A hyperspectral, remote-sensing approach to spectral discrimination of marine habitats at Ningaloo Reef, Western Australia
Pinnel, N., Kobryn, H., Heege, T., Harvey, M. and Beckley, L. (2008) A hyperspectral, remote-sensing approach to spectral discrimination of marine habitats at Ningaloo Reef, Western Australia. In: 11th International Coral Reef Symposium, 7 - 11 July, Fort Lauderale.
Optical remote-sensing techniques, especially hyperspectral sensors, provide an non-invasive and cost-effective approach to mapping and monitoring the condition of reefs over large areas because of their capability to identify reef components on the basis of their spectral response. The aim of this study is to develop a reliable and repeatable procedure for mapping submerged coral reefs using airborne hyperspectral data. Spectral reflectance of corals, macro-algae and sediment were measured underwater with an OceanOptics2000 spectro-radiometer. These spectra were used for development of algorithms for automated applications to image classification. A genetic algorithm technique was used to determine optimal waveband combinations (including derivatives) for identification of substrate types. Initial results show that in situ reflectance spectra of reef substrates were significantly different for various spectral wavelengths. Using a linear discriminant analysis, the in situ spectra of six benthic groups (branching, digitate and tabulate Acropora, massive corals (e.g. Porites), submassive corals (e.g. Pocillopora) and macro-algae) could be classified to 90 % accuracy with as few as six bands. A classification of major habitat groups was applied to airborne hyperspectral data from HyMap acquired in November 2005 and April 2006 over the Yardie Creek area at Ningaloo Reef. The images were corrected for atmospheric, air-water interface and water column effects using the Modular Inversion & Processing System. This removes subjectivity from the classification and approaches an automated classification allowing for improved transferability to other reefs and monitoring applications. The retrieved bottom albedo image was used to classify the benthos, generating a detailed map of benthic habitats, followed by accuracy assessment. The outputs of multi-temporal image analysis contain percent cover of corals, macro-algae and sediment. Results indicate that the spectral response of corals can be determined to 10 m depth and shows that hyperspectral remote sensing techniques offer great potential in mapping coral reefs.
|Publication Type:||Conference Item|
|Murdoch Affiliation:||School of Environmental Science|
|Item Control Page|