Rebecca M. Rogers
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Rebecca M. Rogers has a total of 20 co-authors. Here are the top co-authors by number of shared publications:
- Hazel Parry
CSIRO
2 shared publications - Joel Janek Dabrowski
Data61 CSIRO Australia
2 shared publications - Mubin Ul Haque
CSIRO
2 shared publications Aliesha Hvala
1 shared publication- Amélie Corriveau
Research Institute for Environment and Livelihoods Charles Darwin University Casuarina Northern Territory Australia
1 shared publication David A. Loewensteiner
Research Institute for the Environment and Livelihoods, Charles Darwin University, Darwin, Australia
1 shared publication
Latest Publications
- 24 June 2025
- Published by Springer Nature
- p. 63-75
Detection of Animal Movement from Weather Radar Using Self-supervised Learning
- 24 June 2025
- Published by Springer Nature
- p. 63-75
Detection of Animal Movement from Weather Radar Using Self-supervised Learning
- Open Access
Detection of Animal Movement from Weather Radar using Self-Supervised Learning
Detecting flying animals (e.g., birds, bats, and insects) using weather radar helps gain insights into animal movement and migration patterns, aids in management efforts (such as biosecurity) and enhances our understanding of the ecosystem.The conventional approach to detecting animals in weather radar involves thresholding: defining and applying thresholds for the radar variables, based on expert opinion. More recently, Deep Learning approaches have been shown to provide improved performance in detection. However, obtaining sufficient labelled weather radar data for flying animals to build learning-based models is time-consuming and labor-intensive. To address the challenge of data labelling, we propose a self-supervised learning method for detecting animal movement. In our proposed method, we pre-train our model on a large dataset with noisy labels produced by a threshold approach. The key advantage is that the pre-trained dataset size is limited only by the number of radar images available. We then fine-tune the model on a small human-labelled dataset. Our experiments on Australian weather radar data for waterbird segmentation show that the proposed method outperforms the current state-of-the art approach by 43.53% in the dice co-efficient statistic.
Open AccessDetection of Animal Movement from Weather Radar using Self-Supervised Learning
Detecting flying animals (e.g., birds, bats, and insects) using weather radar helps gain insights into animal movement and migration patterns, aids in management efforts (such as biosecurity) and enhances our understanding of the ecosystem.The conventional approach to detecting animals in weather radar involves thresholding: defining and applying thresholds for the radar variables, based on expert opinion. More recently, Deep Learning approaches have been shown to provide improved performance in detection. However, obtaining sufficient labelled weather radar data for flying animals to build learning-based models is time-consuming and labor-intensive. To address the challenge of data labelling, we propose a self-supervised learning method for detecting animal movement. In our proposed method, we pre-train our model on a large dataset with noisy labels produced by a threshold approach. The key advantage is that the pre-trained dataset size is limited only by the number of radar images available. We then fine-tune the model on a small human-labelled dataset. Our experiments on Australian weather radar data for waterbird segmentation show that the proposed method outperforms the current state-of-the art approach by 43.53% in the dice co-efficient statistic.
- Open Access
- 25 August 2023
- Published by Frontiers Media SA in Frontiers in Conservation Science
- Vol. 4, 1203736
Supplementing aerial drone surveys with biotelemetry data validates wildlife detection probabilities
Aerial drone systems are now widely used to survey wildlife, but validation in the detectability of individuals is rarely assessed. This knowledge gap is critical, given the influence of local environments on wildlife detectability from the air. In this study, we integrated Animal Biotelemetry technology with aerial drones to evaluate the temporal and environmental factors influencing animal detection probability and subsequent population estimates. Wild-caught feral pigs (Sus scrofa) were fitted with GPS tracking collars and releasing them into a large natural habitat enclosure in northern Australia. Utilizing a fixed-wing drone equipped with a dual camera (thermal infrared and RGB), we conducted multiple flights over the study area during both wet and dry seasons, from sunrise to sunset. The study found that the probability that a pig was visible in aerial imagery was highly variable depending on the timing of the aerial survey. Detection probability was at its lowest during mid-afternoon (5 to 20%), while the early evening yielded the highest detection probability (50 to 75%). We observed seasonal differences, with detection probabilities exceeding 50% in the mornings of the wet season, in contrast to less than 30% during the dry season. Temporal trends in detection probability were similar in both thermal infrared and RGB imagery. The GPS location data enabled us to assess how localized factors (canopy cover, land cover, ambient temperature) altered animal detection probability. This information facilitated the identification of survey times to maximize feral pig detection and the development of a correction factor to account for non-detected individuals at specific times and locations. The study demonstrates the value of integrating Animal Biotelemetry technology and aerial drones to account for variations in detection probability when undertaking wildlife aerial surveys. Insights gained from this approach have implications for enhancing the accuracy of population assessments and contributing to more effective wildlife management and conservation.
