The ACCESS-ENSO-Recipes package is a collection of Jupyter notebooks designed to reproduce the metrics described in the CLIVAR 2020 ENSO Metrics Package. These metrics are based on the methodology outlined by Planton et al. (2021) (DOI:10.1175/BAMS-D-19-0337.1) and implemented following the instructions provided by Yann Planton.
This package integrates with ESMValTool, utilising its preprocessors to streamline analysis. Users can explore the ENSO metrics interactively in Jupyter notebooks or execute them as an ESMValTool recipe for batch processing.
- Reproduction of ENSO metrics from the CLIVAR 2020 ENSO Metrics Package.
- Simplified diagnostics for integration into ESMValTool.
- Support for interactive exploration in Jupyter notebooks or execution via ESMValTool recipes.
- Configured for use on NCI Gadi with the ACCESS-NRI conda environment.
The Earth System Model Evaluation Tool (ESMValTool) is a community-developed software package for the evaluation of Earth System Models (ESMs).
For more information, visit the official ESMValTool website.
To use the ACCESS-ENSO-Recipes, ensure the following:
- Access to NCI Gadi.
- The ACCESS-NRI conda environment pre-configured with the esmvaltool-workflow.
Go to the Australian Research Environment website and login with your NCI username and password. If you don't have an NCI account, you can sign up for one at the MyNCI website.
Click on JupyterLab under Featured Apps to configure a new JupyterLab instance. This option is also available under the All Apps section at the bottom of the page and the Interactive Apps dropdown located in the top menu.
You will now be presented with the main JupyterLab instance configuration form. Please complete only the fields below - leave all other fields blank or to their default values.
- 3.1 Walltime: The number of hours the JupyterLab instance will run. For the hackathon, please insert a walltime of
4hours.
- 3.2 Compute Size: Select
Medium (4 cpus, 18G mem)from the dropdown menu.
- 3.3 Project: Please enter
nf33. This will allocate SU usage to the workshop project.
- 3.4 Storage: In ARE, storage locations need to be explicitly defined to access these data from within a JupyterLab instance. Please copy and paste the following string in its entirety into the storage input field:
gdata/xp65+gdata/fs38+gdata/oi10+gdata/al33+gdata/rr3+gdata/rt52+gdata/zz93+gdata/ct11+gdata/zv30
-
3.5 Click
Advanced options ...- Optional: You can check the box here to receive an email notification when your JupyterLab instance starts, but as we are running a relatively small instance, it will likely spin up quickly so this probably isn't necessary.
-
3.6 Module directories: To load the required environment modules, please copy and paste the following. This is equivalent to using
module useon the command line.
/g/data/xp65/public/modules
- 3.7 Modules To load the ESMValTool-workflow environment, please copy and paste the following. This is equivalent to using
module loadon the command line.
esmvaltool
- 3.7 Click
Launchto start your JupyterLab instance.
Once you have clicked Launch the browser will redirect to the 'interactive sessions' page where you will see your JupyterLab instance details and current status which will look something like this:
Once the JupyterLab instance has started (this usually takes around 30 seconds), this status window should update and look something like the following, reporting that the instance has started and the time remaining. More detailed information on the instance can be accessed by clicking the Session ID link.
Click Open JupyterLab. This opens the instance in a new browser window where you can navigate to the location of the files.
This package is maintained by the ACCESS-NRI Model Evaluation and Diagnostics Team. For issues, suggestions, or assistance, please contact ACCESS-NRI support.
We welcome contributions! Please follow the contribution guidelines to submit enhancements or bug fixes.
- CLIVAR 2020 ENSO Metrics Package: CLIVAR Website
- Planton et al. (2021): DOI:10.1175/BAMS-D-19-0337.1
- ESMValTool: ESMValTool Official Website