How to run a simple test with ORCHIDEE offline with libIGCM

Author: J. Ghattas Last revision: 2023/11/30, J. Ghattas

This is a first test for a couple of years using the experiment set up in FG_CRUJRA_HIST. This experiment FG_CRUJRA_HIST is a global historical set up using the full ORCHIDEE model in offline mode. It starts using no restart files by default (to be changed by adding a restart file from a spinup for important simulations). This setup uses the driver "orchideedriver" also called ”the new driver”.

Prepare the experiment folder and create the job

Do as follow:

cd modipsl/config/ORCHIDEE_OL
cp -r FG_CRUJRA_HIST MyFirstTest
cd MyFirstTest

The file config.card contains the main settings for the experiment. Change this file and create the job. You can use vi, emacs or an editor of your choice. Do as follow:

vi config.card  => Change to have the following:
  JobName=MyFirstTest
  DateEnd=1903-12-31
  PeriodLenght=1Y

By default, monthly mean output are set. If you want to change the frequency or add high frequency output, read comments and modify in COMP/sechiba.card and COMP/stomate.card.

Now create the main job (you'll have to answer questions which depend on the machine):

../../../libIGCM/ins_job

A job called Job_MyFirstTest has now been created.

Launch the job

Launch the job on the batch system:

sbatch Job_MyFirstTest      # At jean-zay or spirit/spiritx
ccc_msub Job_MyFirstTest    # At irene
qsub Job_MyFirstTest        # At obelix

You'll see in run.card the proceeding of the job. In file Script_Output_MyFirstTest.000001 you'll see all the text output from the scripts in libIGCM.

Find output

In the Script_Output_MyFirstTest.000001, look for the variable IGCM_OUT to see where the output are stored.

If SpaceName?=TEST, the output will stay in this folder (which is a temporary folder at scratch). But if SpaceName?=PROD or DEVT, the output will be moved to a permanent storage at irene or jean-zay. This will be done in a post-processing job launched automatically at a certain frequency or at least in the end of the simulation.