Murdoch University Research Repository

Welcome to the Murdoch University Research Repository

The Murdoch University Research Repository is an open access digital collection of research
created by Murdoch University staff, researchers and postgraduate students.

Learn more

Detectability of dolphins and turtles from Unoccupied Aerial Vehicle (UAV) survey imagery

Lloyd, Brooke (2021) Detectability of dolphins and turtles from Unoccupied Aerial Vehicle (UAV) survey imagery. Honours thesis, Murdoch University.

[img]
Preview
PDF - Whole Thesis
Download (1MB) | Preview

Abstract

For many decades occupied aircraft with trained observers have conducted aerial surveys of marine megafauna to estimate population size and dynamics. Recent technological advances mean that unoccupied aerial vehicles (UAVs) now provide a potential alternative to occupied surveys, eliminating some of the disadvantages of occupied surveys such as risk to human life, weather constraints and cost. In this study, data collected from an occupied aircraft (at 500 ft) and a UAV (at 1400 ft) flown at the same time, deployed for counting dugongs, were compared for detecting dolphins and turtles within Shark Bay, Western Australia. The UAV images were manually reviewed post hoc to count the animals sighted and the environmental conditions (visibility, sea state, cloud cover and glare) had been classified by the occupied teams’ data for each image. The UAV captured more sightings (174 dolphins and 368 turtles) than were recorded by the flight team (93 dolphins and 312 turtles). Larger aggregations (>10 animals) were also found in the UAV images (5 aggregations of dolphins and turtles) compared to the occupied teams sightings (0 dolphins and 3 aggregations of turtles). A generalised linear mixed model determined that turtle detection was significantly affected by visibility, while cloud cover, sea state and visibility significantly affected dolphin detection in both platforms. An expert survey of 120 images was also conducted to determine the image ground sampling distance (GSD; four levels from 1.7 to 3.5 cm/pixel) needed to identify dolphin and turtles to species. At 3 cm/pixel only 40% of the dolphins and turtles were identified to species with a reasonable level of certainty (>75% certainty). This study demonstrated that UAVs can be successfully deployed for detecting dolphins and turtles and that a GSD of 1.7 – 3cm/pixel is too low resolution to effectively identify dolphin and turtle species. Overcoming the limitations imposed on UAVs such as aviator regulatory bodies and payload capabilities will make UAVs a pivotal tool for future research, conservation, and management.

Item Type: Thesis (Honours)
Murdoch Affiliation(s): Environmental and Conservation Sciences
Supervisor(s): Hodgson, Amanda and Loneragan, Neil
URI: http://researchrepository.murdoch.edu.au/id/eprint/63695
Item Control Page Item Control Page

Downloads

Downloads per month over past year