[Paper Review] Surface Waves Prediction Based on Long-Range Acoustic Backscattering in a Mid-Frequency Range

A. V. Ermoshkin, Dmitry A. Kosteev+3 more|arXiv (Cornell University)|May 25, 2022

Underwater Acoustics Research37 references10 citations

TL;DR

This study demonstrates the feasibility of predicting significant wave height and peak wave frequency using long-range acoustic backscattering in the 1–3 kHz band. By analyzing Doppler spectra of reverberation signals from a mid-frequency sonar experiment in the Black Sea, the authors employ a XGBoost-based machine learning model trained on physical features to estimate surface wave parameters with good agreement to in-situ measurements, marking a novel approach to remote ocean surface monitoring using underwater acoustics.

ABSTRACT

Underwater acoustic echosounding for surface roughness parameters retrieval is studied in a frequency band that is relatively new for such purposes. During the described 2-weeks sea experiment, 1–3 kHz tonal pulses were emitted from an oceanographic platform, located on the northern Black Sea shelf. Doppler spectra of the resulting reverberation were studied. The frequency band of the acoustic system, selected for this study, is chosen due to the fact that the sound propagation range is large enough for remote sensing in a coastal zone, and the resolution cell size does not limit the research. Backscattering of acoustical signals was received for distances around two nautical miles. However, it turned to be quite difficult to interpret the obtained data since backscattering spectrum shape was influenced by a series of effects, resulting in a complicated link to wind waves and currents’ parameters. Significant wave height and dominant wave frequency were estimated as the result of such signals processed with the use of machine learning tools. A decision-tree-based mathematical regression model was trained to solve the inverse problem. Wind waves prediction is in a good agreement with direct measurements, made on the platform, and machine learning results allow physical interpretation.

Motivation & Objective

• To investigate the potential of mid-frequency (1–3 kHz) underwater acoustic backscattering for remote sensing of surface wave parameters.
• To address the inverse problem of estimating significant wave height and peak wave frequency from long-range reverberation signals.
• To evaluate the effectiveness of machine learning in interpreting complex acoustic Doppler spectra influenced by multiple physical scattering mechanisms.
• To validate the model’s predictions against direct in-situ measurements collected during a 14-day sea experiment.

Proposed method

• A continuous sonar system transmitted 1–3 kHz tonal pulses from a platform on the northern Black Sea shelf.
• Doppler spectra of the backscattered reverberation signals were recorded at ranges of approximately 2 nautical miles.
• Physical features derived from the Doppler spectra—such as Bragg scattering strength, spectral skewness, and time delays—were extracted as input for machine learning.
• A XGBoost-based regression model was trained to predict significant wave height and peak wave frequency from the acoustic features.
• Feature importance analysis was performed to identify the most influential physical parameters in the prediction model.
• Model calibration was conducted using local in-situ measurements of wind speed, wave height, and current velocity.

Experimental results

Research questions

• RQ1 Can long-range acoustic backscattering in the 1–3 kHz band be used to estimate significant wave height and peak wave frequency?
• RQ2 How do physical features of the Doppler backscattering spectrum correlate with actual surface wave parameters?
• RQ3 To what extent can machine learning models like XGBoost accurately reconstruct wave parameters from complex, noisy acoustic reverberation data?
• RQ4 What role do bubble scattering and Bragg resonance play in shaping the observed Doppler spectra at mid-frequencies?
• RQ5 How does the performance of the model depend on feature selection grounded in physical understanding of scattering mechanisms?

Key findings

• The XGBoost model achieved good agreement between predicted and in-situ measured significant wave height and peak wave frequency, validating the approach for remote wave monitoring.
• The most important features for prediction were identified as spectral skewness (sk_le) and Bragg scattering strength (lv_brag_le), particularly at specific angles and frequencies.
• Spectral features from signals arriving with small time delays (e.g., 0–30 ms) were prioritized for estimating surface state at the center of the coordinate system.
• The model's performance was highly sensitive to time synchronization; incorrect alignment led to meaningless results, underscoring the importance of precise data alignment.
• Despite strong wind conditions, no significant increase in scattering strength was observed, suggesting that bubble scattering may not dominate in the 1–3 kHz band under the studied conditions.
• The absence of clear Bragg peaks in spectra under moderate to rough sea states indicates that Bragg resonance alone cannot explain the full scattering behavior, and other mechanisms may play a role.