The effect of spatial resolution on measurement of vegetation cover in three Mojave Desert shrub communities

Moderate resolution remotely sensed imagery (20–30 m) has lacked sufficient spatial resolution to accurately measure total vegetative cover in arid environments because the density of plant cover is generally too low to significantly influence spectral reflectance. New and emerging imagery is now available at significantly higher spatial resolutions, resulting in pixel sizes equal to or smaller than many dominant desert shrubs. This had led to an opposing problem, where desert shrubs now exhibit multi-modal probability functions associated with the individual elements of the plant, such as foliage, woody stems and branches, and shadows within and around shrub canopies. High resolution imagery was collected at three spatial resolutions for study sites in the south-central Mojave Desert of California to (1) determine the effect of spatial resolution on the detection of four common species of shrubs (Chilopsis linearis, Psorothamnus spinosus, Larrea tridentata, and Ambrosia dumosa), (2) assess the affect of increasing pixel size on the accuracy of area estimates of shrub cover, and (3) assess the accuracy of vegetation cover measurements when compared to vegetation cover measured in the field. Spatial resolution greater than 1.0 m was not capable of discerning a large percentage of the shrubs in this region of the Mojave Desert. Yet, high correlation (r2=0.71–0.93r2=0.71–0.93) was found between field measurements of percent cover and those derived from imagery at spatial resolutions ranging from 0.6 to 1.0 m. The results indicated that a spatial resolution of 1.0 m or smaller was necessary to estimate percent cover and the area of individual shrubs from high-resolution imagery. Spatial resolution is dependent on the abundance and size of species of desert shrubs within plant communities, however 1.0 m resolution was found to be adequate for this region of the Mojave Desert.