It is an important task to automatically extract water areas from various synthetic aperture radar (SAR) images for hydrological monitoring activities, such as water resource management, flood risk assessment, and drought detection. However, it is challenging to achieve accurate water area extraction, due to image noise, mountain shadows, and land cover interference. To address these limitations, this paper proposes an accurate segmentation network for water area extraction based on the U-Net framework called WASNet. It first integrates Transformer and spatial convolutional networks to cooperatively learn features from both global and local perspectives, which enables effective combination of multiscale features to achieve accurate segmentation of water areas in SAR images. Subsequently, we propose large-kernel generalized attention (LKGA) in each Transformer block of its decoder, to expand its receptive field and enhance global context understanding, for capturing a wide range of feature-associated regions. Furthermore, we design extended local positional encoding (ELPE) to enhance the effectiveness of positional embedding, and add it into the computation of the attention mechanism for each Transformer block, which reinforces the ability to capture water area edge details. Additionally, we devise a multiscale enhanced feed-forward network (MSEFF) to dynamically balance the importance of feature information across different branches, strengthening the multiscale contour details of water areas. Eventually, we verify the performance of our proposed WASNet, by quantitative and qualitative experiments on a representative subset with 1,600 Sentinel-1 SAR image pairs extracted from the publicly available GID-15 data set. Extensive experimental results demonstrate that the WASNet significantly outperforms state-of-the-art water area segmentation networks, particularly in the identification of small branches and edge details, enabling accurate extraction of water areas for understanding climate-driven migration.

Near-real-time quantitative precipitation estimation (QPE) is crucial for flood modeling and disaster early warning. This study proposes a three-stage machine learning framework (TSMLF) for near-real-time QPE using Himawari-8 satellite data and rain gauge observations. The framework introduces a deep imbalance regression (DIR) algorithm to handle the imbalance in precipitation data. Building upon the existing two-stage "classification-estimation" QPE framework, a rainfall intensity classification model is incorporated. Furthermore, 27 combined QPE schemes are designed based on the TSMLF. This novel method is applied in the Xiangjiang River Basin, southern China. The performance of the different QPE schemes is evaluated. Results show that the DIR algorithm is effective for handling precipitation data imbalance, achieving a lower false alarm rate and a higher critical success index compared to the synthetic minority oversampling technique algorithm. Among the QPE schemes designed based on TSMLF, the S1-C41 scheme [i.e., using a rainfall intensity threshold of 4.1 mm/h (90th percentile)] achieves the best performance, with a correlation coefficient (CC) of 0.77 and a mean absolute error of 0.14 mm/h. In the evaluation of a representative rainstorm event, the S1-C41 scheme achieves a CSI of 0.767 and a CC of 0.779, demonstrating superior performance in capturing spatiotemporal precipitation features and identifying small-scale precipitation. The TSMLF significantly outperforms both traditional two-stage QPE models and existing near-real-time precipitation products such as IMERG-E and GSMap_NRT. This study provides new insights into handling precipitation data imbalance and constructing novel QPE structures.

Offshore wind turbines are subjected to complex environmental loads throughout their operational life. Accurately quantifying the contribution of each load component and assessing the inherent randomness of the marine environment are crucial for structural design and reliability assessment. This study establishes a fully coupled numerical model to investigate the dynamic response of a monopile-supported OWT under combined wind, wave, and current loading. A systematic series of load cases, including four control scenarios and eighteen singlevariable environmental conditions, are designed to decouple the effects of individual load components. The results demonstrate that wind load is the dominant factor governing the global dynamic response, while wave load significantly influences the fore-aft mudline loads. Current load, in contrast, exhibits a negligible impact. Furthermore, the influence of stochastic wind-wave processes is rigorously evaluated through sixty realizations under various mean wind speeds. The analysis reveals substantial variability in extreme structural responses induced by this randomness. Notably, at a mean wind speed of 17 m/s, the minimum fore-aft mudline shear force can be as low as 58.2 % of the maximum value observed across different stochastic realizations. These findings underscore the potential risks associated with deterministic design approaches and highlight the necessity of incorporating environmental load stochasticity in the structural design and safety assessment of offshore wind turbines.

