CCSM4_0_RACM27 (Raptor/Garnet)
-------------------------------------
***************************************
2011-12-19: Waiting for WRF input/bdy data for post 2007 - involved with putting together WRF input/bdy data for 2007-2010+
2011-12-16: RACM batch submit script (batch_submit.sh)
2011-12-16: Running long runs on Garnet (September 2006 - January 2007 complete)
2011-11-15: Rerunning RACM on Garnet due to restart issue per Andrew's 11/15 email - changed STOP_OPTION value to 'nmonths' and STOP_N value to '1' - turned off HIST_OPTION and turned on AVGHIST_OPTION with value 'ndays'
2011-11-14: Begin post-processing
2011-11-14: RACM Garnet restart simulation completed from September 1 - October 31, 1989 restarting for October
2011-11-10: Successful simulation runs on both Garnet and Raptor - Submitting 30-day restart run with CONTINUE_RUN set to 'TRUE' and RESUBMIT set to '0'
2011-11-08: Submitted 30-day simulation RACM tag 27 RB on Garnet with same options as set on Raptor
2011-11-08: Submitted 5-day simulation RACM tag 27 RB on Raptor with HIST_OPTION set to 'ndays' and HIST_N set to '1' to take snapshots of coupler I/O every day
2011-11-07: Set-up RACM tag 27 using RASM quick start guide
2011-11-03: Have checked out RACM tag 27
2011-10-31: Setting up RACM tag 27 from the repository - don't have the right credentials - checking with Andrew
-------------------------------------
***************************************
CCSM4_0_RACM26 (fully coupled - Chugach)
-------------------------------------
***************************************
Version copied from: see README
Most changes are located in: /models/atm/wrf/main/atm_comp_mct.F
***************************************
Step 1
Per Tony's June 24 email: change config_grid.xml (/scripts/ccsm_utils/Case.template/config_grid.xml)
"GREGORIAN" entries to "NO_LEAP".
Step 2
Download Matt's time averaging flux fix (atm_comp_mct.F) to /models/atm/wrf/main/.
Uncomment out u* averaging from /models/atm/wrf/phys/module_sf_sfclay.F after line "To prevent oscillations..."
Step 3
Changed /scripts/ccsm_utils/Machines/mkbatch.chugach to include:
default run time from 4 hours to 8 hours for a monthly run
Step 4
cd scripts/
./create_newcase -case test1 -res wr50a_ar9v2 -compset RB -mach chugach -skip_rundb
Creates new case called "test1" using wrf,vic,pop,cice
Step 5
cd test1/
./configure -case
Change /scripts/test1/Buildconf/wrf.buildnml.csh line 18 February from 29 to 28 if
compiling for NO_LEAP option
Step 6
./test1.chugach.build
Step 7
Edit env_run.xml
default value="5"
default value="never"
default value="-999"
default value="never"
default value="-999"
Step 8
qsub test1.chugach.run
-------------------------------------
Useful Info-
Input data (via env_run.xml - line 195)
Output data $WORKDIR/$CASE/test1/run/
Namelist found $WORKDIR/$CASE/test1/run/
Timing found $WORKDIR/$CASE/test1/run/timing/
-------------------------------------
Run log:
2011-07-13
Job submitted. Job killed due to system failure.
2011-07-14
Job submitted. Job completed. Run time 20112 seconds. 8 hour wall-time.
2011-07-15
Restart run submitted and will resubmit 3 times. Job completed. Run time 20367 seconds.
Next simulation will have a longer wall-time and Matt's u* fix.
2011-07-18
Job submitted. Job completed 30 days with a run time of 20112 seconds and a wall time
of 8 hours.
2011-07-19
Restart run submitted. Job will run for 120 days bringing final integration time to the
end of January. Not sure what resubmit does - it does not do what I thought it would do.
Wall time of 28 hours. Job did not complete successfully. The CCSM date and WRF date
for the restart run are not equal. Do you have to configure a new case for a restart
run?
-------------------------------------
Restarts info-
Edit env_run.xml
First simulation (30-days, no restart)
STOP_OPTION='ndays'
STOP_N='30'
Second simulation (120-days, restart, continue for a year)
CONTINUE_RUN='TRUE'