AUTHOR=Li Hongchen , Liu Yuhang , Li Ming , Wang Penghao , Zhu Yuhang , Mao Kefeng , Chen Xi 

TITLE=A deep learning-based reconstruction model for 3D sound speed field combining underwater vertical information

JOURNAL=Frontiers in Marine Science

VOLUME=Volume 12 - 2025

YEAR=2025

URL=https://www.frontiersin.org/journals/marine-science/articles/10.3389/fmars.2025.1551823

DOI=10.3389/fmars.2025.1551823

ISSN=2296-7745

ABSTRACT=IntroductionThe sound speed in the ocean significantly influences the propagation characteristics of underwater acoustic signals. Rapid acquisition of underwater three-dimensional (3D) sound speed fields is essential for target detection, acoustic communication, and underwater navigation. The usual used single empirical orthogonal function (sEOF) method, which reconstructs sound speed profiles (SSP) by establishing statistical relationships between empirical orthogonal coefficients of SSP and sea surface environmental factors, may have several limitations: (1) The principal modes extracted by the EOF method may lose some sound speed information, resulting in low reconstruction accuracy; (2) The grid-by-grid inversion of SSP is computationally inefficient for acquiring large-scale 3D sound speed fields; (3) Oceanic dynamic activities cause disturbances in the sound speed field, and relying solely on sea surface environmental information can limit the accuracy of full-ocean-depth sound speed inversion.MethodIn this paper, we propose a region-oriented reconstruction model named 3dCNN-DEN for 3D sound speed fields using the Convolutional Neural Network (CNN). The model utilizes multi-source satellite remote sensing data and CMEMS temperature-salinity reanalysis data, simultaneously incorporating sea surface environmental factors (SST, SLA, and EKE) and underwater information (average density) as inputs. The key innovation lies in integrating both sea surface and underwater vertical density information to enhance the accuracy of 3D sound speed field reconstruction.ResultThe results showed that the 3dCNN-DEN achieves an average root mean square error (RMSE) of 0.7572 m/s and an average mean absolute error (MAE) of 0.5759 m/s, significantly outperforming conventional EOF-based methods. Incorporating underwater average density improves reconstruction accuracy, showing a 77.1% and 60.3% improvement over the sEOF-r and sEOF-CNN methods, respectively.DiscussionThe 3dCNN-DEN model significantly improves the accuracy and computational efficiency of sound speed reconstruction by fully leveraging the vertical structural characteristics of the marine environment. Unlike the EOF method, it avoids information loss caused by mode truncation. These advancements provide a novel perspective and technical approach for achieving more accurate 3D sound speed field reconstruction.