MIT Integrated Global System Model (IGSM)

This file contains some information on building the IGSM from source. For more on the contents of the source tree, see the commit history using git log, or at http://github.com/mit-jp/igsm.

Contents

• data/ Common data files (grids, etc.) for running the model.
  • clm/ Data for the Community Land Model (CLM).
  • data/ Data for the atmospheric chemistry module (chem).
  • emimed/ Predicted emissions data.
  • init/ General initialization data.
    • 46lat/ Data specific to a 46-latitude-band grid.
    • forcing/ Forcing data.
  • tem/ Data for the Terrestrial Ecosystem Model (TEM).
• src/ Source code for the modules of the model, including…
  • atm/ atmosphere model.
  • chem/ atmospheric chemistry model.
  • clm/ CLM version 3.5.
  • inc/ INClude files common to all modules.
  • meta/ urban METAmodel.
  • ocn_3d/ the MIT General Circulation Model
  • ocn_ml/ two-dimensional, Multi-Layer (ML) OCeaN model.
  • ocm/ Ocean Carbon Model.
  • tem/ TEM version 4.4c.
• util/ Utilities.
  • analysis/ Extract data from output files.
  • eppa/ Preprocessor for emissions data from the MIT Economic Projection & Policy Analysis (EPPA) model, version 5.
• tests/ Example input files for common use cases.
  • run_fullchem/ 1990–2100 predicted-emissions case, full model, 2D ocean.

Build

The following process has only been tested on the MIT svante cluster:

• Originally, a Fedora Core 14 x86_64 system with GNU Make 3.82; PGI 9.0.4–13.8; and NetCDF 3.6.2–4.2.
• Currently, a Fedora Core 24 x86_64 system with make 4.1, and various versions of PGI and NetCDF (see "Testing the build").

1. If building directly from version control:

    $ autoreconf --install
   

   to regenerate the configure script.

2. Load the PGI compilers and the NetCDF libraries. This command uses the Lmod module system on svante, but may not be required on other systems.

    $ module load pgi netcdf
   

3. Create a directory in which to build the model, enter it, configure, and build:

    $ mkdir -p build
    $ cd build
    $ ../configure
    $ make
   

   An executable named igsm22 is produced.

Some basic documentation of options (for example, choosing between historical or predictive models) can be obtained by typing ../configure --help; these should be given when invoking configure in step 3.

Multiple builds can be produced from the same source code in this way, for example in directories named build1, build2 (or something more informative). These directories may even be outside the directory containing this README.md file; this would require giving a relative path when invoking configure (e.g. ../igsm/configure or similar). To see what options were passed to configure, see the file config.log in any build directory.

Testing the build

util/build-test.sh contains a Slurm job script to test the building of the model using a variety of compilers and libraries. Invoke using:

$ sbatch util/build-test.sh ./configure

This produces the log file build-test.log and the directory build-test. The directory has a series of numbered subdirectories; each one corresponds to a different combination of modules (see the script for which ones exactly).

The code is configure'd and, if that succeeds, make'd in each of these subdirectories. Very verbose command-line output from configure and make, along with the output of module list, is sent to a file named build-test.log.

Execute

N.B. Before running any of the experiments in the tests/ directory, copy the entire directory out of the source tree into a separate location. This avoids confusion when the experiments alter the files or create new ones, or updates to the source also change the experiment input files.

Preprocessing

The emiprep binary reads a file called eppa5chm.put from the current directory and generates several files that are read by the IGSM. eppa5chm.put can be obtained by copying the chm.put file produced by EPPA.

Only the EPPA 5 preprocessor is currently included. It is uncertain whether this preprocessor also works for output from EPPA 4 or EPPA 6.

Running

TODO add remaining instructions