Salt-marsh Fairy circles (FC) are enigmatic, quasi-circular structures linked to interacting biogeophysical processes, yet they remain difficult to detect and quantify at scale from conventional RGB imagery. Limited labeled data, transient and variable FC appearance, and severe class-imbalance make single-model machine learning (ML) unreliable for quantitative monitoring. We propose a framework for automatic FC recognition and enumeration on 3-band imagery. A zero-shot foundation model (SAM) segments images into instance-level blocks. Novel distribution-pattern and geometric features, class-equalized losses, weighted resampling, and augmentation are applied within deep-learning (U-Net, Attention-U-Net, Swin-Unet) and ensemble-learning (Random Forest, XGBoost) models. The key innovation is an imbalance-aware Bayesian method that fuses pixel-wise probabilities across models; a counting algorithm then tallies FC instances. We evaluate eight pan-sharpened scenes covering four sites along China’s coast. No individual ML model or standard Bayesian fusion is fully satisfactory. The imbalance-aware Bayesian method improves over the best single model: tight scheme: κ rises from 0.69 to 0.76, F1-score from 70.9% to 75.8% (Class 1) and from 63.5% to 68.2% (Class 2), and AUC from 84.8% to 93.1% and from 78.5% to 84.8%; loose scheme: κ increases from 0.74 to 0.79, AUC from 85.1% to 90.3%, F1-score from 74.3% to 78.6%. The counting algorithm achieves RMSE 1.62 and MAPE 0.33% over 1,135 instances, outperforming DBSCAN. A 22-month case study on Chongming Island captures marsh expansion and dieback dynamics through shifts between FC classes. Our framework delivers reliable FC recognition and enumeration on a small dataset with severe class-imbalance, generalizing across salt-marsh types.

Tidal flat topography is a fundamental attribute affecting inundation dynamics, sediment transport, and ecosystem functioning, yet accurate and spatially consistent large-scale monitoring remains challenging. Here, we leveraged satellite altimetry from the Surface Water and Ocean Topography (SWOT) mission to develop a novel, large-scale framework for deriving tidal flat topography from SWOT data, and demonstrated its capability by generating a high-accuracy, national-scale elevation dataset for China. By combining a percentile-based aggregation of multi-temporal water-surface elevation observations with a tide-constrained, adaptive best-quantile (best-q) reconstruction strategy, followed by linear interpolation for gap filling, we improved both vertical accuracy and spatial completeness. Validation against airborne LiDAR, GNSS-RTK surveys, and ICESat-2 photon data demonstrates robust performance across diverse coastal settings, achieving RMSE = 0.34-0.47 m and R2 = 0.81-0.88 at a horizontal resolution of 100 m. Compared with existing large-scale digital elevation models (DEMs), the SWOT-derived topography not only improves vertical accuracy by over 80% but also providing substantially more complete spatial coverage of tidal flat elevations. Spatial analyses reveal pronounced latitudinal gradients, with higher tidal flats concentrated in low-latitude regions and extensive low-lying flats dominating northern estuarine and deltaic systems. This study establishes a scalable framework for tidal-flat elevation retrieval and provides a foundational dataset to support coastal monitoring and sustainable management.

Reliable tidal flat elevation data with a resolution better than 30 m are vital for understanding macroscale sedimentation processes and storm surge erosion in estuarine deltas. However, the highly dynamic and turbid environments of global estuaries pose significant challenges for traditional manual or uncrewed aerial vehicle (UAV) methods in delta-scale monitoring. Insufficient elevation data currently hinder sea-level rise modeling and coastal management for the world's densely populated, flood-prone river deltas. There is an urgent need to characterize and reconstruct the vertical geomorphology of tidal flats through improved observations and modeling. In this study, we propose a robust tidal flat topography mapping framework using freely available satellite imagery and deep learning (DL). First, we developed an enhanced U-Net model to delineate land-water boundaries and generate inundation frequency maps from time-series Sentinel-2 imagery. A regionally adaptive nonlinear inversion model was then constructed using inundation frequency and ICESat-2 elevation data, allowing us to generate a 10-m resolution tidal flat digital elevation model (DEM) of the Yellow River estuary. The DL model achieved an $F1$ -score of 0.96, and the resulting DEM showed strong agreement with UAV LiDAR data, yielding a root mean square error (RMSE) of 0.18 m. The reconstructed elevations range from -0.51 to 0.96 m, exhibiting a spatial gradient of lower elevations in the north and higher elevations in the southern and eastern sectors. To assess transferability, we applied the method to the radial sand ridge system in the South Yellow Sea. The model achieved a comparable RMSE of 0.52 m without retraining, demonstrating robust generalization across different estuarine environments. This study provides a cost-effective framework for monitoring coastal geomorphological changes and helps fill the data gap in delta-scale topographic mapping for global intertidal zones.

