Kathryn Elmer

McGill University
M.Sc. candidate

Supervisor: Margaret Kalacska
Start: 2018-09-03


Utilization of Airborne Hyperspectral Imagery to Identify and Map Phragmites australis subsp. australis within Îles-de-Boucherville National Park
Other than habitat loss, invasive species currently pose one of the greatest threats to global biodiversity. Therefore, the early detection and identification of species of concern is essential for the creation of prevention and management strategies. By utilizing hyperspectral imagery (HSI) from an airborne platform, a methodology was established to identify and map the prolific invasive reed Phragmites australis subsp. australis throughout Îles-de-Boucherville National Park (Montreal, Quebec), which is one of the five studies sites for the Canadian Airborne Biodiversity Observatory (CABO). Within the context of CABO, no research has previously been conducted to map Phragmites across such a large scale within this study site. Using a Twin Otter aircraft as the sensor platform, five flight lines of hyperspectral imagery (HSI) were acquired at a spatial resolution of 2 m over the entire extent of the park in July 2019. Additionally, a high-accuracy Global Navigation Satellite System (GNSS) ground truth dataset containing positional data of Phragmites and other vegetation within the park was collected. This dataset consists of 319 ground truth points to serve as training and validation data for the HSI and target detection results. Using the Spectral Angle Mapper (SAM) target detection algorithm, a map of detected Phragmites was generated for the entire park area. A total of 2,507 individual Phragmites stands of various sizes were detected, covering an area of 27.42 hectares (0.274 km2), which represents approximately 3.37% of the total park area (814 hectares, 8.14 km2). When accounting for areas of uncertainty due to the inherent spatial error of the HSI, the total area of detected Phragmites was found to be 60.25 hectares (0.602 km2), or approximately 7.41% of the entire park area. The overall accuracy of the map was 84.27%, with a sensitivity of 76.31% and a specificity of 91.56%. In order to compare the accuracy of the SAM target detection method to traditional methods of human visual interpretation, 10 interpreters assessed the same set of 159 validation ground truth points used to determine the accuracy of the Phragmites map. Interpreters viewed high-resolution (60 cm spatial accuracy) satellite images of each validation point, and selected the images that contained Phragmites within a 15-meter buffer around the point. The overall accuracy of the visual interpretation was 66.47%, with a sensitivity of 59.21% and a specificity of 73.13%. These results indicate that human interpretation of the validation points does not yield as accurate results as the algorithm-based target detection methodology, which could have serious implications when planning and collecting data for similar studies regarding invasive species identification and management. The results provided by the HSI target detection methodology could well be worth the investment needed to procure and analyze the data, especially when compared to human interpretation of more readily available data. Overall, the map of detected Phragmites represents one of the first efforts to utilize airborne HSI and algorithm-based target detection methods to detect and map Phragmites over a moderately large scale within Îles-de-Boucherville National Park. The Phragmites map will provide a useful tool for the park’s land managers, as the identification of various sizes of Phragmites is critical for management and eradication of the invasive reed. By establishing the uses and limitations of airborne HSI for the detection of Phragmites, this study seeks to define a methodology that could be adapted in future remote sensing studies of various other invasive vegetation species across a wide range of ecosystems.


remote sensing, invasive species, common reed, ecological management, target detection, CASI-1500, GNSS


1- ASDToolkit: A Novel MATLAB Processing Toolbox for ASD Field Spectroscopy Data
Elmer, Kathryn, Raymond J. Soffer, J. Pablo Arroyo-Mora, Margaret Kalacska
2020 Data

2- A High-Accuracy GNSS Dataset of Ground Truth Points Collected within Îles-de-Boucherville National Park, Quebec, Canada
Elmer, Kathryn, Margaret Kalacska
2021 Data

3- Implementation of a UAV–Hyperspectral Pushbroom Imager for Ecological Monitoring
Arroyo-Mora, J., Margaret Kalacska, Deep Inamdar, Raymond Soffer, Oliver Lucanus, Janine Gorman, Tomas Naprstek, Erica Schaaf, Gabriela Ifimov, Kathryn Elmer, George Leblanc
2019 Drones

4- Accuracy of 3D Landscape Reconstruction without Ground Control Points Using Different UAS Platforms
Kalacska, Margaret, Oliver Lucanus, J. Arroyo-Mora, Étienne Laliberté, Kathryn Elmer, George Leblanc, Andrew Groves
2020 Drones