Thermal Imaging Rock Wallabies
The Problem
Until recently, very little was known of the Kimberley race of the Black-flanked Rock Wallaby (Petrogale lateralis West Kimberley) by western science except that it is locally distributed in the southwest Kimberley on Nyikina Mangala country (where it is known as Wiliji by Traditional Owners) and is genetically distinct from other races of the species. WWF-Aus staff started working with the Nyikina Mangala Rangers in 2012 and over the last 5 years have shown that the species is far rarer (possibly < 300 individuals) than is formally estimated in the Action Plan for Australian Mammals (which estimates the population at ~2500). We have found Wiliji to occur on three distinct ranges. Major threats to the species may include fire management around their core habitat (e.g., dens, foraging habitat), and predation by feral cats and dogs. Counting Rock Wallabies accurately is notoriously difficult, and requires capturing, tagging and re-capturing individuals over time. Sensor cameras can be useful for identifying whether rock wallabies occupy certain areas, detecting predators and even counting pouch-young. However, we have found that it is not possible to differentiate between individuals using sensor cameras. Currently, WWF staff are working with the Rangers and Kimberley Land Council fire ecologists on implementing “right way” fire strategies to improve habitat. We aim to develop technology and methods to assess populations of this species as conservation measures proceed.
Our Proposal
We aim to investigate whether it is possible to use aerial thermal imagery to count Wiliji and other species (predators, other macropod species) on sections of the Erskine Range, an accessible area that has relatively high numbers of Wiliji. Using unmanned aerial vehicles (UAVs), is an inexpensive way to accurately survey populations of elusive species which have large geographic ranges and low population densities. Recent advances in camera technology mean that UAVs now have the capability of providing high resolution imagery across a broad spectrum of wavelengths, including infrared radiation (IR) imagery. IR cameras sense infrared radiation and can discern fine-scale variation in temperature. As a result, acquiring high resolution imagery of mammals may be possible, especially those with nocturnal or crepuscular behaviour patterns where standard cameras may be of limited utility. Wiliji habitat is a remote, topographically complex, and harsh environment, so developing and using technology that can withstand the challenges presented by this environment is of utmost importance. We propose using a robust hexacopter drone fitted with dual cameras, a high resolution radiometric infrared camera and a high resolution RGB camera. The challenge of this project will be marrying thermal imagery with object classification software routines in order to accurately identify Wiliji from other mammal species. We are currently developing this capacity.
We Assume that...
We are assuming that we will be able to acquire imagery of the appropriate resolution to identify Wiliji from other mammal species.
Constraints to Overcome
Obtaining high quality imagery from a topographically complex environment and then processing that imagery into suitable 3-D models is possible. The challenge to overcome is to overlay lower resolution thermal imagery on to that 3-D model and to accurately assess Wiliji (and other mammal) numbers. Prior thermal imaging assessments have worked on more uniform terrain, and often on known target species. Our on-going sensor camera work will help us by providing a context of relative abundance of various mammal species across the landscape.
Current Work
Our goal is to develop the capacity to assess the abundance and distribution of threatened mammal species in remote, topographically complex environments. Specifically, we aim to use the information gathered in order to identify threats to Wiliji, develop management strategies, and conserve this species.
Current Needs
We need to finalize purchase of an appropriate thermal camera gimbal to integrate onto a hexacopter drone that we already own. From there, we need to collect imagery and data to develop our object classification software routines in order to determine if the resolution of the thermal imagery is capable of identifying Wiliji from other mammals. Test flights will need to be conducted at various angles and altitudes to determine the correct ground sampling distance for acquiring suitable imagery. Additionally, we need funding for travel to the remote location of the Erskine Ranges.