Submitted to: Journal of Applied Ecology
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 12/1/1996
Publication Date: N/A
Citation: Interpretive Summary: We studied the feasibility of using satellite imagery to map ecosystem condition on a regional scale. Fifteen images of southern New Mexico spread over the 1989 growing season were used to develop vegetation descriptors which were classified and then confirmed with extensive ground measurements. Our methods of comparing images taken at various times allowed us to accurately identify vegetation types even where vegetation was sparse.
Technical Abstract: The study of a desert landscape consisting of C3 shrublands and C4 grasslands demonstrated the feasibility of mapping desertification processes on a regional scale using time-series satellite images and demonstrated the potential for monitoring desertification processes through time. Fifteen meteorological satellite images of southern New Mexico, spread over the growing season of 1989, were processed to yield images of normalized difference vegetation indices. Computer classification of these images and comparison with extensive ground measurements confirmed the identification of the major classes of shrubland, grassland and mixed shrub and grass areas. Our high-resolution temporal approach is shown to permit the accurate identification of vegetation types even under conditions of sparse vegetation cover. The phenologies of grassland and shrubland were revealed in the time series of satellite images. We used this feature of the data to overcome the tendency of the strong radiation signal from bare soil to mask the biomass signal. Comprehensive comparisons with ground measurements confirmed the accuracy of the biomass identification for local areas within the Jornada Experimental Range and justified extrapolation of the classification to a larger adjacent area.