ICESat-2 provides near-global, high-precision elevation observations for intertidal mapping and time-series analysis. However, photon heterogeneity across ground, vegetation, water surfaces, and noise, together with tidal dynamics and dense vegetation, limits the effectiveness of conventional filtering and classification methods. To address this, we propose Lsr-RF, a multi-level adaptive framework that integrates the local sparsity rate with a random forest classifier. The method fuses ICESat-2 multi-scale geometric features with Sentinel-1 polarization and Sentinel-2 spectral features, applying RF-based feature selection and classification to distinguish noise, bare tidal-flat, vegetation, and ground-under-vegetation photons. Evaluated across strong/weak-beam and day/night conditions, Lsr-RF was compared with denoising-oriented baselines (ATL08, DBSCAN, OPTICS) and XGBoost. Lsr-RF improved overall accuracy by 0.52-28.06 percentage points and the Kappa coefficient by 0.02-0.61, and achieved a multi-class classification overall accuracy of 0.99 with a Kappa coefficient of 0.99. These results demonstrate its potential for accurate photon-level classification and broader intertidal wetland mapping applications.

High-resolution topographic mapping of intertidal wetlands is essential for geomorphic analysis, yet existing remote-sensing methods often struggle with vegetation interference, dependence on dense time-series data, and limited representation of fine geomorphic features. We propose a canopy-height-constrained stratified cooperative inversion framework for the entire intertidal wetland, integrating single-phase sub-meter optical imagery (Jilin-1), spaceborne photon-counting LiDAR (ICESat-2), and machine learning. To accurately construct DEM and canopy height model (CHM) training samples in salt-marsh environments, we developed an ATL03 photon-classification workflow combining histogram-based control-point extraction and morphological refinement to generate these samples directly from ICESat-2 ATL03 photons. The retrieved CHM was then introduced as a structural constraint in the DEM retrieval model to support canopy-terrain signal decoupling in vegetated salt-marsh areas. A case study on Chongming Island, Shanghai, China, demonstrated that the DEM retrieval achieved high accuracy on the test set (R² = 0.94, RMSE = 0.28 m) and maintained consistent performance against independent UAV-LiDAR validation data (R² = 0.53-0.77, RMSE = 0.34-0.53 m). The retrieved 0.5 m DEM reproduced regional elevation gradients, tidal-creek networks, and micro-topographic variations across bare flats and vegetated marshes. SHAP analysis showed that elevation retrieval over bare mudflats relied mainly on spectral predictors, whereas vegetated areas exhibited a complementary spectral-texture-CHM structure, with CHM consistently ranking as a mid-to-high predictor (4th-7th). This further supports the role of CHM as an effective structural constraint. By using only single-phase imagery and ATL03-derived DEM/CHM samples, the framework enables intertidal topographic retrieval that includes vegetated areas. It therefore provides an efficient and low-cost pathway for high-accuracy intertidal topographic monitoring under complex environmental conditions and limited image availability.

The mobile laser scanning (MLS) systems enable rapid acquisition of high-definition 3-D point clouds for urban digitization, topographic mapping, and infrastructure inspection. Despite the critical role of point cloud density in quantifying data fidelity and object discriminability, its inherent spatiotemporal variability-arising from the nonlinear interplay of scanning geometry, platform dynamics, and surface topology-has remained inadequately addressed in current metrological frameworks. This study establishes a rigorous mathematical model that quantifies MLS density variations through the interdependent variables: density search radius, scanning distance, angular resolution, platform velocity, pulse repetition frequency, and three angles defining the spatial orientation of the local infinitesimal plane at the target point. Building upon this formulation, we propose the first MLS point cloud density correction method to mitigate heterogeneity caused by varying influencing factors and to derive a new corrected density value for each point that serves as an indicator of target geometry attribute. Experiments conducted across different platforms and environments demonstrate that the proposed method effectively eliminates inhomogeneity in density. The correction procedure achieves an average 61% decrease in the density coefficient of variation (cv) over homogeneous surfaces. The proposed method exhibits strong performance regarding feasibility and generality, offering significant application value in enhancing MLS data interpretation and understanding spatial distribution patterns of point clouds under various circumstances.