- 25 August 2023
- Published by Frontiers Media SA in Frontiers in Conservation Science
- Vol. 4, 1203736
Open AccessSupplementing aerial drone surveys with biotelemetry data validates wildlife detection probabilities
Aerial drone systems are now widely used to survey wildlife, but validation in the detectability of individuals is rarely assessed. This knowledge gap is critical, given the influence of local environments on wildlife detectability from the air. In this study, we integrated Animal Biotelemetry technology with aerial drones to evaluate the temporal and environmental factors influencing animal detection probability and subsequent population estimates. Wild-caught feral pigs (Sus scrofa) were fitted with GPS tracking collars and releasing them into a large natural habitat enclosure in northern Australia. Utilizing a fixed-wing drone equipped with a dual camera (thermal infrared and RGB), we conducted multiple flights over the study area during both wet and dry seasons, from sunrise to sunset. The study found that the probability that a pig was visible in aerial imagery was highly variable depending on the timing of the aerial survey. Detection probability was at its lowest during mid-afternoon (5 to 20%), while the early evening yielded the highest detection probability (50 to 75%). We observed seasonal differences, with detection probabilities exceeding 50% in the mornings of the wet season, in contrast to less than 30% during the dry season. Temporal trends in detection probability were similar in both thermal infrared and RGB imagery. The GPS location data enabled us to assess how localized factors (canopy cover, land cover, ambient temperature) altered animal detection probability. This information facilitated the identification of survey times to maximize feral pig detection and the development of a correction factor to account for non-detected individuals at specific times and locations. The study demonstrates the value of integrating Animal Biotelemetry technology and aerial drones to account for variations in detection probability when undertaking wildlife aerial surveys. Insights gained from this approach have implications for enhancing the accuracy of population assessments and contributing to more effective wildlife management and conservation.
- Open Access
- 10 July 2022
- Published by Wiley in Ecological Solutions and Evidence
- Vol. 3 (3)
Repurposing open‐source data from weather radars to reduce the costs of aerial waterbird surveys
Aerial counts are the primary means of monitoring waterbird populations. A valid population assessment requires a significant proportion of the population to be surveyed. For broad‐ranging species, this requires costly reconnaissance flights and surveys over large areas of potential habitat. Here, we assess whether free, autonomously collected weather radar reflectivity data can identify waterbird aggregations over broad areas of potential habitat, thereby providing a low‐cost means of directing the timing and location of aerial survey counts. Our study was undertaken over a large flood plain known to be a key seasonal nesting area of the Magpie Goose (Anseranas semipalmata). A standardized aerial survey technique was employed, firstly using reconnaissance flights to determine when the birds were nesting and then 400‐m wide transects 3 km apart over approximately 1100 km, visually counting the geese. Radar reflectivity data from an operational weather radar in Darwin, Australia, 2 weeks prior to and during the aerial survey, were processed to link averaged radar reflectivity to terrestrial habitat at a spatial polar resolution of 250 m × 0.5°. A binary classification approach evaluated the accuracy, precision, sensitivity and specificity of reflectivity data in identifying areas where the aerial surveys recorded geese. Aerial surveys of Magpie Geese are most effective when the geese are nesting, which is usually identified by reconnaissance flights. Here, we show nesting correlated with a decline in radar reflectivity and a concentration of hotspots. By achieving reasonable accuracy and precision (77% and 67%, respectively), the weather radar reflectivity identified high aggregations of Magpie Geese and demonstrated high efficacy when identifying areas where Magpie Geese did not occur (specificity, 94%). We conclude that using weather radar data to guide aerial survey timing would reduce reconnaissance flight needs, and enable aerial surveys to focus on areas with a large proportion of the target waterbird population.
