You must have a javascript-enabled browser and javacript and stylesheets must be enabled to use some of the functions on this site.


Application of Clustering Techniques to Multispectral Optical Data Over the Ocean

Margaret Joseph(1) , Ray Harris(1) , Colette Robertson(2) , Valborg Byfield(2) , and Graham Quartly(2)

(1) University College London, 26 Bedford Way, WC1H 0AP London, United Kingdom
(2) National Oceanography Centre, Empress Dock, SO14 3ZH Southampton, United Kingdom


Unsupervised classification is commonly used in land-based remote sensing. Here we transpose this approach to the marine environment and use clustering techniques to investigate whether multispectral optical data can give us biological information beyond a simple measure of chlorophyll concentration. The investigation uses various combinations of bands within the optical spectrum, and in particular we look at the contribution of the red bands in MERIS to both the estimation of chlorophyll concentration and the discrimination of different phytoplankton groups. A number of high-resolution radiance datasets from MERIS and MODIS are analysed for two test regions. The first is part of the southwest Indian Ocean surveyed in February 2005 during the MadEx cruise. The second study features an area in the English Channel just off Plymouth, for which a long time series of observations are available.


Full paper