[MITgcm-support] Re: No Subject

[ahoward]

Dimitris,

	Thank you. The experiment I am using is global_ocean.90x40x15 and
there is indeed a POLY3.COEFFS file, I was just looking in the wrong
place,
I found it quite logically in the inputs directory. I still have to
figure out what went wrong.

With respect to both Mellor-Yamada and our model, there is more than one
level. At level 2 1/2 one retains a prognostic differential equation
for the kinetic energy, at level 2 the turbulence models are fully
algebraic. I bring this up because I don't know whether the expensive
Mellor Yamada that your wife is reporting on was level 2 1/2. My
coworker Ye Cheng has applied level 2 1/2 in the atmosphere, but
so far I have only applied level 2.
	For level 2 our model (the GISS mixing model) is no more
expensive than KPP, in my runs in the NCAR CSM Ocean Model (1997 version)
and in a test in a preliminary version of the GISS coupled climate model
and in the Naval Research Lab's versionof HYCOM our model was never
more than 10% more time-consuming than KPP and in some cases was
slightly faster than it. This was after I did some tinkering with
the look-up tables used in my model were read, an earlier version
of my model had run slower simply because they were not being
accessed efficiently.  My current version of the GISS model does
involve pretty large look-up tables, which can create stack-size
problems if your stacksize limit is small, but I think I may be
able to decrease the look-up table size significantly without
degrading performance, however I am still working on this.

I've been working with Peter Stone at MIT. He is a consultant at
GISS and so was familiar with our work and interested in
giving it a try.

Armando

On Fri, 30 Apr 2004, Dimitris Menemenlis wrote:

> Armando, where did you get the set-up you are using.  If it's one of the
> standard verification experiments, it should already include the
> appropriate POLY3.COEFFS file.  If you have configured it yourself, then
> you need to use routines in subdirectory utils/knudsen2 to generate the
> POLY3.COEFFS file.  I highly recommend that you move to a newer version
> of the code as Ed suggested, especially if you are developing a new
> mixed layer parameterization.
> 
> >> in the kpp_calc subroutine comments you have
> OK, thanks.  I have changed comments for consistency.
> 
> How expensive is your turbulence closure model?  I have no personal
> experience, but my wife tells me that Mellor Yamada is very expensive.
> Is your scheme more efficient?
> 
> Also who are you working with at MIT?  Is it with John Marshall?
> 
> Cheers, Dimitris
> 
> -- 
> Dimitris Menemenlis <menemenlis at jpl.nasa.gov>
> Jet Propulsion Lab, California Insitute of Technology
> MS 300-323, 4800 Oak Grove Dr, Pasadena CA 91109-8099
> tel: 818-354-1656;  fax: 818-393-6720
>