CLASSIFICATION OF SEA-FLOOR GEOLOGY USING MULTIBEAM SONAR DATA FROM THE SCOTIAN SHELF

We describe a method for classifying multibeam sonar data, and illustr ate the method using data collected with a Simrad EM1000 sonar on the Scotian Shelf, Canada. The method involves comparing various attribute s of the bathymetry and backscatter with typical examples of specified seafloor types and computing their statistical degree of similarity. Sediment ponds are identified as areas of low echo amplitude and low t opographic gradient and curvature. Ridges and troughs are identified b y fitting a paraboloid to patches of the bathymetry. Once the sediment s, ridges and troughs are located, we are able to use the database to extract orientations and other characteristics of these features. For example, the orientation of the topography can be computed from the pa raboloid surface, and rose diagrams of the ridge and trough orientatio ns reflect the fabric of up-turned Cambro-Ordovician sedimentary beds outcropping in this area. The unconsolidated sediments ponded within s mall basins have low topographic gradients with consistent tilts towar ds the southeast (mean 133 degrees), which we interpret as due to offs hore transport of Holocene sediments, possibly in response to storms.