The CFA conventions for the storage of aggregated fields (|version|) #################################################################### David Hassell, `NCAS `_ With thanks to Phil Bentley, John Caron, Jonathan Gregory, Stephen Pascoe and Ag Stephens. Abstract ======== The `proposed CF aggregation rules `_ describe how multiple CF fields may be combined into one, larger field. These rules are fully general and are based purely on CF metadata. To make full use of such an algorithm, it is necessary that such an aggregated field may be stored within software memory (for field creation and manipulation) and on disk (for long term storage). This document proposes the abstract **CFA framework** for the storage of CF-aggregated CF fields, and the **CFA-netCDF conventions** for their efficient storage in netCDF files. As with other similar file storage conventions, on-disk storage is efficient because, wherever possible, the component fields from which an aggregated field is composed are stored as references to the files containing the original datasets. Both the abstract framework and the ability to read and write NCA files are implemented in the `cf-python `_ software library. Contents ======== .. toctree:: :maxdepth: 1 :numbered: introduction framework cfa cfa_attributes Search ====== * :ref:`search`