Submitted to: American Society for Photogrammetry and Remote Sensing Proceedings
Publication Type: Proceedings
Publication Acceptance Date: August 23, 2004
Publication Date: March 7, 2005
Citation: Laliberte, A., Rango, A., Fredrickson, E.L. 2005. Classification of arid rangelands using an object-oriented and multi-scale approach with Quickbird imagery. Proceedings of the American Society for Photogrammetry and Remote Sensing Annual Conference, March 7-11, 2005, Baltimore, Maryland. 2005 CDROM. Interpretive Summary: Interpretive summary not required.
Technical Abstract: The classification of arid rangelands often presents unique problems due to the high reflectance of the soil background, a mixture of green and senescent grasses, and the prevalence of shrubs in grasslands, which can make it difficult to determine the proportion of grass cover from high resolution imagery. On the Jornada Experimental Range (JER), operated by the USDA Agricultural Research Service, ongoing research is aimed at determining the relationship between ground-based observations and remotely sensed data. Specific objectives of this study were to use near-earth photography and QuickBird satellite imagery to develop a detailed vegetation classification of a 1200 ha pasture in order to ascertain the extent of grassland and identify locations, extent and percent cover values for several grass species. We conducted extensive field sampling by taking photos of the ground vegetation from a height of 2.8 m and used thresholding techniques to determine percent vegetation cover and percent bare soil. The QuickBird image was analyzed with an object-oriented approach using the software eCognition. The segmentation was performed at 2 different scales, which was used to construct a hierarchical network of image objects representing the image information in different spatial resolutions simultaneously. This allowed for differentiation of individual shrubs on the lower level, and delineation of broader landscape classes on a higher level. For each image object containing the field plot, several spectral, spatial, and texture characteristics were extracted from the image. Ancillary information included soils, elevation, aspect and slope layers. The data was analyzed using classification and regression trees to determine correlations between features of the segmented image objects and the measured field plot parameters. The object-oriented classification of the QuickBird image worked favorably in this study, because shrubs could be classified separately at a finer scale, while the shrub-interspace vegetation could be analyzed at a coarser scale. This allowed us to get a reliable estimate of grass cover and shrub density in the pasture as well as shrub density within different grass species. The rule base derived from the decision tree proved to be successful at differentiating between the dominant grass species as well as defining several classes of percent grass cover. This approach shows promise for identifying detailed vegetation characteristics on arid rangelands. 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.