Increasing anthropogenic engineering activities on coral reefs have accelerated changes in their topographic and geomorphic features, such as artificial island construction, which contribute to an uneven reef flat and layout of coastal protection structures. Previous research of coral reef wave hydrodynamics frequently simplified coral reef terrain as a step-like structure, overlooking the effects of uneven reef flats and neglecting the influence of human-engineered structures on wave hydrodynamics at coral reefs. As a necessary supplement, this study analyzed wave hydrodynamic properties of tsunami-like waves over an uneven coral reef with a seawall based on the physical experiments. Hydrodynamic influence of wave height and submergence water depth were analyzed in detail based on physical experiments. To further evaluate the hydrodynamic loads acting on the seawall and investigate the influence of reef-step dimensions on wave hydrodynamics, a series of numerical simulations was conducted. Research findings of present study indicate that tsunami-like waves will undergo complex transforming and breaking at the uneven reef flat. The tsunami-like waves will experience a surging breaker at the outer reef flat and a plunging breaker at the inner reef flat. Moreover, backwash water can be generated at the vertical seawall, resulting in additional wave reflection in addition to that occurring at the forereef slope.

Vertical-slit fishways can often serve as the critical ecological passages for the fish migration past instream barriers like dams.However,its operational performance is severely compromised in the sediment-laden rivers.Siltation within pool chambers can obstruct the migration routes to degrade the essential hydraulic habitats,and cause functional failure.Previous research has focused on the sediment reduction and fish passage facilitation as conflicting objectives.Therefore,this study aimed to design,evaluate,compare and integrate the structural modifications for sediment reduction within a vertical-slit fishway.The explicit goal was set to synergistically improve both sediment management and fish passage performance.A systematic investigation was made to quantify the influence of these structures on internal hydraulic features,sediment deposition,and the upstream passage of fish.A three-dimensional(3D)numerical hydrodynamic model was constructed to integrate with a 1∶6-scale physical hydraulic model using Froude similarity.The physical model also replicated a typical vertical-slit fishway prototype.The numerical model was developed using Flow-3D software.The standard k-epsilon turbulence model and the Volume of Fluid were employed for free-surface tracking.The validation test was conducted against velocity measurements from the physical model.The high accuracy was achieved to capture the complex flow patterns.Five sediment-reduction structures were tested:three variants in the position and aspect ratio of a bottom orifice into the baffle,and two variants in the placement of a cylindrical element in the pool chamber.All experiments maintained the identical upstream flow discharge,water level,and sediment concentration boundary conditions.Sediment deposition morphology and volume were measured after a 1.5-hour test duration.Concurrently,the live-fish trials were conducted on the juvenile common carp to assess passage effectiveness;The upstream success rate was calculated,where the fish successfully navigated from the release point past the first upstream slit within a 10-minute period.The experimental results demonstrated that there was a causal relationship between the modified hydraulic environment induced by the structures and the sediment deposition.In the standard fishway design(control case),the deposition was concentrated in the low-velocity recirculation zones adjacent to the main flow jet,with a total sediment mass of 4.7 kg in the pools.Five sediment-reduction configurations successfully modified the flow field to reduce the siltation,compared with the control.A dual-main-jet flow pattern was obtained in the bottom-orifice structures to lower the peak velocity and turbulent kinetic energy near the vertical slit.The vorticity maxima were redistributed to the sides of the jets,and the characteristic distribution of velocity was altered the locations of the sediment deposition.In contrast,the cylindrical structures were induced the prominent periodic vortices with the Ω-shaped flow distribution.The local vorticity and turbulent kinetic energy were significantly enhanced around the cylinder to intensify the sediment suspension and transport capacity.Among all configurations,the most substantial reduction of the sediment was found in the cylindrical structure with a diameter of 0.05 m,which was positioned with its centroid 0.10 m from the upstream baffle and 0.15 m from the right sidewall.The total deposition mass decreased by 48.9％to 2.4 kg.Fish passage data revealed that the structural modifications also influenced the fish behavior and success rates.The optimal cylindrical configuration increased the upstream passage success rate of test fish by 16 percentage points,from 64％(control)to 80％.Fish trajectory analysis via hotspot maps indicated that both bottom orifices and cylinders also provided the additional and alternative migration pathways.The spatial utilization then increased in the pool chambers.The Ω-shaped flow generated by the optimal cylinder into the hydraulically diverse conditions,thus reducing energy expenditure during ascent.In conclusion,the sediment-reduction structures were strategically integrated into the vertical-slit fishways.An effective strategy was provided to concurrently manage the siltation for the high fish passage.The superior performance of the cylindrical elements was achieved to reduce the sediment deposition in the fish passage,compared with the bottom-orifice types.The optimal cylindrical configuration was modified the internal hydraulic regime for the Ω-shaped flow pattern.Sediment transport was promoted for the favorable migration conditions.These findings can offer valuable practical insights for the engineering design of sustainable and efficient fishways in sediment-prone river systems.The parametric optimization of such structures,particularly cylinders,can be extended to long-term monitor the prototype in the durability and ecological benefits.

