The goal of this project is to advance the detection and mapping of aquatic invasive species using UAV-based hyperspectral remote sensing combined with machine learning analysis. The study focuses on freshwater systems in central New York, including Skaneateles Lake and Canandaigua Lake, where invasive and noninvasive aquatic vegetation coexist.
The project has two primary objectives. The first objective is data collection: to acquire UAV-based hyperspectral imagery over selected areas of Skaneateles and Canandaigua Lakes where invasive Eurasian watermilfoil and noninvasive curly-leaf pondweed are present. The second objective is data analysis: to apply machine learning techniques to hyperspectral imagery in conjunction with field-based observations in order to characterize species-specific spectral signatures and produce detailed maps of aquatic vegetation distribution.
Several key milestones have been achieved. The first major milestone was the acquisition and integration of a UAV-compatible hyperspectral imaging system. A Nano hyperspectral sensor was purchased in October 2024. Because the sensor was not originally designed for the project’s UAV platform, custom modifications to the UAV mounting system were required to ensure compatibility, stability, and safe flight operation.
In May 2025, project personnel completed a comprehensive training program at the sensor manufacturer’s headquarters in Boston, Massachusetts. The training covered sensor operation, calibration, and data acquisition workflows. Following this training, initial sensor testing was conducted from May through June 2025 under a variety of water and environmental conditions to assess system performance and data quality.
Field data collection was carried out in late summer and early fall 2025, corresponding to peak aquatic vegetation growth. Test flights using the UAV-mounted hyperspectral sensor were conducted over the Erie Canal in the Camillus area to validate data collection protocols. Additional UAV hyperspectral flights were conducted over the northern portion of Skaneateles Lake in coordination with the Skaneateles Lake Association.
From September 8–19, 2025, the project team also participated in the ROCX 2025 remote sensing data collection campaign in Rochester, New York. During this campaign, multiple days of UAV hyperspectral data were collected over the Tait Preserve, the ROCX project site, in coordination with nearly 50 participating organizations deploying a wide range of complementary sensors. In addition, airborne HySpex hyperspectral surveys were conducted over both Canandaigua Lake and Skaneateles Lake, providing full-lake coverage concurrent with in situ water sample collection by boat.
Processing and analysis of the collected hyperspectral datasets, including machine learning-based species classification and mapping, are planned for 2026.