[MITgcm-support] MITgcm

Alistair Adcroft adcroft at MIT.EDU
Mon Jan 5 08:45:21 EST 2004


It is clear the model is numerically unstable for you set up and most likely
the problem is in the vertical direction: assuming monitorFreq was sensible
(in data:PARM01) there should be lines in the standard output of the form
"(PID.TID 0000.0001) %MON advcfl_*" which are the maximum local CFL values.
"advcfl_W_hf_max" is the vertical CFL number including the variable grid
resolution and most likely to be above 0.2 . If any of these numbers are
above 0.2 the model is close to or past being stable. In which case, you
need to reduce the time-step.

A.

-----Original Message-----
From: mitgcm-support-bounces at mitgcm.org
[mailto:mitgcm-support-bounces at mitgcm.org] On Behalf Of Zhaipu Ma
Sent: Wednesday, December 24, 2003 9:43 AM
To: MITgcm-support at mitgcm.org
Subject: [MITgcm-support] MITgcm


Dear Sir,
 I have a questions to ask about MITgcm.
In verification/exp4/, I adapt the topography to my own, ny=49,nx=61, nr=10;
delX=delY=9260m, delZ=3m~10m, deltaT=300, with no other changes, I find that
if my delZ=6*3,2*5,10,27, (total 10 layers),  mitgcmuv runs for nrarly 2300
iterations, and then  comes  NAN  for rhsMAx, I change the delZ to 10*10,
then runs 5400  itertions, If I change 10*12 for delZ, then runs for 9600
iterations, Can mitgcmuv run without stop,unless it meet nTimeStep? If the
max depth of shelf sea is less than 100m,say, 55m, could MITgcm be use to
the shelf sea? I find it is always come a NAN when delZ is less than 10m.
and if I change the delZ to more than 10m, say,12m, then it runs smoothly. I
check the Mw, Sl,Si Sa, they are all under the limit for stability.

mazhaipu
sincerely.

_______________________________________________
MITgcm-support mailing list
MITgcm-support at mitgcm.org
http://dev.mitgcm.org/mailman/listinfo/mitgcm-support





More information about the MITgcm-support mailing list