Full-waveform airborne LiDAR bathymetry (ALB), which provides waveforms and point clouds, has become an essential technology for shallow water surveys. However, weak seafloor echoes are challenging to detect accurately because of waveform feature confusion caused by the complex measurement environments. To address this issue, waveform feature importances, feature histograms, and feature spaces of 14-dimensional waveform features are conducted to analyze the waveform feature confusion. Then, a random forest with optimized thresholds (RFOT) is proposed to detect normal seafloor echoes and weak seafloor echoes. Finally, waveform sharpening and condition screening are used to extract the seafloor echoes for overlapping waveforms in very shallow waters. The proposed method was verified with 14 swaths obtained by the Optech Aquarius system around Wuzhizhou Island. The results show that the energy features (area under curve, amplitude, etc.) can better discriminate the difference between weak seafloor echoes and noise than the shape features (RL area ratio, kurtosis, etc.). The number of seafloor echoes detected by the proposed method increased by 148.86% compared with the Aquarius system. The reference data prove that seafloor points detected by the proposed method are accurate and effective. Thus, this contribution effectively improves the bathymetric performance of the ALB system.

Seafloor classification and mapping are crucial for understanding the composition and spatial distribution of submarine materials. Currently, data-driven modeling of underwater acoustic remote sensing data is widely adopted in seafloor classification. However, as research and applications expand, limitations of data-driven approaches—such as the lack of theoretical support, weak interpretability, and poor transferability—have become evident. In contrast, inherent geoacoustic properties of the seafloor, such as roughness and acoustic impedance, can be quantitatively inverted using acoustic scattering models. These properties exhibit objectivity, determinism, and interpretability, thereby providing robust constraints for data-driven modeling. Consequently, a novel seafloor classification strategy is proposed that integrates acoustic inversion with data-driven modeling. Firstly, a geoacoustic inversion model based on the angular response (AR) curve is established to obtain seafloor roughness and acoustic impedance as property features. Then, these features are fused with data features derived from acoustic backscatter mosaics and seafloor digital elevation models (DEMs). Finally, a deep neural network (DNN) is constructed for training, testing, and classification. To evaluate performance, the proposed method was applied to three survey areas with varying seafloor terrain complexity and validated by field data (seafloor photographs). Compared to purely data-driven approaches, integrating property features improved the overall classification accuracy by 4.6%, 3.9%, and 5.4%, respectively. Experimental results demonstrate that the proposed method effectively exploits seafloor geoacoustic properties, exhibiting strong generalization and transferability across diverse marine environments. This study provides a valuable reference for integrating acoustics theory into seafloor classification methods.

Marine transportation constitutes a vital segment of international trade logistics. Recognizing marine carrier ship types is essential for the governance and efficiency of the marine transportation sector. Recognizing ship types by analyzing Automatic Identification System (AIS) trajectory data represents a fundamental aspect of trajectory classification within Intelligent Transportation Systems (ITS). Due to their data representation limitations, traditional CNNs and RNNs struggle to simultaneously process the topological relationships and the attributes of trajectory points. This restriction significantly hinders the learning of micro-activity behaviors of ships. To overcome this, we introduce Graph Isomorphic Network (GIN), a new variant of graphical neural networks with enhanced representation capabilities. Therefore, this study develops a GIN-based framework for graph representation learning. Initially, ship trajectories are represented as vector graph structures, with both original and derived features incorporated into the node embeddings. Subsequently, a GIN-based model for graph representation learning is employed to extract features necessary for ship type recognition. The model’s performance is evaluated through experiments on ground truth data from the Gulf of Mexico and the New Jersey Bight, achieving recognition accuracies of 93.95% and 92.33%, respectively. The high accuracy and robustness of the model substantiate the effectiveness of the proposed method in ship type recognition. This research illustrates that GIN-based graph representation learning can efficiently capture trajectory features, offering a valuable reference for motion trajectory representation extraction and pattern recognition in diverse geographical contexts.

