Spectral rank-abundance for measuring landscape diversity

Investigation of the diversity of a landscape implies finding appropriate measures coupling information on richness and equitability. Most of the papers dealing with remotely sensed images have relied on the richness of digital numbers (DNs) or on Shannon entropy or Pielou evenness indices for measuring their heterogeneity. Instead, based on ecological theory, we will show that rank-abundance diagrams may be profitably used in remote sensing to take into account both spectral richness and spectral equitability at the same time, by using a unique approach. After a theoretical introduction to the problem, we will empirically test the proposed method by extracting DN abundances derived from a Landsat Enhanced Thematic Mapper Plus (ETM+) image representing contrasting landscapes (test sites), plotting the relative abundance of each DN value versus its rank (rank-abundance diagrams) and interpreting statistically and ecologically the achieved results. We do not propose rank-abundance diagrams as a replacement of existing measures of spectral diversity, but as a parallel method to encompass (at the same time) both richness and evenness of remotely sensed images.