Large-scale Acoustic Classification of Multibeam Echosounder Datasets within the INFOMAR Programme

Modern multibeam sonar systems, in addition to measuring water-depth, also provide measurements of the intensity of the returned acoustic pulse, termed backscatter. While several factors influence backscatter strength, seafloor composition is often the most significant contributor. Once appropriate corrections compensating for geometric losses and system artefacts are applied to such data, images outlining seafloor areas of similar acoustic response are readily produced. Statistical image segmentation methods are used to subdivide these images into a number of discrete ‘seabed classes’. Correlation of these remotely-sensed ground-types with sediment samples and underwater-imagery allows the assignment of descriptive labels to such ‘seabed classes’, and provides a cost effective means of producing baseline maps delineating seafloor conditions. Such products have applications in the fields of coastal zone management, habitat and fisheries studies, marine research and offshore engineering projects.