once the python dependencies are properly installed you can then install ClimateTools in Julia.
pkg> add ClimateTools # Tagged release
Plotting data and results is done through the ClimatePlots.jl package.
Under the hood, ClimatePlots.jl requires some python packages for mapping purpose and is currently using Cartopy. To ensure that ClimatePlots works properly, it is recommended to use Julia's built-in Python distribution which can properly load the required python packages.
Hence, the best approach is to simply configure PyCall as follows:
julia> using Pkg; Pkg.add("PyCall") julia> ENV["PYTHON"]="" julia> Pkg.build("PyCall")
Then, once PyCall configured, installation is the standard command.
pkg> add ClimatePlots # Tagged release
It is also possible to use your own Python distribution instead of Julia's python.
The first approach is to use Anaconda or the system python distribution and ensure that you can import the following packages. By far, the easiest solution is to use Anaconda, which can easily install the Basemap dependency.
1.1 Python dependencies
1.2 Testing Python installation
No matter the approach used, it is recommended to test the python environment and see if you can import the required python packages.
#python >>> import cartopy >>> import matplotlib.pyplot as plt >>> import cmocean as cm
1.2 Building PyCall After the confirmation that the Python dependencies can be loaded in Python, the user needs to build PyCall with the same Python version. Alternatively, if PyCall is already built, it may be only a matter of installing the Python dependencies with the PyCall's Python version by using
ENV["PYTHON"]="path_to_python_distribution" pkg> build PyCall
If the main system python distribution is not set-up properly, the recommended approach is to use a Conda environment. Cartopy is much easier to install through this approach. The is simply a matter of creating a virtual python environment with conda and installing the required packages and then linking Julia's PyCall.jl package to this virtual conda environment. More information can be found in PyCall documentation for custom Conda environment.
Create a virtual environment with Conda
$ conda create -name ClimateTools python=3.6 $ conda activate ClimateTools $ conda install -c conda-forge matplotlib --strict-channel-priority $ conda install -c conda-forge cartopy --strict-channel-priority $ conda install -c conda-forge cmocean --strict-channel-priority $ which python # gives you the path of Conda virtual environment to use in the next steps.
Once those packages are installed, you need to tell PyCall.jl (in Julia) to use this conda environement in Julia.
julia> using PyCall julia> ENV["PYTHON"]="...PATH TO CONDA PYTHON..." # find the path with "which python" at previous step julia> using Pkg; Pkg.build("PyCall")