[MITgcm-support] Re: No Subject

[ahoward]

Thanks for your prompt response.
Could you tell me where one sets the reference profiles?
I can't find a file named POLY3.COEFFS so that may be
the whole problem though.

The JMD95Z eostype doesn't seem to be in the r1p8 model
version (last stable frozen according to the website)
I'm using, so maybe I really do need to go to a later
version as Ed Hill suggested.



With respect to the diffusivities in KPP
what confused me is that although the code does indeed have

 KPPdiffKzS = vddiff(...,2) = diffus (imt,2) = scalar = salt/tracer
 KPPdiffKzT = vddiff(...,3) = diffus (imt,3) = temperature = temperature

in the kpp_calc subroutine comments you have
c     vddiff (nx,ny,Nrp2,1)- vertical viscosity on 
"t-grid"           (m^2/s)
c     vddiff (nx,ny,Nrp2,2)- vert. diff. on next row for temperature
(m^2/s)
c     vddiff (nx,ny,Nrp2,3)- vert. diff. on next row for salt&tracers
(m^2/s)

which says that "2" is for temperature and "3" is for salt not
vice-a-versa. 


        The vertical mixing parameterization that I am implementing is the
one described in "Ocean Turbulence. Part II: Vertical diffusivities - 
Momentum, Heat, Salt, Mass and Passive Scalars" by V. Canuto, A. Howard,
Y. Cheng & M. Dubovikov in JPO Volume 30,p.240, 2002. It is more
theory-based than KPP and has less to tune in it. It may be thought of
as a generalization of the Mellor Yamada model, but because our turbulence
closure is more complete turbulence exists up to a Richardson number of
~1, instead of dying at Ri~0.2 as in Mellor Yamada. Salt was treated on
an equal basis with heat in deriving the equations so that
double-diffusion is incorporated into the theory naturally rather than
being dealt with by adding on separate parameterizations. We also 
apply the same turbulence model to the background mixing due to
internal waves, although there we need to assume wave dissipation
as an input, taken as a constant/N^2 in the above paper but
modulated by the latitude dependence of Gregg et al.'s recent 
nature paper in my current work.

	I have not looked at the performance of our turbulence model
in the antarctic, although it would be interesting to do so. In a
paper that recently appeared in Ocean Modelling, "Modeling Ocean
Deep Convection" by Canuto et al., we looked at how it behaved
in the Labrador Sea in a coarse resolution OGCM and found that
the pattern of deep convection looked quite different from,
and fairly better than, KPP in the same code. 

I came out here to MIT to insert our mixing in the MIT OGCM,
but the task turned out to  be bigger than I realized, and
my allotted time here is up now. I will, however, continue
working on this from New York, where I am based, at 
NASA-GISS.

Your advice, assistance is greatly appreciated.