Beach surface moisture (BSM) is crucial to studying coastal aeolian sand transport processes. However, traditional measurement techniques fail to accurately monitor moisture distribution with high spatiotemporal resolution. Remote sensing technologies have garnered widespread attention for providing rapid and non-contact moisture measurements, but a single method has inherent limitations. Passive remote sensing is challenged by complex beach illumination and sediment grain size variability. Active remote sensing represented by LiDAR (light detection and ranging) exhibits high sensitivity to moisture, but requires cumbersome intensity correction and may leave data holes in high-moisture areas. Using machine learning, this research proposes a BSM inversion method that fuses UAV (unmanned aerial vehicle) orthophoto brightness with intensity recorded by TLSs (terrestrial laser scanners). First, a back propagation (BP) network rapidly corrects original intensity with in situ scanning data. Second, beach sand grain size is estimated based on the characteristics of the grain size distribution. Then, by applying nearest point matching, intensity and brightness data are fused at the point cloud level. Finally, a new BP network coupled with the fusion data and grain size information enables automatic brightness correction and BSM inversion. A field experiment at Baicheng Beach in Xiamen, China, confirms that this multi-source data fusion strategy effectively integrates key features from diverse sources, enhancing the BP network predictive performance. This method demonstrates robust predictive accuracy in complex beach environments, with an RMSE of 2.63% across 40 samples, efficiently producing high-resolution BSM maps that offer values in studying aeolian sand transport mechanisms.

Coastal inshore areas, recognized as invaluable yet vulnerable, are experiencing shifts between various states due to gradual environmental changes and artificial disturbance. These transitions, however, are often imperceptible with large-scale mapping or through on regional in situ surveying when using traditional techniques. Advanced 2D and 3D technologies, particularly high-resolution remote sensing (HRRS) and LiDAR, offer novel perspectives that unveil fine details and precise vertical 3D structure of coastal ingredients. These technologies enable early, rapid, and accurate identification of significant transient or persistent patterns. Additionally, machine learning (ML), encompassing parametrized algorithms, ensemble learning (EL), and deep learning (DL), provides a unique advantage for automated observation. This work aims to advance the observation of key fine components in coastal inshore areas by designing automated methods and frameworks. It considers both natural and human-made sources as targets. with the focus of Poaceae and marine debris. First, an automated 3D recognition of stalks and leaves for Poaceae in coastal mudflats. Poaceae species (Giant reed and reed) in coastal mudflats hold ecological importance and serve as indicators. However, obtaining their phenotypic parameters like stalks and leaves is challenging. Our new automated, parametrized algorithm recognizes stalks and leaves of individual Poaceae plants in coastal wetlands using terrestrial LiDAR point clouds, leveraging radiometric and geometric features. Second, a new framework for comprehensive surveying of coastal Fairy Circles (FCs). FCs, predominantly formed by Poaceae, are self-organized patterns linked to recovery processes and salt-marsh resilience. Our new framework aims for automated surveying of coastal FCs, utilizing ML methods (which includes state-of-the-art foundation model, EL, and DL methods) on 2D and 3D data (satellite-borne and airborne). It is grounded in clear principles of FCs' definition and dynamics, potentially revolutionizing our understanding of coastal FCs behavior. Third, an automated method for 2D and 3D recognition of marine debris across complex scenarios. Marine debris in coastal environments poses significant ecological and environmental issues and has garnered widespread concern. Our new method detects and extracts marine debris from terrestrial LiDAR point clouds or UAV HRRS imagery, combining calibrated radiometric data with geometric features. Fourth, we have developed a series of mathematical models for instrumentation and data processing to achieve these goals. We proposed generalized rigorous model to mathematically correct the density variation in terrestrial LiDAR point clouds, the novel distribution pattern features, and a model to eliminate the specular effect on UAV LiDAR point cloud intensity.

