Date of Award

Summer 2019

Project Type


Program or Major

Natural Resources

Degree Name

Master of Science

First Advisor

Jenica M Allen

Second Advisor

Heather M Grybas


Invasive (alien) plants are introduced species that can cause harm to native ecosystems, industries, or human health. Managing invasive species requires knowing where they are, and early detection of new populations increases the likelihood of local eradication. Unmanned aerial systems (UAS) are an emerging remote sensing technology that can capture very high spatial resolution imagery, are easily deployed, and may offer a more efficient alternative to extensive ground surveys to locate invasive plants. Imagery collected with UAS has been used to map invasive plants in open canopy habitats, but has yet to be tested for mapping invasive plants in forest understories. My aim was to explore the feasibility of UAS as an understory invasion monitoring tool, including tests of season, sensor type, and image classification method for reliable invasive detection. I collected imagery from a 21-hectare mixed and deciduous New Hampshire forest during spring and fall periods of phenology mismatch between native vegetation and two focal invasive plants, Berberis thunbergii (Japanese barberry) and Rosa multiflora (multiflora rose). I achieved up to 82% classification accuracy by grouping B. thunbergii and R. multiflora as an Invasive class. There were no significant differences in invasive detectability between sensors or classification methods, but spring imagery yielded the highest accuracies overall. Simpler pixel-based classifications are sufficient for achieving over 70% classification accuracy, though object-based segmentation can improve accuracy. UAS are promising technology with potential to reduce and target invasive plant ground surveys for temperate forest management.