Документ взят из кэша поисковой машины. Адрес оригинального документа : http://www.sevin.ru/seminars/argos2/22_20100204_biologging.pdf Дата изменения: Mon Feb 8 12:54:00 2010 Дата индексирования: Tue Oct 2 20:12:27 2012 Кодировка:

BIOLOGGING

Page 1

The challenge of tracking animals in the wild

Workshop IPEE/CLS - 3-FEB-2010


Tracking and modeling
Page 2

OBSERVE UNDERSTANDMODELFORECAST
Biologging data are indispensable to: Observe movements : velocities, migration routes, orientation mechanisms, energy budgets....

Identify habitats: foraging & breeding grounds, temperature-depth habitats.....
Study how animals exploit or are constrained by their environment (requires simultaneous environmental observations)


Tracking animals « above water »
Page 3

Satellite tracking (ARGOS) of animals is always a challenge : Beacons with weak output power & in difficult environmental conditions Majority of low accuracy positions (class A or B) with no error estimation


Tracking animals underwater
Page 4

Internal tag

Pop-up tag
Light-based geolocation is presently the only operational solution for large scale long-term tracking BUT obtaining accurate positions is extremely difficult


Improving « above water » tracking
Page 5

Present ARGOS positioning algorithm:
4 msg : Sequential estimation (least-square) of position + error estimation (loc 0,1,2,3) 2 or 3 msg : geometric estimation of positions, error not estimated (loc A, B)

New algorithm (on-going PhD thesis)
Unscented Kalman Filtering of Argos (doppler shift) measurements assuming a random walk model for animals Slightly less « robust » ( more subtle initialization) Position with error estimates obtained even with 1 msg per pass General reduction in localization error No more « image positions »


An Example
Page 6

The case of Munaroh, a female olive ridley tagged May 20, 2009 on Prancak beach (Bali).


Light-based underwater geolocation
Page 7

· Estimation of sunrise & sunset times allows geolocation of the tag

· Precision in longitude ~ constant ~
· Precision in latitude drops near the equinoxes


Error pattern
Page 8

In practice, a 3 minutes error on sunrise/sunset time is very good !


Example : Atlantic bluefin tuna
Page 9

Large scatter in light-based position estimates


Combining information
Page 10

Typical tag data include the following measurements: Light level Depth (pressure) Temperature


26°

25°

24°
Maximum distance traveled in one day D

D-1 100 m 200 m 300 m


Bluefin tuna
Page 12

Royer & Lutcavage, 2006


Basking shark
Page 13