[MITgcm-support] odd diagnostics package output

[Daniel Enderton]

Hello Folks,

I was curious about switching over to using the diagnostics package 
for my output when I noticed some odd behavior.  For a held-saurez 
simulation I checked the state output using the diagnostics package 
versus the normal state.*.nc that is ouput.  For the diagnostics 
package I set

   frequency(1) = -600.,
   filename(1) = 'surface',
   levels(1,1) = 1.,
   fields(1,1) = 'ETAN    ','ETANSQ  ','DETADT2 ',

and set the  dumpFreq=600 in the data file.  The output for the 
diagnostics package was phase shifted.  The state.*.nc file had

  T = 0, 600, 1200, 1800, 2400, 3000 ;

  Eta =
  0, 0, ...
  [more zeros corresponding to T=0, as desired for this simulation]
  0.0126337883550606, 0.0113340932414461, ...
  [The rest of the values for T=600 and beyond]

While the diagnostics package output read

  T = 600, 1200, 1800, 2400, 3000 ;
  ETAN =
  0, 0, ...
  [more zeros, here corresponding to T=600]
  0.01263379, 0.01133409, ...
  [more non-zero values corresponding to T=1200 and beyond]

I tried playing with the timephase setting to try and get the 
snap-shot output from the diagnostics package to start at T = 0 
(example:  frequency(1) = -600., \ timephase(1) =  -600.,) but I 
could not get it to do so.  Further, this apparent phase shift goofs 
up the diagnostics package time average data and monitor statistics.

Is there something that I am missing about how to use the diagnostics 
package or is this phase shift a problem in the code?  Any 
suggestions would be appreciated!

Daniel