Using remote sensing to detect Eurasian Watermilfoil

Part of the University of Michigan-Dearborn's campus since its inception in 1959, the Environmental Study Area (ESA) is an important natural oasis in a dense urban setting. However, In such a confined natural area, small changes can have large impacts on the overall health of the ecosystem. In the past, invasive species have left their mark on the landscape, such as the emerald ash borer and garlic mustard, which have destroyed or crowded out native species.  In recent years, staff at the Environmental Interpretive Center have noticed concentrations of Eurasian Watermilfoil, an invasive aquatic plant, within Fair Lane Lake at the center of the study area. Because Fair Lane Lake is an important urban stopover for migratory and rare native birds, invasive species can impact local species connections. One of the foci of this study is to visualize, using an unmanned aerial vehicle, how much this invasive species has taken over the lake.

Henry and Clara Ford built their 31,000 sqft dream home, Fair Lane, on 1,300 acres of farmland just a couple of miles from where they were born. The grounds and gardens were designed by the famous landscape architect Jens Jensen who transformed them from farmland to a natural native landscape. In 1959, the Ford Motor Company donated the residence, powerhouse and around 200 acres to the University of Michigan Dearborn campus. Today, the Environmental Study Area (ESA) on campus features many natural habitats used by researchers and for community education.

Mr. Ford was at the forefront of early American bird conservation and he intentionally transformed his estate into habitats suitable for native birds. Henry Ford had hundreds of bird houses installed and his estate became informally known as the "Bird Farm." A wetland area, formerly known as “the swamp,” is located at the northwest corner of the ESA and is one of very few remaining floodplain wetlands situated along the Rouge River. Written accounts from around 1911 suggest that Henry Ford had the size of this wetland increased to about 30 acres through flood management. His goal was to create more “marshland” for nesting and migrating birds (Simek 2015).

Fair Lane Lake in the center of the ESA is a manmade lake designed by Jens Jensen to provide benefits to the surrounding wildlife, including native aquatic plants, waterfowl, and fish. Jensen’s plan called for planting water hyacinth, a sweet-scented water lily, spike rush, cat-tail, arrowhead, native grasses; and wild rice within the shallow areas and shoreline edges. This vegetation plan shows that the lake was also intended to provide habitat for birds, and the EIC staff has made efforts to maintain the marshy habitat at the north end of the lake (Simek 2015). Originally constructed with an area of 11 acres, the size of the lake was reduced during construction of the UM-Dearborn and Henry Ford College campuses. In 2014, a student researcher mapped the depth of the entire lake (Keeling 2014). He found that the shallow lake has a maximum depth of less than 2.5 m. This would make it an ideal habitat for Eurasian Watermilfoil, an invasive aquatic plant.

Fair Lane Lake has a mud bottom that both native and nonnative species thrive in. Eurasian Watermilfoil (EWM) is a submerged aquatic plant that forms a thick canopy which blocks light penetration into the water, allowing it to outcompete native plants. Its predominance reduces the diversity and abundance of native aquatic plant species in the lake. Eurasian Watermilfoil is also difficult to remove using traditional physical means because the plant can regrow from broken pieces of stem and leaflets.

EWM also has adverse effects on water quality and recreational activities. It can lower overall dissolved oxygen concentrations and increase the mosquito population since the plant creates a safe foliage habitat for mosquito larvae. Some studies have shown that native waterfowl tend to avoid this invasive species compared to native milfoils, and in natural connected lakes EWM can interrupt fish spawning.

Being man made and disconnected from other water resources, Fair Lane Lake offers a unique opportunity for environmental research into how invasive species can affect different areas of an ecosystem without risking the exposure of outside waterways.

Remote sensing allows for the monitoring of physical characteristics of an area from a distance by detecting emitted or reflected radiation. Satellites, aircraft, and unmanned aerial vehicles (UAV) are widely used. All three platforms use cameras and sensors to collect data. Each sensor detects different wavelengths (i.e., bands) within the visible light, ultraviolet, or infrared portions of the electromagnetic spectrum. This allows for the detection of characteristics beyond what can be seen with the human eye.

There are many advantages to collecting data using UAVs. While UAV images cover a smaller area than aircraft or satellite images, they still offer a larger perspective than single points from field surveys with less time and energy spent by researchers. Drones also have operational flexibility and are capable of accessing hard to reach areas, such as wetlands. They are also relatively low cost when compared to manned flights or extensive field surveys and allow for more frequent flyovers to detect changes. 

The drone models used in this project were the Phantom 3 and the DJI Mavic 2 Enterprise Advanced. Both UAV’s are developed by DJI Technology co. This allowed the team to see the advantages and disadvantages of both low budget and medium budget UAVs commonly used for research.

The individual images taken from the Phantom 3 were stitched into a single mosaic of the entire lake using Pix4D Mapper. This natural color mosaic clearly shows the level of aquatic vegetation in the lake, although from this sensor height, it is impossible to tell the dominant species of plant. Note that the drone aerial image is much more detailed than the background map, which is comprised of some of the highest resolution satellite imagery publicly available.

