An Accuracy Assessment of Global Navigation Satellite System Wildlife-Tracking Collars in the Southern Chilean Patagonia

In short: Can wildlife collars be trusted in the rugged mountains of Torres del Paine? Kersey Lawrence (published as Kirstin Lawrence-Apfel) and colleagues tested three brands of GPS collar against survey-grade control points in Chilean Patagonia. On open ground the collars came close to the mark, but in canyons and forest they missed more positions, and a few were off by 400 to 500 m.

Summary by Kersey Lawrence, Original Wisdom · Updated 9 November 2026

Citation and access

Authors: Kirstin Lawrence-Apfel (now Dr. Kersey Lawrence), Thomas H. Meyer, Kazi Arifuzzaman, Isaac M. Ortega
Year published: 2012
Journal or source: Anales Instituto Patagonia (Chile) 40(1): 105–113
Full citation: Lawrence-Apfel, K., Meyer, T.H., Arifuzzaman, K., and Ortega, I.M. 2012. An accuracy assessment of global navigation satellite system wildlife-tracking collars in the southern Chilean Patagonia. Anales Instituto Patagonia (Chile) 40(1): 105–113.
DOI: https://doi.org/10.4067/S0718-686X2012000100014 (opens in a new tab)
Access status: Hosted with permission

Download PDF

Detailed keywords

Species: None studied. The corridor test sites were chosen as places that mountain lions (pumas, Puma concolor), guanacos (Lama guanicoe) or huemuls (Hippocamelus bisulcus) could reasonably be expected to use.
Location and habitat: Torres del Paine National Park, Magallanes, Patagonia, Chile (about 2,300 km², 60 m to 3,000 m elevation, on the eastern edge of the Andes; 50°45´ to 51°20´ S, 72°31´ to 73°22´ W). Plains, cliffs, glaciers, deep valleys, shrubland, grassland, beech deciduous forest and desert. Control markers TDP1, TDP2 and TDP4: aluminium caps set in bedrock with open sky, in the Laguna Amarga sector and elsewhere in the park. Corridors: SMAC1, Camp Valley (open grassland and small shrubs in a wide valley); SMAC2, Vega Puma (scrubland and patches of 9 to 15 m deciduous trees in a wide canyon with steep cliffs); SMAC3, Lago Grey (mature forest of 10 to 20 m trees with spreading canopy, near a glacial ice field, below steep, narrow cliffs).
Topics: Author note: Kirstin Lawrence-Apfel is a previous name of Dr. Kersey Lawrence, founder of Original Wisdom. GPS collar positioning accuracy; fix failures (scheduled positions not collected); differences between collar manufacturers; directional bias; antenna orientation and animal posture; canopy and topographic sky obstruction; outliers; manufacturer accuracy claims; geodetic control; rapid static positioning; what collar error means for home range, movement and habitat-use analyses and for species with small, specialised home ranges; how animal activity and habitat choice affect fixes; Landsat image resolution (30 m pixels) and habitat or corridor mapping.
Statistics or models: Analysis of variance (ANOVA) with Tukey post hoc tests for differences between collars and stations; paired t-tests for antenna orientation; descriptive statistics. Errors were computed as radial Euclidean distances in a local topocentric coordinate system in Mathematica v.8. Sample sizes were unequal because of preset schedules: ATS n=13, BlueSky n=81, SirTrack n=66. Inconsistency inside the paper: the results give the SirTrack collar's mean error across all corridors, outliers included, as 61 m (standard deviation 133 m), while the discussion refers to "a mean 101-m error".
Data type: New field data, collected in December and January in both 2005 and 2006. This is the published version of Chapter 3 of Lawrence-Apfel's 2009 master's thesis, which has its own entry.

Synopsis

1. What they did

Global Navigation Satellite System (GNSS) collars, commonly called Global Positioning System (GPS) collars, are meant to give accurate positions for wildlife anywhere and in any weather. Torres del Paine National Park in Chilean Patagonia is rugged country of cliffs, canyons, forest and glaciers, where satellite signals can be blocked. The researchers wanted to know whether collars would work well there before they were put on animals; this was the first such test in the park.

They asked 15 collar manufacturers to take part. BlueSky, Advanced Telemetry Systems (ATS) and SirTrack each lent a collar (only SirTrack could in the first season).

Stationary test: the collars were placed side by side on three survey markers whose positions were known to within 5 cm, with open sky all round. The antennas pointed straight up for two hours, then sideways for two hours. Each collar ran on its manufacturer's preset schedule.
Simulated moving animal corridors: field staff carried a SirTrack collar along routes in three habitats, from open steppe to forested cliffs, with a survey-grade receiver alongside, stopping every 10 minutes to collect 20 minutes of survey observations and one collar position.

