Skip to content

EOPF xarray backend

xarray-eopf is a Python package that enhances xarray by a new backend named "eopf-zarr". This backend allows for reading the ESA EOPF data products in Zarr format and representing them using analysis ready data models.

Overview

After installation of the package, you can open EOPF data products using the usual xarray top-level functions open_dataset() that provides a Dataset object:

dataset = xr.open_dataset(url_or_path, engine="eopf-zarr")

and open_datatree() that provides a DataTree object including groups:

datatree = xr.open_datatree(url_or_path, engine="eopf-zarr")

Features

IMPORTANT
The xarray-eopf package is in a preliminary development state. The following features are only partly or not at all implemented yet.

The backend supports two distinct modes of operation, "native" and "analysis":

  • op_mode="analysis" - using the analysis mode, the returned Dataset or DataTree objects attempt to be user-friendly and analysis-ready to a maximum extend. Certain preprocessing steps will be applied depending on the specific Sentinel product towards an analysis-ready data model and content. For example, for Sentinel-2, we will support the selection of specific bands and resampling to a desired spatial resolution. For Sentinel-3 we will allow for rectifying the images in satellite coordinates to a user-provided, common CRS. It can be used as shown in the examples below: 

    dataset = xr.open_dataset(url_or_path, engine="eopf-zarr", op_mode="analysis")
datatree = xr.open_datatree(url_or_path, engine="eopf-zarr", op_mode="analysis")

  • op_mode="native" - using the native mode, the returned Dataset or DataTree objects try to serve as a 1:1 representation of the actual Zarr product structure and content with either none or minimal preprocessing applied. It can be used as shown in the examples below: 

    dataset = xr.open_dataset(url_or_path, engine="eopf-zarr", op_mode="native")
datatree = xr.open_datatree(url_or_path, engine="eopf-zarr", op_mode="native")

More information on the two modes are given in the User Guide

Data variables will always be represented as chunked Dask arrays for efficient out-of core computations and visualisations.

The package has minimal core dependencies: xarray, zarr, and dask. Packages for accessing remote filesystems are optional, e.g., you will need s3fs if you need to access EOPF data products in S3-compatible remote object storages.

License

The package is open source and released under the Apache 2.0 license license. ❤