prognostic & diagnostic

[susonic]

Hi all

It might sound basic thing but I can't find a definition of prognostic run and
diagnostic run.
Is there anyone who can define those two methods and explain what should I do
if I want to run by those two separate methods with ROMS?
For example, turn on specific cpp flag.

Thank you so much!

Best

[susonic]

As far as I know in ROMS, if you want to do a prognostic running you need to run the model with
undef xx_CLIMATOLOGY in contrast, if you want do a diagnostic running you need to run the model with
define xx_CLIMATOLOGY in your header file. Am I right?
Is there anyone who can tell me so that I can figure it out clearly? Any tips would be appreciated.

Regards,

-Peter

[arango]

Sometimes people use the prognostic and diagnostic adjectives very loosely when describing their computations.

A prognostic run implies to predict a fluid state into the future (forecast) using the model equations. This is what we usually do in ROMS. We initialize the model with the initial conditions and predict (interpolate via the primitive Navier-Stokes equations) the ocean state into the future subject to the boundary conditions (surface/bottom forcing and lateral boundary conditions).

Contrairily, a diagnostic run does not implies prediction of the fluid state. There is not a time-evolution due to the model equations; the computations are done based on the present fluid state. Sometimes a diagnostic run is done to equillibrate the fluid initial and boundary conditions to the first order dynamic balances implicit in the continuum governing equations but affected by their discretization (bathymetry, topography, land/sea mask, grid resolution, etc). It can also used to extract a particular (diagnostic) information, or to filter dynamics to slow manifolds by removing unrepresented processes. In ROMS, this can be done by activating TS_FIXED to supress the time-evolution of density (modeled in terms of potential temperature and salinity). That is, the initial temperature and salinity is hold constant (initial state) for the entire simulation. So you can initialize the model with just zero free-surface, zero momentum, and a given potential temperature and salinity fields. Then, the model will compute the free-surface and momentum that balances the frozen density state subject to the boundary conditions. This must be a short run (usually, 7 to 14 days). This should not be confused with a spin-up of a climatology to seasonal forcing. There is a lot of literature about this (check Haidvogel et al, 2000). We usually use climatological forcing (say montly) and run the model for several years by cycling the same forcing every year. In spin-up runs we check the kinetic energy of the system. We consider the system fully spin-up when the kinetic energy is no longer growing other than to the seasonal signal.

[susonic]

Hi, Mr.Arango.
First of all, happy new year(sorry, it's kind of late greeting)

Thank you so much for your clear explanation.
It's a really specific information and also thank you for the link .
It helped me out to understand better about those two method.
Actually, I've been researching with those topics throughout the papers
but I don't know which is right or wrong and also couldn't figure it out well.
But now I understand.
Thank you again, sir.

Regards,

-Peter