Characterization of the seafloor using normal incidence acoustic backscattered time domain signals from a parametric sonar

For both military and civilian applications, a detailed knowledge of the upper part of the seabed is important. Remote sensing using sonar systems is one way to obtain this knowledge. In order to carry out a robust characterization or classification, we argue that a model based scheme should be used. This way the environmental conditions and the actual sonar systems are part of the processing chain, and parameters like water depth, beam-width and frequency can be properly accounted for. The utilized characterization method is based on properties of the coherent backscattered time signal from the seafloor surface. The analytical background for the approach is described. Estimates of the seafloor roughness can be found from the apparent frequency shift in the data. Estimates of the impedance of the seafloor can then be found, by using the energy of the signal and the estimated roughness. The coherent part of the signal is improved by stacking the time series, and further suppressing of the volume part of the return is done by use of a window. Results from applying the method to both synthetic and field data are presented, as well as sensitivity analysis for roughness.