GIScience 2016 Short Paper Proceedings An Algorithm for Empirically Informed Random Trajectory Generation Between Two Endpoints G. Technitis 1 , R. Weibel 1 , B. Kranstauber 2,3 , K. Safi 2,3 ! Department of Geography, University of Zurich, Winterthurerstrasse 190, 8057 Zurich, Switzerland Email: georgios.technitis@geo.uzh.ch ! Department of Migration and Immuno-ecology, Max Planck Institute for Ornithology, Am Obstberg 1, 78315 Radolfzell, Germany ! Department of Biology, University of Konstanz, Konstanz, Germany Abstract We present a method for enabling the researcher to create empirically informed, and thus realistic, random trajectories between two endpoints. The method used relies on empirical distribution functions, which define a dynamic drift expressed in a stepwise joint probability surface. We create random discrete time-step trajectories that connect spatiotemporal points while maintaining a predefined geometry, often based on real observed trajectory. The resulting trajectories can be used a)to generate null models for hypotheses testing, b)as a basis for resource selection models, through the integration of spatial context and c)to quantify space use intensity. 1.! Introduction Random trajectories have been increasingly used in movement ecology since their introduction in the early 1980s (Kareiva and Shigesada 1983), gaining significant popularity in the last two decades (Turchin 1998). A wide range of case studies have used the concept, addressing multiple questions related to movement and space use. The majority of the examples found in the literature, however, share one characteristic: the movement has only one restrictive point, the start. Consequently, the simulation is forced to start at a specific location, but can then move according to the set conditions in the given space. In the real world however, this is not always useful: when studying migration patterns (Codling et al. 2010), nest borrowing (Waldeck et al. 2008), or fusion of high and low frequency GPS points, etc. the ability to specify the ending point is crucial. Technitis et al. (2015) introduced RTG, an algorithm that enables the user to create randomly varying, possible trajectories between endpoints, based on principles of Time Geography. In this paper we substantially extend this algorithm. We present a methodology to connect two endpoints by generating empirically informed random trajectories, respecting characteristics of the moving object. Our approach is based on core theoretical concepts of Time Geography in combination with the Random Walk movement model, and most importantly, we use empirical data to inform our modelling process. 2.! Background Space-time prisms (STP) assist us in calculating the points accessible in space, given the time budget and the maximum speed of an agent (Kuijpers, et al. 2010). The calculated path space (in three dimensions defined by x,y and t), and more specifically its 2-D spatial projection, also known as potential path area (PPA), is a homogenous area within which the trajectory lies. The concept of the STP is very intuitive, although it accounts only for the maximum speed of the mover, gives no information regarding the preference of the mover within the given boundaries, and the result is an area, not an individual trajectory.
Read full abstract
Jun 22, 2023
Barbara Martin 
Open Access