Comparison of Regularization Methods for Multipath Mitigation in Mirror- Type Coastal Acoustic Tomography

This study addresses a significant challenge in mirror-type coastal acoustic tomographic (MCAT) systems: the accurate extraction of travel-time information from signals af-fected by multipath interference. While MCAT enables real-time monitoring of ocean currents, multipath propagation in coastal environments can severely reduce measurement accuracy, particularly for non-primary acoustic arrivals. We investigate two regularization methods- Tikhonov (TK) and total variation (TV)-to mitigate this interference and enhance the precision of travel-time estimation. Through systematic numerical simulations and Monte Carlo analysis, we evaluate these methods under varying signal-to-noise ratio (SNR) conditions ranging from −5 to 5 dB. Our results reveal distinct performance characteristics for each approach. TK regularization demonstrates superior accuracy in moderate to high SNR environments (greater than 3 dB), achieving correlation coefficients exceeding 0.93 compared to ground truth measurements. This makes TK particularly suitable for typical MCAT operational conditions, where SNR often exceeds 10 dB. In contrast, TV regularization maintains consis-tent performance across all SNR levels, although its correlation coefficients range from 0.58 to 0.73. This provides reliable measurements even in challenging noisy conditions. These findings offer important methodological guidance for signal processing in MCAT systems, allowing for optimal technique selection based on environmental conditions. For standard operations with good signal quality, TK regularization provides superior precision, while TV regularization serves as a robust alternative in more challenging environments. This research enhances ocean monitoring capabilities by establishing a framework to mitigate multipath interference in MCAT systems.