Installation#

Anaconda/Conda#

In order to run \(\omega radlib\), you need to have a Python interpreter installed on your local computer, as well as a number of Python packages. We recommend installing Anaconda as it includes Python, numerous required packages, and other useful tools (e.g. Spyder).

Using Anaconda the installation process is harmonised across platforms. Download and install the latest Anaconda distribution for your specific OS. We recommend using the minimal distributions Miniconda or Miniforge/Mambaforge if you do not want to install a full scientific python stack.

We are constantly performing tests with conda-forge community channel (for the most recent 3 python versions).

If your Python installation is working, the following command (in a console) should work:

$ python --version
Python 3.11.0

Now you can use the conda/mamba package and environment manager (conda documentation / mamba documenation) to setup your \(\omega radlib\) installation.

#. Add the conda-forge channel, where \(\omega radlib\) and its dependencies are located. Read more about the community effort conda-forge:

$ conda config --add channels conda-forge

#. Use strict channel priority to prevent channel clashes:

$ conda config --set channel_priority strict

#. Create a new environment from scratch:

$ conda create --name wradlib python=3.11

#. Activate the \(\omega radlib\) environment:

$ conda activate wradlib

#. Install \(\omega radlib\) and its dependencies:

(wradlib) $ conda install wradlib

Now you have a conda environment with a working \(\omega radlib\) installation.

Test the integrity of your \(\omega radlib\) installation by opening a console window and typing calling the python interpreter:

$ python
Python 3.11.0 | packaged by conda-forge | (main, Oct 25 2022, 06:24:40) [GCC 10.4.0] on linux
Type "help", "copyright", "credits" or "license" for more information.

The Python prompt should appear. Then type:

>>> import wradlib
>>> wradlib.__version__
'2.0.0'

If everything is ok, this will show the running \(\omega radlib\) version. If the \(\omega radlib\) package is not found by the interpreter, you will get::

>>> import wradlib
ImportError: No module named wradlib

Alternatively, you can install the Dependencies, but you have to keep track of \(\omega radlib\)’s dependencies yourself.

Bleeding edge code#

Warning

The \(\omega radlib\) version on PyPI might lag behind the actual developments. You can use the bleeding edge code from the \(\omega radlib\) repository. Note, however, that you need to make sure yourself that all Dependencies are met (see below).

Download the source, unzip, and run:

$ python -m pip install .

Alternatively, you can add the \(\omega radlib\) directory to your environment variable PYTHONPATH.

Installing via pip#

Although we recommend using a conda Python Environment you can install \(\omega radlib\) from PyPi via pip.

Open a terminal and run:

$ python -m pip install wradlib

Depending on your system you might need to be root (or sudo the above command) for this to work or use --user to install into user directory. pip will then fetch the source distribution from the Python Package Index and run the installation.

Afterwards it will check for any dependencies not met, yet.

Be aware that using pip we can only look for python-module dependencies. For example the numpy module itself depends on some other libraries, which need to be present in order for the module to compile properly after being downloaded by pip. We have no control over these dependencies and it is rather hard to give a complete overview.

Therefore we recommend trying to satisfy the dependencies using your favorite package management system.

Installing via pip tries to install all dependencies, but be sure to have all depending non-python libraries installed. Wheels are not available for all dependencies (eg. GDAL).

Dependencies#

\(\omega radlib\) was not designed to be a self-contained library. Besides extensive use of Numpy and Scipy, \(\omega radlib\) uses additional libraries, which you might need to install before you can use \(\omega radlib\) depending on your system and installation procedure.

Package

min

recommended

numpy

>= 1.9

>= latest

scipy

>= 1.0

>= latest

matplotlib

>= 3

>= latest

xarray

>= 0.17

>= latest

xradar

>= 0.3

>= latest

You can check whether the required are available on your computer by opening a Python console and enter:

>>> import <package_name>
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
ModuleNotFoundError: No module named 'package_name'

This will be the response in case the package is not available.

In case the import is successful, you should also check the version number:

>>> package_name.__version__
`some version number`

The version number should be consistent with the above .

Optional Dependencies#

Apart from the obligatory , some dependencies in \(\omega radlib\) are optional. This is because the installation of these dependencies can be somewhat tedious while many \(\omega radlib\) users will not need them anyway. In case users use a \(\omega radlib\) function that requires an optional dependency, and this dependency is not satisfied in the local environment, \(\omega radlib\) will raise an exception.

As for now, the following dependencies are defined as optional:

Package

min

recommended

cartopy

>= 0.22

>= latest

dask

>= 2.20

>= latest

gdal

>= 3.0

>= latest

h5py

>= 3.0.0

>= latest

h5netcdf

>= 0.8.0

>= latest

netCDF4

>= 1.0

>= latest

requests

>= 2.23.0

>= latest

xmltodict

>= 0.12

>= latest

The following libraries are used by netCDF4, h5py/h5netcdf and gdal packages and should apply to these requirements:

Library

min

recommended

used by

geos

>= 3.7.0

>= latest

gdal

hdf5

>= 1.9.0

>= latest

h5py

libnetcdf

>= 4.7.3

>= latest

netCDF4

proj

>= 5.2.0

>= latest

gdal

The speedup module

The speedup module is intended as a collection of Fortran code in order to speed up specific \(\omega radlib\) function that are critical for performance. In order to build the speedup module as a shared library, you need to use f2py. f2py usually ships with numpy and should be available via the command line. To test whether f2py is available on your system, execute f2py on the system console. Or, alternatively, f2py.py. If it is available, you should get a bunch of help instructions. Now change to the \(\omega radlib\) module directory and execute on the system console:

$ f2py.py -c -m speedup speedup.f

Now the speedup module should be available.

Known Issues#

Depending on your OS and installation method you may encounter different problems. Here are some guidelines for attacking them.

We strongly recommend using the Anaconda conda package and environment manager (see ). Using conda-forge we will maintain the wradlib-feedstock for constant availability of recent \(\omega radlib\) versions.

If you can’t use Anaconda/Miniconda, it is generally a good idea to use your systems package manager to install dependencies. This will also take account for other needed bindings, libs etc.

If you encounter problems installing \(\omega radlib\), check on your favorite search engine or create an issue here with details on the problem or open a discussion topic on the openradar-discourse.