Location: Invasive Plant Research Laboratory
Project Number: 6032-22000-013-097-S
Project Type: Non-Assistance Cooperative Agreement
Start Date: Sep 1, 2022
End Date: Oct 1, 2025
1. Establish cooperative agreement to support one student to develop a machine based learning approach using UAVs at low altitudes to detect waterhyacinth and its herbivores along with any other potential targets. 2. The student will work with USDA to design experiments and be part of the team to implement the experimental design. 3. The student will present any findings in written reports, refereed articles and scientific meetings.
Remote sensing of invasive weeds and biological control agents is an emerging technology in weed science. Acquiring, processing and accurately interpreting this data is a major area of research and a focus of Waterhyacinth Areawide Pest Management Project. Much time, effort and resources have been placed into training technicians and scientists on this project (and others in the unit) on flying UAS’s and acquiring images for analysis, but we do not have the time or expertise to refine analysis for the images obtained and test whether biological control agents, or herbivory more generally, can be detected remotely. This is particularly important for wildland weeds that invade difficult terrain such as aquatics, marshes, wetlands and extremely remote areas. We propose to work with the University of Florida at the Indian River Research and Education Center to provide support for a research assistantship for a graduate student to investigate and develop analytic tools to detect weeds and their herbivores remotely, including from images and spectral data provided by USDA UAS. This could have wide-reaching implications for other systems including large-scale insect invasions (e.g., EAB, spotted lantern fly, hemlock woolly adelgid, etc.), and invasive plants in other ecosystems (e.g., grasslands, rangelands) that have implications for fire and other ecological processes. Arthropod damage can trigger host plant signaling that may be exploited by novel remote sensing of invasive pests in agricultural and forest ecosystems. Detecting changes in host plant signaling may improve the ability to accurately locate new invasions and detect them at lower levels of infestation, increasing early detection and rapid response management options. These technologies can be used to assess the presence/efficacy of biocontrol agents and guide management decisions. We will develop novel, plant-signature based remote sensing technologies to facilitate early detection, monitoring, and impact assessment of invasive pests and their biocontrol agents by collecting hyperspectral, LIDAR, and thermal data with unmanned aerial systems and ground-based sensors across in the waterhyacinth system. We will use hyperspectral imagery to track changes in plant chemistry, LIDAR data to monitor changes in foliar biomass, and thermal data to track canopy water stress via reductions in canopy evapotranspiration. We will develop algorithms that accurately detect new populations and assess impact biocontrol agents.