Description and aim
Research into movement of all sorts is an area which has seen increasing interest in recent years. On the one hand, researchers in diverse domains such as transportation research, animal ecology and epidemiology are faced with ever increasing possibilities to collect large volumes of data, often in real time. On the other hand, researchers in domains such as computing science, geographic information science and visualization are actively developing methods to analyze such data and extract useful in-formation. The workshop’s key aim was to bring together researchers from different domains, and in so doing provide a forum for discussion of advances in research on movement data. In so doing, the workshop sought to build on advances made through a range of recent initiatives, including the momentum created by the MOVE1 action of the COST Programme which part-funded the workshop.
The main goals of the workshop were threefold. Firstly, we wanted to facilitate the exchange of knowledge among quite diverse disciplines. Secondly, we aimed to identify important cross-cutting concerns and research questions culminating in the formulation of a research agenda for the coming years. Finally, we hoped to stimulate and encourage new research directions and collaborations.
The workshop was organized to encourage and enable discussion, and as such had three distinct strands of activity. The first strand of activity took the form of six keynote lectures were presented over the course of the week, three from researchers active in tracking a wide variety of animals and three from those involved in different domains of methodological research. Each talk was scheduled for around 45 minutes, allowing ample time for discussion, and the speakers were explicitly asked to formulate what they felt were key research challenges for the field in the coming years.
The second key strand of activity involved a data challenge, which participants were invited to complete in advance of the workshop. Participants were provided with access to data describing the movement of gulls by one of the workshop organizers, and given a set of so-called basic questions as well as more advanced interpretative challenges. Examples of basic questions included the identifying the number and length of trips taken by individual gulls, while more complex tasks involved, for ex-ample, the identification of common shared paths used by the gulls. A total of nine groups took part in the data challenge and submitted results before the workshop, and on the first day these results were presented to all participants. The data challenge proved particularly useful in providing a common ground for the discussions which followed, with the majority of workshop participants having, either through their own research or the challenge, the opportunity to explore real movement data and discuss these with domain experts.
The third strand of activity took the form of group discussions and work which was carried out throughout the workshop. An initial discussion following the presentation of the data challenge identified five themes:
§ Uncertainty and granularity: this group discussed the influence of, for example, sampling interval on the information derived from individual tracks, and explored how previously unused GPS data might be exploited to help in the extraction of behavioral data.
§ Integrating context: Contextual information, such as that relating environmental parameters (e.g. wind speed) to individual behavior is key to expert interpretation of movement data. However, methods to sensibly up and downscale environmental data, and other contextual clues, to the scale of individual tracks are still in their infancy.
§ Scale and scalability: Typical representations of movement focus on traditional geometric re-presentations of space, which typically result in over plotting and limited ability to explore individual events in very large volumes of data. This group developed a prototype visualization, using techniques from visual analytics, to explore in more detail stopovers in space and time during goose migration.
§ Space and time: Many approaches to analysis of movement data privilege either space or time, leaving an increasing need for methods which seek to both quantify and visualize behaviors in space and time simultaneously.
§ Grouping and clustering: Much methodological research has been expended on the development of methods to extract interesting and relevant patterns from movement data, and to cluster similar patterns. However, typically such research is carried out in isolation, and the group discussed how applications could be developed which would be used by much larger communities, for example through the development of web services for Movebank2.
Each group reported twice to the whole workshop during the week, and all groups had the opportunity to work intensively on real data.
The data challenge participants have agreed to work together to prepare and draft a paper on the results of this exercise, most likely for submission to Methods in Ecology and Evolution. A meeting re-port has also been submitted to Biology Letters for review and potential publication3. A number of the working groups agreed to continue working on the problems identified, and more papers are likely to result from these new links. Through MOVE funding it is likely that researchers involved in the work-shop will undertake short terms scientific missions to continue collaboration. Furthermore, a set of key references suggested in the keynote lectures will be summarized and added to MOVE’s bibliographic database4.
The workshop broadly succeeded in all of its goals, bringing together a diverse group of researchers, providing an opportunity to identify cross-cutting concerns and encouraging new research directions. The data challenge and generous time allocated to plenary and group discussions ensured that participants had ample time to familiarize themselves with and explore real movement data, in an enjoyable and stimulating atmosphere.
The Lorentz Center is thanked for facilitating this workshop, and the staff of the Lorentz Center for their assistance in arranging and running of the workshop. The workshop was partially supported by the ESF COST action IC0903 MOVE (Knowledge Discovery from Moving Objects).
3 Shamoun-Baranes J., E.E. van Loon, R. Purves, B. Speckmann, D. Weiskopf, K. Camphuysen. 2011. Analysis and visualisation of animal movement. Biology Letters (submitted).