These tutorials and projects all use model2obs, and are expected to run on NCAR’s Casper machine. If you want to run them outside of Casper, see below.
Installing model2obs¶
To install model2obs, open a terminal session and run
git clone https://github.com/CROCODILE-CESM/model2obs.git
cd model2obs
./install.sh
conda activate model2obsNote that the CrocoLake’s project does not need model2obs: CrocoLake is a dataset, not a python package, and installing dask, cartopy, and matplotlib is sufficient to read it and make maps. However, model2obs comes with those packages, so if you have installed it already during the workshop, you can directly use the model2obs environment to run CrocoLake’s project too.
model2obs outside of Casper¶
model2obs uses DART under the hood, and for this workshop we have set up a DART executable on Casper. If you need to run model2obs outside of Casper, you basically need to compile DART on the machine where you want to run it.
On Derecho¶
Make sure the following modules are loaded (use module load to load them):
ncarenv
craype
intel
ncarcompilers
cray-mpich
hdf5
netcdf
Build DART (I suggest in the work folder):
cd $WORK/
git clone https://github.com/CROCODILE-CESM/DART.git DART-derecho
cd DART-derecho
cd build_templates
cp mkmf.template.intel.linux mkmf.template
cd ../models/MOM6/work
./quickbuild.shhttps://
You can now set up your config.yaml files to point to the correct path, replacing:
perfect_model_obs_dir: /path/to/DART/models/MOM6/work/with
perfect_model_obs_dir: $WORK/DART-derecho/models/MOM6/work/or the path you chose.
On your machine¶
Installing DART differs from machine to machine and we refer you to DART’s installation guide.
Once you installed DART succesfully, you can set perfect_model_obs_dir in your config file to point to models/MOM6/work/ from the DART root directory.