Mapping of marine benthic habitats using image processing techniques within a raster-based geographic information system

The present paper describes a method where maps of marine benthic communities in the littoral and sublittoral zones are produced using image processing techniques within a raster-based geographic information system (GIS). Maps of the littoral zones were based on digital raster images derived from aerial photographs. For the sublittoral zone, a ‘raster image’ was generated from acoustic data, recorded by a RoxAnn™ system. The image processing technique of unsupervised classification divided the images into areas which possess similar physical environmental characteristics. Images were validated with biological information collected from dedicated visual (littoral) and video (sublittoral) surveys. Ground validation data were overlain onto the image to enable a supervised classification. Each area was assigned to a biotope thus creating biotope maps for the littoral and sublittoral zones. Using image processing for the analysis of RoxAnn™ data is a new approach to benthic mapping because it treats acoustic data as continuous variables rather than categorical data as used by traditional methods. The methods described here allow remotely sensed data from tile sublittoral and littoral environments to be processed using similar analyses. However, prior to producing a composite map of both these environments, problems of scale and spatial resolution need to be resolved. These issues are discussed.