gmacgilchrist/esm4ppe

esm4ppe

Python package for working with ESM4 perfect predictability experiments. This is predominantly a light wrapper around climpred, with some additional functionality to work efficiently with the PP/AN file structure.

Notes on installation and package requirements

Most of the functionality relies only on xarray, so will work in any environment with a reasonably up-to-date version of this. You will also need to have some back-end packages installed, including netcdf and zarr, as well as cftime for handling calendars. To use the climpred functionality, you need to have this installed. Depending on your environment set-up, you may also need to install jupyterlab or ipykernel.

Additionally, esm4ppe depends on the gfdl_utils package, which is a basic package for navigating the filestructure on PP/AN. This package can be found here. Clone that repository to your local machine, and install it in your environment by issuing pip install -e . from within the repository.

Installing the esm4ppe package

1. Clone this repository
2. In the esm4ppe subfolder, edit the version.py file to point towards directories that you have write permissions for. I recommend creating a directory on the /work filesystem on PP/AN. This is the location where esm4ppe can save raw and processed data, as well as check to see if those data are already saved. See the settings in the original version.py for an idea of this file structure. Note that you will have to make these directories locally (esm4ppe will not do that for you).
3. In the main esm4ppe repository folder (and while the climpred_clean environment is activated) issue pip install -e ..
4. In a jupyter notebook, you should now be able to import the esm4ppe module.