[MITgcm-devel] negative diagnostics freq

[Baylor Fox-Kemper]

Hi Dimitri,
   The idea was that the time-mean should come out at frequencies that 
are halfway through the interval of averaging.  I wasn't aware that the 
snapshots were then similarly located, but if so, it should be the case 
that the first snapshot occurs halfway through the first period, and 
then at the period after that.  I.e.,

Times (T=1/Freq)

T/2, 3T/2, 5T/2,...

Note that although the first snapshot is only T/2 , subsequent snaps 
should  be T apart.
should produce averages and the snapshots I suppose should match.
   Cheers,
      -Baylor


On Apr 29, 2005, at 11:02 AM, Jean-Michel Campin wrote:

> Hi Dimitri,
>
>> Well, I'll be damned.  It is actually coded so that freq=-20
>> produces snapshot at timestep 10.  Why is that?  D.
>
> Baylor had good arguments in favor of this ; I personally
> think it's not a bad idea. It is also complementary
> to the snapshot output (dumpFreq=20*dT => snapshot at it=0,20,...)
> and, from my point of view, avoid the problem (iter 20 or 21 ?)
> for state variables (T,S) with stagger-timestep.
>
> Jean-Michel
> _______________________________________________
> MITgcm-devel mailing list
> MITgcm-devel at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-devel