Remotely sensed images can be used to determine the health, abundance, and type of vegetation. For example, the health of vegetation can be calculated using the Normalized Difference Vegetation Index (NDVI), which measures the difference in reflectivity between the red portion of the visible light and the infrared band. The chlorophyll in healthy vegetation reflects more near-infrared (NIR) and green light compared to other wavelengths, but it absorbs more red and blue light. This is why our eyes see vegetation in green. The NDVI of Fair Lane Lake shows the thick vegetation of the tree canopy (in green) surrounding the lake, and areas of low vegetation (red/orange) around the north of the lake. Notice that there are lighter green and yellow areas within the center of the lake representing abundance of vegetation near the water’s surface. These green spaces are more concentrated near the center of the lake.

The drone used for these images did not have an available sensor which would have enabled deeper penetration into the water in order to better distinguish aquatic species at varying depths. However, with the visible images taken by the UAV the student team was able to determine the growth pattern of EWM that was confirmed during the field site survey. EWM can be seen along the surface of the water more frequently compared to Coontail, and, by using the drone aerial image, surface vegetation could be separated from subsurface vegetation.

There are other variations of NDVI that are useful when looking at plants in different stages of growth, like the Red-Edge NDVI (NDVIre) method, which focuses more on estimating vegetation health by using the red-edge band of light. This can help distinguish between plants with different chlorophyll levels, such as Coontail, which has a darker color than milfoil. In order to better view the differences in surface vegetation, this NDVIre (which focuses on the red areas of an image) was de-colorized in a close up cross section of Fair Lane lake.

Remotely sensed images can be used to identify types of surfaces based on their reflectivity. Pixels in close proximity to each other that have similar characteristics are grouped together using a “supervised classification.” These classes represent different types of surfaces, such as clear water, darker water (i.e., any water that cannot be seen through), algae, aquatic vegetation, forest, shadow, and ground. With these classifications the aquatic vegetation can be compared to algal growths within the lake, and areas of darker water can be separated. The light green areas of the lake are leafy aquatic vegetation, which is located throughout the center of the lake.

These classifications allow one to distinguish between algal growth and other aquatic vegetation as well as darker water. The light green areas in the image represent leafy aquatic vegetation, which is located throughout the center of the lake. Darker water is found closer to the shoreline.The north end of the lake displays a concentration of algae, common in marshland. Algae are important to a wetland ecosystem, and the north end of the lake was originally intended as home for marsh birds, many of which rely on algae for foraging. The aquatic vegetation that covers large areas along the southern and western edge of the lake was identified as Coontail. It secretes a chemical that inhibits the growth of blue green algae (GISD, 2021).

A classifed map also allows the reasercher to calculate the relative area of a surface. Within Fair Lane Lake, there is 43% more area classified as aquatic vegetation compared to darker water. The darker water areas, based on visual determination during the field site survey, tended to be deeper water vegetation and not surface vegetation. This is more likely the Coontail or other native vegetation because Eurasian Watermilfoil creates a dense canopy at the water’s surface. This image also shows the low amount of water within the lake that is clear of vegetation, algae, or debris.

This study has shown there is a problem with EWM in Fair Lane Lake. Thanks to the images and maps shown here, the EIC staff can zero in on the aquatic vegetation in the lake and develop management plans using different techniques. EWM can be difficult to remove because it can regrow from broken stem and leaf fragments, so many techniques need to be considered.

Remote sensing (RS) and geographic information system (GIS) technologies can support efforts to control invasive species by facilitating the mapping of actual invader distributions or areas at risk of invasion. The above results show that multispectral drone imagery can capture the extent of the aquatic vegetation within a body of water and distinguishing between this invasive species and many other aquatic vegetation species. According to Dr. Brooks, “Moving from analyzing spectral profiles to imaging EWM and other SAV [submerged aquatic vegetation] in the field is the next step in a process for using remote sensing as a practical tool in invasive SAV management” (Brooks, 2020).

This is just the first step in what should be regular monitoring of the lake, and more data is needed to understand the health of the lake over time. The next step for the University’s Environmental Interpretive Center would be determining the feasibility of using higher end equipment with better modules, like cameras capable of using a wider range of bands that could penetrate deeper water, compared to less expensive UAVs.

While it is uncertain that the EIC can acquire the higher end UAV used in this study, a less expensive UAV with interchangeable modules could still facilitate a wider range of project based learning opportunities for students to understand different applications of remote sensing that could be implemented within the natural study area. This could also increase campus collaboration with off-campus experts, who can help create important connections for students in their future careers, and hands-on training. This is why the use of UAVs by the Environmental Interpretive Center to facilitate research in the Environmental Study Area could be an important endeavor for the future of environmental education and stewardship on the University of Michigan-Dearborn campus.

Finally, over a decade ago physical removal was unsuccessful, however, newer techniques and better methods have been developed since then that have yet to be tried within the lake. The UAVs can determine long term effectivness of removal of this invasive species using various methods.