[MITgcm-devel] diagnostics package and mnc format

[Dimitris Menemenlis]

Hi Ed, I am leaving for China Sunday night for ten days.
Best to coordinate with Chris Hill about the hi-res cube
integration.  I am appending the instructions for running
the cube that I just sent to Jim Taft.  Also Benny Cheng is
supposed to babysit a 12-year integration with NCEP.

Tuesday 11AM is the teleconference: 650-604-3395
I will e-mail Bill Thigpen and ask him to include you
on list and to help with accounts. 

D.







Jim, I have finally prepared a tarball of updated MIT GCM code, with which to
aim for the 10-year/day milestone.  I am off on vacation for next ten days
but Chris said that he will be in touch with you to try to work on the 
10-day/year milestone, so that we can submit the EOS article and to speed
up the data assimilation efforts.  There are a number of changes which have
been implemented in the code which should help for the 10-year/day milestone:

1) bug fix to advection scheme and new horizontal viscosity scheme permit
a less viscous solution than what we had before yet maintaining the 1200-s
time step.

2) code includes capability for calling KPP and seaice packages less
frequently to reduce cost associated with these packages.

3) size of pickup files has been reduced from 7.5 GB to 5 GB

the tarballs are on kalpana:
/nobackup1/menemenl/cube/MITgcm_checkpoint54b_post.tar
/nobackup1/menemenl/cube/MITgcm_contrib.tar

Instructions for setting up and running the code are in
MITgcm_contrib/high_res_cube/README_ice
For example, for a 216 setup you need to:

tar xvf MITgcm_checkpoint54b_post.tar
tar xvf MITgcm_contrib.tar
cd MITgcm/verification/global_ocean.cs32x15/code
\cp ../../../../MITgcm_contrib/high_res_cube/code-mods/* .
\cp ../../../../MITgcm_contrib/high_res_cube/code-mods/s216t_85x85/* .
\cp SIZE.h_216 SIZE.h
\cp packages.conf_hr packages.conf
cd ../build
\rm *
../../../tools/genmake2 -mods=../code 
-of ../../../tools/build_options/linux_ia64_efc+mpi_altix
make depend
make -j
cd ..
mkdir run
cd run
\rm *
\cp ../input/* .
\cp ../../../../MITgcm_contrib/high_res_cube/input/* .
qsub job_altix92

If I have not screwed up, the only file you need to modify is "job_altix92"
to specify the correct runtime directory.    Also the "data" file is set
up to run this for 1 year.  You can change this by modifying
"endTime=31881600" in the data file.  Example output for this setup is in
/nobackup1/menemenl/cube/MITgcm/verification/global_ocean.cs32x15/run




-- 
Dimitris Menemenlis <menemenlis at jpl.nasa.gov>
Jet Propulsion Lab, California Insitute of Technology
MS 300-323, 4800 Oak Grove Dr, Pasadena CA 91109-8099
tel: 818-354-1656;  fax: 818-393-6720