NEW TECHNIQUE FOR ENHANCING THE DETECTION AND CLASSIFICATION OF SHALLOW MARINE HABITATS

A new method is described for enhancing the visual interpretation and quantitative analysis of shallow marine habitats. The method is based on the principle of exploiting the critical depth range of each habita t using a Degital Elevation Model (DEM) generated from a bathymetric m ap of the study area. The DEM is used as an input into a mathematical model derived by regressing known depth data with their thematic mappe r (TM) spectral values over the dominating depth-variable sandy substr ate. The resulting depth reflectance image is then combined with the T M bands to minimize the effect of depth variation within the image. Th e new method proved to be effective for enhancing the visual interpret ation of submerged habitats and particularly sea-grass beds off the co ast of Bahrain which could not be detected with conventional image pro cessing techniques. The technique also proved to be useful for enhanci ng areas of high suspended sediments in the water. In the quantitative analysis stage, the technique was found to increase the final classif ication accuracy of the subscene substantially. The slope and aspect v ariation of the topography caused confusion in classification between some of the habitats. Ways of improving the technique are suggested fo r future investigations.