Data are beautiful. They describe the state of the world around us without us having to rely on subjective feelings or judgements about it. But data can also be tricky – especially when there is lots of it. How do we cope with masses of data and how do we make sense of it all?
We encountered this problem recently when studying the synchronous calls of cicadas in the Australian bush. We set out listening stations that recorded the calls of male cicadas over 24 hours and over approximately half a square kilometre. We wanted to understand when and where the cicadas were calling and whether we could discern any patterns within this mass of noisy data.
To visualise the data, we first plotted circles representing each one of the recording stations. The intensity of the circle’s colour and its size is proportional to the volume of sound in that area of the forest at that time (the video plays 15 times faster than real time). But of course, this approach does not allow us to hear the interactions between cicadas in different areas of the forest. We could play the audio from the original sound recordings over the top of this video – but needless to say, this would sound horrific. Therefore, we took another approach – we decided to convert the cicadas' calls into music.
Each one of the four different coloured block of recorders plays a different chord (we chose the standard I–V–vi–IV progression in the key of C major). By doing this, you can now not only see, but hear when cicadas start to sing, when others cease singing, and listen to the additive effect of all individuals singing together across large swathes of the forest.
This data segment represents the cicadas' 'morning chorus’. The segment starts at around 5.30 am when the sun starts to rise and finishes 50 minutes later. You’ll notice that the cicadas on the right hand side of the video start to call first. This area of forest is the first area to be hit by the sun's morning rays, which likely warms the insects leading to the initiation of calling. Gradually more and more cicadas join the chorus, and the calls get louder and louder. You’ll also notice at the start of the video, the cicadas oscillate their calls; increasing and then decreasing in volume in about 22 second bouts (in real time). But over time, these oscillations are lost and the cicadas sing at a deafening 'full' volume. Half an hour later, towards the end of the morning chorus, the oscillations return before the cicadas eventually cease calling.
Visualising and ‘audiolising ’ data in this way allows us to discover interesting patterns and processes that would have been difficult to detect otherwise. If you were to stand in one spot in the forest, you would never be privy to the complexity of this calling behaviour around you. Data visualisation techniques such as this allows us to see the trees through the forest and bridges the gap between art and science.
S. Thomas Carmichael said, I am writing a paper about emergent properties in brain repair after injury (stroke, head trauma). In looking for a compelling way to introduce the term "emergent property" I ran across your work with cicadas in Australia. This is really interesting. I think most of us with an experience of mid summer in a woodsy place can identify with your work. I would like to cite this in my paper, and cannot locate an abstract or paper. Do you have a research reference?
Cordially,
Tom Carmichael