Submitted to: Environmental Monitoring and Assessment
Publication Type: Peer reviewed journal
Publication Acceptance Date: 12/19/2007
Publication Date: 1/22/2008
Citation: Ge, S., Carruthers, R.I., Spencer, D.F., Yu, Q. 2008. Canopy assessment of biochemical features by ground-based hyperspectral data for an invasive species, giant reed (Arundo donax). Environmental Monitoring and Assessment. 147:271-278. Interpretive Summary: Arundo donax in an invasive grass similar in stature to bamboo. It is very aggressive and grows in moist areas adjacent to streams and other wetlands. It damages the environment by displacing other desired plant and by causing severe erosion, even breaking levees through its vigorous root growth. Understanding its basic biology and means of vegetative reproduction will allow land and water way managers to better control this weed over wide areas. Assessing its vegetative condition through the use of remote sensing will allow managers to both assess this weed's distribution and impacts. Hyperspectral remote sensing further allows the determination of its vigor, growth potential and the risk of spread when conducted through areawide assessment. This study provided a means of assessing nutrient levels within Arundo donax canopies that can be used to interpret remote sensed data providing measures of chlorophyll, nitrogen and carbon contents in Arundo leaves. These measures help determine its growth stature and potential risk of spread.
Technical Abstract: This study explored the potential use of hyperspectral data in the non-destructive assessment of chlorophyll, carbon, and nitrogen content of giant reed at the canopy level. We found that pseudoabsorption and derivatives of original hyperspectral data were able to enhance the relationship between spectral data and measured biochemical characteristics. Based on correlogram analyses of ground-based hyperspectral data, we found that derivatives of pseudoabsorption were the best predictors of chlorophyll, carbon, and nitrogen content of giant canopies. Within the visible region, spectral data significantly correlated with chlorophyll content at both 461 nm and 693 nm wavelengths. Within the near-infrared region, carbon levels correlated with hyperspectral data at five additional wavelengths: 1038 nm, 1945 nm, 1132 nm, 1525 nm, and 1704 nm. The best spectral wavelength for estimating nitrogen content was 1542 nm. Such relationships between nutrient content and spectral data were best represented by exponential functions in most situations.