Multi-scale image segmentation of multibeam backscatter data for benthic monitoring

In the last decade seabed classification and mapping has developed as one of the most active areas of marine science. However, the translation of acoustic data into spatial products has been challenged by two important concepts: scale and habitat classification accuracy. Although acknowledged in the habitat mapping literature as being important, these two issues are receiving surprisingly little attention. Relationships between seabed habitat classes (as objects) and the scales they can be accurately are identified at are explored in this research in the context of mapping for marine protected area assessment where specific seabed geomorphometry classes are required to be accurately identified. The benefits for coupling multi-scale analysis and seabed habitat classes delineated using geomorphometry, are discussed.