Machine-specific Install

NCAR HPC Cheyenne/Casper

Below are specific recipes for getting started with MELODIES MONET on the NCAR HPC, Cheyenne/Casper.

Official NCAR JupyterHub kernel

One option is to use the “melodies-monet” kernel 1 on the NCAR JupyterHub.

  1. Navigate to https://jupyterhub.hpc.ucar.edu in your web browser.

  2. Select “Production” and log in with your NCAR CIT credentials.

  3. Start a server.

  4. In the Notebook section of the Launcher, select “melodies-monet” to create a new notebook using the kernel.

Since you don’t have to install anything yourself, this is the easiest way to get started.

Personal conda environment

Another option is to use conda to create your own MELODIES MONET installation in a conda environment. This creates a “stand-alone” instance of interdependent packages that will not interfere with your access to the main installation of Python on the system. You can use the NCAR-maintained conda installation to get access to conda by invoking:

$ module load conda/latest

Or, install your own copy of Miniconda or Anaconda.

Once you have conda, you are ready to install MELODIES MONET. If you plan to modify the MELODIES MONET or MONET/MONETIO codebases, follow the dev install instructions. Otherwise, follow the user install instructions.

NOAA HPC Hera

Below is a specific recipe for how to set up all the necessary Python dependencies on the NOAA Hera machine. There are three steps to complete before you can use and develop MELODIES MONET on hera: Step 1. Install the conda package manager Anaconda/Miniconda, Step 2. Install MELODIES MONET, and Step 3. link cartopy files.

We will use the conda package manager system to create a contained Python environment for running and developing MELODIES MONET.

  1. Install Anaconda/Miniconda: Follow the instructions on the RDHPCS wiki to download and run Anaconda/Miniconda on Hera. Tips for success:

    • You will need a NOAA HPC account to access the RDHPCS wiki link above.

    • Both Anaconda/Miniconda will work well for MELODIES MONET. See conda instructions to determine, which is the best option for you.

    • Pick a directory for your download and run the following wget command with modifications if needed:

      $ wget -nH -m -nd https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh
      
    • Follow the instructions on conda’s website, which is generally described below:

      $ bash Miniconda3-latest-Linux-x86_64.sh
      
      • Follow all prompts. Installing Anaconda/Miniconda on scratch is recommended due to the limited space available on your home directory. Make sure you change the default location.

      • Unless you want to initialize Anaconda/Miniconda yourself select “yes” when asked “Do you wish the installer to initialize Miniconda3 by running conda init?”

  2. Install MELODIES MONET: If you plan to modify the MELODIES MONET or MONET/MONETIO codebases, follow the dev install instructions. Otherwise, follow the user install instructions.

  3. Link the cartopy shapefiles: Hera has download restrictions, so link the required cartopy shapefiles for plotting by running the link_cartopy_files.sh script.

    If you have cloned the repo (e.g. following the dev install instructions):

    $ cd MELODIES-MONET/python_env_ymls/hera
    $ ./link_cartopy_files.sh
    

    If you didn’t clone the repo and don’t want to:

    $ wget -O - https://raw.githubusercontent.com/NOAA-CSL/MELODIES-MONET/main/python_env_ymls/hera/link_cartopy_files.sh | bash
    

You are ready to start using and developing MELODIES MONET!

Note

In the recent past 3, Hera did not allow downloading from PyPI. As a result, such pip installs commands failed since pip was not able to download setuptools from PyPI. As a (reluctant) workaround, python setup.py develop can be used instead for editable (development) installs of MELODIES MONET and MONET/MONETIO.

Note

In the recent past, downloading a lot of dependent packages at once with conda on Hera led to stalling. To overcome this challenge, try installing packages individually:

$ conda create --name melodies-monet python=3.9
$ conda activate melodies-monet
$ conda install -c conda-forge jupyterlab
$ conda install -c conda-forge netcdf4
$ conda install -c conda-forge wrf-python
$ conda install -c conda-forge cartopy
$ conda install -c conda-forge esmf
$ conda install -c conda-forge monet
$ conda install -c conda-forge monetio
1

Maintained by NCAR CISL and members of the MELODIES MONET development team.

2

See the cloning notes if you have trouble cloning the repositories this way.

3

Recent as of 12-Apr-2022. See GH79.