fremor CMORizes FRE output with CMOR. It is a conda package and it's documentation can be found on
readthedocs.
Apache License 2.0 — see LICENSE.md
fremor is a model output rewriter (CMORizer) for FRE/FMS based models and output. It is specifically geared for standardizing
NOAA-GFDL datasets for further quality control checks, assessments and data publishing pipelines in the context of CMIP7
using the CMOR library.
fremor was originally the fre.cmor submodule of NOAA-GFDL/fre-cli and so stands
on the shoulders of it's contributors, retaining it's general structure and lessons learned from it. Future re-integrations
back into fre-cli, as a formal package dependency, are being assessed.
AI was heavily used in the creation of this repository, primarily github's copilot with Claude (opus4.6, sonnet4.6,
and haiku), and Gemini and Chat-GPT models to a lesser extent, in agent mode. Claude and Codex agents have also
contributed.
For an overview of required inputs and sample commands, see the CMOR Quickstart.
python>=3.11click>=8.2cmor>=3.15.0netCDF4>=1.7numpy>=2pyyaml
For development and testing, pylint, pytest, and pytest-cov are all highly recommended as helpful additions.
If you're trying to gain access to fremor functionality as quickly as possible:
# the current post-release in main
module load fremor/test
# a tagged version of fremor, post-releases will never be named modules
module load fremor/X.Y.ZIf you have a path to a fremor environment you can activate it like so:
conda activate some/path/to/fremor_envIf you want your own fremor environment:
# the environment will be named fremor_en
conda create -n fremor_env conda-forge::fremor
# see fremor_env in the list --> activate it by name
conda env list
conda activate fremor_envor, if you've already activated a conda environment
conda create -n empty_env
conda activate empty_env
conda install -c conda-forge fremor
# equivalent syntax
conda install conda-forge::fremorIf you're trying to develop fremor capabilities, or edit the code to your liking in either a big or small way,
this is for you. This checks out the code, creates and activates an environment, installs into the environment,
and runs all unit-tests and pylint checks:
# omit --recursive if you don't want tables as submodules
git clone --recursive https://github.com/NOAA-GFDL/fremor.git
cd fremor
# create an environment and install the local checkout
conda env create -f environment.yaml
conda activate fremor
pip install -e .
# Run tests
pytest fremor/tests/
# Run linter
pylint --rcfile pylintrc fremor/The CLI entry point is fremor, currently a suite of (currently) six routines for facilitating data preparation for
CMIP7.
# The full list of subcommands
fremor init # Initialize CMOR configuration resources: generate template user config, fetch tables
fremor find # Find and print variables in MIP tables according to your variable lists or other input
fremor varlist # Create a simple variable list of netCDF files in a directory
fremor config # Generate a basic CMOR YAML configuration from a pp directory tree
fremor yaml # Bulk routine for processing data based on a CMOR YAML config, calls fremor run many times
fremor run # Lowest-level routine, no CMOR YAML needed, rewrites output files in a directory with CMORThe CLI offers full logging and verbosity control independent of the command chosen:
# verbosity and logging
fremor -v ... # INFO level logging
fremor -vv ... # DEBUG level logging
fremor -q ... # ERROR level only (quiet)
fremor -l mylog.txt ... # Log to file (appends)If you've used the previous fre cmor command, there is a direct mapping of syntax:
# past fre-cli command
fre -vv -l logfile.txt cmor <COMMAND> [OPTIONS]
# fremor equivalent
fremor -vv -l logfile.txt <COMMAND> [OPTIONS]Each CLI subcommand (run, yaml, etc.) maps to an API under under fremor, so the CLI functionality
is equivalently available via import in scripts as a proper python module
The wcrp_compliance_check workflow validates CMORized NetCDF outputs against WCRP project
specifications using cc-plugin-wcrp, a plugin for
the IOOS compliance-checker. This pipeline:
- Runs automatically on pull requests and via manual dispatch
- Executes unit tests to generate CMORized output files
- Gathers and categorizes outputs by CMIP version (CMIP6, CMIP7)
- Validates outputs using the
wcrp_cmip6compliance checker - Uploads compliance reports as workflow artifacts (retained for 30 days)
To view compliance results from a workflow/CI run:
- Navigate to the Actions tab in GitHub
- Select the
wcrp_compliance_checkworkflow run - Download the
wcrp-compliance-reportsartifact
fremor uses a post-release scheme to identify development beyond the latest tagged version. To avoid confusion
with fre-workflows and fre-cli, which often demand that the version tags match, fremor's version format is
X.Y.Z[.post].
This procedure is being actively tested and checked now, and may change in the future. Document any deviations taken from this guide!
To publish new release, cease merging new PRs to main, and carefully follow the below procedure:
- create a new branch off of
main, which should be the previous tagged version +.post, give it a name different than the exact tag you are creating - edit the version number in
fremor/_version.pyfrom the current one, to the desired version tag, remove.post, then open a PR tomainin this repository. - confirm the branch is functional by letting workflows finish, if you see green checks only, proceed. otherwise, stop and debug.
- at this point, light clean up style edits are OK, but functional edits are not. Do so until happy and keep the checks passing.
- now create the tag from the branch at this point locally in your terminal with
git tag X.Y.Z;
WARNING: any problems or mistakes after the next step are irreversible due to package immutability so make sure things are working before continuing
- now push your tag you created locally with
git checkout X.Y.Z; git push origin HEAD:refs/tags/X.Y.Z - workflow checks are triggered by the new push to the new tag
X.Y.Z, and one of the workflows will to publish the built package toPyPI, check the actions tab - on github, create a new release from the new tag, generate release notes automatically comparing to the previous
tag, and upload the built
X.Y.Zpackage downloaded fromPyPIyou cannot do this later due to immutability of releases
- use (create if needed) a fork (e.g. https://github.com/ilaflott/fremor-feedstock) to create a new branch called
fremorX.Y.Z, e.g. https://github.com/ilaflott/fremor-feedstock/tree/fremor0.9.3 - adjust the version to
X.Y.Zand update thesha256to what it says on PyPI inrecipe.yaml - open a PR to the
conda-forge/fremor-feedstocke.g. conda-forge/fremor-feedstock#3 - once checks pass, a reviewer with access to
conda-forge/fremor-feedstockcan approve and merge, kicking off the rest of the publishing pipeline toconda-forge
- back to the
fremorPR we opened intially. - edit the version number in
fremor/_version.pytoX.Y.Z.post, let the checks pass - merge the PR branch you used for creating the release to
main
what a published release on PyPI looks like: https://pypi.org/project/fremor/0.9.3/#fremor-0.9.3.tar.gz
what a published PyPI package on github looks like: https://github.com/NOAA-GFDL/fremor/releases/tag/0.9.3
what a published PyPI package on conda-forge looks like: https://anaconda.org/channels/conda-forge/packages/fremor/files