A quick summary of how to download, compile, and run CM1¶

Valid for cm1r18 and later versions

Step 1: Download the code.¶

1. Read through the license.
2. If you agree to the terms in the license, they you may download and use CM1.
3. CM1 is distributed as a gzipped tar file that can be downloaded from this page.
4. Place the file on the disk that you will be running the model. Since the output from CM1 can be large, you should place the file on a disk with a substantial amount of free space (i.e., several GB of disk space should be available).
5. To uncompress and extract the files, type: gunzip cm1r18.tar.gz followed by tar xvf cm1r18.tar
6. Proceed to Step 2 below.

Step 2: Setup the code for your study.¶

1. Edit Makefile: cd into the src directory.
2. Edit base.F (if necessary): Modify the base-state conditions, as appropriate. There are two sections: one for the hydrostatic pressure, temperature, and moisture sounding (see "isnd" section); and one for the initial winds (u and v components) (see "iwnd" section).
3. Edit init3d.F (if necessary): In init3d.F, you can add perturbations to the base state. Several default options are available.
4. Edit init_terrain.F (if necessary): If you are using terrain, you will have to specify the terrain via the "zs" array in the file init_terrain.F.
5. Edit init_surface.F (if necessary): If you are using surface fluxes of heat/moisture/momentum, then you might have to specify the horizontal distribution of several variables in the file init_surface.F. See the param12 section in README.namelist for more information.
6. Choose arguments for make based on operating system and parallelization method appropriate for your hardware. See README.compile.md.
7. Compile the code: Type make [argu_list] within the "src" directory. On some machines, you may need to use "gmake" instead.

For example, to compile on NCAR's derecho (compile with 4 parallel threads, use shared + distributed memory + netCDF) make -j4 USE_OPENMP=true USE_MPI=true USE_NETCDF=true

1. Edit namelist.input: If the code compiled without error, cd into the "run" directory and edit "namelist.input". See README.namelist for guidance. Here, you set the domain dimensions, as well as the number of processors (using "nodex" and "nodey"). See the README files in the main directory for more information.
2. Place the input_sounding file in the same directory as cm1.exe (if necessary): If you are supplying an external sounding file, make sure it is called input_sounding and place it in the same directory as cm1.exe. See "Soundings for idealized simulations" for more information.
3. Place the LANDUSE.TBL file in the same directory as cm1.exe (if necessary): If you are using surface fluxes of heat/momentum/moisture, or if you are using the atmospheric radiation scheme, then you need to specify the surface conditions. (See the param12 section of README.namelist for more information.) The LANDUSE.TBL file comes with cm1 in the run directory.

Step 3: Run CM1.¶

1. Run the code: Type ./cm1.exe to run the code. You will probably need a different command for some MPI applications ... check the documentation for your supercomputer for more information.
2. stout and sterr info: To redirect all standard output and standard error information ... i.e., all the stuff that prints to screen when running cm1 ... use this command: ./cm1.exe >&! cm1.print.out &
3. Output: All output is placed in the "run" directory, by default. You can move the output files to another directory when the code is finished.

For more information ...¶

For more information about model settings, see README.namelist.

For more information about running on distributed memory computers (with MPI), see README.parallel.

For more information about running with terrain, see README.terrain.

For more information about running the axisymmetric version of the model, see README.axisymm.

For more information about running the model with stretched grids, see README.stretch.