This study examines the interaction between internal solitary waves (ISWs) and an elliptical submersible based on the eKdV theory. The numerical model’s accuracy is validated through comparison of force calculations on a single cylinder with experimental data. Analysis reveals significant hydrodynamic characteristics around the elliptical submersible. The findings demonstrate that vertical positioning and elevation angle substantially influence wave loads on the submersible. At various vertical positions, uniform distribution of elevated hydrodynamic pressure may result in submersible disintegration. Submersibles with ±30° elevation angles experience greater total moments, increasing their susceptibility to capsizing. The maximum horizontal force increases by factors ranging from 1.46 to 8.81 when the submersible is inclined between −30° and 30°. The interface exhibits negative vorticity accumulation due to shear effects between upper and lower fluids. Submersible inclination leads to increased surrounding velocity and vorticity.

Denil fishways exhibit limited passage efficiency for weak-swimming and benthic species, partly due to severe near-bed hydrodynamics generated by the sharp V-notch apex of conventional baffles. Modifying bottom geometry is a promising optimization pathway, but previous studies often lack rigorous comparison under constrained baffle openness ratios. This study employed CFD with the RNG k-epsilon turbulence model to evaluate conventional V-shaped (TDF), equivalent U-shaped (SCDF), and rectangular (RDF) baffles under a unified openness ratio. A layered hydrodynamic evaluation framework demarcated by the effective blocking height was developed to distinguish flow responses in the upper jet-dominated and lower baffle-controlled layers. Results show that the upper-layer conveyance indicators remain broadly comparable across configurations, whereas the lower-layer indicators show configuration-related differences within the tested discharge range. The RDF and SCDF reduce lower-layer mean velocity and TKE relative to the TDF baseline across the tested discharge range, with the RDF achieving the larger velocity reduction and the SCDF the larger TKE reduction. The maximum relative reduction in lower-layer TKE, approximately 22%, occurs under intermediate discharge. These results suggest that bottom baffle geometry can provide a potential means of adjusting near-bed hydraulic conditions in Denil fishways, although the ecological consequences require further verification.

