Determination of the separability of coastal community assemblages of the Al Wajh Barrier Reef, Red Sea, from hyperspectral data

Remote sensing provides a practical means by which coral reefs and their associated communities are commonly mapped. The availability of spectral information is a key determinant of the detail discernable in the mapping process and consequent detail presented in output maps. Testament to this is the increasing utility of hyperspectral sensors, which typically yield datasets of higher resolution, spectrally continuous wavebands. Image classification algorithms distinguish between the different and unique reflectance characteristics of target features. While the availability of more wavebands provides the opportunity to apply analysis techniques that treat the data as spectrally continuous, such a large number of data dimensions also present a considerable computing burden. Through multiple discriminant function analysis, this paper identifies an optimal subset of wavelengths for resolving the reflectance of key terrestrial and marine coverages at the Al Wajh Barrier reef system, Saudi Arabia, Red Sea. The goal of such analysis is to facilitate the processing of high resolution, spectrally continuous remote sensing data of coastal landscapes.