[Mitgcm-support] RE: Summary of model implementation problems

[mitgcm-support at dev.mitgcm.org]

Hi Sai,

 Sorry its being a pain. I'm pretty confident if you
start from Helen's setup (code version etc...) it has
to work. That seems like best way to go for now.
We can worry about why the c46 code doesn't cut it
later.

Chris

-----Original Message-----
From: Sai Ravela [mailto:ravela at MIT.EDU]
Sent: Wednesday, October 30, 2002 1:16 PM
To: marshall at plume.mit.edu; cnh at plume.mit.edu
Subject: Summary of model implementation problems


hi-

	a bit of a disappointing week, I have not made much progress. The basic
problem here is that the circulation does not set itself up and, instead,
other types of solutions with macabre temperature gradients or surface waves
get generated. With some help from Jean-Michel (sp?) I seemed to make
progress, but his suggestions did not work either. Now, here are the
suggestions on plate:

1. Repeat Helen's experiment
2. Aspect ratio may be a problem. Helen used 1cmx1cmx1/2cm - or so the data
file says - whereas I have 1/2cm x 1/2cm x 1cm.
3. Randomization of initial conditions, limiting solver to few iterations,
Forcing boundary to have negative temperature gradient (time), playing with
the time stepping parameters. These have been tried with success in the
sense that the system runs longer and, in some cases, what might look like a
reasonable circulation for a short period (minutes).
4. Removing preconditioning, non-hydrostatic mode, adding a diffusive term
\lambda(T_can - T), increasing resolution at the center, gradation of
molecular viscosity are all things that have not been tried. We also don't
know what if there is some numerical issue as the delZ integrations to
surface - bottom do not add up to zero but have a small residual left.

Per John's suggestion I will start with Helen's experiment when Chris comes
back to see if we can reproduce everything there and then move back to
Hadley. In the meanwhile, I am switching over to getting the framegrabber up
and running with the box.

a bit frustrated...Sai