Detecting whales from drones

New study highlights how environmental conditions and technological performance may affect detection

Norut's fixed wing UAV in Rystraumen. Photo: Ronald Smit

New study published in the journal Ecosphere: Whale detection certainty in images collected using unmanned aerial vehicles.

Ana Sofia Aniceto

Unmanned aerial vehicles, or drones, can be a useful tool to detect whales and other marine mammals in remote marine areas. The use of this technology has been heavily debated in the last decades, with many scientists defending that drones have the potential to revolutionize spatial ecology and replace current survey methods. Using drones to detect marine mammals has potential to improve data collection, by reducing costs and safety risks associated with ship or manned aircraft surveys.

A team of scientists tested how this equipment can perform in Arctic conditions by reviewing and assessing the images acquired by the UAVs for environmental conditions, changes in resolution, animal presence, and presence/absence certainty. They found that the presence of waves (water turbulence) and image brightness were the key factors affecting the certainty of whale detections.

The findings are based on 12 fixed-wing UAV flights focusing on humpback, killer whales, and harbor porpoises, during summer and polar night in Northern Norway. The drones followed pre-designed tracks, except during take-off and landing, and took consecutive photos throughout the entire duration of the flights. A total of 288 sightings were found, of which 50 where humpback whales, 63 were killer whales, 57 where harbor porpoises, and 118 were unidentified species. The research team found that darker images (low levels of brightness) have a positive effect on observer’s certainty of detection, whether the whales were breaking the water surface or were present just below the water. Sea conditions were found to have a negative effect on observer certainty, which shows that drones can be affected by environmental conditions in a similar way as traditional aerial surveys.

“While UAVs (Unmanned Aerial Vehicles) may eliminate observer bias in the data collection phase, the behavior of the animals and the probability of detection when reexamining the images still pose some challenges,” says lead author of the study, Ana Sofia Aniceto, from Akvaplan-niva. “Particularly at high latitudes, whether during summer or winter, environmental conditions can be limiting and should be incorporated into statistical models as well as UAV aircraft features that could change pixel resolution, and therefore affect detection probability”.

Aircraft stability can affect pixel size in aerial images, which in turn can affect the scientists’ ability to detect and correctly identify a species. The conditions used for the surveys in Norway were not extreme to the point of having a significant effect on certainty of detections. Still, such sources of bias in surveys should not be neglected, as they can affect estimates of whale abundance used for conservation, management, and mitigation.

Drones are generally more efficient than other types of surveys, though their effectiveness remains under debate. The study emphasizes the potential effectively collecting marine mammal data using unmanned aircraft, though more research is still necessary for the development of appropriate methodologies that can further optimize the detection of marine mammals.

Norut’s fixed wing UAV in Rystraumen. Photo: Ronald Smit