The amplitude of internal solitary waves (ISWs) is a crucial parameter characterizing their properties. Leveraging machine learning and optical remote sensing images for ISW amplitude inversion has proven highly efficient. However, determining the best input features in the inversion model is often overlooked. This study addresses the feature selection problem in ISW amplitude inversion using a random forest (RF) method. The peak-to-peak distance and relative grayscale differences significantly influence ISW amplitude inversion. When solely using imaging features for ISW amplitude inversion, more significant errors are observed for low-amplitude ISWs due to their weak modulation. In amplitude inversion, selecting the dimensionless ISW amplitude for the output is necessary because it better represents the amplitude magnitude. We find that adding stratification parameters improves the inversion effect, especially the depth ratio. Thus, the impact of physical mechanisms on ISW amplitude inversion is pivotal, and incorporating more hydrological parameters as inputs would lead to further improvements in ISW amplitude inversion.

Achieving high-precision, high-resolution monitoring of nearshore water depth is essential for addressing marine disasters and environmental variations. Synthetic Aperture Radar (SAR) imaging offers the advantage of all-day, all-weather observations of coastlines, and imaging is unaffected by water quality. The current depth inversion methods typically exhibit an MRE of around 10%, with spatial resolution typically ranging from hundreds of meters to kilometers. However, Random Forest(RF) can leverage extensive data and complex algorithms to integrate the high resolution of SAR images and the high precision of in-situ data into the inversion model. To address this, we have employed ETOPO2022, multibeam bathymetric, and SAR images to create a depth inversion dataset comprising 542588 data points. In order to leverage this dataset effectively, we implemented an RF model for depth inversion from satellite images. During the model establishment process, ETOPO2022 data served as the primary training dataset, while high-precision multibeam data compensated for the limitations of low spatial resolution and low accuracy in shallow depths. The inversion model achieved a mean relative error (MRE) of 3.72% and a root mean square error (RMSE) of 2.28m on an independent dataset. When the model is applied to a larger area, the overall trend of the inversion results is accurate. Compared to reanalysis data, the inversion model exhibits higher spatial resolution, approaching 20m×22m. It is worth noting that the model demonstrates a strong inversion capability, especially in challenging shallow water areas. When accounting for extraction errors, the model demonstrated a considerable tolerance for errors.

Seafloor sediment surveys play a crucial role in providing invaluable information and guidance for ocean monitoring and management. In recent years, the combination of multibeam echosounder (MBES) data and field samples has become one of the most widespread approaches for seafloor sediment supervised classification. However, owing to the limitations of low efficiency, high complexity and high cost of seabed in situ sampling, ground truth sample data tend to have a small size, which impedes the training and deployment of classifiers. In light of the aforementioned issues, this paper proposed a seabed sediment sample enhancement method on the basis of superpixel segmentation and active learning. First, we introduce superpixel multi-resolution segmentation to effectively expand the training sample set, which takes into account the different weights of the input features. Subsequently, poor samples in the new sample set are removed using an active learning technique to improve the quality of labels. In the end, 33 original sample points are augmented to a total of 6097 valid sample points. To validate the effectiveness of our strategy, the new sample set is employed for the supervised classification of MBES data in the Southern North Sea, UK. The experimental results show that the classification accuracy of our sample enhancement method reaches 86.67%, which is a significant improvement over that of the original sample set and traditional sample enhancement methods.

