Terminology

DataArray

A multi-dimensional array with labeled or named dimensions. DataArray objects add metadata such as dimension names, coordinates, and attributes to underlying “unlabeled” data structures such as numpy and Dask arrays. If its optional name property is set, it is a named DataArray.

Dataset

A dict-like collection of DataArray objects with aligned dimensions. Thus, most operations that can be performed on the dimensions of a single DataArray can be performed on a dataset. Datasets have data variables, dimensions, coordinates, and attributes.

DataTree

A tree-like collection of Dataset objects. A tree is made up of one or more nodes, each of which can store the same information as a single Dataset (accessed via .dataset). This data is stored in the same way as in a Dataset, i.e. in the form of data variables, dimensions, coordinates, and attributes.

NEMODataTree

A hierarchical DataTree object designed to organise NEMO ocean model outputs, such that each domain is represented as a level in the DataTree, while the grids comprising the NEMO ocean mesh (gridT, gridU, gridV, gridW, gridF, etc.) are stored as nodes within that domain.

NEMODataArray

An extension of the DataArray object which stores a multi-dimensional array defined on a NEMO model grid. Each NEMODataArray supports grid-aware computation by interfacing with its parent NEMODataTree to provide useful properties, discrete operators and statistics, alongside utility methods to interpolate variables onto neighbouring NEMO model grids.