|Hunt, Earle - Ray|
Submitted to: Journal of Range Management
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 3/30/2003
Publication Date: 1/1/2004
Citation: Parker-Williams, A., Hunt, E.R. 2004. Accuracy assessment of leafy spurge detection with hyperspectral remotely sensed data. Journal of Range Management. 56(1):106-112. Interpretive Summary: Leafy spurge (Euphorbia esula L.) is a noxious perennial weed infesting 1,200,000 hectares of land in North America and causes severe economic impact, especially in North Dakota, South Dakota, Montana and Wyoming. When flowering, leafy spurge has distinctive yellow-green bracts that are spectrally distinct from other vegetation, which can be detected with hyperspectral remote sensing. NASA's hyperspectral sensor, the Airborne Infrared Imaging Spectrometer (AVIRIS) was flown over a study area encompassing Devils Tower National Monument in northeastern Wyoming. We tested a new relatively new technique called Mixture Tuned Matched Filtering, to detect leafy spurge even in pixels that are dominated by other land cover types. The technique had an overall 87% to 95% accuracy in detecting leafy spurge, with best results in prairies and riparian areas. The results were worse for conifer woodlands, where leafy spurge was detected 56% of the time when it was on the ground, which is good considering the leafy spurge was hidden in shadows and under tree crowns. The increasing availability of other airborne sensors, such as the Compact Airborne Spectrographic Imager and the AISA airborne imaging sensor, will allow this technique to be used in an automated operational program at lower cost compared to the AVIRIS imagery.
Technical Abstract: When flowering, leafy spurge (Euphorbia esula L.) has distinctive yellow-green bracts that are spectrally distinct from other vegetation. Airborne Visible Infrared Imaging Spectrometer (AVIRIS) data were acquired in northwestern Wyoming, near Devils Tower National Monument. Using the reflectance spectrum of flowering leafy spurge, the presence and absence of fleafy spurge was mapped with a relatively new method of spectral mixture analysis, called mixture tuned matched filtering (MTMF). Ground reference data were obtained to assess the accuracy of the presence/absence map. The MTMF technique mapped spurge very well with overall accuracies ranging from 87% to 95% depending on the classification criteria employed. Differences in classification criteria result in a trade-off between false positives, pixels that were mapped as leafy spurge but did not contain leafy spurge on the ground, and false negatives, areas that had leafy spurge on the ground but were not mapped as leafy spurge. The classifications of leafy spurge presence or absence were best for mixed prairie and riparian cover types, and poorest for conifer woodlands because of the interference from tree crowns and their shadows. MTMF was successful in accurately mapping leafy spurge because of the technique's power and the unique reflectance spectrum of leafy spurge.