jimcircadian / icenet-notebooks Goto Github PK

This repository contains instructional notebooks that describe how to use the IceNet library and pipelining utilities built around it.

Currently, notebooks 01, 02, 03, 04 are completely tested

The notebooks 01-05 are intended for instruction and contain output from example runs.

Notebooks xx.* are illustrative and exemplify/contain runs from the operational infrastructure, but rely on the model run(s) they were run against. These are easily adaptable for your own local executions with IceNet.

A brief description of the different notebooks included in this repository:

• 01.cli_demonstration.ipynb: Shows the usage of command line interfaces (with IceNet library installed) for end-to-end runs to forecast sea-ice concentration.

• 02.pipeline_demonstration.ipynb: Shows the usage of the IceNet pipeline) for end-to-end runs and ensemble modelling, as deployed to a SLURM cluster (A cluster is not mandatory, it can also be run locally).

• 03.data_and_forecasts.ipynb: Shows the the various data sources, intermediaries and products that arise from both of the first two notebooks CLI demonstrator notebook and Pipeline demonstrator notebook activities.

• 04.data_and_forecasts.ipynb: Shows the usage of the IceNet library via its Python API.

These are compatible at time of writing with the 0.2.* version of the IceNet library.

Please ensure you have installed netCDF4<=1.6 in your Python environment. You might run into issues on netCDF outputs with newer versions.

pip install -U netCDF4<=1.6.0

If you are running into setup issues, please use the conda environment.yml file from icenet-pipeline corresponding to the tagged version of this repository to create a python conda environment.

All three main icenet-related repositories (icenet, icenet-notebooks and icenet-pipeline) share a common version tag.

Many of the other notebooks are scratch or potentially useful (or were once useful) when the codebase was being refactored. These are named scratch_ to show that they're only potentially useful if you want to explore or adapt them.

Any other notebook is likely related to events or presentations that have been given.

Please fork and raise PRs, contributions are most welcome!