DeBruin, Darryl L.

In this dissertation, a fuzzy logic impedance inversion model is developed to classify marine sediments. Expert knowledge and fuzzy decision making constrain the inversion procedures to the resolving ability of the transmitted. The model is validated by comparing the estimated impedance profile with the measured impedance profile. A coherent surface scattering and incoherent volume scattering model are incorporated into a single geoacoustic scattering model that is applied to acoustic subbottom measurements. The reflected signal is modeled as the convolution of the transmitted processed wavelet and the impulse response of the sea bottom. The impedance of the acoustic return is inverted at the layer interfaces and the volume scattering strength is measured between layer interfaces. The model is applied to acoustic subbottom measurements obtained by an X-STAR subbottom profiler sonar system. The inversion techniques are developed for a 2-10 kHz 20 msec swept FM pulse. A fuzzy logic layer tracking procedure identifies the coherent surface scattering layer interfaces in a subbottom profile image. The peak amplitudes and locations are used as fuzzy inputs in the layer tracking rule base. The rule base determines which peak is assigned to the layer when two peaks compete for assignment or which layer is assigned to the peak when two layers compete for assignment. The fuzzy event detection algorithm estimates the impulse response of the acoustic return by complex least squares fitting parts of the transmitted wavelet with sections of the acoustic return. Reflectors are iteratively identified and removed from the return and the residual return is reprocessed. The detection procedure is constrained by the resolving ability of the matching signals and the peak envelope shape of the acoustic return. A genetic algorithm allows up to five low error reflector estimates to be processed until converging on the correct estimated impulse response (the tree branch whose summed error is minimized). The impedance is correlated with sediment bulk density by empirical relation. Experimental results validate that the fuzzy logic impedance inversion model reliably estimates the impedance of the sea bottom. The estimated impedance profiles of fifty acoustic returns are averaged and compared with measured impedance values.

An acoustic measurement system is developed to estimate the compressional wave attenuation of marine sediments in real time. A chirp sonar transmits filtered digital reflection data to a signal processing computer that processes the data on an AT&T DSP32C chip. The signal processing computer estimates and displays the center frequency of the processed pulse as it is attenuated by the ocean sediments. Wavelet modelling establishes the relationship between the center frequency shift and relaxation time, from which the sediment type and compressional wave attenuation are determined. Frequency contours from two different data sites demonstrate that the system is able to reliably estimate sediment type and compressional wave attenuation. Error introduced by noise is below 1% for noise levels less than 0.1 of the normalized processed signal. Random error in the estimates is minimized by determining reliable frequency values and by ensemble averaging the values.