Date of Award
1-1-2025
Document Type
Dissertation
Degree Name
Ph.D. in Physics
First Advisor
Likun Zhang
Second Advisor
Roger Waxler
Third Advisor
Joel Mobley
School
University of Mississippi
Relational Format
dissertation/thesis
Abstract
Underwater acoustics has substantial implications for climate science, marine ecosystems, environmental monitoring, mineral exploration, and oceanography studies. Accurate underwater sound speed data is crucial for acoustic propagation modeling and applications in ocean environments, including sonar systems. However, due to limited data availability and computational constraints, conventional methods often struggle to deliver real-time, high-resolution mappings of three-dimensional (3D) sound speed fields. This interdisciplinary study integrates remote sensing data, machine learning models, and underwater acoustics to create a comprehensive tool for mapping sound speed across vast oceanic regions. By leveraging readily available sea surface temperature and salinity data from satellite observations, we can rapidly and accurately estimate 3D underwater sound speed. We examine the relationships between surface data and subsurface sound speed and capture these relationships through machine learning model training. We enable real-time, high-resolution sound speed mapping by incorporating spatial and temporal variables. Validation against in-situ profiling and Argo float measurements demonstrates our method’s capability to produce precise 3D sound speed maps across various seasons, regions, and timeframes. This approach may significantly advance underwater sound speed prediction, overcoming traditional limitations. Acoustic propagation modeling further underscores the model’s applicability for various underwater operations, including detection, communication, and noise propagation.
Climate change significantly impacts underwater sound propagation through alterations in ocean temperature and salinity distributions, yet the spatial and temporal evolution of these changes remains poorly understood. This study employs a comprehensive analysis of historical data (1960- 2023) using the IAPv4 dataset and GSW toolbox to identify regions experiencing significant sound speed variations. The results reveal pronounced changes in the Arctic region, particularly near Svalbard, with increased sound speed up to 11 m/s (0.7%) over six decades. The location at the Barents Sea (28° E, 74° N) was selected for further analysis since this region shows significant sound speed changes over seasonal variations. Polynomial regression was applied to historical surface temperature and salinity data to extrapolate oceanic surface conditions for the year 2060 at the selected location. Using these future projections, our machine learning model predicted the corresponding sound speed profiles (SSPs). Comparing past (1960) and future (2060) SSPs revealed significant changes up to 20 m/s (1.3% ), particularly in the upper ocean regions.
Transmission loss simulations were conducted based on the SSPs for both past and future scenarios. In the 1960 data, acoustic waves propagated seamlessly throughout the ocean depth, ensuring uniform coverage. However, in the 2060 scenario, projections indicated the formation of a shadow layer in the upper 100 meters of the ocean. These shadow layers limit sound propagation to the surface layers, while sound below 100 meters continues to propagate effectively. This phenomenon imposes constraints on low- and mid-frequency acoustic applications, altering wave propagation patterns and disrupting the broader uniform coverage observed in the past. These findings offer critical insights for managing marine ecosystems, designing underwater communication systems, and planning naval operations. Additionally, they establish a predictive framework for assessing future changes in underwater acoustic environments, reflecting the impacts of ongoing climate change.
Recommended Citation
Madiligama, Madusanka, "Harnessing AI for Underwater Sound Speed Prediction and Climate Change Variations" (2025). Electronic Theses and Dissertations. 3325.
https://egrove.olemiss.edu/etd/3325