Monitoring invasive, aquatic weeds with multispectral imagery

Yellow floating heart (Nymphoides peltata) is an invasive aquatic weed from southeast Asia that was introduced to the US for use in ornamental ponds. Since its introduction, the Yellow floating heart (YFH) has become an issue affecting many lakes, as the aquatic weed spreads quickly and outcompetes many native species. YFH creates a dense mat of leaves that block light to native species, it also reduces oxygen levels in the water, creates a prime breeding ground for mosquitoes, and impedes water activities at lakes (i.e., boating, fishing, swimming).

The effects of this weed have led states like Michigan, Minnesota, and Wisconsin to ban the species.

In Oklahoma, Lake Carl Blackwell is the only lake known to have a YFH issue. Lake Carl Blackwell is a mid-sized reservoir maintained by Oklahoma State University (OSU) and used for campus drinking water. Researchers at OSU know that the longer YFH is present at Lake Carl Blackwell, the more chance it will spread to other surrounding water bodies. Therefore, efforts were taken quickly to address this issue.

Since the weed is very difficult to eradicate, an herbicide application was the only option. However, herbicide applications need to be selective and sparing as to not harm the water quality of the lake or the health of the native, aquatic plants. A selective, systemic herbicide was chosen which has been approved by the EPA after 8 years of testing. It is rated safe for drinking water and has been used in other states to control YFH.

In order to monitor the effectiveness of this herbicide application, researchers at OSU wanted to use remote sensing methods; a combination of multispectral, drone-based imagery and satellite imagery from Sentinel-2. For the drone-based imagery, they chose a DJI drone paired with a MicaSense series RedEdge-MX sensor. One of the main reasons they chose RedEdge-MX was for the DLS2 and calibration panel, to ensure a strong radiometric calibration. They needed data repeatability as they were flying 13 times across the season, from April to October.

Victoria Natalie and Abby McCrea at OSU lead the efforts to collect the RedEdge-MX imagery. They knew that capturing multispectral imagery over water would prove to be tricky as the water surface is generally uniform, which can make stitching the imagery difficult. To help ensure good outputs, the team increased their overlap to 80%, to ensure that enough points of commonality between captures were identified in order for the software to stitch. Using this increased overlap for their flight paths, they were able to get good results.

They flew after the initial herbicide application, which covered about 1% of the lake. The RedEdge-MX provided insight into the weed health; essential for monitoring the selective herbicide applications made on this environmentally sensitive area. RGB and false-color composite (FCC) from Sentinel-2 satellite were then used to measure on a larger scale, what the composition of the YFH looked like post-herbicide application. A courser resolution was sufficient for this measurement, and the 10m spatial resolution imagery showed that the application was effective, with 96% dieback of YFH.

The battle against YFH is not quite over yet though. Regrowth is expected and efforts will be ongoing to monitor the lake. YFH propagates by producing ramets, which break away from the main plant and float away to grow in a new location, or sink to the bottom of the lake and remain dormant. Either way, the ramets mean that some of the YFH can evade foliar herbicide applications. Remote sensing efforts will be ongoing to protect Lake Carl Blackwell and prevent the spread of YFH to other lakes and ponds in the state.