The transport processes of coastal sediments play a critical role in shaping coastal geomorphology, with sediment properties-such as grain size-being fundamental to understanding morphodynamics. However, the field collection and laboratory analysis of sediments are time-consuming and labour-intensive, posing great challenges for large-scale and rapid monitoring of sediment spatiotemporal variations. Unmanned aerial vehicle platforms, combined with machine learning techniques, offer a promising solution for efficiently capturing and analysing sediment characteristics. In this study, surface sediment samples were collected from Dasha Beach, a sandy beach located along the East China Sea, and a sediment type coding scheme was established to convert text-based sediment types into digitized codes. Using 10 spatial and spectral unmanned aerial vehicle datasets, along with machine learning models and traditional mathematical methods, we predicted five sediment characteristics: sediment types, sediment water content, mean grain size, sorting coefficient and skewness. Among the models tested, Random Forest demonstrated superior performance, achieving an overall accuracy of 95.65% and a Kappa coefficient of 0.78 for sediment type. For the other four continuous variables, the Random Forest model yielded an average R2 of 0.86 and 0.82 on the validation and test sets, respectively, significantly outperforming traditional multiple linear regression. The study revealed five key predictors: near-infrared, red edge, digital surface model, red and slope, underscoring the necessity of integrating spatial and spectral data for accurate predictions. In contrast, variables like intensity, green and NDVI were less relevant in predicting sediment characteristics, particularly in unvegetated areas like beaches. This study highlights an efficient and accurate approach to obtaining high-resolution sediment characteristics, addressing the limitations of traditional sampling and laboratory methods while significantly reducing labour and financial costs. Its application holds considerable potential in diverse coastal environments, including remote or inaccessible regions, offering a robust framework for future sedimentological studies.

High-precision elevation mapping is essential for ecological restoration, marine disaster assessment, and morphodynamic simulation in intertidal zones. Current methodologies are often impeded by an over-reliance on extensive in situ measurements and are typically applicable only to regions devoid of vegetation. In this study, we first propose a novel method for spatially continuous elevation mapping of large-scale muddy intertidal zones within highly turbid estuaries, utilizing features at pixel, neighborhood, and temporal scales from satellite multispectral images. This method utilizes a random forest (RF) to model the relationships between elevations from Ice, Cloud, and Elevation Satellite 2 (ICESat-2) and band, texture, and index features from Sentinel-2, without relying on any supplementary in situ measurements. The innovation and strength of the proposed method lie in the simultaneous incorporation of two temporal features: vegetation occurrence frequency and water inundation frequency. These two features effectively utilize the variations observed in different regions and land covers within the Sentinel-2 image series caused by the unique tide periodic fluctuation phenomenon and elevation trend law in intertidal zones, thereby rendering the method applicable to elevation prediction across the entire spatial range of intertidal zones, rather than being limited to nonvegetated regions. A case study conducted on the muddy intertidal zones of the islands in the Yangtze River Estuary from 2019 to 2023 reveals that the average root mean square errors are 0.33 and 0.69 m for high-mid and mid-low intertidal zones, respectively. The proposed method demonstrates superior performance in terms of vertical accuracy, spatiotemporal resolution, and spatial continuity in comparison to the state-of-the-art waterline detection and inundation frequency methods.

Quantitative estimation and spatial mapping of aboveground biomass (AGB) for salt marsh vegetation are crucial for modeling biogeochemical cycles and assessing wetland carbon stocks. Remote sensing offers a noninvasive method for monitoring vegetation traits over large areas. However, a single technique often cannot simultaneously capture both spectral information and the vertical structure of vegetation, greatly hindering its capabilities in classifying and estimating AGB of salt marsh vegetation. This work introduces a machine learning and heterogeneous data-based approach for AGB estimation by integrating passive multispectral two-dimensional imagery and active light detection and ranging (LiDAR) three-dimensional point clouds acquired from a drone platform. Vegetation indices along with texture features are extracted from multispectral imagery, whereas intensity values and height attributes are derived from LiDAR data. Four machine learning methods, namely, extreme gradient boosting (XGBoost), random forest (RF), support vector machine (SVM), and light gradient boosting machine (LightGBM), are employed to classify vegetation and estimate AGB through the strategic utilization of a carefully derived multispectral-LiDAR feature set. A comprehensive case study of a coastal salt marsh in Chongming Island, China, reveals that (1) No noticeable discrepancies are found for different machine learning models in AGB estimation, with XGBoost achieving the highest accuracy (R2 = 0.9207, MAE = 0.2835 kg/m2, RMSE = 0.3229 kg/m2); (2) Feature types and sensors considerably affect AGB estimation accuracy, with height and texture features having greater influence than intensity features and vegetation indices, and LiDAR outperforms multispectral data; and (3) AGB varies remarkably among species and environments, with Spartina alterniflora being more sensitive to changes in soil moisture and nutrients, while Phragmites australis maintains higher AGB even under unfavorable conditions. The proposed approach offers an alternative and effective strategy for AGB estimation and shows strong potential in quantitatively characterizing the ecological processes of salt marsh vegetation.

