How to run with multiple forcing files (one per year) in data_table

[cspencerjones]

I am running OM4_025.JRA (the exact run from the MOM6-examples repo). I would like to extend beyond 1958. I know that I need to download the JRA data and pad the files using GitHub - adcroft/pad_JRA: Makefile to create "padded" versions of the JRA-55-do forcing files, and to regrid friver and licalvf onto the model grid using GitHub - adcroft/regrid_runoff: Conservatively re-grids runoff data to the nearest coastal wet-point of a MOM6 ocean grid. But how do I format the data_table so it knows to use multiple files (one for each year of the run)?

Thanks for your continued patience and help!

 

[Raphael Dussin]

I create a template data_table with <YEAR> in place of the year in filenames and my script update it.
This might be useful:

cat mom.sub


#!/bin/bash
#SBATCH -n 252
#SBATCH -J OM4p5_bp
#SBATCH --error=OM4p5.err
#SBATCH --output=OM4p5.out
#SBATCH --exclusive
#SBATCH --time=10:00:00
#SBATCH --qos=urgent
#SBATCH --partition=batch
#SBATCH --clusters=c4
## obviously use your group account:
#SBATCH --account=gfdl_o

njobs=40

#--------------------------------- system settings ----- ---------------------------
module load cray-netcdf

export NC_BLKSZ=1M
ulimit -s unlimited

# setup the run directory
if [ ! -d RESTART ] ; then mkdir RESTART ; fi
if [ ! -d outputs_raw ] ; then mkdir outputs_raw ; fi
if [ ! -d restarts_raw ] ; then mkdir restarts_raw ; fi
if [ ! -d logs ] ; then mkdir logs ; fi

#--------------------------------- prepare input files ---------------------------

ctrldir=$( pwd )
subscript=mom.sub

if [ ! -f jobscompleted ] ; then touch jobscompleted ; fi

lastjob=$( tail -1 jobscompleted )
thisjob=$(( $lastjob + 1 )) # if file empty, takes job number one

echo 'starting job #' $thisjob

# at first job, replace restart by cold start
if [[ $thisjob == 1 ]] ; then
sed -i -e "s/input_filename = 'r'/input_filename = 'n'/g" input.nml
# at second job, replace init by restart
elif [[ $thisjob == 2 ]] ; then
sed -i -e "s/input_filename = 'n'/input_filename = 'r'/g" input.nml
fi

# grep the first year of the run
yearbeg=$( grep "current_date" input.nml | sed -e "s/,/ /g" | awk '{print $3}' )
thisyear=$(( $yearbeg + $thisjob -1 ))

cat data_table.template | sed -e "s/<YEAR>/$thisyear/g" > data_table

#--------------------------------- run the model -----------------------------------

srun --cpu_bind=rank -n 252 ./MOM6

#--------------------------------- check status of run -----------------------------

runok=$( tail -200 OM4p5.out | grep -i "Total runtime" )
if [[ $runok != '' ]] ; then
# move outputs
mv *.nc ./outputs_raw/.
mv *.nc.???? ./outputs_raw/.
# initiate transfer
#sbatch transfer.sub $thisyear
# save restarts and move to input
tar -cvf restarts.$thisjob RESTART/*
mv restarts.$thisjob ./restarts_raw
mv RESTART/* INPUT/.
# move logs
tar -cvf logs.tar.$thisjob MOM_parameter_doc.* SIS_parameter_doc.* OM4p5.err OM4p5.out ocean.stats* logfile.000000.out available_diags.000000 seaice.stats SIS.available_diags SIS_fast.available_diags ocean_stats*
mv logs.tar.$thisjob ./logs/.

# notify completion
echo $thisjob >> $ctrldir/jobscompleted
# test for resubmission
if (( $thisjob < $njobs )) ; then
cd $ctrldir ; sbatch ./$subscript
else
# final job
echo this is the last job
fi
else
# run blew up
echo "run blew up"
exit 1
fi

 

[cspencerjones]

OK great: I guess I will run to the end of the year and then update the data table using your script.