Influence of endmember spectra on scaling effect of vegetation cover estimationwith NDVI

摘要

Fraction of vegetation cover (FVC) retrieved by NDVI-isoline based linear mixture model (LMM) suffers from systematic biases due to difference in spatial resolution among sensors, known as ‘scaling effect’. Scaling effect of the FVC has been investigated analytically in a previous work under the two-endmember assumption. It implies that the scaling effect (systematic bias) will be minimized by using a certain combinations of endmember spectra. The objective of this study is to examine influence of endmember spectra on the scaling effect of FVC with LANDSAT7/ETM+ data. The results demonstrated that there exists endmember spectrum of vegetation and non-vegetation which mitigate the scaling effect. Findings in this study should contribute to cross-sensor calibration of FVC estimation necessary to improve consistency and accuracy in monitoring of vegetation cover.