Benthic Habitats and the Effects of Fishing

Symposium Abstract: Seabed Classification with Multibeam Sonars for Mapping Benthic Habitat

J. M. Preston, A. C. Christney, W. T. Collins, and B. D. Bornhold

doi: https://doi.org/10.47886/9781888569605.ch31

Seabed images, from multibeam systems or sidescans, convey a lot of information about seabed type. Large-scale rocky relief often gives dramatic images, and morphology such as sand waves can be very evident. Fine-grained sediments affect images in less obvious ways. Statistical processing of the backscatter amplitudes generate features adequate for seabed classification that agree with both largescale interpretation and fine-grained details. Before calculating features, it is essential to precondition the image by compensating for artifacts due to range and grazing angle. Useful features include ratios of integrals of the power spectrum over various frequency bands, descriptors of grey-level co-occurrence matrices and histograms, means and higher order moments, and fractal dimension. Generating many features and then using multivariate statistical techniques to select the linear combinations that capture most of the variance in the dataset improves the quality and usefulness of the resulting classifications by adapting the classification to each set of images. To complete the classification process, records are assigned to classes by the same clustering process used in the existing Quester Tangent classification products. Maps of these acoustic classes show regions of distinct acoustic character, thus of distinct sediment type in some sense. To make all this useful for benthic studies, one must understand how this acoustic diversity correlates with the distribution of species of interest. Various spatial analysis techniques are available to accomplish this, and several examples of the integration of acoustic and benthic information will be presented.