First published online December 14, 2005
Journal of Experimental Biology 209, 128-140 (2006)
Published by The Company of Biologists 2006
doi: 10.1242/jeb.01970
Interpolation of animal tracking data in a fluid environment
Yann Tremblay1,*,
Scott A. Shaffer1,
Shannon L. Fowler1,
Carey E. Kuhn1,
Birgitte I. McDonald1,
Michael J. Weise1,
Charle-André Bost2,
Henri Weimerskirch2,
Daniel E. Crocker3,
Michael E. Goebel4 and
Daniel P. Costa1
1 University of California, Santa Cruz, Long Marine Laboratory, Center for
Ocean Health, 100 Shaffer Road, Santa Cruz, CA 95060, USA
2 Centre d'Etude Biologiques de Chizé, 79360 Villiers en Bois,
France
3 Department of Biology, Sonoma State University, Rohnert Park, CA 94928,
USA
4 NOAA, National Marine Fisheries, 8604 La Jolla Shores Drive, La Jolla, CA
92038, USA

View larger version (40K):
[in a new window]
|
Fig. 1. Laysan albatross Argos track (A), and selected examples of linear (B) and
Bézier (C) interpolation of this track (every 10 min). In B and C, the
circled cross represents an Argos position that was removed to use as a
reference position. The distance between this position and the corresponding
interpolated location was calculated for each mathematical algorithm that we
used (see Materials and methods). Note the possibility to visualize transit
speed in interpolated tracks.
|
|

View larger version (33K):
[in a new window]
|
Fig. 2. Selected example of a black-footed albatross track, illustrating some of
the various versions of the track. In this example, the geolocation-like Argos
track was interpolated using the Bézier algorithm with µ=0.3 (dashed
line).
|
|

View larger version (28K):
[in a new window]
|
Fig. 3. Occurrence of more accurate locations when using curvilinear algorithm
compared with a linear interpolation method for each species, each curvilinear
algorithm and each track (i.e. each individual). Dots (representing
individuals/tracks) are alternately shown in black and grey for clarity. Dots
to the right of the 50% line represent tracks in which the curvilinear method
yields more accurate locations than the linear interpolation method.
|
|

View larger version (23K):
[in a new window]
|
Fig. 4. Female northern elephant seal Argos tracks (A) and enlargement of a portion
of a track (B), illustrating Runge's oscillation (overshoot) of the cubic
spline (squares) interpolation (every 10 min) compared with the Bézier
algorithm (circles).
|
|
© The Company of Biologists Ltd 2006