ARS | Publication request: Using Quickbird Satellite Imagery to Distinguish Two Noxious Weeds in Southern Texas

Submitted to: Forest Service Remote Sensing Conference Proceedings
Publication Acceptance Date: April 15, 2004
Publication Date: October 20, 2004
Citation: Everitt, J.H., Yang, C. 2004. Using Quickbird Satellite Imagery To Distinguish Two Noxious Weeds In Southern Texas. Forest Service Remote Sensing Conference Proceedings.

Interpretive Summary: The invasion and spread of noxious weeds in riparian and rangeland areas is a major deterrent to management of these areas. Improved methods are needed to assess the impact of weedy species in riparian and rangeland areas. QuickBird high resolution (2.8 m) satellite imagery was evaluated for detecting infestations of the invasive weeds, giant reed and spiny aster, in riparian and rangeland areas in south Texas. Both species could be readily distinguished in satellite images. An accuracy assessment performed on a computer classified image of a spiny aster study site had both a producer's and user's accuracy of 90%, whereas an accuracy assessment performed on a classified image of a giant reed study site had a producer's accuracy of 94% and a user's accuracy of 100%. These results should be of interest to rangeland and natural resource managers interested in mapping and controlling noxious weed infestations.

Technical Abstract: Giant reed (Arundo donax L.) and spiny aster [Leucosyris spinosus (Beaty.) Greene] are noxious weeds that invade riparian and rangeland areas. QuickBird high resolution (2.8 m) satellite imagery was evaluated for distinguishing spiny aster and giant reed infestations in southern Texas. Unsupervised classification techniques were used to classify false color composite (green, red, and near-infrared ) images of study sites. An accuracy assessment performed on a classified map of the spiny aster study site had both a producer's and user's accuracy of 90%, whereas an accuracy assessment performed on a classified map of the giant reed study site had a producer's accuracy of 94.4% and a user's accuracy of 100%.