Tools to support analysis of CESM/MOM6 model solutions. See documentation for more information.

Installation

Clone the repository from github:

git clone https://github.com/NCAR/mom6-tools.git

Install required packages and mom6-tools in a brand new conda environment:
```
cd mom6-tools
conda env create --file environment.yml
```

Register mom6-tools in ipykernel:

conda run -n mom6-tools python -m ipykernel install --user --name mom6-tools

Notes

get_cluster(n_workers) runs a dask.distributed.LocalCluster on the node it is called from. Batch workers are opt-in: to submit them you must name the class, either with parallel, cluster, client = get_cluster(n_workers, cluster_class='PBSCluster'), with --cluster-class PBSCluster on the command line, or with cluster_class: PBSCluster in the Jobqueue: block of diag_config.yml. Resource settings on their own (--queue, a Jobqueue: block without cluster_class, …) will not do it – they are ignored, with a warning naming them.
If running on HPC, make sure the project account is added to ~/.config/dask/jobqueue.yaml. Or set a default account as an environment variable that your scheduler can access. Anything not set in the Jobqueue: block or on the command line falls back to that dask configuration.

User guide

Reference