Seabed characterization based on the statistical classification using the seabed reflection amplitudes of sub-bottom profiler data

Abstract

The seabed reflection amplitudes (SRAs) extracted from the sub-bottom profile have a strong correlation with the types and physical properties of the seabed sediments. In this paper, the SRAs distribution of classified seabed sediments is statistically obtained by calibration with seabed sampling results, discovering that SRAs on different seafloor sediment types exhibit Rayleigh distributions with varying parameters. Firstly, SRAs are compensated and enhanced, to improve their identification. Then, a novel classification method based on K-S test was proposed. This method measures the maximum distance between the cumulative distribution functions (CDF) of the unknown seabed and the calibrated sediment SRAs to check whether unknown samples belong to any of the known types. This proposed method only requires a small amount of seabed samples to automatically classify the seabed with high accuracy, and the model is simple, robust, and provides classification confidence.