History files

Suggest improvements/optimizations to the ROMS code.

Moderators: arango, robertson

Post Reply
Message
Author
mathieu
Posts: 74
Joined: Fri Sep 17, 2004 2:22 pm
Location: Institut Rudjer Boskovic

History files

#1 Unread post by mathieu »

Dear all,

I recently worked on running ROMS coupled with SWAN and I wanted
to put the wave informations (Pwave, Dwave, etc.) as output in history file.
This faciliity is not available with ROMS 2.2 since those variables were forcing
informations put into the input file.
What about having the ability, in general, of putting forcing data into history files?

Another problem that we face is the size of history files if one puts 3D arrays.
What about having multiple history files? For example, one could
set up one series of history record every hour containing only zeta, ubar, vbar fields
and another series every six hours containing the full set of variable for a restart.

Thank you

User avatar
kate
Posts: 4088
Joined: Wed Jul 02, 2003 5:29 pm
Location: CFOS/UAF, USA

#2 Unread post by kate »

This is the Rutgers ROMS? It has long had the ability to make multiple history/averages files.

Code: Select all

!  NDEFHIS     Number of time-steps between the creation of new history file.
!              If NDEFHIS=0, the model will only process one history file.
!              This feature is useful for long simulations when history files
!              get too large; it creates a new file every NDEFHIS time-steps.
!
As for the forcing fields, it can now write out some of them, surface stress, for instance. If you need to be writing new ones, copy/paste/edit some similar code. Make sure to look in varinfo.dat for the relevant idBlah codes. You will need to edit def_his.F and wrt_his.F, or the similar def_avg.F and wrt_avg.F. If you want the averages, you will have to keep the running means, with changes to mod_average.F and set_avg.F. Then send the patch file to Hernan and keep track of what you change so it gets into your code the next time you update.

mathieu
Posts: 74
Joined: Fri Sep 17, 2004 2:22 pm
Location: Institut Rudjer Boskovic

#3 Unread post by mathieu »

Yes this is ROMS rutgers and yes I did not explained myself correctly.
What I would like is that the file ocean.in contain:

Code: Select all

    NHISFILES == 2  
     HISNAMES == ocean_his_complete_6hour.in \
ocean_his_partial_30min.in
Then the file ocean_his_complete_6hour.in would contain:

Code: Select all

        NHIS == 54
     NDEFHIS == 1728            ! 6 hour outputs
     HISNAME == out_his_6hour.nc
Hout(...)= ...
By doing that, one would be able to have different sequences
of history files and then better manage the total size problem.
Separating the history files declarations from the model declarations
is a good thing in itself I think.

I know that we can reduce the size of history files by splitting
them into several ones with NDEFHIS but there is no way
at present to write the 3D arrays only once every 12 histories
and so split or no split we have huge disk storage expenses.

Post Reply