Movement beyond the snapshot – Dynamic analysis of geospatial lifelines☆
Introduction
Opportunities to trace individual movement have grown in tandem with the development of electronic transaction networks, location-aware devices and surveillance systems, all capable of tracking people (Mountain & Raper, 2001b), animals (Hulbert, 2001) or vehicles (Wolfson, Sistla, Chamberlain, & Yesha, 1999). Generically, this development has offered an opportunity to move ‘beyond the snapshot’ (Chrisman, 1998) with respect to our understanding of processes involving individual movement. Specifically, the recent arrival of devices capable of the low cost capture of high resolution locational data now allows the widespread construction of individually-based geospatial lifelines (Mark, 1998). Such individual lifelines presage a new era of movement analysis (Eagle & Pentland, 2005) in which scientists from various research fields previously hampered by sparse and random movement observations can now be hard on the heels of their subjects as they move in space and time.
Yet while there is a growing commitment of resources to the large-scale recording of paths, the analysis commonly conducted with trajectory data remains fairly limited in scope and sophistication (Wolfer, Madani, Valenti, & Lipp, 2001). In disciplines outside of geography which do not commonly use geospatial methods or theory this may be due to a lack of awareness and understanding of the power of spatial analysis and GISystems, and within geography GIScience’s’ fetish for the static may be a factor (Raper, 2002). What ever the cause, GIScience faces a challenge to develop sophisticated analytical tools that integrate geography’s spatial awareness with its long-term experience in processing large spatio-temporal data bases. This paper discusses opportunities and shortcomings of analysing lifeline data from a GIScience perspective, specifically in the situation where three spatial dimensions are involved and where movement is largely unfettered.
Clearly, the representation and analysis of geospatial lifelines challenge the GIScience community with respect to procedures for aggregation, generalisation, uncertainty and interpolation. Aggregation and generalisation of lifeline data can be considered to be important tools for coping with the voluminous outputs of movement-related agent-based simulations, for example with respect to emergency planning (Batty, Desyllas, & Duxbury, 2003) or transportation planning (Nagel, Esser, & Rickert, 2000). Uncertainty and generalisation of lifeline data is of interest for designers of location based services (LBS) analysing the lifelines of people tracked by location aware devices (Duckham et al., 2003, Mountain and Raper, 2001a, Smyth, 2001). The more basic derivation of actual motion descriptors, such as speed, motion azimuth, or path sinuosity, also merits attention. Such descriptors build, the underlying basis for attempts to investigate the similarity of trajectories (Sinha & Mark, 2005), which is an important task in spatio-temporal data mining and geographic knowledge discovery (Miller & Han, 2001), and may well pay dividends in the fields of spatialisation (Skupin & Fabrikant, 2003), eye-movement analysis (Fabrikant, 2005) and the evolution of semantic relationships in cognitive spaces (Pike & Gahegan, 2003).
Lifeline data analysis is also relevant to a range of applied research fields outside of geography itself or in cognate areas, such as animal biology and biogeography. In behavioural ecology, the key factors in avian navigation are still not completely understood (Bonadonna et al., 2005, Wiltschko and Wiltschko, 2003), but Steiner et al. (2000) identify the analysis of the homing routes of racing pigeons as the optimal method for almost any study in this field. Advanced path analysis is furthermore considered to be a crucial obligation for the interpretation of behavioural experiments conducted with genetically modified animals, for example for water maze experiments with mice exploring spatial learning (Wolfer et al., 2001). Similar analysis is also of increasing interest in agricultural science, contributing to the development, for example, of optimal grazing strategies for cattle with respect to livestock management (Ganskopp, 2001). Research involving video surveillance (Ng, 2001, Porikli, 2004, Shim and Chang, 2003) or sports scene analysis (Moore, Whigham, Holt, Aldridge, & Hodge, 2003) are further examples where disciplines exhibit deep interests in individual trajectories.
The distinctive feature of high resolution tracking data is that they allow the tracking of individuals along an actual movement path, leaving little need for interpretation between sparse observation points. Thus, at almost every instant along the lifeline we can robustly determine the individual’s current movement properties, such as speed, acceleration, motion azimuth, path sinuosity, as well as generate even more complex motion properties. Section 2 reviews work in this area, and elaborates the nature of the existing analytical tool set. In Section 3.1 we propose lifeline context operators to derive motion descriptors along dense trajectories, and adopt a dynamic analysis perspective for this. Section 3.2 rehearses a number of well known motion descriptors such as speed or motion azimuth, and introduces additional measures before discussing their computation as lifeline context operators. Section 3.3 explores the wide variety of analytical options that can be deployed by applying various standardisations for statistical analysis or using different aggregations of the tracking data. To illustrate these possibilities the trajectories of homing pigeons are identified as a research milieu (Section 4), and data from this area are utilised to demonstrate and critique the techniques (Section 5). The paper concludes in Section 6 with a brief review of prospects.
Section snippets
Related work
There is ample research on deriving overall descriptors of movement trajectories, such as time total, flying time, airline distance, net displacement, flight path, flight speed, bias from airline (e.g. Berger et al., 1999, Steiner et al., 2000, Turchin, 1998, Wolfer et al., 2001). In avian navigation research, where detailed tracking was impossible until recently, the recording of the vanishing bearing of homing pigeons was used as a ‘whole flight’ indicator, since this direction was believed
Methods
This section introduces a set of analytical approaches for lifeline data. Mark (1998, p. 12) defines a geospatial lifeline as a “continuous set of positions occupied in space over some time period. Geospatial lifeline data consists of discrete space–time observations of a geospatial lifeline, describing an individual’s location in geographic space at regular or irregular intervals”. All methods introduced in this section apply for a data set consisting of n individuals, mn fixes of the form (x,
Examples
This section illustrates the methodology introduced above using data emerging from an interdisciplinary study in behavioural ecology, investigating the navigation of homing pigeons. Given a data set of a total of approximately mn = 30,000 fixes of n = 54 individual pigeon flight trajectories (Guilbert, Dennis, & Walker, unpublished).
Discussion
This paper has identified a framework for enriching the toolset available for analysing densely sampled lifelines, and has explored ways to implement a growing range of movement parameters within that broad framework. These have covered movement properties of lifelines already deployed in other research, such as speed, movement azimuth, turning angles, or sinuosity, as well as additional measures and standardisations. In Section 3 we showed that there are, for many movement descriptors, more
Conclusions and outlook
Acknowledging the emerging opportunities for the mass analysis of individual movement data, an eclectic set of disciplines interested in movement have recently shown a growing interest in dynamic spatio-temporal analytical methods for trajectory data. These disciplines include geography, GIScience, data base research, animal behaviour research, surveillance and security analysis, transport analysis and market research.
In this paper we have adopted the concept of spatial context operators, often
Acknowledgements
Author’s current work is funded by the Swiss National Science Foundation, grant no. PBZH2-110315. The authors wish to thank David O’Sullivan who provided valuable comments on the presented research and previous drafts of the manuscript.
References (62)
How to reliably estimate the tortuosity of an animal’s path: Straightness, sinuosity, or fractal dimension?
Journal of Theoretical Biology
(2004)- et al.
Virtual biologists observe virtual grasshoppers: An assessment of different mobility parameters for the analysis of movement patterns
Ecological Modelling
(1999) - et al.
Spatial-analysis of animals movements using a correlated random-walk model
Journal of Theoretical Biology
(1988) - et al.
An exploratory data analysis (EDA) of the paths of moving animals
Journal of Statistical Planning and Inference
(2004) - et al.
Further evidence for visual landmark involvement in the pigeon’s familiar area map
Animal Behaviour
(1997) - et al.
Seeking structure in records of spatio-temporal behaviour: Visualization issues, efforts and application
Computational Statistics and Data Analysis
(2003) - et al.
Analysis of the movement and use of space of animals in confinement: The effect of sampling effort
Applied Animal Behaviour Science
(2006) Manipulating cattle distribution with salt and water in large arid-land pastures: A GPS/GIS assessment
Applied Animal Behaviour Science
(2001)- et al.
Positional entropy during pigeon homing II: Navigational interpretation of Bayesian latent state models
Journal of Theoretical Biology
(2004) Interactive geovisualization of activity-travel patterns using three dimensional geographical information systems: a methodological exploration with a large data set
Transportation Research Part C
(2000)
An edge-detection approach to investigating pigeon navigation
Journal of Theoretical Biology
Egocentric path integration models and their application to desert arthropods
Journal of Theoretical Biology
Positional entropy during pigeon homing I: Application of Bayesian latent state modelling
Journal of Theoretical Biology
A GPS logger and software for analysis of homing in pigeons and small mammals
Physiology and Behavior
Avian navigation: From historical to modern concepts
Animal Behaviour
Extended analysis of path data from mutant mice using the public domain software Wintrack
Physiology and Behavior
The discrete dynamics of small-scale spatial events: Agent-based models of mobility in carnivals and street parades
International Journal of Geographical Information Science
Orientation in the wandering albatross: Interfering with magnetic perception does not affect orientation performance
Proceedings of the Royal Society B
Beyond the snapshot: changing the approach to change, error, and process
Thread trailing of turtles: Methods for evaluating spatial movements and pathway structure
Canadian Journal of Zoology – Revue Canadienne De Zoologie
GIS modeling in raster
Using fractal dimensions for characterizing tortuosity of animal trails
Physiological Entomology
Imprecise navigation
GeoInformatica
Scale, sinuosity and point selection in digital line generalisation
Cartography and Geographic Information Systems
Reality mining: Sensing complex social systems
Journal of Personal and Ubiquitous Computing
Temporal zooming
Transactions in GIS
Modeling moving objects over multiple granularities
Annals of Mathematics and Artificial Intelligence
Cited by (73)
How regularly do people visit service places?
2023, Computers, Environment and Urban SystemsCitation Excerpt :The weekly service place visit dataset provides highly-resolved place types, including the name, coordinates, and type of up to five service places visited during the week. According to the literature, people visit five or fewer service places in a week, with all service types counted (Laube et al., 2007; Gonzalez et al., 2008; Gambs et al., 2012; Long & Nelson, 2013; Chen et al., 2018; Cui et al., 2021). The service place visit data are collected via the weekly visit survey, totaling 39,278 visits to 12,985 unique service places.
Patterns in Mongolian nomadic household movement derived from GPS trajectories
2020, Applied GeographyA feature extraction based trajectory segmentation approach based on multiple movement parameters
2020, Engineering Applications of Artificial IntelligenceFrom edit distance to augmented space-time-weighted edit distance: Detecting and clustering patterns of human activities in Puget Sound region
2019, Journal of Transport GeographyCalibrating spatial interaction models from GPS tracking data: An example of retail behaviour
2019, Computers, Environment and Urban SystemsCitation Excerpt :In recent years increasing effort has been put into the analysis of mobile phone data (recording movements between GSM towers) showing the potential of these data in identifying fine-grained variations in urban flows over time, for estimating movements in urban spaces, and identifying potential social interactions and significant places for individuals (inter aliaRatti, Frenchman, Pulselli, & Williams, 2006; Kwan, 2007; Ratti et al., 2010; Calabrese, Diao, Di Lorenzo, Ferreira, & Ratti, 2013; Calabrese, Ferrari, & Blondel, 2014; Ahas et al., 2015; Behadili 2016). Currently, however, these advances in movement data collection technologies are well ahead of the existing methods for extracting meaningful information from such data (Laube, Dennis, Forer, & Walker, 2007; J. Long & Nelson, 2012). Furthermore, there have been very few studies that have tried to analyse decision-making processes related to mobility using data from emerging technologies.
Space-Time GIS and Its Evolution
2017, Comprehensive Geographic Information Systems
- ☆
This paper is an enhanced and fully reviewed version of a paper that initially appeared in Proceedings of the ICA Workshop on Geospatial Analysis and Modelling, 8 July, Vienna, Austria.