- 10 July 2022
- Published by Wiley in Ecological Solutions and Evidence
- Vol. 3 (3)
Open AccessRepurposing open‐source data from weather radars to reduce the costs of aerial waterbird surveys
Aerial counts are the primary means of monitoring waterbird populations. A valid population assessment requires a significant proportion of the population to be surveyed. For broad‐ranging species, this requires costly reconnaissance flights and surveys over large areas of potential habitat. Here, we assess whether free, autonomously collected weather radar reflectivity data can identify waterbird aggregations over broad areas of potential habitat, thereby providing a low‐cost means of directing the timing and location of aerial survey counts. Our study was undertaken over a large flood plain known to be a key seasonal nesting area of the Magpie Goose (Anseranas semipalmata). A standardized aerial survey technique was employed, firstly using reconnaissance flights to determine when the birds were nesting and then 400‐m wide transects 3 km apart over approximately 1100 km, visually counting the geese. Radar reflectivity data from an operational weather radar in Darwin, Australia, 2 weeks prior to and during the aerial survey, were processed to link averaged radar reflectivity to terrestrial habitat at a spatial polar resolution of 250 m × 0.5°. A binary classification approach evaluated the accuracy, precision, sensitivity and specificity of reflectivity data in identifying areas where the aerial surveys recorded geese. Aerial surveys of Magpie Geese are most effective when the geese are nesting, which is usually identified by reconnaissance flights. Here, we show nesting correlated with a decline in radar reflectivity and a concentration of hotspots. By achieving reasonable accuracy and precision (77% and 67%, respectively), the weather radar reflectivity identified high aggregations of Magpie Geese and demonstrated high efficacy when identifying areas where Magpie Geese did not occur (specificity, 94%). We conclude that using weather radar data to guide aerial survey timing would reduce reconnaissance flight needs, and enable aerial surveys to focus on areas with a large proportion of the target waterbird population.
- 1 October 2020
- Published by Taylor & Francis in Emu - Austral Ornithology
- Vol. 120 (4) , 343-354
Broad-scale opportunistic movements in the tropical waterbird Anseranas semipalmata: implications for human-wildlife conflicts
Waterbirds commonly undertake extensive movements in search of resources. While much is known about waterbird movements within temperate regions where resource distribution is seasonally predictable, less is understood about waterbird movements in tropical and subtropical environments where the temporal distribution of resource is less predictable. This knowledge is critical to understanding a species’ response to environmental and anthropogenic changes to natural habitats. Here, we investigated the movement behaviour of a tropical waterbird of northern Australia, the Magpie Goose (Anseranas semipalmata), seasonally utilising a recently developed mixed natural-anthropogenic landscape. Using satellite telemetry, we assessed the daily movements and space use of 18 individuals (4673 goose-days) to determine whether these individuals spent the year locally, migrated seasonally from further areas, or were transitory individuals from across northern Australia. Geese released from peri-urban and rural areas in the late dry season dispersed to both local (≤ 50 km) and distant (≤ 566 km) floodplain systems over the wet season. Tracked individuals returned to sites within the same floodplain systems in the wet season and within the same anthropogenic areas in the late dry season over consecutive years. Our results suggest that anthropogenic environments may provide predictable resources for Magpie Geese during a seasonal bottleneck of natural resources. We highlight the need to learn more about waterbird movement behaviour within tropical human-dominated landscapes to evaluate the likelihood that predictable resources could exacerbate human–wildlife conflicts, and the possible long-term effects on species population dynamics, behaviour and distribution.
- 1 October 2020
- Published by Taylor & Francis in Emu - Austral Ornithology
- Vol. 120 (4) , 343-354
Broad-scale opportunistic movements in the tropical waterbird Anseranas semipalmata: implications for human-wildlife conflicts
Waterbirds commonly undertake extensive movements in search of resources. While much is known about waterbird movements within temperate regions where resource distribution is seasonally predictable, less is understood about waterbird movements in tropical and subtropical environments where the temporal distribution of resource is less predictable. This knowledge is critical to understanding a species’ response to environmental and anthropogenic changes to natural habitats. Here, we investigated the movement behaviour of a tropical waterbird of northern Australia, the Magpie Goose (Anseranas semipalmata), seasonally utilising a recently developed mixed natural-anthropogenic landscape. Using satellite telemetry, we assessed the daily movements and space use of 18 individuals (4673 goose-days) to determine whether these individuals spent the year locally, migrated seasonally from further areas, or were transitory individuals from across northern Australia. Geese released from peri-urban and rural areas in the late dry season dispersed to both local (≤ 50 km) and distant (≤ 566 km) floodplain systems over the wet season. Tracked individuals returned to sites within the same floodplain systems in the wet season and within the same anthropogenic areas in the late dry season over consecutive years. Our results suggest that anthropogenic environments may provide predictable resources for Magpie Geese during a seasonal bottleneck of natural resources. We highlight the need to learn more about waterbird movement behaviour within tropical human-dominated landscapes to evaluate the likelihood that predictable resources could exacerbate human–wildlife conflicts, and the possible long-term effects on species population dynamics, behaviour and distribution.
