Publications
Broadband Modal Coherence and Beamforming at Megameter Ranges
Abstract
This thesis develops a method for estimating the normal mode decomposition of broadband signals and uses it to analyze data from the Acoustic Thermometry of Ocean Climate (ATOC) experiment. Normal modes are the eigenfunctions of the ocean waveguide, derived from the frequency-domain wave equation. They are useful in underwater acoustics, particularly matched field processing and tomography, because the lowest modes provide an efficient description of the most energetic arrivals at long ranges. Extracting source or environmental information from the mode signals depends on understanding the effects of internal waves on coherence and the validity of adiabatic approximation. While much theoretical research has been done on long-range propagation of modes in deep water, there have been few opportunities to compare theoretical predictions with experimental measurements.
The first contribution of this thesis is a short-time Fourier framework for estimating broadband signals propagating in the lowest modes of the ocean waveguide. Since previous research has focused primarily on narrowband sources, this work concentrates on broadband processing issues. Specifically, it addresses the fundamental issue of frequency resolution required for mode estimation, analyzes the performance characteristics of two modal beamforming algorithms and explores the time/frequency tradeoffs inherent in STFT mode processing.
The second contribution of this research is a detailed analysis of the low-mode arrivals at megameter ranges using five months of data from the ATOC vertical line array at Hawaii (3515 km range). Short-time Fourier processing of these receptions revealed that each low mode contains a series of arrivals, rather than the single dispersive arrival that would characterize adiabatic propagation. Average coherence times of the mode signals are on the order of 6-8 minutes. The multipath structure changes significantly between receptions at 4-hour intervals, indicating that stochastic methods are required for mode tomography at megameter ranges. A statistical analysis found that modes do retain travel-time information at megameter ranges, e.g., the centroids show the expected dispersion characteristics of a deep water channel. The centroids show statistically significant trends in mode arrival time over the period of the experiment.
© 2000 Massachusetts Institute of Technology and Woods Hole Oceanographic Institution.