A Functional Introduction to GDAL, Part 0: Installing GDAL

Welcome to GDAL!

This is the first of what should be several posts on GDAL, the bread-and-butter library and toolset for playing with raster data. There are several introductions to GDAL on the web already that eloquently explain what GDAL is and why it is useful—for example, Robert Simmon's 4-part series on Medium.

This series will be geared towards working GIS professionals: people who understand projections, have used the raster tools in ArcMap or QGIS, but may be a little hesitant to dip into the command line. It will be geared around the tools available and how they're used for common tasks—an elaboration on the official docs with an emphasis on practical examples.

Installing GDAL On Windows

Installing GDAL used to be either really simple or really annoying, depending on your computing environment. GDAL is written and actively maintained by developers: people who are comfortable compiling from source and managing environment variables and installations. As such, it works great in their world but requires a little massaging for everyday GISers, especially those of us coming from a Windows-centric ArcMap world.

Then came Conda.

Conda is amazing. Conda is package and environment management for python and python-related packages. Conda makes installing and running GDAL easy.

So, how do you get it?

First, download and install Anaconda. The Anaconda Platform is the support structure for the Conda command line tools and includes helpful utilities and installers.

Next, we'll use the Anaconda Navigator to create a new environment. To do this, go to the Environments pane and select "Create" at the bottom. Give it a name—I call mine "gdal" for clarity's sake—and select the latest 3.x version of Python (unless you've got a good reason to use 2.7). Click "Create" to finish it off.

Note: It doesn't matter what python is currently installed on your computer (like if you've got python 2.7 from ArcGIS Desktop) because Conda creates its own sandbox for each environment. These sandboxes are completely separate from each other and from any other Python installation on your computer—that's the beauty of Conda!

Next, we're going to add the conda-forge repository ("repo"), as it usually has a more up-to-date version of GDAL. To do this, click on the "Channels" button and then select "Add." Type "conda-forge" into the text box and hit Enter.

Now, you can search for and add the gdal packages to your new environment. With your environment selected, change the dropdown menu at the top of the packages pane to "All" and type "gdal" (w/o quotes) into the search box. Select the checkbox next to the entry for gdal when it appears and click "Apply" in the bottom right corner. It will work out what other packages are needed and install everything.

Another Note: Getting the latest version of GDAL gets you the latest updates and bug fixes. For example, GDAL 2.1.3 has a bug that does not allow you to set the altitude parameter when calling DEMProcessing() via the Python bindings (see this bug report). This bug was fixed in 2.2.0.

More Notes! As of this writing (end of 2017), I've noticed that installing the SciPy libraries with the Anaconda Navigator downgrades GDAL to 2.1.3. After you install SciPy, you should see a little green arrow in the checkbox for GDAL in the installed packages view, indicating an update is available. Go ahead and update GDAL back to 2.1.3 by clicking this green arrow and then selecting "Apply" at the bottom.   

Switching Environment
To play in your new environment, navigate to the Anaconda folder group in the Start Menu and select Anaconda Prompt. This opens a new command prompt that is aware of our Conda environments.

Like I mentioned, Conda gives us a separate sandbox for each environment we create. We need to switch to our new environment by typing activate gdal (replacing gdal with whatever environment name you chose earlier). After a pause, the command prompt will change to something like (gdal) C:\Users\jadams. This means that you're now in the gdal sandbox.

To verify GDAL is installed and working properly, type gdalinfo --version and hit Enter. You should get output that looks like GDAL 2.2.2, released 2017/09/15. 

If you create other environments, you can switch between them by activating the other environment with activate foo, just like you'd do to activate your first environment.

Conda on Other OSes

macOS
Installation on macOS is similar to Windows, and Anaconda Navigator works the same in both OSes. The only difference is that we'll use the standard Terminal app instead of the Anaconda Prompt. In addition, we have to type source activate gdal instead of just activate gdal. Beyond that, enjoy the flexibility of bash (or csh, or whatever shell you're using).

Linux
Conda on Linux requires more CLI-based magic than Windows or MacOS. I've not tried it myself, but a quick search for "install conda ubuntu" gives us these instructions, and it should work just the same for other distros as well.

Linux also shares the same source activate requirement as macOS to switch between environments.

Coming Up Next

The next post will focus on some common file creation considerations that you'll use whenever your GDAL commands result in a new file.

Resources

GDAL Homepage: http://www.gdal.org
Conda User Guide: https://conda.io/docs/user-guide/index.html
Anaconda User Guide: https://docs.anaconda.com/anaconda/user-guide/
GISInterals pre-built GDAL binaries for Windows: http://www.gisinternals.com/release.php
Alternate Windows Installation: https://sandbox.idre.ucla.edu/sandbox/tutorials/installing-gdal-for-windows