Skip to content

Julia for Geospatial Programming part 1 – setting up

  • blog

update Dec 2022 – This has all be checked for Julia 1.8.3

The top six programming languages for Data Science, according to the first links returned to me from google are:

  1. Python
  2. R
  3. MATLAB
  4. Java
  5. Julia
  6. Scala

When I searched for the top five programming languages for GIS, the top results returned this list:

  1. Python
  2. JavaScript
  3. R
  4. SQL (the most undervalued skill in Geospatial today)
  5. Java
  6. C++/C
  7. C#

Pretty hard to argue with the results here. I saw no mention of Julia or Scala in the top five search results for GIS programming languages.

I will use this series of posts to investigate the Julia language and its use for Geospatial Programming. In part 2 I will investigate the speed vs Python and in part 3 I will look at reading and writing Satellite data.

Why not just stick with Python?

I struggle to answer this question. Python is excellent for rapidly prototyping, has hundreds of libraries that you can draw on, it has a large and expanding community; the language of GIS has arguably become Python and it is not going away any time soon. In 2018 Python went past Javascript as the language with the highest percentage of questions asked about it on Stackoverflow. Python is a good choice.

If you are new to Geospatial programming, choose Python (or R or SQL or a mixture) – it is well supported for Geospatial development.

Why Julia?

Julia is fast.

“The release of Julia 1.0 signals that Julia is now ready to change the technical world by combining the high-level productivity and ease of use of Python and R with the lightning-fast speed of C++,”

The benchmark tests seem to back this up, with speeds comparable to C.

https://julialang.org/benchmarks/

The most common complaint I have heard about Python is that it is slow. For me, I’ve never really felt the pain of the slowness, it’s always done what I have wanted it to do. There is plenty of discussion about nested loops in Python and Julia, here for example. However we are in a data deluge today, and making code run faster without investing more into hardware is worth a look, I think.

This is all brand new to me, so I’ll step through what I did.

Julia setup

Download and install Julia from here:

https://julialang.org/downloads/

At the time of writing I am installing the current stable release (v1.1.0). I found the installation of Julia pretty straightforward. There are several options for running Julia described in the link above. I am just going to run from the command line on my machine to start with. If installed correctly, when you run for the first time you should get a screen similar to this:

I found on my computer that opening Julia via Run as administrator worked best for me. At the moment that is what I would recommend you doing.

We are up and running now. Below I did a few tests on the command line:

We are going to need some packages to do the Geospatial work. In no particular order run these commands and install:

using Pkg
Pkg.add("GDAL")
Pkg.build("GDAL") 
Pkg.add("ArchGDAL") 
Pkg.add("ImageView")
Pkg.add("Images")
Pkg.add("Colors")
Pkg.add("GeoArrays")

I found from testing these were the packages I needed initially. BenchmarkTools is for working out how long code took to run. The majority of the work I will show in these posts will utilise the ArchGDAL package. You can read more about it here.

https://github.com/yeesian/ArchGDAL.jl

Finally, let’s install the IJulia package so we can run our code in Jupyter Notebooks and, frankly, make life super easy for ourselves.

using Pkg
Pkg.add("IJulia")

After this has installed, type:

using IJulia
notebook(dir="D:/Julia_work/")

This will open the Jupyter Notebooks in D:/Julia_work (change to fit your drive and folder)

I can see a Python 3 and a Julia 1.1.0 kernel. I have even found that I can shut everything down and open Anaconda run:

jupyter notebook

and it allows access both my Python 3 anaconda environment and Julia environment. Meaning that I can move between languages – should I need to. This would be great for presentations. I imagine that is true for R kernels as well.

I have put together a very brief notebook on some of the basic intro ‘things’ I have been using in Julia so far. You can access it here.

If you want to go deeper I think this looks like a great place to start with Julia.

A Comprehensive Tutorial to Learn Data Science with Julia from Scratch

Recap

In this introductory post we have looked at the common data science languages and the common GIS languages. We have:

  • Learnt how to install and get set up.
  • Learnt how Julia can sit in your programming environment alongside Python.
  • Plus, I’ve shared a brief notebook on some basic functions of the Julia language.

In part two I’ll show you how to load an image into an array and compare the speed with Python. In part three I’ll show you how to read and write raster data using Julia.

I am a freelancer able to help you with your projects. I offer consultancy, training and writing. I’d be delighted to hear from you.

I have grouped all my previous blogs (technical stuff / tutorials / opinions / ideas) at http://gis.acgeospatial.co.uk.

Feel free to connect or follow me; I am always keen to talk about Earth Observation.

I am @map_andrew on twitter

 

Tags: