Setting up ADF#

Instructions on how to build the necessary environment for running the ADF.

Setting up the environment and getting the ADF will be done in the command line and with git/GitHub; this tutorial assumes basic working knowledge of both.

There are two steps (in this tutorial) to get the ADF up and running on your CISL machine:

  1. Loading up NCAR’s necessary modules for the required ADF packages/environment

  2. Using git to clone the main branch of the ADF to your machine

Attention

For this tutorial you must be on an NCAR CISL machine

There are some instructions for non-CISL machines, but there can be many obstacles such as not being able to have access to the data, the correct libraries, insufficient computer resources, etc.

For simplicity, this tutorial will only be geared towards users who have a CISL account and access to CISL machines, ie Derecho/Casper/Cheyenne and NCAR JupyterHub.

Setting up ADF environment#

CISL Computers#

Setting up the ADF environment on CISL (Derecho/Cheyenne/Casper) machines

Loading modules#

CISL machines have a cluster of modules that are available to load/unload. Such modules include: conda, nco, python, etc.

Load conda#

To make the ADF as flexible as possible, it is recommended that you run in an NCAR precompiled conda environment called npl

First load conda via module command

module load conda

Attention

You may encounter a module load error.

Try unloading the python module: module unload python
and re-try.

You can unload conda and load python again when you’re done with this session.

Activate npl#

Now you will want to activate NCAR's npl environment

conda activate npl

Load nco#

In addition to conda package, the ncrcat NetCDF Operator (NCO) is needed

module load nco

Load ncl#

And if the CVDP is desired, NCL is also needed

module load ncl

That’s it, all done. Wash up and go help with dinner

Cloning the ADF#

It is fairly straight forward to get the ADF using GitHub. You will need to clone the NCAR/ADF repository to your machine.

Navigate to some where you want to house the ADF root directory. For this example we will put the ADF in your work drive on glade

cd /glade/work/{user}

Next, just a simple git clone to copy the ADF to your current location:

git clone https://github.com/NCAR/ADF.git

This will create a new folder called ADF and will be the root directory: /glade/work/{user}/ADF/ This will house all the necessary scripts for the ADF to run and where custom scripts would be added if desired!