The Effect Of Bottom Attenuation On High Frequency Source Localization In Shallow Water

High frequency signals are more difficult to model due to sound scattering by the random media and rough surface,consequently source localization using matched filed processing(MFP)becomes more and more difficult as the frequency increases.The frequency difference MFP was originally proposed by Worthmann et al,can effectively downshift the signal and compensating the random medium effect,and since then there have been many papers published on this subject with some degrees of success with experimental data.It is noted that the frequency-difference signal is in theory not the same as a low frequency signal at the difference frequency.In the original paper,it was argued that the frequency-difference signals may approach the low frequency signal when the frequency-difference signals are averaged over a wide band signal.The difference between the two signals(the mismatch)and the conditions under which the difference might diminish have not been studied(systematically)before.In this thesis we investigate the effect of bottom attenuation on the mismatch between the two(frequency-difference and low frequency)signals.It is observed that bottom attenuations will affect the signal modal/ray distributions differently at different frequencies and consequently the goodness of source localizationWe first study the multipath structure of the frequency-difference and low frequency signals.For this purpose,we study the channel impulse responses(CIRs)of the frequency-difference signals and the low frequency signals for a fixed bandwidth.We conduct this study first using the KAM11 environment similar to that used by Worthmann et al.,and second using the Florida experimental environment.(The reason for considering the Florida experiment is that there exist actual measurement data for both low and high frequencies,making the comparison of simulation and data meaningful.)For these two environments,we calculate the CIRs for different bottom attenuations.It is well known that multipath arrivals suffer higher loss at higher frequencies and the loss increases with increasing bottom attenuations,Consequently,the multipath arrival structure or the CIRs is expected to be different between the frequency-difference and low frequency signals.It is found that the mismatch becomes smaller as bottom attenuation increases but there still exist some differences.Next,we study source location using the frequency-difference MFP with different bottom attenuations.To understand the localization results,we study the mode content(structure)of the frequency-difference and low frequency signals.One finds that there exist many more modes in the low frequency signals than in the frequency-difference signals.The reason is that higher modes are attenuated more at higher frequencies.This presents a mode mismatch between the two signals.Again,the mismatch decreases as bottom attenuation increases.For source localization,Worthmann et al.used only three modes in the replica field,because there is less mismatch between the low order modes.The consequence is that the main lobe resolution is poor using only three modes.Using more modes,say 10 modes,it is found that main lobe is sharper but there exist high level sidelobes,and hence sometimes incorrect source localization,due to the mismatch between high order modes.The effect of mismatch on source location is less for bottom with higher attenuation.In view of the mismatch problem,we propose a match frequency-difference field processing(matching two frequency-difference fields),where both the data field and replica(or modeled)field is consisted of the frequency-difference field,hence there exists in principle no mismatch problem.We show that the performance of matched frequency-difference field is much better than frequency-difference MFP.Specifically the main lobe with is ten times smaller and the side lobe level 10dB smaller.