2. What they found

Stationary test

Under these ideal conditions, mean errors were 4 m for ATS (largest 7 m), 8 m for BlueSky (largest 27 m) and 16 m for SirTrack (largest 84 m).
The SirTrack collar had four outlying positions, all at one marker (TDP4), and its east-west readings were all shifted to one side and differed significantly from the other two collars. The authors found no cause.
Turning the antenna sideways made no significant difference to error.

Corridors

The SirTrack collar recorded 30 of 45 attempted positions (67%): 70% on open steppe, 86% in the scrubby canyon and only 38% (3 of 8) in the cliff forest.
No error was under 2 m, and 26 were under 45 m. The largest errors were 84 m and 396 m (steppe), 414 m (canyon) and 528 m (cliff forest).
Mean error, outliers included, was 38 m, 80 m and 186 m for the three corridors; without the three largest errors it was 20 m, 14 m and 14 m. Across all corridors it was 61 m with the outliers and 19 m without.

3. What they concluded

Even under ideal conditions the collars differed significantly from each other, but the differences were small compared with the home ranges of the animals that would wear them.
Mean error and missed positions generally increased with canopy and steep terrain.
Large outliers, obvious here because the true positions were known, may not be easy to spot in data from wide-ranging animals and could suggest habitat use that never happened. Species with small, specialised home ranges are most at risk of being misread.
Mapping on 30 m Landsat images could overestimate critical habitat and corridors for wide-ranging species, and get them entirely wrong for highly localised species.
Although the results were disappointing, the errors are small compared with the daily movements of large mammals, and an occasional undetected error would likely not invalidate a whole study. The authors recommend the collars for wildlife research in Torres del Paine, with care over the scale of analysis and the species.

5. Tracker or observer certification

Not needed.

6. Other technologies, skills or innovations

BlueSky, ATS and SirTrack GPS collars.
Javad Legacy and Odyssey dual-frequency, dual-constellation (GPS and GLONASS) receivers, tied to International GNSS Service (IGS) reference stations.
Rapid static positioning with a base and a rover, giving centimetre-accurate check points along the corridors.
TOPCON PC-CDU and Pinnacle software, and Wolfram Mathematica v.8.
A Landsat 7 satellite image of the study area.

7. Implications for conservation (from the paper)

Undetected collar errors could be misread as habitat or landscape use, which matters most for species with small home ranges and critical habitats.
Collars should be tested where they will be used, before they go on animals; this is rarely done.

8. Implications for tracking (Original Wisdom's observations)

These are Original Wisdom's observations, not the authors'.

Collar points in canyons and forest can be hundreds of metres off, and that kind of error can look like real habitat use. Tracks and signs on the ground can show whether an animal was really there.

Questions and answers

What did Lawrence-Apfel and Meyer et al. (2012) study, and how?

Can wildlife collars be trusted in the rugged mountains of Torres del Paine? Kersey Lawrence (published as Kirstin Lawrence-Apfel) and colleagues tested three brands of GPS collar against survey-grade control points in Chilean Patagonia. On open ground the collars came close to the mark, but in canyons and forest they missed more positions, and a few were off by 400 to 500 m. Read the full section

What did Lawrence-Apfel and Meyer et al. (2012) find?

Under these ideal conditions, mean errors were 4 m for ATS (largest 7 m), 8 m for BlueSky (largest 27 m) and 16 m for SirTrack (largest 84 m). The SirTrack collar had four outlying positions, all at one marker (TDP4), and its east-west readings were all shifted to one side and differed significantly from the other two collars. Read the full section

What did Lawrence-Apfel and Meyer et al. (2012) conclude?

Even under ideal conditions the collars differed significantly from each other, but the differences were small compared with the home ranges of the animals that would wear them. Mean error and missed positions generally increased with canopy and steep terrain. Read the full section

Who did the tracking in Lawrence-Apfel and Meyer et al. (2012), and were they certified?

Not needed. Read the full section

Cite this summary

Lawrence, K. 2026. Summary of: An Accuracy Assessment of Global Navigation Satellite System Wildlife-Tracking Collars in the Southern Chilean Patagonia (Lawrence-Apfel and Meyer et al., 2012). Original Wisdom, Tracking Literature. https://www.originalwisdom.com/tracking-literature/2012-lawrence-apfel-meyer-gnss-collar-accuracy-patagonia/

For the paper itself, cite the full citation above.

← Back to Tracking Literature