[MITgcm-devel] diagnostics

[Jean-Michel Campin]

Hi Patrick,

there are good reasons not to have DIAGNOSTICS_SWITCH_ONOFF
in thermodynamics (it really needs to be called very early in
the time-step), for sure before DO_ATMOSPHERIC_PHYS, but probably
also before SEAICE, BULKF_FORCING ?, and in fact before any S/R
that calls diagnostics_fill.

It might be different regarding DO_STATEVARS_DIAGS, but 
moving the call inside thermodynamics is not enough, and you
will need a 2nd call anyway (hidden in DYNAMICS might be a solution 
for DIAGNOSTICS_FILL_STATE, but it's too late for THSICE_DIAGNOSTICS_FILL
and LAND_DIAGNOSTICS_FILL). 
Now, the more "elegant" solution for TAF might become
less pretty for the pkg-diagnostics itself ...

This 1rst call to DO_STATEVARS_DIAGS, at the beginning of forward_step,
makes it clear that it cannot be used for non-state variables (since nothing
has been computed yet); It's also easy to figure out what is 
the status of the state-variable (& hFac) at this point of the time-step,
and I kind of like this way.

Cheers,
Jean-Michel

On Mon, Oct 17, 2005 at 09:11:44PM -0400, Patrick Heimbach wrote:
> 
> Hi there,
> 
> the routine forward_step.F isn't a very good place
> for calls to
>         CALL DIAGNOSTICS_SWITCH_ONOFF
> and
>         CALL DO_STATEVARS_DIAGS
> at least from an AD point of view
> (TAF likes to kick out "passive" routines;
> there's a work-around, but the more elegant solution 
> is to hide passive calls from TAF alltogether).
> 
> Can these calls go somewhere else, e.g. at the beginning 
> of thermodynamics?
> 
> I guess, the case (.NOT.staggerTimeStep) would cover
>  CALL DIAGNOSTICS_SWITCH_ONOFF( myTime, 0, myIter, myThid )
> and the case (staggerTimeStep) would cover
>  CALL DIAGNOSTICS_SWITCH_ONOFF( myTime, 1, myIter, myThid )
> but maybe I missed something(?)
> 
> -Patrick
> 
> -- 
> --------------------------------------------------------
> Patrick Heimbach   Massachusetts Institute of Technology
> FON: +1/617/253-5259                  EAPS, Room 54-1518
> FAX: +1/617/253-4464             77 Massachusetts Avenue
> mailto:heimbach at mit.edu               Cambridge MA 02139
> http://www.mit.edu/~heimbach/                        USA
> 
> _______________________________________________
> MITgcm-devel mailing list
> MITgcm-devel at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-devel