NEMO_project_template

Repository containing the recommended folder structure for an AMM7 NEMO-FABM-BGC project, including scripts to handle automatic run cycling.

Repository structure/workflow

Clone repository. It is recommended that you clone for a specific project, for example, NECCTON rather than NEMO_project_template.

git clone -b AMM7 https://github.com/pmlmodelling/NEMO_project_template.git <PROJECT_NAME>

Set Environment

First, navigate to the scripts/ directory and run setup_<SYSTEM>.sh, depending on if you are using scylla or archer2. This will set the module environment, architecture files for compilation and link the inputs to <PROJECT_NAME>/INPUTS, with the folder structure detailed below.

Next edit scripts/config.sh. Most settings here can be left as standard, with the exception of:

1. EXP_NAME - Name for your experiment. If you run multiple experiments under the same project, this can be changed between runs to keep them separate
2. START_YEAR - Beginning year of the experiment. Please ensure restart files exist in INPUTS/DOM/. If using the existing restarts please keep this as 1993. You can start from a later time by adding the current_date file (see below)
3. END_YEAR - Final year of experiment (inclusive)

Compiling code

To compile the code needed to run, you can execute scripts/compile_code.sh with various flags. The first time you need to include the -c flag to also clone the code, subsequent recompiles will not need to include this:

• XIOS

  To compile xios, execute `compile_code.sh -xc`. Note Scylla does not have access to svn, so the base code gets linked through the setup and you just need to execute `compile_code.sh -x`

• FABM/ERSEM

  To compile fabm (with ersem), execute `compile_code.sh -fc`

• NEMO

  To compile nemo, execute `compile_code.sh -nc`. This must be done after compiling XIOS and FABM.

The executables for xios and nemo should now be available in code/executables

RUN

Before kickstarting a run, its a good idea to check everything is setup correctly. To help with this the script scripts/dry_run.sh is provided. This script creates the run folder RUN/EXP_NAME, links all the files needed to run from START_YEAR and sets a cfg files as needed. Please check this is as expected. A dry run with the flag -c will do perform a clean instance if you make any changes.

To perform a single cycle, navigate to RUN/<EXP_NAME> and submit the job using the testing scripts

sbatch runscript_testing.slurm

To start a cycle run, navigate to scripts and submit the run:

sbatch runscript.slurm

If you want to start from a date other than the January of START_YEAR, or the run needs to be restarted you can do so by creating/editing the file scripts/current_date with the following:

export year=XXXX
export month=Y

Note that if you change this before running to start from a date other than 1/1/1993, you will need to change scripts/core-scripts/setup_initial.sh to link the correct restart file.

The cycle run uses the run configuration files in RUN/EXP00. Please make any changes here, or alternatively set a new DEFAULT_RUN_DIR in config.sh before running if you want to change the configuration.

OUTPUTS

By default, output will be moved to the OUTPUTS/ directory using the EXP_NAME/YYYY/MM/ folder structure.

INPUTS

Once linked, the INPUTS directory should point to files with the following folder structure:

• DOM

  Includes domain files, coordinates, bathymetry and initial conditions

• BDY

  Folders for physics and BGC Lateral/open boundary conditions

• RIV

  Includes river forcing files

• SBC

  Folders for atmospheric forcing and BGC surface forcing

• TIDE

  Includes tidal forcing files

BGC_setup

The scripts used to create the biogeochemical inputs: initial conditions, lateral and surface boundary files are found here should you wish to use them to create any alternative input files.