The application of a geometric optical canopy reflectance model to semiarid shrub vegetation

Abstract

The Li-Strahler (1985) canopy model was tested, using SPOT HRV XS imagery, for semiarid shrub vegetation, based on 26 small (1-ha) sites in five classes of shrub vegetation, two dominated by tarbush, one by creosote bush, and two by mesquite. The model was driven by reflectance values derived from June and September imagery. While predictions of crown size and density for individual sites had a large average error of 35%, the predictions of shrub size and density were reasonably accurate when grouped by shrub class. The aggregated predictions for a number of stands within a class were accurate to within one or two standard errors of the observed average value. Accuracy was highest but predictions were biased for some classes (size was underestimated) when the nonrandom shrub pattern was characterized for the class based on the average coefficient of determination of density. >