The Python cf package is an Earth Science data analysis library that is built on a complete implementation of the CF data model.

From version 3.14.0 the cf package uses Dask for all of its data manipulations.

http://ncas-cms.github.io/cf-python

http://ncas-cms.github.io/cf-python/installation.html

https://ncas-cms.github.io/cf-python/cheat_sheet.html

https://ncas-cms.github.io/cf-python/recipes

https://ncas-cms.github.io/cf-python/tutorial.html

The cf package implements the CF data model for its internal data structures and so is able to process any CF-compliant dataset. It is not strict about CF-compliance, however, so that partially conformant datasets may be ingested from existing datasets and written to new datasets. This is so that datasets which are partially conformant may nonetheless be modified in memory.

A simple example of reading a field construct from a file and inspecting it:

>>> import cf
>>> f = cf.read('file.nc')
>>> print(f[0])
Field: air_temperature (ncvar%tas)
----------------------------------
Data            : air_temperature(time(12), latitude(64), longitude(128)) K
Cell methods    : time(12): mean (interval: 1.0 month)
Dimension coords: time(12) = [1991-11-16 00:00:00, ..., 1991-10-16 12:00:00] noleap
                : latitude(64) = [-87.8638, ..., 87.8638] degrees_north
                : longitude(128) = [0.0, ..., 357.1875] degrees_east
                : height(1) = [2.0] m

The cf package uses Dask for all of its array manipulation and can:

• read field and domain constructs from netCDF, CDL, Zarr, Kerchunk, PP and UM datasets with a choice of netCDF backends, including from local, HTTP, OPeNDAP and S3 locations,
• be fully flexible with respect to dataset storage chunking,
• create new field and domain constructs in memory,
• write and append field and domain constructs to netCDF and Zarr v3 datasets on disk, with control over HDF5 internal file metadata,
• read, write, and manipulate UGRID mesh topologies,
• read, write, and manipulate HEALPix grids,
• read, write, and create coordinates defined by geometry cells,
• read and write netCDF4 string data-type variables,
• read, write, and create netCDF and CDL datasets containing hierarchical groups,
• read, write, and create data that have been compressed by convention (i.e. ragged or gathered arrays, or coordinate arrays compressed by subsampling), whilst presenting a view of the data in its uncompressed form,
• read and write data that are quantized to eliminate false precision,
• convert field and domain constructs to xarray datasets in memory,
• inspect field constructs,
• test whether two field constructs are the same,
• modify field construct metadata and data,
• create subspaces of field constructs,
• incorporate, and create, metadata stored in external files,
• combine field constructs arithmetically,
• manipulate field construct data by arithmetical and trigonometrical operations,
• perform weighted statistical collapses on field constructs, including those with geometry cells, UGRID mesh topologies, and HEALPix grids,
• perform histogram, percentile and binning operations on field constructs,
• regrid structured grid, UGRID, HEALPix, and DSG field constructs with (multi-)linear, nearest neighbour, first- and second-order conservative and higher order patch recovery methods, including 3-d regridding, and large-grid support,
• apply convolution filters to field constructs,
• create running means from field constructs,
• apply differential operators to field constructs, and
• create derived quantities (such as relative vorticity).

Powerful and flexible visualizations of cf field constructs, designed to be produced and configured in as few lines of code as possible, are available with the cf-plot package, which needs to be installed separately to the cf package.

During installation the cfa command line utility is also installed, which

• generates text descriptions of field constructs contained in files, and

• creates new datasets aggregated from existing files.

Tests are run from within the cf/test directory:

python run_tests.py

Training material on cf-python and cf-plot i.e. the CF Data Tools, in the form of Jupyter Notebooks, is openly-accessible and available (with instructions for set-up provided) at: https://github.com/NCAS-CMS/cf-tools-training.