Mapping a Riparian Weed with SPOT 5 Imagery and Image Analysis

Authors: Everitt, James; Yang, Chenghai; Fletcher, Reginald; Deloach Jr, Culver

Submitted to: Biennial Workshop on Aerial Photography, Videography, and High Resolution Digital Imagery for Resource Assessment Proceedings
Publication Type: Proceedings
Publication Acceptance Date: 7/12/2007
Publication Date: 2/15/2008
Citation: Everitt, J.H., Yang, C., Fletcher, R.S., Deloach Jr, C.J. 2008. Mapping a Riparian weed with SPOT 5 imagery and image analysis. Biennial Workshop on Aerial Photography, Videography, and High Resolution Digital Imagery for Resource Assessment Proceedings. CDROM.

Interpretive Summary: The invasion and spread of undesirable plant species present serious problems for resource managers. Giant reed is an invasive perennial grass that invades riparian sites in many areas of the world. It reduces the diversity of native vegetation and consumes excessive amounts of water. A study was conducted along the Rio Grande in southwest Texas to determine the potential of SPOT 5 (10 m resolution) satellite imagery for distinguishing giant reed infestations. Three subset images were subjected to supervised and unsupervised image analysis. Accuracy assessments performed on supervised classification maps of the images from the three sites had poducer’s and user’s accuracies ranging from 75.7% to 93.3%, while accuracy assessments performed on unsupervised classification maps from the three sites had producer’s and user’s accuracies ranging from 67.7% to 97.3%. Both supervised and unsupervised techniques did a good job in identifying giant reed and were deemed equal. These results should be of interest to weed scientists and riparian resource managers.

Technical Abstract: SPOT 5 (10 m resolution) multi-spectral satellite imagery was evaluated for mapping infestations of the invasive grass giant reed (Arundo donax L.) along the Rio Grande in southwest Texas. The imagery had three bands (green, red, and near-infrared). Three subsets from the SPOT 5 image were extracted and used as study sites. The subset images were subjected to supervised and unsupervised image analysis. Accuracy assessments performed on supervised classification maps of images from the three sites had producer’s and user’s accuracies for giant reed ranging from 75.7% to 93.3%, while accuracy assessments performed on unsupervised classification maps of images from the three sites had producer’s and user’s accuracies for giant reed ranging from 67.7% to 97.3%. These results indicate SPOT 5 imagery coupled with image analysis techniques can be used successfully for detecting and mapping giant reed infestations.