Page Banner

United States Department of Agriculture

Agricultural Research Service

Research Project: USING REMOTE SENSING AND GIS FOR DETECTING AND MAPPING INVASIVE WEEDS IN RIPARIAN AND WETLAND ECOSYSTEMS Title: Evaluation of hyperspectral reflectance data for discriminating among six aquatic weeds

Authors
item Everitt, James -
item Yang, Chenghai
item Summy, K. -
item Glomski, L. -
item Owens, C. -

Submitted to: Journal of Aquatic Plant Management
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: February 18, 2011
Publication Date: December 30, 2011
Citation: Everitt, J.H., Yang, C., Summy, K.R., Glomski, L.M., Owens, C.S. 2011. Evaluation of hyperspectral reflectance data for discriminating among six aquatic weeds. Journal of Aquatic Plant Management. 49:94-100.

Interpretive Summary: The invasion of aquatic ecosystems by noxious plant species presents a serious problem to management of water bodies. Field hyperspectral reflectance data were studied at 50 wavebands (10 nm bandwidth) over the 400 to 900 nm spectral range to determine their potential for distinguishing among six aquatic weed species: curly-leaf pondweed, hydrilla, Eurasian watermilfoil, northern milfoil, hybrid milfoil, and parrotfeather. The species were studied on three dates: May 11, May 30, and July 1, 2009. Two procedures were studied to determine the optimum bands for discriminating among the species: multiple comparison range test and stepwise discriminant analusis. Multiple comparison range test results for both May dates showed the most separations among the species occurred at bands in the green-red edge, red, and red-near-infrared (NIR) edge spectral regions. For the July date, the most separations among species occurred at all the green and most of the red bands, as well as some of the red-NIR edge bands. Using stepwise discriminant analysis, nine bands on May 11 and 10 bands on May 30 in the blue to NIR spectral regions had the highest power for discriminating among the species, while seven bands in the red-NIR edge and NIR regions were useful for discriminating among the species on the July date. These results provide insight for determining the optimum bands when using hyperspectral imagery captured from aircraft or satellite platforms for identifying the aquatic weeds studied here and should be useful to aquatic weed specialists’ and wetland resource managers.

Technical Abstract: In situ hyperspectral reflectance data were studied at 50 wavebands (10 nm bandwidth) over the 400 to 900 nm spectral range to determine their potential for discriminating among six aquatic weed species: curly-leaf pondweed (Potamogeton crispus L.), hydrilla [Hydrilla verticillata (L. F.) Royle], Eurasian watermilfoil (Myriophyllum spicatum L.), northern milfoil (Myriophyllum sibiricum Kom.), hybrid milfoil (Myriophyllum spicatum x Myriophyllum sibiricum), and parrotfeather [Myriophyllum aquaticum (J. M. da Conceicao) Vellozo]. The species were studied on three dates: May 11, May 30, and July 1, 2009. All six species were studied on the two May dates, while for the July date only four species (hydrilla, Eurasian watermilfoil, hybrid milfoil, and parrotfeather) were studied. To determine the optimum bands for discriminating among the species, two procedures were used: multiple comparison range test and stepwise discriminant analysis. Multiple comparison range test results for both May dates showed that most separations among species occurred at bands in the green-red edge, red, and red-near-infrared (NIR) edge spectral regions. For the July date, the most separations among species occurred at all the green and most of the red bands, as well as some of the red-NIR edge and NIR bands. Using stepwise discriminant analysis, nine bands for May 11 and 10 bands for May 30 in the blue to NIR spectral regions had the highest power for discriminating among the six species, while seven bands in the red-NIR edge and NIR regions were useful for discriminating among the four species on the July date.

Last Modified: 7/24/2014
Footer Content Back to Top of Page