1 December 2007 Classification of benthic composition in a coral reef environment using spectral unmixing
Author Affiliations +
Remote sensing is being applied with increasing success in the evaluation and management of coral ecosystems. We demonstrate a successful application of hyperspectral image analysis of the benthic composition in Kaneohe Bay, Hawaii using data acquired from NASA's Airborne Visible Infrared Imaging Spectrometer. We employ a multi-level approach, combining a semi-analytical inversion model with linear spectral unmixing, to extract information on the coral, algae and sand composition of each pixel. The unmixing model is based on the spectral characteristics of the dominant species and substrate types in Kaneohe Bay, and uses an optimization routine to mathematically invert the relationship of how each component spectrally interacts and mixes. The functional result is the ability to quantitatively classify individual pixel composition according to the percent contribution from each of three main reef components. Output compares favorably with available field measurements and habitat information for Kaneohe Bay, and the overall analysis illustrates the capacity to simultaneously derive information on water properties, bathymetry and habitat composition from hyperspectral remote sensing data. Further, the resulting spatial analysis capacity contributes an improved capability for monitoring coral ecosystems and an important basis for resource management decisions.
James A. Goodman, James A. Goodman, Susan L. Ustin, Susan L. Ustin, } "Classification of benthic composition in a coral reef environment using spectral unmixing," Journal of Applied Remote Sensing 1(1), 011501 (1 December 2007). https://doi.org/10.1117/1.2815907 . Submission:

Back to Top