Accurate mapping of terrain elevations at a large scale and fine resolution can characterize the detailed surface height and geomorphic changes and is very critical for the studies of the internal motions and external forces of the earth. The emergence of the Ice, Cloud, and Land Elevation Satellite-2 (ICESat-2) offers unprecedented possibilities for global elevation mapping with high vertical accuracy using three-dimensional photon points. However, the ICESat-2 photon points are still sparse in terms of spatial/horizontal resolution, making it unable to satisfy the high-resolution demand of terrain elevation mapping and digital elevation model production. A few previous studies have attempted to estimate elevations/topography in regions with single landscape and landcover (e.g., forest, shallow water, and polar regions) by combining ICESat-2 data with other passive satellite remotely sensed data. However, the potential and capability of ICESat-2 for mapping elevations for spatially continuous large regions with multiple complicated land cover types remains unknown. In this study, a spatially continuous large-scale terrain elevation estimation method is developed under multiple land covers based on the random forest model and the freely accessed satellite data of ICESat-2, Sentinel-1, and Sentinel-2. The core principle is to construct a random forest model that can characterize the complicated relationships of the ICESat-2 ATL03 terrain elevations and their corresponding land cover related polarization characteristics and spectral variables from Sentinel-1 and Sentinel-2, respectively. Integrating the superiorities of the data of these three different satellites enables the proposed method to extrapolate the terrain elevations with decimeter-level vertical accuracy and 10 m spatial/horizontal resolution simultaneously without any prior in situ data or manually set parameters. The proposed method is tested using the elevations from 2021 to 2022 at the third largest island (Chongming Island, Shanghai) in China. The estimated terrain elevations are locally validated with the airborne LiDAR-derived elevations. Moreover, they are compared with the ICESat-2 ATL08 height_terrain_bestfit data and Global Ecosystem Dynamics Investigation L2A elev_lowestmode data from the global perspectives. The predicted elevations exhibit a high correlation with the measured elevations from the two airborne LiDAR validation regions with root mean square errors (RMSE) of 0.34 and 0.59 m. The averaged RMSEs of the predicted elevations at different land covers are 1.26 and 1.18 m when compared with those derived from ATL08 and GEDI L2A, respectively. No remarkable abnormal predicted elevations are observed. This finding suggests the satisfactory robustness performance of the proposed method under different land covers and a relatively good consistency between the predicted elevations and the actual terrain of the entire island. As far as we know, the present work is the first to map elevations at 10 m resolution based only on the newly available satellite active and passive remotely sensed data without any ground truth surveys, manual intervention, and prior knowledge. Different with existing studies for terrain elevation mapping only at single landcovers, the proposed method demonstrates the capability and effectiveness of ICESat-2 for any landforms and landcovers and shows great potential for high-accuracy and high-resolution time-series terrain elevation estimation and updating at regional/national/global scales.

