Skip to main content
ARS Home » Research » Publications at this Location » Publication #186564
Title: DETECTING RED HARVESTER ANT MOUNDS WITH PANCHROMATIC QUICKBIRD IMAGERY

Author
item Fletcher, Reginald
item Everitt, James
item DRAWE, LYNN - WELDER WILDLIFE REFUGE

Submitted to: Journal of Applied Remote Sensing (JARS)
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 10/30/2007
Publication Date: 12/17/2007
Citation: Fletcher, R.S., Everitt, J.H., Drawe, L. 2007. Detecting red harvester ant mounds with panchromatic QuickBird Imagery. Journal of Applied Remote Sensing. 1:013556

Interpretive Summary: Natural resource managers have an interest in locating red harvester ant mounds because of the impact that heavy infestations have on pastures and because of the importance of the ants to the survival of the threatened Texas horned lizard. The objective of this study was to evaluate panchromatic QuickBird satellite imagery subjected to computer classification as a tool for detecting red harvester ant mounds. Quantitative analyses of maps derived from the classification indicated that the imagery provided adequate information for separating the mounds (accuracy greater than or equal to 94%) from mixed woody vegetation and mixed grass. The methods used in this study should have application for natural resource managers interested in detecting red harvester ant mounds with high-resolution satellite imagery.

Technical Abstract: Natural resource managers have an interest in locating red harvester ant (Pogonomyrex barbatus) mounds because of the impact that heavy infestations have on pastures and because of the importance of the ants to the survival of the threatened Texas horned lizard (Phrynosoma cornutum). This study evaluated panchromatic QuickBird imagery (450-900 nm; 0.6 m spatial resolution) subjected to computer classification as a tool for detecting red harvester ant mounds. The project focused on two sites on the Welder Wildlife Refuge (28.1225 N, 97.3641 W). Prior to image classification, we masked out trail roads and/or large bare areas that were not ant mounds because they had similar spectral values to the ant mounds. Using an unsupervised classification approach based on K-means clustering, image processing software separated the images into spectral clusters, which we assigned to the mixed woody vegetation, the mixed grass, or the ant mound class. User’s accuracy and producer’s accuracy of the thematic maps were greater than or equal to 94.0% for the ant mound class, indicating that it is possible to use thematic maps generated from panchromatic QuickBird imagery and computer classification to detect red harvester ant mounds.