Objective As important hydraulic structures of hydropower station systems, flood discharge tunnels are characterized by high operating water head and frequent working condition transitions. Their safe and stable operation is crucial to the functional realization and overall operational performance of hydropower stations. Sufficient aeration of flood discharge tunnels is the key to ensuring engineering safety, improving flow pattern, and preventing cavitation erosion and vibration. Unreasonable layout may lead to negative pressure formation, blocked mass transport and increased energy loss. In addition to aeration problems, long flood discharge tunnels involve complex hydraulic phenomena during gate opening and closing, such as the transition between pressurized flow and free-surface flow, enhanced local aeration, development of negative pressure, and variation of gate pressure. In view of the above issues, a three-dimensional numerical model of a high-head long flood discharge tunnel is established in this study. The evolution law of gas-water two-phase flow inside the long flood discharge tunnel during the dynamic opening and closing of the gate is systematically revealed, with focus on analyzing the transient characteristics of aeration parameters (e.g., gas velocity, ventilation rate) and hydraulic characteristics (e.g., flow pattern variation, pressure fluctuation).Methods Firstly, a full-scale three-dimensional numerical model is constructed for a high-head long spillway tunnel in a pumped storage power station, encompassing the downstream reservoir, gate chamber, ventilation shaft, and a 287 m-long spillway tunnel with a 1.5% slope. This model targets the key region of gas-water two-phase flow evolution during dynamic gate operation to capture transient hydraulic and ventilation responses precisely. Numerical simulations are implemented in ANSYS FLUENT based on unsteady incompressible viscous flow theory, with the Reynolds-Averaged Navier-Stokes (RANS) equations serving as the governing equations to satisfy mass and momentum conservation laws. The deferred correction scheme is adopted for convective term discretization, while the second-order central difference scheme is used for pressure gradient and diffusion terms; the PISO (Pressure-Implicit with Splitting of Operators) algorithm is applied for pressure-velocity coupling to guarantee computational accuracy and stability. Secondly, the Wilcox k-ω turbulence model is utilized to calculate turbulent dynamic viscosity, which improves prediction performance for wall effects, free-surface aeration, vortices, and secondary flows with superior near-wall accuracy. The Volume of Fluid (VOF) model is employed as the interface-capturing approach to identify gas-water phases via the water volume fraction function, and the mixed density and viscosity of the two phases are computed accordingly to characterize gas-water interactions. Thirdly, three-dimensional structured meshes are generated for the computational domain, with local refinement at the gate bottom edge, gate slot, ventilation hole, and spillway inlet, plus boundary layer meshing near solid walls; the dynamic layering mesh technique and user-defined function (UDF) are combined to simulate the vertical uniform movement of the gate at 1.92 m/min, thus avoiding mesh distortion and maintaining convergence. Boundary conditions adopt a pressure outlet and no-slip wall boundary conditions, with initial conditions determined from the reservoir water level. Finally, model validation is completed by comparing simulated velocity distributions with measured data from similar studies, verifying the reliability of the numerical framework in simulating gas-water two-phase flow. This integrated method realizes refined simulation of transient ventilation and hydraulic characteristics during gate operation under high-head conditions, supporting the revelation of coupled response mechanisms.Results and Discussions Gate opening degree and closing speed are the dominant factors affecting the aeration response of long flood discharge tunnels. Under steady working conditions, the gas velocity in the ventilation hole remains generally low at a fixed gate opening, below 3 m/s. During the transient closing process at the designed closing speed of 1.92 m/min, the peak gas velocity in the ventilation hole exceeds 40 m/s. When the closing speed is reduced to 0.064 m/min, the peak gas velocity can be controlled below 40 m/s. The aeration system presents obvious staged response characteristics during transient gate closing. When the gate is closed from full opening, as the gate opening decreases from approximately 2.40 m to 1.96 m, the ventilation rate of the ventilation hole rapidly rises from 0.07 m³/s to a peak value of 6.04 m³/s, and then decreases to 0.248 m³/s as the gate further closes to 1.80 m. In terms of hydraulic characteristics, gate opening degree exerts a significant influence on the internal flow pattern and pressure distribution of the tunnel. With the decrease of gate opening, the flow pattern at the outlet flip bucket gradually transforms from jet flow to weir flow. During the transient closing process, the scope of the negative pressure zone inside the tunnel and the pressure characteristics of the gate change remarkably with the reduction of gate opening. When the gate closes gradually from full opening to half opening, a continuous negative pressure zone exists in the downstream slope section of the gate, with local negative pressure reaching the order of thousands of pascals. Meanwhile, the average pressure on the gate surface generally increases first and then tends to be stable; after the gate is fully closed, the pressure distribution gradually becomes uniform along the height direction. Cavitation number analysis shows that the cavitation number of the negative pressure zone in the flood discharge tunnel ranges from 0.23 to 0.28 under steady and transient conditions. Meanwhile, the cavitation number at the gate decreases with the reduction of opening degree. The possibility of cavitation is low under large opening conditions (1.0~0.8), while the cavitation number drops to 0.45~0.49 under small opening conditions (0.40~0.20), indicating a potential risk of cavitation occurrence.Conclusions This study further reveals the synergistic variation of aeration and hydraulic responses in high-head long flood discharge tunnels during dynamic gate closing, and identifies critical gate opening intervals associated with excessive ventilation velocity and unfavorable flow patterns. The findings provide a reference for aeration system design, gate operation control, and structural safety protection of high-head flood discharge tunnels. Limited by research conditions, the influence of different ventilation hole layouts on air supply uniformity and hydraulic characteristics was not considered. Future research will be further conducted in combination with physical model tests.

Large-scale wave prediction is essential for wave energy development, site selection, and operations and maintenance. In this study, a hybrid deep learning model is proposed for the accurate two-dimensional prediction of Significant Wave Height (SWH) over large spatial domains. Multi-scale convolution is employed to extract spatial features, while a convolution-based multi-head attention mechanism is utilized to capture global temporal dependencies. Through end-to-end training, the nonlinear spatiotemporal coupling between wind and wave fields is effectively captured. ECMWF wind and wave reanalysis datasets from 2000 to 2019 over the Northwest Pacific (0-50 degrees N, 100-150 degrees E) were utilized for model validation. Cross-seasonal wave patterns were accurately reproduced, and stable spatial performance was maintained in nearshore, offshore, and complex terrain regions. Furthermore, wave height trends were tracked accurately over time. Reliable predictions were achieved for forecast horizons within 48 h. Under typhoon-driven high-energy sea states, details of wind-wave coupling were represented accurately, and the maximum Root Mean Square Error (RMSE) was limited to 0.384 m. For rough sea states with SWH greater than 4 m, the absolute error of the predicted occurrence probability was found to be below 0.1% in more than 90% of the domain. These results indicate that strong potential for large-scale wave prediction across various scenarios is demonstrated. Consequently, technical support can be provided for wave energy planning, sea state warnings, and routing decisions.

