Which open licence should I choose?

/

I’ve written about open data a few times recently. And not-so-recently. And there’s been quite a bit of chat about open subsurface benchmarks in the Software Underground recently. As more people consider openly releasing data — or code, or other content — one question comes up fairly often is: Which licence should I choose?

I’ll start at the beginning, and I am not a lawyer, but this is going to be very high level. So do click on the links to read more.

What is copyright?

You automatically own the copyright to anything original that you create. You don’t have to register it, but the thing you made — and it must be a thing, you can’t copyright ideas — must be original. It could be a photo, a song, or a seismic interpretation. Physical measurements with no creative input, such as well logs, are not copyrightable… but a database consisting of such data is (so-called database rights). Your rights are exclusive, worldwide, and last until some years after you die (it varies).

If someone wants to use your work, even if they just found it on the Internet, they must either claim Fair Use, or seek permission from you. Giving permission means granting a licence; it can be as restrictive and arcane as you want.

If you don’t want people bothering you about licences, or if you want to actively encourage people to use and adapt your work, you can preemptively grant an open licence.

What is openness?

Before you start thinking about licences, there are two more big things to learn about:

  1. What is open? Not all licences, not even all Creative Commons licences, meet the Open Definition. In brief, this states that “Open data and content can be freely used, modified, and shared by anyone for any purpose” — you can’t restrict people based on their use case or location. So licences that forbid commercial application are not open.

  2. What is permissiveness? Once you’ve decided to go open, you need to decide where you stand on permissiveness. Some licences, notably those advocated by the GNU Free Software Movement, compel licensees (users) to preserve the openness of the work in any future redistribution. This ‘viral’ condition is sometimes called copyleft.

In some circles, a near-religious war smoulders on the permissiveness issue. You need to make up your own mind where you stand, or at least understand the issues.

By the way, granting a licence does not mean giving up your rights. In fact, you must own the copyright in order to grant the licence. Many scientists don’t realize we’ve been giving away the copyright in our work for decades, as a (completely unnecessary and made up) condition of publication.

Another source of confusion: open licences are also not the same thing as public domain. Public domain means that the work is free from copyright restrictions. In general though, it cannot be applied to a copyrighted work (though CC0 tries to relinquish copyright where possible). For example, On The Origin of Species is public domain, as is most work produced by the United States government (for example, by the USGS).

One last thing: an often overlooked aspect of licensing is protection for you, the licensor. All common licences include language that indemnifies you from misuse or misinterpretation of your work. So be careful about putting your stuff ‘out there’ with anything other than a standard licence: you may be leaving yourself open to liability issues later.

Open licences

Rather than writing a lot of stuff that’s been written by smarter people than me, I thought I’d draw a diagram to try to explain the differences between some common licences (there are certainly a lot more than the ones I mention here).

Just to re-iterate: there are a lot more licences than the ones mentioned here, these are just examples.

What do I recommend?

For content, my personal belief is that CC-BY most aptly captures the way science works. Scientists 'build on the shoulders of giants' by re-using the work of others with fastidious attribution, usually by citation. Accordingly, the CC-BY protects the licensor, ensures attribution, and that's it. If you prefer copyleft licences, the equivalent licence is CC-BY-SA.

But Creative Commons recommend against using CC licences for source code, so what should you do then?

For code, the permissive licence closest to CC-BY is the MIT/BSD/Apache family of licences, of which only the Apache 2.0 licence offers some specific protections with respect to patents (in particular, it protects licensees from ‘upstream’ patent infringements). The equivalent copyleft licences are the GPL (for applications) and LGPL (for libraries).

For data, I tend to use CC-BY, but there are some specialist data licences (beware, they are poorly named in my opinion: the seemingly ‘vanilla’ ODbL is copyleft; the permissive equivalent is ODC-By).

What about mixed content, like a Jupyter Notebook? You have to be practical; maybe it depends on whether you consider your notebooks to be 'content' or 'source code'. I sometimes put at the bottom of a notebook something like Open source content. Text is CC-BY, code is Apache 2.0 and I think this makes my intent clear.

Tools

There are some tools around to help you make a choice of licence:

Last thing

Note that open licences are just one piece of the jigsaw puzzle of reproducible science and reusable content. You also need to think about open and accessible data formats (e.g. CSV not XLS), accessible content (DOIs and open indexes), and documentation.

Although insufficient, open licences are a necessary component though. And while licences can be changed, they cannot be revoked… so it’s worth putting some thought into your choices before you start pushing your content out into the world.