- 16 November 2017
- Published by CSIRO Publishing in Wildlife Research
- Vol. 44 (7) , 558-564
An inexpensive satellite-download GPS receiver for wildlife: field trial on black swans
Context: Knowledge of the movements and space-use patterns of wildlife is crucial for implementation of effective conservation and management actions. Such information can be difficult to obtain, especially from hard-to-capture or far-ranging taxa. Global-positioning-system (GPS) receivers that have remote data-acquisition capability via satellite-communication networks offer an effective means of tracking highly mobile animals; however, often the equipment and operational costs of these devices are prohibitive. Aim: To evaluate the operational performance and potential of an inexpensive, commercially available satellite-download GPS receiver that can be easily modified for use on wildlife, and to demonstrate its utility for revealing novel information on individual animal movements in an important wildlife management context. Methods: We added external batteries and supplementary waterproofing to ‘off-the-shelf’ SPOT Trace® GPS receivers (~$120 USD; Spot LLC, Covington, LA) that transmit recorded location data to users through the Globalstar satellite network. We then deployed these devices on eight wild black swans (Cygnus atratus) inhabiting a large inland marine harbor adjacent to Auckland International Airport and assessed the GPS units’ operational performance and overall cost effectiveness. Key results: During field trials with the tracking devices lasting up to 25 days we were able to collect thousands of precision location observations from the swans without the need for recapture. Cost-per-fix was only ~$0.04 USD – up to an order of magnitude lower than some satellite-download GPS receivers manufactured specifically for wildlife. We provide novel information about the swans’ patterns of activity, movement, space-use, habitat preferences and responses to actions taken by airport staff to scare problem individuals from threat zones. Conclusions: User-modified satellite-download GPS devices can be a highly cost-effective means of tracking the movements of animals. Implications: Knowledge of where and when swans present the greatest likely threat to air traffic and how they respond to displacement from restricted areas near the airport will be useful in devising threat-management plans. The ready availability of affordable remote-download GPS receivers will provide many new opportunities to study the movement and/or space-use patterns of wildlife and incorporate such information into a broad range of research and management programs.
- 16 November 2017
- Published by CSIRO Publishing in Wildlife Research
- Vol. 44 (7) , 558-564
An inexpensive satellite-download GPS receiver for wildlife: field trial on black swans
Context: Knowledge of the movements and space-use patterns of wildlife is crucial for implementation of effective conservation and management actions. Such information can be difficult to obtain, especially from hard-to-capture or far-ranging taxa. Global-positioning-system (GPS) receivers that have remote data-acquisition capability via satellite-communication networks offer an effective means of tracking highly mobile animals; however, often the equipment and operational costs of these devices are prohibitive. Aim: To evaluate the operational performance and potential of an inexpensive, commercially available satellite-download GPS receiver that can be easily modified for use on wildlife, and to demonstrate its utility for revealing novel information on individual animal movements in an important wildlife management context. Methods: We added external batteries and supplementary waterproofing to ‘off-the-shelf’ SPOT Trace® GPS receivers (~$120 USD; Spot LLC, Covington, LA) that transmit recorded location data to users through the Globalstar satellite network. We then deployed these devices on eight wild black swans (Cygnus atratus) inhabiting a large inland marine harbor adjacent to Auckland International Airport and assessed the GPS units’ operational performance and overall cost effectiveness. Key results: During field trials with the tracking devices lasting up to 25 days we were able to collect thousands of precision location observations from the swans without the need for recapture. Cost-per-fix was only ~$0.04 USD – up to an order of magnitude lower than some satellite-download GPS receivers manufactured specifically for wildlife. We provide novel information about the swans’ patterns of activity, movement, space-use, habitat preferences and responses to actions taken by airport staff to scare problem individuals from threat zones. Conclusions: User-modified satellite-download GPS devices can be a highly cost-effective means of tracking the movements of animals. Implications: Knowledge of where and when swans present the greatest likely threat to air traffic and how they respond to displacement from restricted areas near the airport will be useful in devising threat-management plans. The ready availability of affordable remote-download GPS receivers will provide many new opportunities to study the movement and/or space-use patterns of wildlife and incorporate such information into a broad range of research and management programs.