search

COSIMA / access-om2

ACCESS-OM2 global ocean - sea ice coupled model configurations.
21 stars 23 forks source link

Use TEOS-10 equation of state? #140

Open aekiss opened 5 years ago

aekiss commented 5 years ago

There is an email thread on this from around 2019-03-28: Re: COSIMA model suite: Questions about conservative & potential temperature | EOS | Salinity

Trevor is encouraging us to use TEOS-10 and says having the model with its S variable being Practical Salinity, while its temperature variable is Conservative Temperature is a weird half-way combination of EOS-80 and TEOS-10

We are currently using the Jackett McDougall Feistel Wright and Griffies (2006) pre-TEOS10 seawater equation of state and freezing temperature at all resolutions in ACCESS-OM2 (eos_preteos10=true and freezing_temp_preteos10=true).

pre-TEOS10 requires that we use practical salinity as the prognostic variable for salt, and gives the option of conservative or potential temperature as the prognostic temperature variable. We use conservative temperature in the 1 deg and 0.25 deg configurations, and potential temperature for the 0.1 deg configuration used in the model announcement paper (this was an oversight; future 0.1 deg configurations will use conservative temperature - e.g. new 0.1deg RYF spinups: https://github.com/OceansAus/access-om2/projects/1).

Can anyone see any reason not to switch to TEOS-10? (I recall we discussed this but can't remember the details...)

If we switched to TEOS-10 we'd need to use absolute salinity instead of practical salinity. WOA13 uses practical salinity - will it be an issue to use this for the initial condition and restoring if the model variable is absolute salinity? I haven't found any documentation specifying what form of salinity WOA18 uses (but it is unitless, suggesting psu).

aekiss commented 5 years ago

@russfiedler's comment:

This is pretty straight forward. Just convert the salinities using the gsw package (python, Fortran, Matlab and C versions exist I think).

MOM5 still uses the original MBFJ (2011?) TEOS-10 formulation rather than the Roquet et al (2015) version. I'll look at updating this.

There's 2 ways of implementing TEOS10 in MOM5.

1) Use preformed salinity and a correction term (Fdelta) as prognostic tracers. Absolute salinity gets updated as Den%salinty = Sp*(1+fdelta). Fdelta is restored towards a prescribed field. 2) Use absolute salinity (aliased to preformed salinity) as a prognostic tracer. This seems to be the way most groups are doing it. This does mean that you need to carry around a dummy prognostic trace Fdelta which is identically zero.

MOM6 seems to have incorporated the GSW package more fully.