Acoustic seabed classification (ASC) is a fast and large-scale seabed sediment survey method. In particular, combining it with an automated classifier can theoretically achieve fast automatic seabed sediment classification. However, owing to the cost of sampling, a lack of labeled data for sediment classification based on seabed acoustic images impedes the training and deployment of classifiers. Herein, we use shallow-water, side-scan sonar images collected from the Pearl River Estuary combined with deep learning to study sediment classification and optimization methods for a small dataset of seabed acoustic images. In this paper, we applied different and deeper convolutional neural networks (CNNs) and used grayscale CIFAR-10 for pretraining to achieve large-span parameter migration and improve model performance. The best result in the experiment is a 3.459% error rate achieved by ResNet after fine tuning, verifying the improvement brought by our fine tuning strategy and the deeper models used in such tasks. The results of data enhancement based on generative adversarial networks (GANs) indicated that this method can improve the accuracy of sediment classification; however, the effects of GANs are limited and they are computationally expensive. Overall, our findings resolve, to an extent, the dilemma of using small datasets of seabed acoustic images for sediment classification and provide a framework for future studies on sediment classification, which has a certain significance in helping people better understand the seabed.

Seabed acoustic images are image data mosaics derived from seafloor acoustic backscattering intensity data, which is related to the type of sediment covering the seabed. Therefore, submarine sediment classification can be realized using seabed acoustic images, and has been studied extensively. Recently, deep learning has also rapidly advanced, in particular, deep convolutional neural networks (CNNs) are now being used to achieve remarkable results in the field of image processing—showing that they are well-suited for image classification tasks. Previous studies have used GoogleNet to classify large-scale side-scan sonar data, with some sediments being well-classified. However, deep learning is data driven and, theoretically, the greater the depth, the stronger is the learning ability of the feature. It is worth noting that the dataset used for sediment classification can sometimes be small. Hitherto, no related research has analyzed the feasibility and applicability of a CNN classifier for a small-sized seabed acoustic image dataset, so we adopted two different CNN classifier models to conduct the classification experiment in this study. As the results show, the CNN classifier can be applied to the classification of sediments based on a small-sized seabed acoustic image dataset, and the classification performance of shallow CNN was found to be better than that of deep CNN on existing side-scan sonar data. In particular, the accuracy obtained from the results of several sediment classification experiments using a shallow CNN classifier ranged between 93.4% (Sand Wave) and 87.54% (Reef).

The bottom tracking is to confirm the first bottom return signal in each ping to divide the data into the seabed area and the water column area, which is an essential part of side-scan sonar (SSS) data processing. More than that, SSS is usually affected by water flow or suspended solids, resulting in serious noises, which make the bottom tracking more difficult and complex. Some existing methods follow ideas such as abnormal point detection or filtering, resulting in unsatisfactory bottom tracking effects. Inspired by the semantic segmentation mission in deep learning, we innovatively transform the bottom tracking into a binary-class semantic segmentation mission on 1-D SSS data. Following this idea, we propose a semantic segmentation-based bottom tracking method and modify two image semantic segmentation models, SegNet and U-Net, so that they can achieve bottom tracking on 1-D SSS data. Benefiting from the large receptive field brought by convolution and down-sampling, our method has made significant progress in accuracy and antinoise performance compared with the existing method. According to the experimental result, our method has obvious improvement compared with the existing method on the testing set; not only that, our method also performs better on the unused data.

Sea ice is a significant factor in influencing environmental change on Earth. Monitoring sea ice is of major importance, and one of the main objectives of this monitoring is sea ice classification. Currently, synthetic aperture radar (SAR) data are primarily used for sea ice classification, with a single polarization band or simple combinations of polarization bands being common choices. While much of the current research has focused on optimizing network structures to achieve high classification accuracy, which requires substantial training resources, we aim to extract more information from the SAR data for classification. Therefore we propose a multi-featured SAR sea ice classification method that combines polarization features calculated by polarization decomposition and spectrogram features calculated by joint time-frequency analysis (JTFA). We built a convolutional neural network (CNN) structure for learning the multi-features of sea ice, which combines spatial features and physical properties, including polarization and spectrogram features of sea ice. In this paper, we utilized ALOS PALSAR SLC data with HH, HV, VH, and VV, four types of polarization for the multi-featured sea ice classification method. We divided the sea ice into new ice (NI), first-year ice (FI), old ice (OI), deformed ice (DI), and open water (OW). Then, the accuracy calculation by confusion matrix and comparative analysis were carried out. Our experimental results demonstrate that the multi-feature method proposed in this paper can achieve high accuracy with a smaller data volume and computational effort. In the four scenes selected for validation, the overall accuracy could reach 95%, 91%, 96%, and 95%, respectively, which represents a significant improvement compared to the single-feature sea ice classification method.

Seafloor topography data are fundamental for marine resource development, oceanographic research, and maritime rights protection. However, approximately 75% of the ocean remains unsurveyed for bathymetry. Sole reliance on shipborne measurements is insufficient for constructing a global bathymetric model within a short timeframe; consequently, satellite altimetry-based inversion techniques are essential for filling data gaps. Recent advancements have improved the variety and quality of satellite altimetry gravity data. To leverage the complementary advantages of multiparameter gravity data, we propose a 1D convolutional neural network based on a convolutional attention module, termed the Attention Channel 1D Convolutional Network (AC1D). Results of a case study of the Mariana Trench indicated that the AC1D grid predictions exhibited improved agreement with single-beam depth checkpoints, with standard deviation reductions of 6.32%, 20.79%, and 36.77% and root mean square error reductions of 7.11%, 22.82%, and 50.99% compared with those of parallel linked backpropagation, the gravity–geological method, and a convolutional neural network, respectively. The AC1D grid demonstrated enhanced stability in multibeam bathymetric validation metrics and exhibited better consistency with multibeam bathymetry data and the GEBCO2023 grid. Power spectral density analysis revealed that AC1D effectively captured rich topographic signals when predicting terrain features with wavelengths below 6.33 km.

Only approximately 20% of the global seafloor topography has been finely modeled. The rest either lacks data or its data are not accurate enough to meet practical requirements. On the one hand, the satellite altimeter has the advantages of large-scale and real-time observation. Therefore, it is widely used to measure bathymetry, the core of seafloor topography. However, there is often room to improve its precision. Multibeam sonar bathymetry is more precise but generally limited to a smaller coverage, so it is in a complementary relationship with the satellite-derived bathymetry. To combine the advantages of satellite altimetry-derived and multibeam sonar-derived bathymetry, we apply deep learning to perform multibeam sonar-based bathymetry correction for satellite altimetry bathymetry data. Specifically, we modify a pretrained VGGNet neural network model to train on three sets of bathymetry data from the West Pacific, Southern Ocean, and East Pacific. Experiments show that the correlation of bathymetry data before and after correction can reach a high level, with the performance of R² being as high as 0.81, and the normalized root-mean-square deviation (NRMSE) improved by over 19% compared with previous research. We then explore the relationship between R² and water depth and conclude that it varies at different depths. Thus, the terrain specificity is a factor that affects the precision of the correction. Finally, we apply the difference in water depth before and after the correction for evaluation and find that our method can improve by more than 17% compared with previous research. The results show that the VGGNet model can perform better correction to the bathymetry data. Hence, we provide a novel method for accurate modeling of the seafloor topography.