The physics of flocking in 3-D

You might think that bird watchers are only amateurs armed with binoculars. But biologists have also been studying flocks of birds for a long time. And now, they're helped by physicists, according to this New Scientist article about how European researchers are capturing flocks in 3-D. With a couple of digital cameras and some special software, these physicists were able to track the position of 2,700 starlings in less than 3 hours. Before this new method, called the epipolar technique, it took 1,000 hours to check only a handful of birds by using automated analysis took. The lead researcher says this project, called STARFLAG, 'may also apply to human economic behaviors, which also exhibit flocking phenomena like passing fads.' But read more...

You might think that bird watchers are only amateurs armed with binoculars. But biologists have also been studying flocks of birds for a long time. And now, they're helped by physicists, according to this New Scientist article about how European researchers are capturing flocks in 3-D. With a couple of digital cameras and some special software, these physicists were able to track the position of 2,700 starlings in less than 3 hours. Before this new method, called the epipolar technique, it took 1,000 hours to check only a handful of birds by using automated analysis took. The lead researcher says this project, called STARFLAG, 'may also apply to human economic behaviours, which also exhibit flocking phenomena like passing fads.' But read more...

A flock under attack

As an example, you can see on the left a flock under attack by a peregrine falcon in the sky of Rome. (Credit: STARFLAG project, INFM-CNR).

The STARFLAG project has been coordinated by Andrea Cavagna, a researcher at the Center for Statistical Mechanics and Complexity (SMC) of the Italian National Institute for Condensed Matter Physics (INFM). Here are two links to his personal contributionss to the project, "STARFLAG: a project on collective animal behaviour" and "Flocking."

The Department of Biological Physics, at Eötvös University in Budapest, Hungary, has clearly defined the goals of the project. Here is the last one. "As far as collective movements are common phenomena also in human behaviour (particular relevant in economics) we would like to explore the possibility of exporting the models and the techniques to economic collective choices. In this way we try to develop ways to investigate the reasons of social events, e.g. fashions, to understand socio-economic herding, and possibly to devise methods to tame dangerous excessive market fluctuations."

A flock under attack

On the left is another photo of several flocks in the sky of Rome. (Credit: STARFLAG project, INFM-CNR). This picture, as well as the one above, are featured on this STARFLAG photo gallery.

Let's return for a moment to the New Scientist article. "The real novelty is the method used to match up the same birds on the two images, says biologist Frank Heppner at the University of Rhode Island, US. "If the individual birds are different colours or sizes, there's no problem," says Heppner. "Alas, starlings are identical."

Cavagna and Heppner both say that this new technique can be rejected by biologists. But as Heppner says, "I cheerfully admit that it would take a lot more distance for me to understand the physics here than for a physicist to understand the biology. On the other hand, I've been watching these damned flocks for almost fifty years, and I can think like a starling. I challenge a physicist to do that."

This research work has been published as a two-part article (see below) under the title "The STARFLAG handbook on collective animal behaviour." Here are two links to the the abstract and a preprint version (PDF format, 45 pages, 3.64 MB) of "Part I, empirical methods."

And here are two links to the the abstract and a preprint version (PDF format, 27 pages, 1.01 MB) of "Part II, three-dimensional analysis." Here is the beginning. " The study of collective animal behaviour must progress through a comparison between the theoretical predictions of numerical models and data coming from empirical observations. To this aim it is important to develop methods of three-dimensional (3D) analysis that are at the same time informative about the structure of the group and suitable to empirical data. In fact, empirical data are considerably noisier than numerical data, and they are subject to several constraints. We review here the tools of analysis used by the STARFLAG project to characterise the 3D structure of large flocks of starlings in the field."

The final versions of these articles are available online from Animal Behaviour, an Elsevier scientific journal, which also carries an additional article about the same subject.

Sources: Colin Barras, New Scientist, May 7, 2008; and various websites

You'll find related stories by following the links below.