Submitted to: American Society for Photogrammetry and Remote Sensing Proceedings
Publication Type: Proceedings
Publication Acceptance Date: June 21, 2005
Publication Date: October 4, 2005
Citation: Laliberte, A.S., Rango, A., Fredrickson, E.L. 2005. Multiscale, object-oriented analysis of quickbird imagery for determining percent cover in arid land vegetation. 20th Biennial Workshop on Aerial Photography, Videography, and High Resolution Digital Imagery for Resource Assessment, October 4-6, 2005, Weslaco, Texas. 2005 CDROM. Interpretive Summary: No interpretive summary required.
Technical Abstract: Efforts to remotely sense arid land vegetation are often encumbered by high reflectance of the soil background, mixtures of green and senescent grasses, and the prevalence of shrubs in grasslands. These issues make it difficult to classify vegetation or estimate percent vegetation cover. Objectives of this study were to derive estimates of percent cover for several vegetation classes in a 1200 ha pasture at the USDA, Agricultural Research Service’s Jornada Experimental Range. A stratified random sample approach was used to determine percent cover for 322 field plots. A QuickBird satellite image was segmented at different scales which resulted in image objects for which a multitude of spectral, spatial, and texture characteristics were extracted. We used regression trees to develop a rule base for image classification and performed discrete and fuzzy accuracy assessments. For classes with discrete boundaries map accuracy was 73%, while accuracy values ranged from 81-86% using a 2.5-5% cover boundary around each class. This object-oriented multi-scale approach allowed us to extract shrubs at a fine scale and determine percent cover values for the shrub-interspace at a coarser scale. In addition, shrub density and percent cover for different grass species could be extracted using this method. The regression tree was an excellent tool for reducing the number of input variables derived from the image. Future research will include refining the predictive ability of the decision tree and determining the possibility of applying this model to other locations and/or to other scales.