Hyperspectral shape-based unmixing to improve intra- and interclass variability for forest and agro-ecosystem monitoring Laurent Tits a , Wanda De Keersmaecker a , Ben Somers b , Gregory P. Asner c , Jamshid Farifteh a , Pol Coppin a a Katholieke Universiteit Leuven, Department of Biosystems, W. De Croylaan 34, BE-3001 Leuven, Belgium b Flemish Institute for Technological Research, Remote Sensing research unit, Boeretang 200, BE-2400, Mol, Belgium c Carnegie Institution for Science, Department of Global Ecology, 260 Panama St., Stanford, CA 94305 Abstract The monitoring of forests and agro-ecosystems often requires the use of a spectral mixture model to provide detailed information on spatial and tem- poral variations in vegetation cover. Two key issues in the mapping of vege- tation cover in complex ecosystems are the high spectral similarity (i.e. low interclass variability) between and the high spectral variability among differ- ent vegetation species (i.e. high intraclass variability), as they impede the performance of the Root Mean Square Error (RMSE) criterion, traditionally used in Spectral Mixture Analyses (SMA) to optimise the fit between mod- elled and measured mixed signal. Shape-based objective functions have been proposed as an alternative. Experiments, based on ray-tracing simulations, indeed demonstrated the added value of implementing shape-based error met- rics in unmixing of vegetation in (i) reducing the effects of intra-class end- Email address: laurent.tits@biw.kuleuven.be (Laurent Tits) Preprint submitted to Elsevier September 3, 2012