by Carolyn Bernhardt
July 10, 2024
European common reed grass (Phragmites australis subsp. australis) may sway slowly amidst tranquil expanses of wetlands, but it is a problematic invader. Known for its rapid growth and dense stands that crowd out native vegetation, this species disrupts delicate wetland ecosystems crucial for wildlife and environmental balance.
When Joe Knight, PhD, head of the Department of Forest Resources and Environmental Conservation at Virginia Tech, was at the University of Minnesota, he and his team tackled this challenge in a MITPPC-funded project. The project ended in 2023 and leveraged remote sensing technologies to track and manage non-native Phragmites across Minnesota. By integrating advanced approaches, the team created 3D maps that helped them identify unfolding infestations with 90% accuracy.
European common reed grass typically grows in wetlands across temperate and tropical regions around the world. "Phragmites spreads rapidly and out-competes native vegetation. It grows in dense, tall stands that essentially choke out other vegetation," says Knight. This process reduces habitat quality for native fish, insects, plants, and birds, decreases biodiversity, and alters a wetland’s water chemistry.
"Phragmites disrupts normal wetland functions. Communities depend on healthy wetlands for wildlife habitat, water filtration, water storage, and carbon sequestration," Knight says. European common reed grass is classified as a restricted noxious weed in Minnesota and beyond.
Mapping a menace
Instead of relying on pixilated aerial imagery alone, the team developed a layered approach, combining Object-Based Image Analysis (OBIA), 3D structural measures from light detection and ranging (lidar) or stereo imagery, and machine learning.
"By combining high-resolution imagery with 3D structural measures derived from lidar and stereo drone data, we can identify and monitor Phragmites stands with much higher accuracy,” says Knight. The team could see how the vegetation was changing in appearance, such as in color, and how it was changing in height, density, and texture.
"When Phragmites infests an area, the existing vegetation is rapidly replaced with a dense, monotypic stand of Phragmites that can grow as quickly as meters per year. That change signal is visible with this combination of data," he says. "Aerial imagery alone gives us the ability to see a change in appearance, but it does not provide structural measures.”
According to Knight, OBIA provides much higher-quality results than pixel-based methods because it breaks down image pixels into groups, such as different types of land cover, and then labels them. OBIA makes it possible to study things that are hard to see with just pixels. OBIA can measure things like how big and what shape objects are, their texture, and how they relate to nearby objects such as how far a plant is from water.
Lidar sends out laser pulses and measures how long they take to bounce back from objects on the ground, like trees or buildings. It helps create very detailed 3D maps and can show things like the height of trees and the shape of the landscape.
Machine learning algorithms are advanced tools for analyzing remote-sensing images. The tool improves by analyzing data repeatedly, which helps find complex patterns that traditional methods may miss. This makes machine learning better at identifying different objects or features in images over time.
Upfront precision, streamlined management
"If cost were not a factor, field monitoring would be the best approach to managing Phragmites,” says Knight. “However, keeping people in the field is expensive, and Minnesota is a large state. The only cost-effective way to study the entire state is with remotely sensed data. Satellite and aerial imagery are routinely collected over Minnesota, allowing us to detect and monitor changes in vegetation such as Phragmites infestation. Field work will always be required to treat infestations and verify remote measurements, but remote sensing allows us to focus that field work where it is most needed."
Knight says the team was “very pleased to see such high accuracy of the results." In time, they plan to package these tools online so that others can use them without needing to develop expertise in lidar, OBIA, and machine learning.
Next, the team is working on scaling their methods to study non-native Phragmites throughout the state. “Our most important goals,” says Knight, “are to be able to monitor known Phragmites stands to determine how they’re spreading, evaluate how they respond to treatment, and detect previously unknown infestations.”
In the meantime, he says, "The most important things people can do are to avoid spreading Phragmites by cleaning their watercraft properly, and to be vigilant for and report Phragmites infestations by contacting their DNR Invasive Species Specialist."
The Minnesota Invasive Terrestrial Plants and Pests Center is supported by the Environment and Natural Resources Trust Fund as recommended by the Legislative-Citizen Commission for Minnesota Resources.
Stay connected by signing up for the MITPPC newsletter. Follow us on LinkedIn and Facebook.
You may also like...
- Non-native subspecies of Phragmites from the Minnesota Department of Natural Resources
- Invasive Phragmites resources and research from the Minnesota Aquatic Invasive Species Research Center
- Verified Phragmites populations map
- Detection and monitoring of invasive Phragmites, research project
- Assessing spatiotemporal dynamics of Phragmites invasion and management in Minnesota, research project