Salt marshes provide diverse habitats for a wide range of creatures and play a key defensive and buffering role in resisting extreme marine hazards for coastal communities. Accurately obtaining the terrains of salt marshes is crucial for the comprehensive management and conservation of coastal resources and ecology. However, dense vegetation coverage, periodic tide inundation, and pervasive ditch distribution create challenges for measuring or estimating salt marsh terrains. These environmental factors make most existing techniques and methods ineffective in terms of data acquisition resolution, accuracy, and efficiency. Drone multi-line light detection and ranging (LiDAR) has offered a fire-new perspective in the 3D point cloud data acquisition and potentially exhibited great superiority in accurately deriving salt marsh terrains. The prerequisite for terrain characterization from drone multi-line LiDAR data is point cloud filtering, which means that ground points must be discriminated from the non-ground points. Existing filtering methods typically rely on either LiDAR geometric or intensity features. These methods may not perform well in salt marshes with dense, diverse, and complex vegetation. This study proposes a new filtering method for drone multi-line LiDAR point clouds in salt marshes based on the artificial neural network (ANN) machine learning model. First, a series of spatial–spectral features at the individual (e.g., elevation, distance, and intensity) and neighborhood (e.g., eigenvalues, linearity, and sphericity) scales are derived from the original data. Then, the derived spatial–spectral features are selected to remove the related and redundant ones for optimizing the performance of the ANN model. Finally, the reserved features are integrated as input variables in the ANN model to characterize their nonlinear relationships with the point categories (ground or non-ground) at different perspectives. A case study of two typical salt marshes at the mouth of the Yangtze River, using a drone 6-line LiDAR, demonstrates the effectiveness and generalization of the proposed filtering method. The average G-mean and AUC achieved were 0.9441 and 0.9450, respectively, outperforming traditional geometric information-based methods and other advanced machine learning methods, as well as the deep learning model (RandLA-Net). Additionally, the integration of spatial–spectral features at individual–neighborhood scales results in better filtering outcomes than using either single-type or single-scale features. The proposed method offers an innovative strategy for drone LiDAR point cloud filtering and salt marsh terrain derivation under the novel solution of deeply integrating geometric and radiometric data.

With the rapid advancement of 3D sensors, there is an increasing demand for 3D scene understanding and an increasing number of 3D deep learning algorithms have been proposed. However, a large-scale and richly annotated 3D point cloud dataset is critical to understanding complicated road and urban scenes. Motivated by the need to bridge the gap between the rising demand for 3D urban scene understanding and limited LiDAR point cloud datasets, this paper proposes a richly annotated WHU-Urban3D dataset and an effective method for semantic instance segmentation. WHU-Urban3D stands out from existing datasets due to its distinctive features: (1) extensive coverage of both Airborne Laser Scanning and Mobile Laser Scanning point clouds, along with panoramic images; (2) containing large-scale road and urban scenes in different cities (over 3.2×106m2 area), with richly point-wise semantic instance labels (over 200 million points); (3) inclusion of particular attributes (e.g., reflected intensity, number of returns) in addition to 3D coordinates. This paper also provides the performance of several representative baseline methods and outlines potential future works and challenges for fully exploiting this dataset. The WHU-Urban3D dataset is publicly accessible at https://whu3d.com/.

Quantitatively characterizing coastal salt-marsh terrains and the corresponding spatiotemporal changes are crucial for formulating comprehensive management plans and clarifying the dynamic carbon evolution. Multiline light detection and ranging (LiDAR) exhibits great capability for terrain measuring for salt marshes with strong penetration performance and a new scanning mode. The prerequisite to obtaining the high-precision terrain requires accurate filtering of the salt-marsh vegetation points from the ground/mudflat ones in the multiline LiDAR data. In this study, a new alternative salt-marsh vegetation point-cloud filtering method is proposed for drone multiline LiDAR based on the extreme gradient boosting (i.e., XGBoost) model. According to the basic principle that vegetation and the ground exhibit different geometric and radiometric characteristics, the XGBoost is constructed to model the relationships of point categories with a series of selected basic geometric and radiometric metrics (i.e., distance, scan angle, elevation, normal vectors, and intensity), where absent instantaneous scan geometry (i.e., distance and scan angle) for each point is accurately estimated according to the scanning principles and point-cloud spatial distribution characteristics of drone multiline LiDAR. Based on the constructed model, the combination of the selected features can accurately and intelligently predict the category of each point. The proposed method is tested in a coastal salt marsh in Shanghai, China by a drone 16-line LiDAR system. The results demonstrate that the averaged AUC and G-mean values of the proposed method are 0.9111 and 0.9063, respectively. The proposed method exhibits enhanced applicability and versatility and outperforms the traditional and other machine-learning methods in different areas with varying topography and vegetation-growth status, which shows promising potential for point-cloud filtering and classification, particularly in extreme environments where the terrains, land covers, and point-cloud distributions are highly complicated.

