Image-based classification of multibeam sonar backscatter data for objective surficial sediment mapping of Georges Bank, Canada

Developments in acoustic survey techniques, in particular multibeam sonar, have revolutionised the way we are able to image, map and understand the seabed environment. It is now cost effective to image large areas of the seafloor using these techniques, and the information from such surveys provides base line data from which thematic maps of the seabed environment, including maps of surficial geology, can be derived when interpreted in conjunction with in-situ ground truthing data. Traditional methods for the interpretation of acoustic backscatter rely on experienced interpretation by eye of grey-scale images produced from the data. However, interpretation of data can be subjective, and new developments in semi-automated backscatter classification software offer an objective method of segmentation of acoustic backscatter data into acoustically similar regions, but are not yet well tested or accepted. A large multibeam sonar data set from Georges Bank, Canada, was classified using the backscatter classification software, QTC-Multiview. Data from 4800 km2 of seabed were classified and results were compared with 110 ground truthing stations to assess the performance of the classification for geological discrimination. The relationship between backscatter metrics derived from the classification software and benthic geological characteristics were explored using statistical methods. Results suggest that image-based backscatter classification shows considerable promise for interpretation of multibeam sonar data for the production of geological maps.