/mlosch/MITgcm/small_vortex
(cvs co -r checkpoint_51f_pre MITgcm_code)
vortex is (100x100x29) and has 5km horizontal resolution
small_vortex is (20x20x10) and has 25km horizontal resolution
the experiment-directories are also in sea:~mlosch/vortex.tgz and 
sea:~mlosch/small_vortex.tgz but in those the .genmakerc-files in the 
code-directory need to be changed (remove "recom" from the disable list 
and unset the ROOTDIR variable)

Experiments start from rest with an intial temperature/salinity 
perturbation (temp.ini,salt.ini), open boundaries with Orlanski 
radiation conditions on all boundaries, I use the standard timestepping 
(see data file), linear free surface. Integration is 3months (the runs 
don't get there) with daily snapshots. Obviously, volume is not 
conserved (mean sea level decreases and then increases in the coarse 
model, oscillates with high amplitudes, .4 meters, increases and then 
goes to negative infinity in the high resolutioin experiment). In the 
high resolution experiment, the high vertical velocities near the 
boundaries are more obvious than in the coarse run.

BTW, different problem: I tried double periodic boundary conditions on 
the same configuration, but then the gravity waves do not damp out in 
the high resolution run, unless I use high laplacian friction 
(1000m^2/s for 5km resolution), and the model explodes, too. Why is 
that? Coriolis constantly adding energy? I tried 
useEnergyConservingCoriolis=.true., but that doesn't help.

Puzzled,

Martin

On Tuesday, December 2, 2003, at 05:20  PM, Alistair Adcroft wrote:

> Do you really mean "only North and East" has problems? Clearly a bug 
> if so -
> make sure your initial conditions are symmetric and beta=0. If the 
> OBCS are
> assymetric then lets find the bug.
>
> A.
> --
> Dr Alistair Adcroft            http://www.mit.edu/~adcroft
> MIT Climate Modeling Initiative        tel: (617) 253-5938
> EAPS 54-1523,  77 Massachusetts Ave,  Cambridge,  MA,  USA
>
> -----Original Message-----
> From: mitgcm-support-bounces at mitgcm.org
> [mailto:mitgcm-support-bounces at mitgcm.org] On Behalf Of Martin Losch
> Sent: Tuesday, December 02, 2003 7:33 AM
> To: MITgcm Support
> Cc: Samar Khatiwala
> Subject: [MITgcm-support] orlanski open boundary conditions
>
>
> Hi,
>
> is there any experience on how well the open boudary conditions with
> orlanski radiation  are working? I would like to study the
> relaxation/spin-down of a solitary geostrophic eddy and from my initial
> adjustment I get all sorts of odd processes near the open boundaries
> (only "north" and "east"). I had hoped that the gravity waves would be
> more or less radiated out of the domain, which they mostly do, but
> there are strong vertical velocities near the open boundaries and
> eventually, the (linear free) sea-surface height is affected (its mean
> goes down by meters) and the solution diverges (explodes).
>
> Any experience with this? Any ideas?
>
> Martin
>
> PS. this is my data.obcs
>   &OBCS_PARM01
>   OB_Jnorth=100*-1,
>   OB_Jsouth=100*1,
>   OB_Ieast=100*-1,
>   OB_Iwest=100*1,
>   useOrlanskiNorth=.TRUE.,
>   useOrlanskiSouth=.TRUE.,
>   useOrlanskiEast =.TRUE.,
>   useOrlanskiWest =.TRUE.,
>   &
>
>   &OBCS_PARM02
>   cmax=0.45D0,
>   cvelTimeScale=500.D0,
>   &
>
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://dev.mitgcm.org/mailman/listinfo/mitgcm-support
>
> _______________________________________________
> MITgcm-support mailing list
> MITgcm-support at mitgcm.org
> http://dev.mitgcm.org/mailman/listinfo/mitgcm-support