Spotlight: Eyes in the sky | Wader Study 130(3)
by Deborah Buehler originally published in Wader Study 130(3)
How do you count something you cannot see? How do you even know what you are missing?
Researchers in conservation and ecology are often faced with these questions because they need to count species and individuals in challenging terrain. Even relatively flat environments like coastal mudflats can be challenging. For example, researchers counting shorebirds from dry land might miss birds tucked in behind elevated ground or just below the horizon of a slanting shoreline. In this issue of Wader Study, Castenschiold and colleagues report on how they confronted the task of counting birds in such blind zones in the Danish Wadden Sea.1 They used drones as their “eyes in the sky” and aimed to: (1) predict, map and validate the locations of the blind zones, (2) use drones to peer into blind zones and determine how many birds were missed from the shore, and (3) examine whether differences between species affected the extent of the problem.
Before drones, the only “eyes in the sky” were those of trained researchers, who were flown over mudflats by fearless pilots, and were tasked with counting birds as they flew up when disturbed by the plane. Those who have performed such counts know that it is exhilarating, but also that counting large numbers of birds from the window of a moving plane is rather difficult. Furthermore, flying low and slow as birds fly up beneath and around the aircraft is dangerous. Pilots must sometimes take evasive measures to avoid collisions with larger birds and there is little room for error when flying so low to the ground. Indeed, such flights carry an unfortunately high risk of mortality for the birds and the people.2 In contrast to planes, drones are small and can fly slowly and safely at lower altitudes. This causes far less disturbance, is safer, and has a much lower cost. Additionally, drone cameras can take pictures, which become a permanent record of the flight.
The Wadden Sea is the largest unbroken system of coastal sand and mud flats in the world, and borders the Netherlands, Germany and Denmark.3 The area is of international importance for migratory shorebirds, which have been monitored there for decades. The researchers worked along the west-facing shore of the Danish Wadden Sea, in the intertidal area where sea and land meet between high and low tides. Their study site encompassed the intertidal area visible from five ground-based vantage points on the Ribe, Rejsby and Ballum seawalls. This site provided high numbers of birds and a topography that made it possible to map blind zones and then use drones to peer within them.
Castenschiold and colleagues began by using a visibility simulation tool in Geographic Information System (GIS) software to predict the location of blind zones. This produced a file with elevation measurements plotted on an evenly spaced grid called a raster. Using this raster and the location of the five observation points on the seawall, the researchers could calculate the vertical distance below the observational line of sight for every given point on the mudflats. This was referred to as vertical invisibility (VI).
The next step was to verify the blind zones predicted by simulation with data from the field. To do this, the researchers had to get out onto the mud. They set up dummy birds, approximating the heights of the most common shorebirds in the area (12, 20, 30 and 40 cm tall), along lines of sight between the mudflats and the vantage points. One observer looked out from the seawall through a telescope, while another moved the dummy bird on the mudflat away from the foreland until only the top 5–10 cm of the dummy was visible to the observer. Because of the way line-of-sight works looking down an uneven slope, the tallest dummy birds were set closer to the obstructing foreland and the smallest were set further out onto the mudflat (Figure 2a from the paper1 is reprinted below for a visualization of how these blind zones work). The blind zone stretched from the edge of the foreland to the 12-cm silhouette.
The blind zones mapped using dummy birds in the field closely matched those simulated using the GIS software. Thus, the visibility simulations proved a reliable tool for finding blind zones, and with those areas mapped, the researchers knew where to conduct the drone surveys. They conducted 22 successful flights between 25 September and 15 November 2019, primarily around high tide. The drone was flown within a “green zone” altitude where the minimum flight height did not cause disturbance to the birds and the maximum flight height still allowed for reliable species identification from the pictures and video taken by the onboard camera.
Using images taken from the drones, the researchers manually identified birds from more than 120,000 positions on the mudflat. These data were manually entered into the map produced by the GIS software, which already contained the simulated VIs for positions on the mudflat. Roosting birds were mainly detected within 300–450 m and 600–750 m from the foreland, likely because sandbanks at these distances provided ideal resting areas during the study period. There were 14 species with numbers exceeding 100 individuals present on three or more of the surveyed high tide roosts, and these species became the focal species for the study. With birds identified to species, the researchers could estimate the standing height of most of the birds using the average height for the species. This allowed them to determine whether the bird, detected by the drone, would have been visible to an observer on the seawall. A bird was deemed visible if VI was less than standing height and invisible if VI was greater than standing height.
Because species differ in their size, habitat preference, and preferred position on the mudflat, some were more likely to go undetected than others. Take the example of Bar-tailed Godwits Limosa lapponica and Eurasian Curlews Numenius arquata, both large species, but one far more visible from the seawall than the other. Because godwits did not roost in areas affected by blind zones, all of the godwits detected in drone images were also deemed visible from the seawall. In contrast, 90.4% of the curlews detected in drone images were not visible from the seawall because they tended to roost close to the foreland. Dunlins Calidris alpina are a smaller species with individuals detected by drone both within and outside blind zones and 78.5% deemed not visible from the seawall. These numbers make it clear that a considerable number of birds go undetected during traditional ground-based counts—indeed some 51 to 61% of all roosting birds in this study.
After more than a decade of civilian use, the idea of drones watching like “unblinking eyes in the sky”4 remains unnerving to some. Indeed, the surveillance of human behaviour has increased with technological advances, and not only from above but also through our devices and data.5 The ethics of technology use are complicated, but when it comes to drones, Castenschiold and colleagues’ work points to their usefulness as relatively safe and low-cost tools for aerial wildlife observation.
1 Castenschiold, J. H. F., D. Bruhn, C. Pertoldi and T. Bregnballe. 2023. Monitoring roosting waterbirds: The use of drones to overcome the challenge of hidden individuals in blind zones on intertidal flats. Wader Study 130(3): 239–253.
2 Sasse, D.B. 2003. Job-related mortality of wildlife workers in the United States, 1937–2000. Wildlife Society Bulletin 31: 1015–1020.
3 UNESCO World Heritage Centre (2023) Wadden Sea. Accessed 29 Oct 2023 at: https://whc.unesco.org/en/list/1314/.
4 Unblinking eyes in the sky (2012) The Economist [US] 3 Mar 2012: 12. Accessed 31 Oct 2023 at: http://www.economist.com/node/21548485.
5 Laidler, J. (2019). High tech is watching you (interview with Shoshana Zuboff author of Surveillance Capitalism). Harvard Gazette. Accessed 21 Nov 2023 https://news.harvard.edu/gazette/story/2019/03/harvard-professor-says-surveillance-capitalism-is-undermining-democracy/
PDF of this article