Location: Range Management ResearchTitle: Remote sensing of threshold conditions in an arid ecosystem Author
Submitted to: American Geophysical Union
Publication Type: Abstract Only
Publication Acceptance Date: 11/15/2007
Publication Date: 12/10/2007
Citation: Steele, C.M., Bestelmeyer, B.T., Rango, A., Smith, P.L., Laliberte, A. 2007. Remote sensing of threshold conditions in an arid ecosystem [abstract]. American Geophysical Union Fall Meeting, December 10-14, 2007, San Francisco, California. Abstract No. B43B-1160. Interpretive Summary:
Technical Abstract: Land management in the arid southwestern USA increasingly addresses thresholds in response to recent concepts adopted by private and public lands agencies and conservation organizations. Vegetation in arid rangelands typically presents as distinctive mosaics of vegetation patches, which persist in dynamic equilibrium with the abiotic environment and facilitative-competitive interactions between organisms. Theory and observation suggest that as an area approaches a threshold in response to disturbance, there is a concomitant change in the spatial arrangement of vegetation patches. This change is readily identifiable on fine spatial resolution aerial photography or satellite sensor imagery. We propose a classification method for identifying threshold-inducing change in vegetation pattern. To illustrate this method, we have applied an object-oriented, supervised classification to subsets of Quickbird imagery (70 cm ground resolution) over the Jornada basin in southern New Mexico. The imagery covers several land management regimes (private, public, federal) and provides spatial variation in ecosystem conditions. Imagery was first segmented to create fine and coarse resolution image objects. Fine resolution image objects are defined as having within-object spectral homogeneity at the scale of the shrub or single patch of grass or soil. Coarse resolution image objects are defined as containing spectral homogeneity at the scale of the vegetation stand. A classification tree was used to classify coarse resolution image objects to high risk of a threshold, low risk of a threshold, or post-threshold according to the content and spatial arrangement of shrub, grass and soil patches within them. Ground-based monitoring to detect localized threshold conditions across broad management areas is intractable so the use of remote sensing is essential to successful prevention of threshold development.