[Mitgcm-support] RE: OTC

Wed Jul 9 15:53:56 EDT 2003

Hi Boyin,

 Your numbers sound very odd! Could it be that the
now four years old and frequently modified code setup 
you are starting from has been broken along the way?

 For the release MITgcm (http://mitgcm.org), a coarse 
resolution climate problem with async. time stepping
works as expected i.e. for a 4 degree run, the async. time
step is around two days for thermodynamics 
(and implicit free surface). The global_ocean.90x40x15 test 
case from the release code illustrates the parameter settings
for Ah, Az, dt for momentum etc...

Chris

-----Original Message-----
From: Boyin Huang [mailto:huang_boyin at yahoo.com]
Sent: Monday, December 02, 2002 1:07 PM
To: Jeff Scott
Cc: cnh at plume.mit.edu; vikram at crces.org
Subject: Re: OTC

Hi, Scott:

Thank you for your concern about my MIT OGCM setting.

The OTC looks OK now except some detailed structure.

The problem now is that the time step of 150 second
for momentum equations is really too small (2 hours
for tracer equations): In the parameters of viscosity
Ah=5.e4 and Az=1.e-4, dt=150 is already the maximum,
although Ah and Az should be smaller due to the
resolution I am using according to the paper Bryan et
al. (1975) in JPO. (the vertical diffusivity is 1.e-5,
and the isopycnal diffusivity is set to 1.e3).

The model brews up when I increase dt for momentum eq
or reduce Ah to 1.e4.

The time step of similar resolution in MOM or other
model is obviously much longer than 150s. I want to
know why. Is this due to that the MIT model does not
separate the barotropic and baroclinic modes?

Comments from you and Chris are very much appreciated.

Boyin

--- Jeff Scott <jscott at wind.mit.edu> wrote:
> 
> Hi Boyin,
> 
> I was away for the past two weeks.... Have you made
> any progress on this
> or are you still having difficulties?
> 
> Jeff
> 
> On Mon, 18 Nov 2002, Boyin Huang wrote:
> 
> > Hi, Scott:
> > 
> > It seems that I have to use the new version of
> > MITOGCM:
> > MITgcm-r1p5. 
> > 
> > Since I will need to set the model for my work
> here, I
> > am setting the model as follows (I hope I will be
> able
> > to do the convection problem with the new version
> of
> > model) but still find the OTC is weak at year 15:
> > 
> > a realistic topography from 80S to 88N. The
> > latitudinal
> > resolution is 0.5 degree near equator, increases
> to 2
> > degrees at 20N and beyond. The longitudinal
> resolution
> > is 2 degrees. The vertical resolution is 10m for
> first
> >  50m, 25m between 50 and 150m, then 50m, 75m ...
> etc. 
> > 
> > The model is initialized by annual Levitus T and
> S,
> > forced by monthly Levitus SST and SSS with
> restoring
> > (don't worry about for the restoring for the time
> > being) time of 10 and 30 days.
> > 
> > I found the OTC is only about 5 Sv at year 15,
> does to
> > have signature to increase.
> > 
> > I suspect the weak OTC may be associated with
> strong
> > viscosity: I set
> > viscAz = 1.E-3,
> > viscAh = 5.E+5,
> > diffKzT= 1.E-5,
> > diffKzS= 1.E-5,
> > ivdc_kappa          = 100.,
> > GM_background_K    = 1.e+3,
> > 
> > To ensure the model stability, I have to set 
> > deltaTmom    =     150.,
> > tauCD        =   36000.,
> > deltaTtracer =    7200.,
> > deltaTClock  =    7200.,
> > abEps        = 0.1,
> > 
> > I am not sure whether I can increase the OTC by
> > reducing viscosity. On the otherhand, the
> deltaTmom is
> > already too small, and there is no possibility to
> > reduce it (3hr cpu time on 2ghz computer for 1 yr
> > integration).
> > 
> > I wish to see your comments.
> > 
> > 
> > Boyin Huang
> > 
> > 
> > 
> > 
> > 

=====
Boyin Huang
The Center for Research on the Changing Earth System (CRCES)
10211 Wincopin Circle, Suite 240
Columbia, Maryland 21044
Tel. 410-992-5704 (O), 410-737-4557(H)
Fax. 410-992-5944, Email. huang_boyin at yahoo.com
http://www.DecVar.org

__________________________________________________
Do you Yahoo!?
Yahoo! Mail Plus - Powerful. Affordable. Sign up now.
http://mailplus.yahoo.com