Full-coverage and high-efficiency seabed sediment detection and identification are critical elements of digital marine construction that support the three-dimensional and thematic development of maritime spatial geographic information systems. With the development of multibeam echo sounder (MBES), the use of MBES backscatter intensity and bathymetry data to extract backscatter angular response (AR) features has increased. Using backscatter intensity features and seabed terrain features for classification is an effective way to achieve large-scale seabed sediment classification. However, it was still limited by small sample size problems and the poor stability of the classification model due to the complexity of performing seabed sediment sampling. In response to the above issues, this article proposes a sample enhancement method based on simple linear iterative clustering (SLIC) superpixel segmentation to address the problems. First, a superpixel-based sample homogeneity expansion method is combined with multibeam backscatter intensity images to achieve adaptive sample range selection. Then, a random forest (RF)-based model is constructed using MBES backscatter intensity and seabed terrain features. To assess the model’s validity, the experiment uses data from an extensive MBES survey and field sampling information from the Celtic Sea, U.K. It achieves accurate predictions for the area’s eight sediment types. The experimental results show that the proposed method expands 60 groups of original sample points to 9293 groups of valid sample points and achieves an overall classification accuracy and kappa coefficient of 84.05% and 0.81 for the seabed sediment, respectively. In addition, the classification accuracy is significantly better compared to the traditional methods.

Airborne LiDAR bathymetry (ALB) is a useful technology for seamless topobathymetric mapping, offering high acquisition rate and point density. However, in very shallow waters (90%). In the simulated dataset, the root mean square error of the laser travel time between the estimated and truth values is 0.22 ns (corresponding to 2.5-cm slant range). The results indicate that the proposed method provides a new solution for filling the bathymetric gap in very shallow water, which is very essential for topobathymetry mapping.

Airborne light detection and ranging (LiDAR) bathymetry (ALB) and multibeam echo sounder (MBES) are both active remote sensing technologies that are complementary in terms of survey scope. The registration of ALB and MBES data can provide complete overwater and underwater geoinformation on a measurement target. However, in the overlapping area of the ALB and MBES data, there are different point densities and few identifiable structure features. Although the existing multiplatform registration strategies can provide good results for overwater datasets, they are difficult to adapt for the registration of ALB and MBES data. Therefore, to address these problems, a new registration method for ALB and MBES datasets is proposed in this letter. First, a triangulated irregular network (TIN) is constructed with control points extracted from the MBES data. Then, the features of the TIN facets are extracted to identify the data gaps. Finally, the transformation parameters are iteratively calculated by minimizing the distances between the ALB points and MBES TIN facets. Five samples with different characteristics captured around Yuanzhi Island in the South China Sea are selected to evaluate the performance of the proposed method. The mean root mean square error (RMSE) of the five samples is approximately 0.2 m. The results indicate that the proposed method performs well for the registration of ALB and MBES datasets, with advantages in accuracy and robustness.

Global Navigation Satellite System (GNSS) buoys were used to retrieve the significant wave heights (SWHs) and dominant periods of waves in the Qingdao coastal waters off China. The precise point positioning (PPP) and postprocessed kinematic (PPK) techniques were used to obtain the absolute motion of the GNSS buoys. Even though PPP has relatively low absolute positioning accuracy, this accuracy has a minimal influence on the spectrum in the wave frequency band. SWH values calculated from the PPK and PPP modes are nearly identical, with correlation coefficients and symmetric regression slopes both higher than 0.99. The SWH values calculated from PPP also show good agreement with a dedicated wave buoy, with a correlation coefficient of 0.935 and a difference of 0.82 cm ± 4.63 cm. A GaoFen-1 satellite image was used to assess the GNSS PPP wave spectrum via the dispersion relationship; both exhibit wave spectra with two peaks, consistent with the wave characteristics in the area.

