|
Counting the polar bears by satellite Researchers test a new way to determine bear populations in inaccessible Arctic regions; might also be usable for other wildlife Alan Bailey Petroleum News
A team of U.S. and Canadian scientists has been investigating the potential to use high resolution satellite imagery for monitoring wildlife in remote regions of the Arctic. An initial study, the results of which have been published in a paper in scientific journal PLOS One, determined that it is possible to count polar bears using the satellite technique, given that modern satellites can achieve image resolutions as fine as 0.5 meters, a resolution sufficient to depict the bears in summer habitat as white specks against a dark background.
“We tested the use of satellite technology from DigitalGlobe to count polar bears by tasking the satellite to collect photos from an area where we were also conducting aerial surveys,” said Dr. Todd Atwood, research leader for the U.S. Geological Survey Polar Bear Research Program. “We then analyzed the satellite and aerial survey data separately and found that the abundance estimates were remarkably similar.”
In addition to applying the satellite technique to obtain polar bear population estimates, the research team thinks that the technique might work for other Arctic mega fauna, with dark animals such as musk oxen or caribou, for example, perhaps being counted against a snow-covered spring landscape.
And, although the satellite technique does not yield the same level of detail as traditional monitoring programs such as aerial surveys, the technique does not disturb the animals and offers the possibility of surveying vast areas of remote territory.
Study location For their study into the effectiveness of the technique, the scientists decided to try counting the polar bear population on an island in the Foxe Basin in the eastern Canadian Arctic. The team conducted the survey in the late summer, at a time when the region was ice free and the bears had retreated from sea ice to become stranded on the island. With no snow cover on the land, the bears were conspicuous against the dark land surface.
Essentially, the team obtained images of the island from early September 2012 from two satellites, one with 0.5-meter resolution and the other with 0.65-meter resolution. The team then compared these target images with reference images obtained in August 2009 and 2010, when no bears were on the island. White specks appearing on the target images but not on the reference images were interpreted as polar bears: Two independent observers identified and counted presumed bear observations, later comparing their results and eliminating questionable observations. The team then surveyed the island by helicopter, to ensure that the observed white specks could not be accounted for as features such as light colored rocks or foaming water.
And, a couple of days before the date of the satellite survey, the team conducted an aerial survey of the bears, to obtain a population estimate that could be compared with the satellite-based assessment.
The result of the satellite survey was the identification of 92 bears on the island, with five occurrences of likely family groups. The aerial survey located just 34 bears on the island, but this extrapolated to a population estimate of 102 after applying an abundance model to the data. The resulting abundance estimates of 92 and 102 were very close, within their margins of error, suggesting that satellite imagery is a promising tool for monitoring wildlife.
However, the independent review of the imagery by two observers proved to be a tedious, 100-hour exercise, making a re-examination of the images impractical and suggesting that application of the technique to a more extensive territory may require an automated image classification process, the researchers reported in their PLOS One paper.
Other lessons learned In addition, the two observers initially came up with markedly different polar bear counts, with these differences appearing to relate to the observers’ previous experience in observing polar bears and to the use of different techniques for locating bears on the images - it appears that the optimum method for locating the bears is to regularly compare the target and reference satellite images, rather than to just use the target image to find the bears and then use the reference image for verification. Application of the satellite technique will likely require specified search protocols, coupled with training for observers, the researchers reported.
Because of uncertainties over the timing with which a satellite will traverse a particular location, it is not currently possible to synchronize surface surveys with satellite imagery, to directly verify the results of satellite observations. And because of the limited ability of the satellite imagery to resolve images of bear cubs, the satellite technique appeared to underestimate the number of bear family groups likely to be present.
Further development Given that the trial survey was conducted at a location where conditions for the survey were ideal, extrapolation of the satellite technique to more challenging places, perhaps with lower population densities and a more variable landscape, will require further technique development, the researchers suggested. And multi-spectral imagery may prove useful in picking up the spectral signature of the surveyed animals, and in conducting surveys in more challenging environments, the researchers wrote.
|
