The primary objective of this study was to assess the condition of a portion of Saudi Arabia's rangelands and evaluate the eVects of grazing by the animal herds of indigenous nomads over the last decade. Because of the desertic condition of these rangelands, changes in vegetation cover are more subtle than would be the case for other, less arid areas. Consequently, a new analytic methodology for the detection of deserti® cation of arid and hyper-arid rangelands was developed speci® cally for this project. The conceptual framework for the analysis is the use of the coe cient of variation (COV) of the monthly Normalized DiVerence Vegetation Index (NDVI, maximum-value composite) as a measure of vegetative biomass change. A higher NDVI COV for a given pixel (excluding cases of changes in land use) represents a greater change in vegetation biomass in the ground area represented by that pixel. Linear regression was used to determine the trend in COV values for each pixel over the 12-year period for which data was available; pixels with a negative slope are considered to represent ground areas with decreasing amounts of vegetation. Results were validated by tests of statistical signi® cance and by comparison of the theoretical results to vegetation change and land-cover data from the remote sensing systems and from reconnaissance¯ights over select areas. These deserti® cation trend results were then combined with land-cover information to provide an assessment of deserti® cation status.

No abstract