Acoustic remote sensing with multibeam echo-sounder systems (MBESs) has been extensively used for coastal and ocean survey works. The imperfect integration of multibeam echo sounder and motion sensor can introduce integration errors, which manifest as high-frequency wobbles in swaths and hinder the accurate expression of high-resolution seabed topographic maps. To address this issue, a precise method is developed to calibrate these integration errors based on a simplified georeferenced model. First, an equivalent attitude coordinate (EAC) system is defined to represent mutual transformation between equivalent attitudes and the beam launch vector; then, with the help of equivalent attitudes, a simplified footprint georeferenced model is deduced that considers the effect of the parameterized integration errors (including time delay, motion scale, yaw misalignment, and lever arm errors); finally, in a selected flat region, integration errors are inverted by regressing the bathymetric data to the corresponding fitted plane via the differential evolution (DE) algorithm. The results indicate that the proposed method can effectively eliminate wobbles in multibeam bathymetries caused by single or multiple integration errors in both shallow- and deep-water areas. By comparing the data before and after calibration with thein situmeasurements, the accuracy of the calibrated shallow- and deep-water data is controlled within approximately 0.15% and 0.06% of the water depth, respectively.

The ability to accurately map the seabed sediments plays an important role in seabed habitat development and stakeholder decision-making. In conventional seabed sediment classification methods, maps of seabed sediment are provided in categorical form (sediment classes). Therefore, the prediction of the sediment compositions in multibeam observational units has become a difficult issue in using conventional methods. To tackle this challenge, a new strategy is developed to realize the subpixel decomposition of seabed sediments. A key attribute of the proposed sediment decomposition model is that it utilizes spatial–spectral information provided by multibeam backscatter angular responses (ARs). First, an AR feature extraction method utilizing a bidirectional sliding window is proposed and a$K$-means clustering algorithm is used for segmentation. Second, a deep-sea sediment decomposition model based on the fuzzy method is constructed by selecting experimental samples that are distributed within a single clustering region. This model inverts the abundance of each sediment composition in the form of membership degrees. Finally, deep-sea multibeam survey data collected from the central Philippine Sea are used for verification. The overall mean square error and coefficient of determination reach 0.043 and 0.856, respectively. The experimental results show that the new method can accurately decompose deep-sea sediment compositions, thus providing a new technique for deep-sea acoustic sediment remote sensing and quantitative analysis.

Very high resolution (VHR) remote sensing images contain various multi-scale objects, such as large-scale buildings and small-scale cars. However, these multi-scale objects cannot be considered simultaneously in the widely used backbones with large downsampling factor (e.g. VGG-like and ResNet-like), resulting in the appearance of various context aggregation approaches such as fusing low-level features and attention-based modules. To alleviate this problem caused by backbones with large downsampling factor, we propose a feature-selection high-resolution network (FSHRNet) based on an observation: if the features maintain high resolution throughout the network, a high precision segmentation result can be obtained by only using a 1×1 convolution layer with no need for complex context aggregation modules. Specifically, the backbone of FSHRNet is a multi-branch structure similar to HRNet where the high-resolution branch is the principal line. Then, a lightweight dynamic weight module, named feature-selection convolution layer (FSConv), is presented to fuse multi-resolution features, allowing adaptively feature selection based on the characteristic of objects. Finally, a specially designed 1×1 convolution layer derived from hypersphere embedding is used to produce the segmentation result. Experiments with other widely used methods show that the proposed FSHRNet obtains competitive performance on ISPRS Vaihingen dataset, ISPRS Potsdam dataset and iSAID dataset.

Bathymetric Multibeam echo sounder systems (MBESs) are the most effective and reliable way to survey the earth’s seafloor. However, multibeam bathymetric data inevitably contain different types of outliers due to measurement characteristics and complex underwater environments. The traditional automatic approach to eliminate outliers may lead to more than one of the questions of reliability, limitation, and efficiency, respectively. This paper offers an algorithm aiming to detect different types of outliers by considering their characteristic of distribution and distance of them, rapidly. First, a coding octree based on Morton code is built to guarantee perfect efficiency and space division. Second, coarse outlier removal is performed by octree-based voxelized representation of the bathymetric data, and outliers far away from the seafloor will be detected and eliminated by connected component labeling. Third, fine outlier removal is employed to delete outliers connected closely to the seafloor by the improved morphological method based on the combination of the k-d tree and octree. Experimental results show that the proposed algorithm can achieve promising results. The percent of outliers that are detected by the hand-edit method that is regarded as a reference result is 5.06%. 4.70% of points in a total number of 3645541 points are detected in our method. Compared with other classical filtering methods, the intelligent method for detecting different types of the outlier from coarse to fine attains favorable performance in a reasonable time, avoiding over-filtering, and demonstrates high reliability for multibeam bathymetric point cloud.

