Member-only story
Using Docker to Overcome Geospatial Package Challenges and Implementing It in HPC (Part 1)
When I first started using Docker images, I wondered how I could apply them to my geospatial work to make my workflow smoother. This question became even more relevant when I began working with HPC (High-Performance Computing) during my PhD project. I needed a way to simplify my environment setup without having to install numerous modules across different environments manually.
Yes, I know! You might say, “Why not just use a requirements.txt file to reinstall the necessary packages?" While that's a valid approach, geospatial packages pose unique challenges. Their dependencies often change, and troubleshooting those changes can consume a lot of time. Additionally, finding the right versions of required packages for a specific operating system can be tricky. For instance, I recently discovered how challenging it is to install GDAL on macOS.
This is where Docker and Docker images come to the rescue! Docker eliminates many of these issues by allowing us to containerize our workflows. I won’t dive into the basics of installing Docker or explaining how it works — I’ll assume you’re already familiar with that and are looking for ways to use Docker in your geospatial workflows.
