How to Get More GPS Fixes While Saving Battery Life
Our very good client in Australia conducted a feral cat project in which he determined that by turning the GPS on very frequently the collars were able to produce more GPS positions before the battery died than if he turned on the GPS less frequently. In other words, asking for GPS more frequently took less power. Thanks to Dr. Hugh McGregor! Here is what he kindly shared with us:
“One of the biggest struggles when designing studies of animal movements using GPS collars is deciding upon the frequency of fixes to set. You essentially have to choose between maximizing battery life and maximizing the detail of information obtained. Long time-intervals between fixes will extend the battery life per deployment of a GPS device, but result in coarse information of movements. Short time-intervals give more detail but will drain the battery faster. This trade-off is complicated by the fact that GPS units can obtain a fix faster if they can remember where they were last, meaning that short time-intervals can actually be more efficient.
We learned just how much more efficient high-frequency fixes can be during our research on feral cat movements in north-west Australia, a project run by the University of Tasmania and the Australian Wildlife Conservancy. We bought Telemetry Solutions Quantum 4000 enhanced collars, with each predicted to generate 1700 fixes. At first, we decided on recording fixes every three hours, which should have given us around 6 months per deployment. It was coarse data, but our battery life was extended. As the habitat in our study area was savanna with an open canopy, the GPS units did not take the full 60 seconds to obtain a fix, and we were getting around 2000 fixes per deployment. Then, about a year into our study we collaborated with another researcher investigating cat – dingo interactions. This required fine-scale movement data, with a fix, collected every 15 minutes to see if these predators interacted. This finer-scale should have given us around 18 days of data. However, when we collected the collars three months later, we were blown away. Each had recorded around 65 days’ worth of data!
When our collars were collecting fixes at three-hourly intervals, evidently the collars had a cold start each time. However, at the finer scale of 15-minute intervals, the GPS units already knew roughly where they should be. The end result was 3.4 times the efficiency in the time it took to obtain a fix.
Not only was this fine-scale data more efficient per battery, but it told so much more about the lives and patterns of these cats. You feel so much more in the mind of the animal when you are watching a track overlaid on a map, as opposed to a pile of dots. You see their decision making in action; following roads, quick forays into open burnt areas, weaving around rivers etc. The majority of cats’ rests were less than three hours long, so it was only with the fine-scale movement data that we were able to differentiate habitat selection between resting and moving. In the end, we were able to write five manuscripts from this fine-scale dataset, with a possibility of two more. At best, we would have only been able to write two had we only collected coarse fixes three-hours apart.
Of course, in terms of overall duration, the 65 days of fine-scale fixes does not compare with the 200 days that would have had resulted from three-hour fixes. For long-term studies, it will always be better to have long time-intervals between fixes. Also, not all other research projects will obtain the same difference in efficiency that we did. Studies in regions with more canopy cover will have a greater average time-until-fix, so are unlikely to get the same extra efficiency we did. In any case, we demonstrate the potential for large gains in efficiency from finer-scale movement data and the richer datasets that result in more possibility for analysis.
Further detail can be found in our publication.
Dr. Hugh McGregor”