Sediment classification using Sub-bottom profiler

Sub bottom profilers are commonly used as mapping tool for the seafloor and sub-bottom structure in the upper few meters of the seafloor. Their recent enhanced performance in terms of resolution adds the potential to classify the sediment layers as well. In this research, the seabed surface and sub layers classification are investigated using model based techniques. The remote sediment classification technique of the seabed surface is achieved by matching the back scatter measurements to the predicted backscatter intensity of the model. The model simulates the returned signals of a monostatic sub bottom profiler operating at 100 kHz. The back scattering strength in the angle domain is estimated using the APL-UW backscattering model. The matching procedure was applied on averaged echo envelopes performed by Hilbert transform. The averaging process is essential to reduce the stochastic variability of the acquired data. The sub layers data was obtained by operating frequencies of (5, 10 and 15 kHz). The layer classification was achieved by estimating the geoacoustic parameters such as reflectivity and impedance contrast. Two techniques were investigated based on a reflectivity model. The first technique is an extension work of D.Simons [11] which aims to estimate the reflection coefficients via the received and transmitted energy ratio. The second technique is an implementation of a similar approach but applies the attenuation on the received frequency components in place of the nominal components. Both models accounted for energy propagation and its corresponding geometrical and sediment attenuation losses. The classification techniques were carried out to a dataset that was acquired in the Baltic Sea near Rostock in 2004. The acquired dataset is characterized by various bottom types such as mud, sand and coarse sediments. The general description of the acquired areas was used as a reference for the final results. Due to the lack of core samples, the classification was evaluated by comparing the results of the backscatter to the energy model. The results were consistent with the general description of the dataset. However, the matching process of the backscatter model is a cumbersome and very sensible to the envelope averaging technique. Averaging the reflected signals from the soft sediments has to ensure to preserve the surficial and volume back scatter information. On the other hand, at rough surfaces, the late arrivals are likely to be irregular reflections or noise that has to be averaged to avoid ambiguous results. The initial results of sub layers reflection models were consistent with the data description. However, due to the high resolution of the sub bottom profile, the computation procedure can easily fail by missing sub layers. In order to reduce the probability of missing layers, an overlapping window concept was implemented, where the reflection coefficients are estimated at shorter intervals. The methods investigated here leaves room for further optimization through model adjustment such as signal interference, backscatters and error propagation.GeomaticsAerospace Engineerin