Software tools developed for seafloor classification

The advent of modern acoustic remote sensing techniques has brought the long sought after goal of acoustic classification of seafloor materials closer to fruition. Quantitative sediment classification is not a trivial task; many of the processes that determine the characteristics of the collected data cannot be adequately or systematically described. This thesis presents the development of software tools and approaches designed to facilitate quantitative seafloor classification.; The TracEd tool is a robust multiple-horizon tracker that can be used for the evaluation of real time seafloor classification algorithms. It has a modular architecture so that various approaches can easily be implemented and tested. Data collected with TracEd can be visualized in both time and geographical reference frames. In addition to seafloor classification, the potential for using this tool to identify both mid-water and sub-bottom features has garnered much interest from biologists and geophysicists.; Almost all analysis of seafloor acoustic data depends on the ability to track the seafloor return and map its spatial distribution. A bottom tracking algorithm that is both very efficient and robust was therefore developed for TracEd; this algorithm has not failed with any of the data processed so far.; A second program, BatCor was developed to correct bathymetry artifacts by removing along-track low frequency trends. BatCor was applied to a data set collected in Western Lake Ontario, providing important new insights into the interpretation of the data.; Acoustic remote sensing data may be classified by isolating descriptive features of the returns and attempting to correlate these to seafloor type. A third program, Lassoo! was developed to group and classify these features in a two-dimensional feature space. Lassoo! can also overlay these classified features over an image of the seafloor in a georeferenced space. Finally the user may select portions of the image of the seafloor and obtain the frequency distribution of the selected data. Lassoo! was used for the classification of vertical incidence sonar data obtained near L'anse au Beaufils, PQ and compared to a commercial seafloor classification system (RoxAnn). Lassoo! was able to redefine the original sediment classes in a way that showed improved correlation with measured changes in sediment properties e.g., grain size. Lassoo! may also be used for any type of georeferenced data.