Tree species mapping in tropical forests using multi-temporal imaging spectroscopy: Wavelength adaptive spectral mixture analysis

• Published in: International journal of applied earth observation and geoinformation 2014-09, Vol.31, p.57-66
• Main Authors: Somers, B., Asner, G.P.
• Format: Article
• Language: English
• Subjects: Applied geophysics; Earth sciences; Earth, ocean, space; Exact sciences and technology; Hyperion; Hyperspectral; Internal geophysics; MESMA; Morella faya; Phenology; Psidium cattleianum
• ISSN: 1569-8432; 1872-826X
• DOI: 10.1016/j.jag.2014.02.006

The use of imaging spectroscopy for florisic mapping of forests is complicated by the spectral similarity among co-existing species. Here we evaluated an alternative spectral unmixing strategy combining a time series of EO-1 Hyperion images and an automated feature selection in Multiple Endmember Spectral Mixture Analysis (MESMA). The temporal analysis provided a way to incorporate species phenology while feature selection indicated the best phenological time and best spectral feature set to optimize the separability between tree species. Instead of using the same set of spectral bands throughout the image which is the standard approach in MESMA, our modified Wavelength Adaptive Spectral Mixture Analysis (WASMA) approach allowed the spectral subsets to vary on a per pixel basis. As such we were able to optimize the spectral separability between the tree species present in each pixel. The potential of the new approach for floristic mapping of tree species in Hawaiian rainforests was quantitatively assessed using both simulated and actual hyperspectral image time-series. With a Cohen's Kappa coefficient of 0.65, WASMA provided a more accurate tree species map compared to conventional MESMA (Kappa=0.54; p-value