If it seems hard to navigate, do get in touch, we’d be happy to help if and where we can (notwithstadning IANAL). If your situation is at all complicated I recommend seeking professional legal advice — but do go out of your way to find one who understands both the motivation for, and the legal issues around, open licensing.

The procedural generation of geology

/

Procedural generation is a way of faking stuff with computers. But by writing code, or otherwise defining algorithms — not by manually choosing or composing or sculpting things. It’s used to produce landscapes and other assets in computer games, or just to make beautiful things. Honestly, I know almost nothing about it, and I don’t play computer games, so I’m really just coming at it from the ‘beautiful things’ side. So let’s just stick to looking at some examples…

Robert Hodgin produces jaw-dropping images, and happily one of his favourite subjects is meandering rivers. Better yet, Harold Fisk’s maps are among his inspirations. The results are mindblowing — just check this animation out:

What’s really remarkable is that everything on that map is procedurally generated: the roadways, the vegetation, the wonderful names.

If you love meanders (who doesn’t love meanders?), then you also need to know about Zoltan Sylvester’s work (not to mention his Etsy store). He produces some great animations, and also maintains some open-source Python projects (e.g. meanderpy) for producing them, so you can get stuck in and make your own.

It’s not just about meanders. Artist Tyler Hobbs has produced some striking images that strongly resemble structural cross-sections. In this thread he mentions that this wasn’t his goal, they just came out that way.

Mattias Herder, a space-obsessed viz wizard, did intend to produce crystals though. He’s using Houdini software, which I believe is also what Robert Hodgin uses for his maps. I wonder if any geologists are using it…

Landscapes are one of the big areas of application of this sort of tech, and while not strictly geological, I love these frozen vistas by French artist Guillaume Cottet:

Finally, this example from digital artist Ian Smith hints at a bit of the creative process. This guy really knows how to make rocks…

This is all so much magic to me, but I’m intrigued. Like the black hole in Interstellar, could this kind of work actually shed light on how dynamic, non-linear natural systems work? Or is it just an illusion?

What is an Ormsby wavelet anyway?

/

If you dabble in reflection seismic analysis, you probably know the Ricker wavelet. We’ve visited it a few times on this blog — Evan once showed how to make and plot one, I looked at some analytic properties of it, and we even played golf with it.

The Ricker is everywhere, but it has an important limitation — bandwidth. Its shape in the frequency domain is roughly Gaussian (below, left), which is the reason it only really has one parameter: the central frequency. This kind of spectrum is sometimes okay for older seismic surveys, especially marine data, because it matches the bandwidth reasonably. But for modern surveys, or even old land data, we often want a broader set of frequencies — more of a trapezoidal spectral shape. We need the Ormsby wavelet:

ricker-vs-ormsby.png

How to make an Ormsby wavelet

The earliest reference I can find to the Ormsby wavelet is in an article by Harold Ryan entitled, Ricker, Ormsby, Klauder, Butterworth — a choice of wavelets, in the September 1994 issue of the CSEG Recorder. It’s not clear at all who Ormsby was, other than “an aeronautical engineer”. And I don’t think anyone outside exploration geophysics knows what an Ormsby is, they just call it a ‘bandpass filter’.

Ryan helpfully provided both a time-domain analytic expression — which turns out to have four typos use the classical definiton of the sinc function — and a plot:

The equation in Ryan, and my modified Figure 3 (right). the result of the equation is in red.

The equation in Ryan, and my modified Figure 3 (right). the result of the equation is in red.

ryan_ormsby-cf-expression-2.png

This equation does not produce the wavelet (black) in the plot, however, it produces the one I’ve added in red. If you find this surprising, you shouldn’t — in my experience, it’s very common for the words and/or maths in a paper not to match its figures. [Edit: as noted above, in this case it’s because of how NumPy’s sinc function is defined; see the comment from Robert Kern, below.] We established this at the SEG Repro Zoo in 2018. If an author is not required to produce code or data, it’s not very surprising; and even if they do, the peer review system is not set up for referees to do this kind of check — apart from anything else, it’s far too onerous. But I digress.

