Methods for Remotely Mapping, Classifying and Verifying Seabed Properties and Fish Habitats Using Fisheries Echosounders

Echosounder data collected using fisheries echosounders during fisheries surveys provides a means for simultaneous seabed mapping, prediction of seabed class, and estimation of fish biomass. The multi-frequency biplanar interferometric imaging (MBI) technique enables remote estimation of seabed depth, slope, roughness and seabed backscattering strength. Using MBI parameters, including the variation of backscattering strength by incidence angle and frequency estimated from vertical split-beam fisheries echosounder data, seabed lithology class may be predicted using a simple model. More than twenty classes of seabed lithology in and around rockfish (Sebastes) habitat were predicted by the acoustic model trained on a subset of image data from a remotely operated vehicle (ROV), and then verified using an independent test image set.  A classifier based on the acoustic model parameters was > 90% accurate in predictions of seabed lithology.