/fcidecomp

This is an fork to open for contributions from the community to the FCIDECOMP

Primary LanguageCApache License 2.0Apache-2.0

EUMETSAT fcidecomp software

The fcidecomp software enables decompression of near real-time spectral imagery data from the EUMETSAT Meteosat Third Generation (MTG) FCI instrument that are disseminated in compressed format (JPEG-LS netCDF). The decompression of files is possible via source code, netCDF-C and netCDF-Java based softwares, and various Python libraries

Supported platforms and installation

The fcidecomp software can be installed on:

  • Linux CentOS 7 64-bit
  • RockyLinux 8 64-bit (also from source code)
  • Linux Ubuntu 18.04 LTS 64-bit
  • Linux Ubuntu 20.04 LTS 64-bit (also from source code)
  • Windows 10 64-bit
  • Windows 10 32-bit

Installing the fcidecomp software

Installation procedures are described in the INSTALL file.

Testing the fcidecomp software (only for the conda-based installation)

A set of Python unit tests is present to ensure the installed software works correctly. They should be run within the Conda environment in which the software has been installed.

Prerequisites

  • pytest, installed in the Conda environment in which the software has been installed as described here

Also, the tests depend on the presence of a set of test data, which can be downloaded here. Test files should be placed in a directories tree structured as follows (replace $EPCT_TEST_DATA_DIR with any chosen name):

|_$EPCT_TEST_DATA_DIR
  |_MTG
    |_MTGFCIL1
      |_<test_file_1>
      |_<test_file_2>
      |_ ...

Once this is done, the environment variable EPCT_TEST_DATA_DIR should be set to the full path to the $EPCT_TEST_DATA_DIR directory.

Running the tests

Tests can be executed running the following command from within the root directory of the fcidecomp software repository:

pytest -vv tests

Using the fcidecomp software

The fcidecomp decompression library can be used in different ways, described in the following sections.

Use with netCDF4-C tools

Prerequisites

To use fcidecomp with netCDF4-C-based tools (e.g. nccopy), two prerequisites are needed:

  • the tools must be installed
  • the HDF5_PLUGIN_PATH environment variable must be set as described in the INSTALL file.

If using the conda installation, simply activate the environment created above and the prerequisites are met.

If the package has been installed from the source code, set the environment variable and install the relevant package (netcdf-4.7.0 on RockyLinux 8, netcdf-bin on Ubuntu Linux 20.04).

Example with nccopy

Once the prerequisites above are met, netCDF4-C tools should be automatically configured to decompress JPEG-LS compressed netCDF files. As an example, to decompress a file using nccopy, run the following line:

nccopy -F none $PATH_TO_COMPRESSED_FILE $PATH_TO_DECOMPRESSED_FILE

where:

  • $PATH_TO_COMPRESSED_FILE is the path to the JPEG-LS compressed file
  • $PATH_TO_DECOMPRESSED_FILE is the path where the decompressed file should be saved

Use with h5py-based Python libraries (conda install only)

Once the fcidecomp Conda package is installed and the Conda environment in which it is installed is activated, use of the fcidecomp decompression libraries should be automatically enabled for h5py-based Python libraries.

To ensure the fcidecomp filter is loaded, in a Python shell execute:

import fcidecomp

Now every h5py-based Python library, such as xarray, will be able to open and read JPEG-LS compressed files without further steps.

Use with netCDF-Java based tools

With netCDF-Java versions greater than 5.5.2, it is possible to open JPEG-LS compressed netCDFs with netCDF-Java based tools, such as toolsUI and Panoply, instructing netCDF-Java to use the netCDF-C library for reading purposes. To enable this feature:

  1. if fcidecomp has been installed from the source, install the netCDF4 library package (netcdf-4.7.0 on RockyLinux 8, libnetcdf-c++4 on UbuntuLinux 20..04)

  2. ensure the file $HOME/.unidata/nj22Config.xml exists (if it doesn't, it should be created) and that it contains the following lines:

     <nj22Config>
   <Netcdf4Clibrary>
     <libraryPath>$PATH_TO_NETCDF_LIB_DIR</libraryPath>
     <libraryName>netcdf</libraryName>
     <useForReading>true</useForReading>
   </Netcdf4Clibrary>
 </nj22Config>

    where $PATH_TO_NETCDF_LIB_DIR is the path to the directory containing the netcdf4 library, which:

    • in Linux (conda install), corresponds to $PATH_TO_CONDA_ENV/lib with $PATH_TO_CONDA_ENV equal to the path to the conda environment in which fcidecomp is installed.
    • in Windows (conda install), corresponds to $PATH_TO_CONDA_ENV\Library\lib with $PATH_TO_CONDA_ENV equal to the path to the conda environment in which fcidecomp is installed.
    • in RockyLinux (install from source), corresponds to /usr/lib64
    • in Ubuntu 20.04 (install from source), corresponds to /usr/lib/x86_64-linux-gnu/

Tested with ToolsUI 5.5.3 on Windows, Panoply 5.1.1 on Linux (known as not working for Panoply for that version in Windows due to a Panoply issue).

Inventory Notices

Licenses and copyright information for software dependencies up to version 2.0.0 is documented within the inventory folder.

Files listed under inventory/items/data_proprietary.ABOUT are licensed under EUMETSAT Proprietary license.

Dependencies

The following dependencies are not included in the package but are required and they will be downloaded at build or compilation time: