Analysing infrequently sampled animal tracking data by incorporating generalized movement trajectories with kernel density estimation

https://doi.org/10.1016/j.compenvurbsys.2011.12.001Get rights and content

Abstract

When analysing the movements of an animal, a common task is to generate a continuous probability density surface that characterises the spatial distribution of its locations, termed a home range. Traditional kernel density estimation (KDE), the Brownian Bridges kernel method, and time-geographic density estimation are all commonly used for this purpose, although their applicability in some practical situations is limited. Other studies have argued that KDE is inappropriate analysing moving objects, while the latter two methods are only suitable for tracking data collected at frequent enough intervals such that an object’s movement pattern can be adequately represented using a space–time path created by connecting consecutive points. This research formulates and evaluates KDE using generalised movement trajectories approximated by Delaunay triangulation (KDE-DT) as a method for analysing infrequently sampled animal tracking data. In this approach, a DT is constructed from a point pattern of tracking data in order to approximate the network of movement trajectories for an animal. This network represents the generalised movement patterns of an animal rather than its specific, individual trajectories between locations. Then, kernel density estimates are calculated with distances measured using that network. First, this paper describes the method and then applies it to generate a probability density surface for a Florida panther from radio-tracking data collected three times per week. Second, the performance of the technique is evaluated in the context of delineating wildlife home ranges and core areas from simulated animal locational data. The results of the simulations suggest that KDE-DT produces more accurate home range estimates than traditional KDE, which was evaluated with the same data in a previous study. In addition to animal home range analysis, the technique may be useful for characterising a variety of spatial point patterns generated by objects that move through continuous space, such as pedestrians or ships.

Highlights

► KDE-DT is used to analyse infrequently sampled animal tracking data. ► KDE-DT produces a continuous probability density surface of an animal’s movement. ► KDE-DT is evaluated in the context of delineating animal home ranges. ► The results suggest KDE-DT produces more accurate home ranges than traditional KDE.

Section snippets

Introduction and background

Advances in global positioning system (GPS) and satellite tracking technologies have enabled unprecedented collection of locational information for mobile objects such as people (Khosravinasr & Zhu, 2010), animals (Amstrup, McDonald, & Durner, 2004), vehicles (Coifman & Kim, 2009), ships (Knauss & Alexander, 2000), and aircraft (Feng, Ochieng, Walsh, & Ioannides, 2006). Likewise, recent efforts in GIScience have contributed numerous strategies for managing (Kollios et al., 2005, Zhang et al.,

KDE using generalised movement trajectories

Trajectory-based density estimators, such as TGDE and the Brownian Bridges KDE method, generate continuous probability density surfaces from tracking data by applying data smoothing techniques to a model of trajectories approximated by straight lines. In cases when individual trajectories cannot be accurately estimated in that matter, a model of approximated or generalised movement trajectories could be used as a substitute. Such an approach was initially proposed by in preliminary conference

Home range analysis

Animal home range analysis involves characterising the spatial intensity of an animal’s movements during a specified time period from sampled point locations (Borger et al., 2008, Johnson et al., 2008, Kernohan et al., 2001). The goal is generally to create a footprint of the animal’s space-use pattern, where the home range delineates the area ‘typically’ occupied by the individual and the core area defines regions of more intensive activity (Worton, 1989). Although home ranges are sometimes

Discussion and conclusions

The application of KDE-DT to panther tracking data illustrated the technique’s potential for generating probability density surfaces from infrequently sampled tracking data, as well as demonstrated its potential for use as a home range estimator in ecology. Likewise, results of its application to simulated animal locational data indicate the method produces reasonably accurate measures of wildlife home range and core area sizes, as quantified from the 95% and 50% contours of the resulting

Acknowledgements

The authors would like to thank Dave Onorato from the Florida Fish and Wildlife Conservation Commission Panther Project for providing the radio-telemetry data used in this research. This research was partially funded by grants made to the authors from the National Science Foundation (BCS- 1062947 [Downs]; BCS- 1062924 [Horner]). The contents of this manuscript are the responsibility of the authors and do not reflect the views of the NSF.

References (64)

  • H.B. Yu

    Visualizing and analyzing activities in an integrated space-time environment – Temporal geographic information system design and implementation

    Transportation Research Record

    (2007)
  • N.J. Aebischer et al.

    Compositional analysis of habitat use from animal radio-tracking data

    Ecology

    (1993)
  • S.C. Amstrup et al.

    Using satellite radiotelemetry data to delineate and manage wildlife populations

    Wildlife Society Bulletin

    (2004)
  • M. Andersson et al.

    Reporting leaders and followers among trajectories of moving point objects

    GeoInformatica

    (2008)
  • G.M. Blundell et al.

    Linear home ranges: Effects of smoothing, sample size, and autocorrelation on kernel estimates

    Ecological Monographs

    (2001)
  • C. Bonhomme et al.

    Metaphors for visual querying of spatio-temporal databases

  • L. Borger et al.

    Are there general mechanisms of animal home range behaviour? A review and prospects for future research

    Ecology Letters

    (2008)
  • G. Borruso

    Network density and the delimitation of urban areas

    Transactions in GIS

    (2003)
  • F. Bullard

    Estimating the home range of an animal: A Brownian bridge approach

    (1991)
  • B. Coifman et al.

    Measuring freeway traffic conditions with transit vehicles

    Transportation Research Record

    (2009)
  • N.A.C. Cressie

    Statistics for spatial data

    (1993)
  • J.A. Downs

    Time-geographic density estimation for moving objects

    Lecture Notes in Computer Science

    (2010)
  • Downs, J.A., Heller, J.H., Loraamm, R., Oppenheim-Stein, D., McDaniel, C., Onorato, D., in press. Accuracy of home...
  • J.A. Downs et al.

    Network-based home range analysis using Delaunay Triangulation

  • J.A. Downs et al.

    Effects of point pattern shape on home-range estimates

    Journal of Wildlife Management

    (2008)
  • J.A. Downs et al.

    A characteristic-hull based method for home range estimation

    Transactions in GIS

    (2009)
  • J.A. Downs et al.

    Time-geographic density estimation for home range analysis

    Annals of GIS

    (2011)
  • S. Feng et al.

    A dynamic sampling scheme for GPS integrity assessment

    Aeronautical Journal

    (2006)
  • J. Fieberg

    Kernel density estimators of home range: Smoothing and the autocorrelation red herring

    Ecology

    (2007)
  • M.-J. Fortin et al.

    Spatial analysis: A guide for ecologists

    (2005)
  • W.M. Getz et al.

    A local nearest-neighbor convex-hull construction of home ranges and utilization distributions

    Ecography

    (2004)
  • R.A. Gitzen et al.

    Evaluation of least squares cross validation bandwidth selection options for kernel estimation

    Wildlife Society Bulletin

    (2003)

    • Cited by (22)

    • Identifying habitat selection via Fauna of Hor Al-Dalmaj and its surrounding terrestrial Areas, Iraq by using ArcGIS

      2023, Materials Today: Proceedings

    • Uncertainty in geospatial health: challenges and opportunities ahead

      2022, Annals of Epidemiology
      Citation Excerpt :

      Researchers need caution when leveraging on massive amount of GPS information, as such data may be incomplete. Although several studies have proposed algorithms to fill in spatiotemporal gaps [108–111], it is not clear how incomplete data may bias measures of health outcomes, such as exposure to air pollution. Furthermore, collecting GPS data to study individual behavior and health outcomes may contain privacy and confidentiality concerns.

    • Biking islands in cities: An analysis combining bike trajectory and percolation theory

      2019, Journal of Transport Geography
      Citation Excerpt :

      For the first question, trips that traversed at least one road segment within a biking island are extracted and deemed as the associated trips for that biking island. Then, the hotspots for the origins and destinations (ODs) of the associated trips are detected using kernel density function (Downs and Horner, 2012). The OD hotspots can identify locations where the majority of associated trips start/end for each biking island.

    • Automatic intersection and traffic rule detection by mining motor-vehicle GPS trajectories

      2017, Computers, Environment and Urban Systems
      Citation Excerpt :

      The presence of small sharp points in the density grid (surface) may result in the detection of false intersections; therefore, kernel density estimation (KDE) is used for data smoothing. Kernel density is widely used in traffic accident identification, bandwidth selection, movement trajectory analysis, data forecasting, etc. (Ivana, Jan, & Kamila, 2013; Joni & Mark, 2012; Michal, Richard, & Zbynek, 2013; Siddharth & James, 2016). The kernel function is expressed as a bivariate probability density function, in which the kernel function and bandwidth determine a bump, as shown in Fig. 4, that controls the smoothness of the resulting density curve (Silverman, 1986).

    • Agent-based simulation of Muscovy duck movements using observed habitat transition and distance frequencies

      2017, Computers, Environment and Urban Systems
      Citation Excerpt :

      A popular reason is to generate a complete baseline dataset of known tracking data to test methods that are useful for analysing samples of VHF, GPS, or satellite data collected in the field. For example, simulated tracking data are routinely used to test methods of home range analysis, where the performance of different techniques is evaluated for samples of different sizes, tracking intervals, or other qualities (Downs & Horner, 2009, 2012; Getz et al., 2007; Girard, Ouellet, Courtois, Dussault, & Breton, 2002; Laver & Kelly, 2008). Simulated animal tracking data are also widely used as reference datasets for the purposes of quantifying animal interactions.

    • Understanding the impact of temporal scale on human movement analytics

      2022, Journal of Geographical Systems
    View all citing articles on Scopus
    View full text
    Copyright © 2011 Elsevier Ltd. All rights reserved.