Terrestrial laser scanning (TLS) can acquire high-precision and high-resolution 3-D point clouds of the scanned targets, and the common technical challenge is how to accurately and efficiently process the massive unorganized point clouds. Density is one of the most fundamental and significant quantities that can be derived for each point by counting the number of neighbors in a given 3-D space. It contains key information of the target geometric features and thus plays an important role in TLS data processing (e.g., classification and feature extraction). However, density usually varies remarkably for different scenes with varied scan geometry (e.g., distance and incidence angle), scan resolution (e.g., horizontal and vertical angular resolutions), and object geometry (e.g., slope, size, and spatial attitude), thereby greatly hindering its full capability in reflecting the geometric discrepancies of different targets. The density must be corrected before reliably used as a proxy for target geometric features. In this study, a generalized rigorous model is proposed to correct the density of single-scan TLS point clouds based on the mathematically deduced relation between density and related influencing factors. The superiority of the proposed model over existing models is that it is rigorously developed based on the instrumental scanning principles and target spatial geometry and can be used for different instruments, scanning scenes, targets, and scanning parameters. The proposed model is verified by a series of indoor quantitative control experiments and three natural scenes with totally different instruments and scanning conditions. The results show that the proposed model can accurately simulate the density variation at different scanning geometries with an average correlation coefficient of 0.9957 between the calculated and actual densities. The effects of all influencing factors on density are significantly removed in complex natural scenes by the proposed model, where the coefficient of variation (CV) of the density from a homogeneous surface after correction is reduced by 72.31%, on the average. The proposed model exhibits good performance in terms of the feasibility, effectiveness, and generality and can derive a corrected density value that is a proxy of target geometric size. Additionally, the proposed model can be conducted on each scan individually before the coregistration in the actual TLS campaign with a number of single scans, which has tremendous application value in facilitating TLS data interpretation.

The increasing inshore marine litters (IML) have been jeopardizing the coastal ecology and environment and have attracted widespread concerns. Nevertheless, the accurate detection and quantitative characterization of IML remain a challenge. In this study, a new method is proposed to automatically detect and extract the IML from terrestrial laser scanning (TLS) 3D point clouds. IML are progressively extracted from the surroundings through four major steps by jointly using the radiometric/intensity information and a series of derived geometric features. First, the intensity data are calibrated by a polynomial model for an initial segmentation according to the spectral differences between the IML and surroundings. Second, a new proposed model is used to calibrate the density data for a further discrimination based on the size discrepancies between the IML and surroundings. Third, a connectivity clustering algorithm is used to group the points into different clusters. Cluster geometric features in terms of the shapes and patterns (i.e., linearity, sizes, and verticality) are constructed to identify the IML. Fourth, a geometric self-repairing procedure is used to retrieve the misclassified IML points. An artificially-arranged scene on a bare mudflat and four natural scenes with different circumstances and IML categories are investigated to validate the proposed method. The overall accuracy and kappa coefficient of the proposed method are averagely 98% and 0.69, respectively. Compared with the classical methods, the proposed method shows good robustness performance in different natural scenes with varied IML categories, vegetation coverages, and environmental disturbances. The proposed method shows great promise in IML spatiotemporal interpretation and provides an alternative tool for the validation of large-scale IML products from space-borne or airborne remote sensing platforms.

