Extracting stay regions with uncertain boundaries from GPS trajectories: a case study in animal ecology

In this paper we present a time-aware, density-based clustering technique for the identification of stay regions in trajectories of low-sampling-rate GPS points, and its application to the study of animal migrations. A stay region is defined as a portion of space which generally does not designate a precise geographical entity and where an object is significantly present for a period of time, in spite of relatively short periods of absence. Stay regions can delimit for example the residence of animals, i.e. the home-range. The proposed technique enables the extraction of stay regions represented by dense and temporally disjoint sub-trajectories, through the specification of a small set of parameters related to density and presence. While this work takes inspiration from the field of animal ecology, we argue that the approach can be of more general concern and used in perspective in different domains, e.g. the study of human mobility over large temporal scales. We experiment with the approach on a case study, regarding the seasonal migration of a group of roe deer.