yeoda
yeoda stands for your earth observation data access and is open source Python tool developed by the GEO Department at TU Wien. It provides transparent, flexible and user-friendly access to work with well-defined and structured earth observation data, keeping a close relation with files on disk. It relies on several in-house developed Python packages wrapping around well-known standard libraries like GDAL, netCDF4, xarray, etc. The overall picture how yeoda connects with these packages to access manifold geospatial datasets in a homogenous manner is shown below.
On the bottom, we have geospade defining geometries, tiles, or mosaics and how they relate to each other. On top of geospade, veranda unites these abstract geometries with I/O classes for dealing with all kinds of data formats. Finally, with the help geopathfinder, yeoda allows to link data managed by veranda with a file naming convention to interpret or build file names, from which file-based dimensions/coordinates are constructed. With this modular setup, yeoda is able to accept either a data object (xarray.Dataset) or a list of files as input, apply spatial sub-setting or filtering based on the pre-defined dimensions, and finally to read or write data from or to different formats. For example, you can read a stack of GeoTIFF files and write a single NetCDF file or vice versa.
yeoda offers two basic datacube classes, DataCubeReader and DataCubeWriter. Both inherit from DataCube, which consists of these essential parts:
A file register represented by a data frame managing a stack/list of files containing the minimum set of columns:
“filepath”: Full system paths to the files.
stack_dimension (defaults to “layer_id”): Specifies an ID to which layer a file belongs to, e.g. a layer counter or a timestamp.
tile_dimension (defaults to “tile_id”): Tile name or ID to which tile a file belongs to.
A mosaic geometry geospade.raster.MosaicGeometry (or a child class) managing the spatial properties or representing the actual mosaic/grid of the files.
The names of the tile (tile_dimension) and stack dimension (stack_dimension)
After initialising one of the datacube classes, you have access to a rich set of spatial selections, e.g. using a bounding box (select_bbox()) or polygon (select_polygon()), dimensional filters, e.g. filtering for file-based coordinates (select_by_dimension) or splitting a datacube in time (split_by_temporal_freq()), or datacube interactions, e.g. intersecting (intersect()) or uniting (unite()) two datacubes. As soon as the data satisfies your data request, you can finally read (read()) data on disk to create an xarray.Dataset or write/export (write()/export()) in a file naming convention and data format of your choice.
Start playing around with yeoda by following the documentation and use case examples and feel free to contribute to the project via GitHub!