For mapping the near-shore seafloor bathymetry, retrieving depth information using multispectral satellite image is highly cost-effective. To effectively detect and characterize the bathymetry variation, accurate and reliable information about the uncertainty of the derived depths is critical. In estimating the uncertainty of the resulted satellite-derived bathymetry (SDB), the conventional homoscedasticity assumption states that the error variance of the observations is constant across different depth ranges. However, this assumption is violated due to the influence of various environmental factors inherently correlated with depth. In this article, we develop a data-driven approach to extract the depth-dependent pattern of observation error. The residual information of the regression is analyzed to model the influence of the depth on the uncertainty of retrieval bathymetry, while nonlinearity and outliers are also considered. This results in a more realistic estimate of SDB accuracy. Our experimental results reveal that the observation uncertainty is significantly correlated with the depth in the bathymetry retrieval process. It is also shown that the presented algorithm effectively captures the depth-dependent pattern of the observation uncertainty, and further provides a more realistic characterization of the uncertainty information of SDB.

To develop the data mining capability of airborne LiDAR bathymetry (ALB) waveform information, the waveform decomposition and feature extraction algorithm of airborne LiDAR bathymetry is proposed in this paper. This method uses the “separate” and “combine” two-step optimized waveform decomposition algorithm to decompose the bottom contribution waveform and extract 7 waveform feature parameters. To verify the effectiveness of this algorithm, it is applied to Yuanzhi Island ALB data. The experimental results showed that the average root mean square error (RMSE) of waveform decomposition is 11.75; the average coefficient of determination (R2) is 0.983, and the extraction time for every 500 sets of data is only 0.24 s. After classing coral reefs using the extracted ALB waveform features, the overall accuracy of coral reef information extraction are 2.27%,3.84%, and 4.02% improved with K-Nearest Neighbor (KNN), Support Vector Machine (SVM), and Random Forest (RF) methods, respectively, compared with the previous waveform features.

This contribution identifies the uncertainty sources influencing the component uncertainties for an airborne LiDAR bathymetry (ALB) measurement and presents the models for various component uncertainties (arising from the device, environment, and target for ALBs). Since various instrumental and environmental factors introduce vertical and horizontal uncertainties in ALB data, these uncertainties should be first analyzed and then precisely modeled to ensure the accuracy of the measurements. For this purpose, ten different effects that influence the accuracy of ALB data are systematically analyzed and modeled for four aspects in this article: the device aspect (laser pointing deflection, trajectory uncertainty, and boresight/lever arm offset), environmental aspect (atmospheric limitation, refraction on the sea surface, refraction in water, scattering in water, and water level fluctuation), target aspect (irregular bottom), and other aspect (accuracy of coordinate transformation model). In addition, the effect of the laser spot size is also discussed. To verify the presented uncertainty models, an ALB survey was operated around Yuanzhi Island in the South China Sea. For a water depth of 10 m, the theoretical overall root-sum-squares (RSS) for ten different effects of an ALB system (approximately 34 cm calculated using the total vertical uncertainty (TVU) models) is generally in accordance with the actual performance of the ALB data (approximately 39 cm) performance. The difference is mainly attributed to the limited accuracy of the ground truth data, and the difference between the water depth and laser ranging is reasonable. In this process, the topography data in the same region captured by a shipborne multibeam echo sounder (MBES) were used as the ground truth. The results indicated that for the typical ALB system, the laser pointing uncertainty and refraction uncertainty on the sea surface are primary uncertainty sources and should be corrected in a higher priority to meet the seafloor topographic accuracy demand of the International Hydrographic Organization (IHO) Standards for Hydrographic Surveys (S-44). The proposed uncertainty models can be used not only to guide the actual measurement of an ALB system but also to provide the uncertainty correction reference for ALB data postprocessing.

