[MITgcm-support] upper ocean temperature drift

[David Wang]

Hi,

Set aside the KE drift, I have been frustrated by not getting the tropical
upper ocean into thermal quasi-equilibrium after a few (*300*) years of
spin-up. The model configuration is similar to the ECCO-JPL 1 x 0.3-1 x L46
setup, and is simply forced by COADS monthly climatological wind stress (da
Silva et al. 1994) and Levitus monthly climatological restoring SST/SSS,
with asynchronous timestepping. Both kpp and gmredi pkg are used. I expected
the tropical upper ocean T/S to be stabilized much sooner, but during the
last 50 years of the 300-year spin-up, the mean upper ocean (0-500 m)
temperature in the tropical Pacific (23.5 N/S) has still increased by about
0.04 degC. The temperature evolution as a function of depth and time over
the last 100 years (model year 201-300) shows that the drift is NOT
negligible (http://fats-raid.ldeo.columbia.edu/~dwang/mcane/theta2.gif gives
the temperature anomaly averaged over two regions during model year 201-300,
relative to the first 20 years of this period. The lower panel shows that
even at the equator, the drift is still evident).

Now I really have no idea how to proceed with this. Should I run a few more
hundred years of spin-up (which I hesitate to do due to computational cost)?
Is background vertical diffusivity too small (5.e-6 m^2/s)? Does
asynchronous timestepping interact with kpp/gmredi? etc, etc...

If anyone out there can help to troubleshot, I will very appreciate it and
provide my data namelist.

Thank you!
David

On Wed, Feb 18, 2009 at 5:16 PM, David Wang <climater at gmail.com> wrote:

> Hi there,
>
> I'm bothered with a 100-year spin-up of an intermediate resolution (about
> 1x1), near global ocean, with a recent version of MITgcm, which keeps losing
> kinetic energy (momKE_ave by the way of diagnostics pkg). I have a rather
> simple setup with horizontal bi-harmonic viscosity, KPP vertical viscosity,
> no slip bottom and free slip side, as well as asynchronous time stepping for
> momentum and tracers (see also the attached "data" namelist). GMREDI pkg is
> also used. The surface mechanical condition is climatological monthly wind
> stresses from da Silva et al (1994), and the surface buoyancy condition is
> simply SST/SSS relaxation to their monthly climatologies. The model started
> from the rest and climatological temperature/salinity. I've attached a plot
> of the global-mean KE which shows a clear downward trend. However, I expect,
> in principle, the global-mean KE will be stabilized after a few decades at
> most. I checked the wind stress diagnostics (oceTAUX, oceTAUY) which do the
> correct job forcing the ocean year after year with the same seasonal cycle.
> The model does have a reasonable general ciculation. But the leaking KE is
> disturbing to me and makes me doubt the integrity of the run. Since I only
> have limited computational resouce, I'd like to see if anyone out there may
> have a guess what could go wrong with my model run.
>
> Thank you!
> David
>
> --
> life grows, death doesn't.
>



-- 
life grows, death doesn't. there is no dress rehearsal.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mitgcm.org/pipermail/mitgcm-support/attachments/20090316/7496a9a9/attachment.htm>