After some fiddling around, I realized that the expression being passed to NumPy’s sinc function should be \(ft\), not \(\pi ft\). This produces a result that matches the figure almost exactly (and, counting wiggles, has the right frequency). So here’s the result of that new expression, shown in green here with the original figure (black) and the same red wavelet as above:

ryan_ormsby-cf-bruges.png

This green thing is the wavelet implemented in bruges so it’s easy to produce it; the arguments are the duration (0.4 seconds), the sample interval dt (4 ms) and the corner frequencies f (5, 10, 40, and 45 Hz respectively):

bruges.filters.ormsby(duration=0.4, dt=0.004, f=[5, 10, 40, 45])

What about other examples from the literature?

Good question! Apart from my own Python code in bruges, I did find one or two other implementations:

So it seems from this tiny experiment that only one of the implementations I found matched the figure in the Ryan article perfectly. The other wavelets are variations on the theme. Which is probably fine — after all, they are all only models for real seismic impulses — but in the interests of scientific reproducibility, I think it underscores the importance of transparent methodology and publishing your code.

Update on 9 Feb: A conversation in Software Underground revealed that Petrel’s version of the Ormsby wavelet matches the bruges implementation — but with a triangular window multiplied in (similar to how a Hamming window is multiplied into the seismic.jl version.

518px-Jupyter_logo.svg.png

I pushed my Python Jupyter Notebook to the new repro-zoo repository on GitHub. Please feel free to fork this project and add your own attempted reproductions of computational geoscience papers.

The original repro-zoo repo from the 2018 event is on SEG’s GitHub.

References

Ryan, H (1994). Ricker, Ormsby, Klauder, Butterworth — a choice of wavelets. CSEG Recorder 19 (7). Available online.

Soo-Kyung Miong, Robert R. Stewart and Joe Wong (2007). Characterizing the near surface with VSP and well logs. CREWES Research Report 19. Available online.

Future proof

/

Last week I wrote about the turmoil many subsurface professionals are experiencing today. There’s no advice that will work for everyone, but one thing that changed my life (ok, my career at least) was learning a programming language. Not only because programming computers is useful and fun, but also because of the technology insights it brings. Whether you’re into data management or machine learning, workflow automation or just being a more rounded professional, there really is no faster way to build digital muscles!

learn_python_thumb_2.png

Six classes

We have six public classes coming up in the next few weeks. But there are thousands of online and virtual classes you can take — what’s different about ours? Here’s what I think:

  • All of the instructors are geoscientists, and we have experience in sedimentology, geophysics, and structural geology. We’ve been programming in Python for years, but we remember how it felt to learn it for the first time.

  • We refer to subsurface data and typical workflows throughout the class. We don’t use abstract or unfamiliar examples. We focus 100% on scientific computing and data visualization. You can get a flavour of our material from the X Lines of Python blog series.

  • We want you to be self-sufficient, so we give you everything you need to start being productive right away. You’ll walk away with the full scientific Python stack on your computer, and dozens of notebooks showing you how to do all sorts of things from loading data to making a synthetic seismogram.

Let’s look at what we have on offer.

python_examples.png

Upcoming classes

We have a total of 6 public classes coming up, in two sets of three classes: one set with timing for North, Central and South America, and one set with timing for Europe, Africa, and the Middle East. Here they are:

  • Intro to Geocomputing, 5 half-days, 15–19 Feb — 🌎 Timing for Americas — 🌍 Timing for Europe & Africa — If you’re just getting started in scientific computing, or are coming to Python from another language, this is the class for you. No prerequisites.

  • Digital Geology with Python, 4 half-days, 22–25 Feb — 🌍 Timing for Europe & Africa — A closer look at geological workflows using Python. This class is for scientists and engineers with some Python experience.

  • Digital Geophysics with Python, 4 half-days, 22–25 Feb — 🌎 Timing for Americas — We get into some geophysical workflows using Python. This class is for quantitative scientists with some Python experience.

  • Machine Learning for Subsurface, 4 half-days in March — 🌎 Timing for Americas (1–4 Mar) — 🌍 Timing for Europe & Africa (8–11 Mar) — The best way into machine learning for earth scientists and subsurface engineers. We give you everything you need to manage your data and start exploring the world of data science and machine learning.

Follow the links above to find out more about each class. We have space for 14 people in each class. You find pricing options for students and those currently out of work. If you are in special circumstances, please get in touch — we don’t want price to be a barrier to these classes.

In-house options

If you have more than about 5 people to train, it might be worth thinking about an in-house class. That way, the class is full of colleagues learning things together — they can speak more openly and share more freely. We can also tailor the content and the examples to your needs more easily.

Get in touch if you want more info about this approach.