|Parker-Williams, A - UNIVERSITY OF WY|
Submitted to: Journal of Range Management
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: February 22, 2006
Publication Date: September 20, 2006
Citation: Hunt, E.R., Parker-Williams, A.E. 2006. Detection of flowering leafy spurge with multispectral imagery. Journal of Range Management. 59:494-499. Interpretive Summary: Leafy spurge is a noxious invasive weed that infests large portions of Northern Great Plains. The conspicuous yellow-green flowers of leaf spurge are highly visible in June and July and large patches are visible in aerial photography. Hyperspectral remote sensing uses many, contiguous, narrow bands which can be used to determine a reflectance spectrum for a pixel; however, these data cover small areas, are difficult to process, and difficult to acquire. Previous work has shown that advanced algorithms developed for hyperspectral remote sensing detects small patches of flowering leafy spurge. Multispectral remote sensing uses a few, broad bands and the data are readily available from many sources. This study was undertaken to determine if the work with hyperspectral remote sensing can be applied to multispectral remote sensing. Whereas bands at green wavelengths are most sensitive to flowering leafy spurge, vegetation indices were only weakly correlated with the cover of leafy spurge. Thus, multispectral satellite sensors such as Landsat and SPOT can not be used to detect leafy spurge reliably. These results were predicted by a computer simulation model of canopy reflectance. Therefore, databases of leaf and flower spectral reflectances and transmittances under specific conditions can be used to predict which methods of remote sensing can be used to detect invasive weeds.
Technical Abstract: The distribution and abundance of leafy spurge (Euphorbia esula L.) can be determined with hyperspectral remote sensing, but the availability of hyperspectral sensors is limited. Hence, the Landsat 7 Enhanced Thematic Mapper Plus (ETM+) and System Pour d'Observation de la Terre (SPOT) 4 imagery were acquired to test the ability of these sensors to detect leafy spurge. The green:red band ratio was the vegetation index with the highest correlations to leafy spurge cover, but the correlations were weak and not useful for predictions. With Airborne Visible Infrared Imaging Spectrometer (AVIRIS) data, the green:red band ratio was also very weakly correlated to spurge cover, whereas fractional abundance of leafy spurge from the Mixed Tuned Matched Filtering algorithm was highly correlated with leafy spurge cover. Canopy reflectance modeling using the Scattering by Arbitrarily Inclined Leaves (SAIL) model suggests the poor correlations were caused by variations in leaf area index. It is important to develop spectral libraries in order to use canopy reflectance models effectively to reduce time and effort testing methods to remotely sense leafy spurge and other invasive weeds.