Objective Characterized by high phenotypic plasticity, salinity tolerance, and metal tolerance, Poaceae in mudflats and wetlands are considered to have great potentials for ecological restoration, coastal risk response, and climate change indication. Accordingly, in the context of severer climate change and fast-risen global mean sea level, there is a strong and urgent requirement of phenotypic traits extraction and growth monitoring for these plants. Terrestrial laser scanner (TLS) is a novel but effective way for retrieving phenotypic, biochemical, and physical parameters of Poaceae plants in intertidal wetlands. Before retrieving these various parameters, intelligent identification and precise separation for stalks and leaves are required. However, Poaceae plants in mudflats and wetlands are densely growing with tangled and complex leaves, making it more challenging to automatically separate the stalks and leaves. With the challenge above, we propose a new separation algorithm for stalks and leaves of individual Poaceae plants in intertidal wetlands using TLS threedimensional point cloud data. Methods In the present algorithm, reflectance information (intensity data) and several spatial geometric characteristics (i. e., density, normal vectors, and spatial connectivity) are employed. Typically, there is an edge loss or edge effect in the laser scanning data of Poaceae in mudflats and wetlands. This results in low intensity for edge parts of stalk and leaves and intensity data errors on these parts. Additionally, differences in geometry and sizes of stalks and leaves can lead to discrepancies in the number of neighborhood points within a given search radius (e., density). Therefore, corrected intensity and density data can initially be used to separate stalks and leaves. Further, individual Poaceae plants are divided into two different types (i. e., upturned leaves and drooping leaves), and separation is continuously conducted from two different routes based on the geometric differences (i. e., density, normal vectors, and spatial connectivity). The specific procedures of the two routes are subtly different. The fundamental principles of the two routes are based on preliminary separation using normal vectors and density data, and stalk, in which leaf points are eventually classified according to the spatial connectivity logic. Results and Discussions Riegl VZ-4000, a long range full -waveform TLS, is used to obtain the point cloud data of a total of 16 Giant Reeds or Reeds from the western of Chongming Island in Shanghai to test and analyze the proposed method (Fig. 4). To assess the predictive performance of the proposed algorithm convincingly, we quantitatively assess all samples' results using the confusion matrix (Table 1). Hence, the manual separation results are taken as truth reference data. By inputting a single parameter r(a) into the entire algorithm, an averaged overall accuracy of 0.87, and an averaged Kappa coefficient of 0.68 are achieved (Table 2 and Fig. 5). ra is empirically determined following the common stalk size of Giant Reed or Reed, is suitable to all samples in this research. However, when given a large number of samples, it is essential to adjust ra to achieve more satisfactory separation results. In addition, future studies are recommended to address the adaptive estimation of ra to improve the proposed method' s automatic and unsupervised performance. Results show that the proposed method has relatively high accuracy and fairly good robustness. However, because of the complexity of Poaceae morphology, surface heterogeneity in the reflectance (usually caused by withered stalks and leaves and speckles of diseases), and dearth of data points (especially for plants far from the instrument due to occlusion effects), the clustering process of intensity can be over-segmented. Therefore, under those circumstances, spatial connectivity of stalks and leaves may be destroyed, and misclassification will be inevitable. The proposed algorithm only uses some fundamental information or characteristics. Thus, it is more efficient and does not require time-consuming work like some existing methods, such as neural network training, regression statistical analysis, or grid construction. Moreover, the proposed method can be extended to stalk and leaf separation for other Poaceae species (e. g., wheat, maize, sorghum, and bamboo). More deep investigations should be conducted to separate stalks and leaves for natural growing Poaceae in mudflats and wetlands. Combining multiplatform and multi-type remote sensing observations may be a potential solution. Conclusions In this study, a novel separation algorithm is exploratively proposed for stalk and leaves of Poaceae (Giant Reed and Reed) in mudflats and wetlands using TLS three-dimensional point cloud data. Accordingly, an overall accuracy of 0.87 is acquired by setting a single parameter. The proposed method succeeds in providing a technical solution for retrieving phenotypic, biochemical, and physical parameters of Poaceae plants in mudflats and wetlands. It is worth mentioning that only very few existing methods can achieve effective stalk and leaf separation of Poaceae in mudflats and wetlands. The major innovation is that different kinds of spectral and geometric information are fully utilized in the proposed method, enabling the providing of an effective remote sensing solution for vegetation monitoring or biomass observation in estuarine and coastal zones.

Discriminating leaf and wood components in terrestrial laser scanning (TLS) point clouds is a prerequisite for accurately estimating 3-D structural and biophysical attributes of both individual trees and entire forests. However, most existing separation methods are conducted at local (i.e., individual or plot) level. The local level separation methods need a presegmentation of the acquired point clouds, and the separation accuracy and reliability are greatly influenced by forest occlusion effect and point cloud qualities. A new generalized method merely based on differences in geometric features, including curvature, density, and salient features, is proposed in this study for separating leaf and wood components at the TLS single-scan level. A preliminary separation is conducted using the quantity of normal change rate (i.e., surface variation) given that leaf points often demonstrate sharp local curvature changes. Then, separation is continually conducted on the basis of calibrated density data (i.e., number of points in a given radius) because of the scattered orientations and small sizes of leaves. Finally, a new self-adjusting connectivity segmentation algorithm is proposed to group remaining points into different clusters. Leaf and wood clusters are separated in accordance with salient features and sizes simultaneously. Results indicate that derived geometric quantities from curvature, density, and salient features of individual points and segmented clusters can be jointly used to discriminate leaf and wood components effectively and robustly in single-scan TLS point clouds with a mean overall accuracy of approximately 93%. In addition, results show good performance in terms of the insensitivity to distance, instrument type, occlusion effect, and forest composition of the proposed method.