A framework for scientific data mining in hydro-acoustic data sets.

A data mining framework for handling large volumes of scientific hydro-acoustic measurements is proposed. The method is applicable to data collected by the new generation of multibeam echosounders, capable of logging acoustic data for the full watercolumn. S; such instruments will be used for fisheries applications in the near future. The data mining technique is based on an inverse modeling of the underlying physics and electronics of a generic multibeam sonar system. A set of tagged soundings is obtained, which serves as a base for further advanced analysis techniques. It is anticipated that the proposed framework will serve as a research tool for the scientific fisheries research community.