Inter-algorithm Relationships for Retrievals of Fraction of Vegetation Cover in a Framework of Linear Mixture Model

Fraction of vegetation cover (FVC) retrieved from remotely sensed reflectance spectra serves as a useful measure of land cover change. Since its retrieval algorithms show variations in assumptions of reflectance models and conditions imposed on the modeled spectra, the retrieved values also show some variations among the algorithms. This study discusses relationships among the FVC retrieval algorithms based on a well-known linear mixture model (LMM). The relationships among the algorithms were derived analytically within a framework of two-endmember LMM. It was clarified that some of the algorithms are equivalent in the sense that a one-to-one relationship exists among the algorithms. Numerical experiments had been conducted to evaluate the differences in error propagation among the algorithms induced by uncertainties in measured reflectance (often represented by a signal-to-noise ratio). The results indicate that the error propagation mechanisms are different in some extent among the algorithms. Moreover, the magnitude of the propagated error depends on the location of a target reflectance spectrum in the red-NIR reflectance space. Although the reflectance model employed in this study is quite limited, the fundamental aspects of the derived relationships would contribute to better understanding of the FVC retrievals.