[MITgcm-devel] speed of sound

[Jean-Michel Campin]

Hi Gael,

Following your email + phone conversation, I would like to add this:

There should be 1 computation of the sound-speed consistent
with each EOS (I think exactly like alpha and beta, it's simply related 
to the 3rd partial derivative of rho(T,S,p) but at constant S and entropy), 
and this should go in model/src with the other EOS calculation.

Now, there are (in MITgcm code) other calculations of EOS terms which are 
not consistent with the EOS the model uses (for diagnostic purpose, like 
in pkg/sbo) and theses should not go in model/src (beacause the one we expect
to find in model/src is the one constistent with the model EOS).

My impression is that the one in pkg/ecco/cost_averagesfields.F
falls into the 2nd category. May be you could take a look at what is 
done in pkg/sbo (e.g., sbo_rho.F). But may be using the sound-speed
which is consistent with the model EOS would be better ?

Cheers,
Jean-Michel

> item 1) sound speed diagnostics and observational constraints.
> ...
> item 1) the code in pkg/ecco/cost_averagesfields.F (under=20
>   ALLOW_IESTAU_COST_CONTRIBUTION) needs to be put
>   in a separate routine. Assuming its diagnostic can be
>   of interest to forward model users, and so that it is easy to=20
>   access by pkg/ecco or pkg/profiles, I would suggest we put=20
>   the code in model/src (as done for phihyd) and add a=20
>   common block array in model/inc. Makes sense?=20
>   Do you care to suggest routine/variables/CPP names?
>