An algorithm is described and tested to provide accurate and robust deep-sea seafloor classification based on the backscatter data derived from a multibeam bathymetry system. This article focuses on significant heterogeneity in the deep-sea backscatter strength (BS) data. The angular response curve information is decomposed into different units, and BS data are grouped on the basis of the incidence angle to address the heterogeneity in the across-ship direction. Subsequently, a sliding window is applied on BS data in each group, and a robust estimation method is used to address the potential heterogeneity in the window during feature extraction. Thereafter, the extracted features are learned by fuzzy c-means (FCM) to obtain a clustering solution. In the learning process, the features of each group are learned by an independent FCM. The modified FCM algorithm is used to cluster each group of data to handle unbalanced backscatter data sets. With this procedure, heterogeneity in BS data can be accounted for, which is universal in deep-sea survey application. Finally, the results of the different groups are merged to obtain a global label set for the survey region. The method was tested on the multibeam data collected from an offshore region around the Kyushu-Palau Ridge. Monte Carlo simulation was performed to evaluate the performance of the robust method. Computational results demonstrate that the improved algorithm can address the heterogeneity in BS data efficiently and provide an accurate classification solution in the deep-sea survey environment.

Geographic information describing coral reefs plays an important role in constructing electronic chart systems and protecting the ecological environment of the ocean. To derive geographic information of coral reefs more effectively, this paper proposes a methodology to detect coral reefs by combining airborne LiDAR bathymetry (ALB) bottom waveform and bathymetric feature data. A feature vector was established by deriving bottom waveform variables (the peak amplitude, pulsewidth, area, skewness, kurtosis, and backscatter cross section) and bathymetric variables (the depth standard deviation, slope, bathymetric position index, Gaussian curvature, mean curvature, and roughness). Using a support vector machine classifier, coral reefs were detected by distinguishing two classes (coral reefs and others) on the seafloor. To evaluate the classification performance of coral reefs, the developed method was applied to Yuanzhi Island, South China Sea surveys, and verified by field data (aerial digital camera images and underwater video images). The results showed that the classification overall accuracy of coral reefs can be greatly improved from 80.59%/90.31% when ALB bottom waveform or bathymetric variables features were used separately to 93.57% when using a combination of ALB bottom waveform and bathymetric features. In addition, the kappa coefficient can also be greatly improved from approximately 0.61/0.80 to 0.87. And the new proposed method performs better compared to the current classification method using ALB data to detect coral reefs with an overall accuracy of 90.92% and Kappa of 0.81. This highlights the potential of ALB data, combining waveform data and bathymetric data, for precisely detecting coral reefs in shallow water areas.

Water depth can be measured using airborne LiDAR bathymetry (ALB). However, when the green laser beam passes through the air-water interface, the sea surface slope greatly affects the laser propagation path, significantly influencing the accuracy of the measured seafloor topography. To reduce its influence, a refraction correction method at the air-water interface based on the sea surface profile and ray tracing is proposed. First, the 3-D sea surface profile is fit based on the least-squares criteria and the wave spectrum, using the laser point data reflected by the sea surface. Then, on the basis of the sea surface slope, the geolocation biases of the laser points are corrected by tracing every laser transmission path at the air-water interface. The developed method is used to correct the ALB data collected in the South China Sea, and verified by the topography data captured by a ship-borne multibeam echo sounder. Before the refraction correction, the mean absolute error (MAE) is 14.2 cm, and the root-mean-square error (RMSE) is 17.5 cm. After the refraction correction, the MAE and RMSE decrease to 7.2 and 8.3 cm, respectively. The developed method can effectively improve the bathymetric accuracy of the ALB data.