Meshless methods have attracted a lot of attention in flow simulations by eliminating the reliance on high-quality grids. This paper presents a novel meshless level set method that integrates a particle migration scheme with an RBF-based surface reconstruction technique to simulate free-surface flows. Utilizing a hybrid particle framework, this method leverages a semi-Lagrangian scheme for control particles (level set carriers) to prioritize robustness, and a fully Lagrangian approach for interface particles to maximize tracking precision. The RBF-FD method is applied to approximate the level set function and its gradient with interface particles, boundary nodes, and control particles. A regularization term is introduced to enhance the stability and smoothness of the RBF interpolation. This method is validated through a series of benchmarks, including Zalesak's disk rotation test and vortex deformation test, demonstrating reliable interface preservation and mass conservation. Further numerical tests on dam-break flows, water entry and sinking cylinder problems confirm its robustness and accuracy. The proposed method offers a flexible, accurate, and efficient alternative for meshless simulation of free-surface flows.

ObjectiveMountainous hydraulic engineering projects are commonly constructed in regions characterized by complex geological structures, steep terrain, and strong hydro-meteorological influences. These conditions create a highly dynamic geomorphic environment where landslides frequently occur and may pose severe threats to infrastructure safety, regional ecological stability, and sustainable socio-economic development. Irrigation districts located in mountainous areas are particularly vulnerable because the construction and operation of hydraulic facilities—such as canals, aqueducts, and slope excavations—often alter the original stress balance of slopes and disturb the geological environment. Under the combined effects of rainfall infiltration, geological structures, and engineering disturbances, slope stability may gradually decrease, leading to the occurrence of landslide hazards. Therefore, accurately identifying landslide-prone areas and quantitatively evaluating landslide susceptibility are essential for hazard prevention, infrastructure protection, and long-term management of hydraulic engineering systems. Against this background, this study takes the Quanmutang Irrigation District in Hunan Province, China, as a representative mountainous hydraulic engineering area. The main objective is to develop a landslide susceptibility assessment framework that integrates time-series InSAR deformation information with machine learning models. Specifically, this research aims to (1) identify potential landslide hazards in the study area using SBAS-InSAR-derived deformation velocity; (2) construct a comprehensive landslide susceptibility evaluation system based on multi-source environmental factors; (3) compare the predictive performance of multiple machine learning models; and (4) investigate the contribution of InSAR-derived deformation factors to improving landslide susceptibility assessment accuracy.MethodsTo achieve the above objectives, a multi-source data-driven landslide susceptibility assessment framework was established by integrating time-series InSAR analysis, information value modeling, and machine learning algorithms. First, surface deformation monitoring was carried out using the SBAS-InSAR technique. A total of 210 Sentinel-1A ascending SAR images covering the period from July 18, 2020 to August 8, 2023 were collected, along with precise orbit ephemerides (POD) data. The SAR data were processed using the ISCE software platform, including image co-registration, interferogram generation, filtering, and phase unwrapping. Small baseline interferometric pairs were constructed using temporal and spatial baseline thresholds of 90 days and 500 meters, respectively, with an average coherence threshold of 0.3. After screening, 321 interferometric pairs were generated. Time-series analysis and error corrections were then applied to obtain the surface deformation velocity field across the study area. Based on the deformation results, potential landslide hazard zones were identified. Second, a comprehensive landslide susceptibility evaluation factor system was constructed. Initially, 16 candidate factors were selected from four categories: topography and geomorphology, geological conditions, natural environment, and human activities. These factors included elevation, slope, aspect, curvature, plan curvature, profile curvature, terrain wetness index, distance to rivers, distance to roads, distance to faults, land use type, geological age, normalized difference vegetation index (NDVI), rainfall, terrain relief, and surface roughness. To avoid multicollinearity among variables, Pearson correlation analysis was conducted. Factors with strong correlations were removed to ensure the independence of input variables. As a result, 14 static evaluation factors were retained. To eliminate the influence of dimensional differences among factors, the information value method was employed for normalization. The information value model is a statistical approach based on information theory, which quantifies the contribution of each factor category to landslide occurrence by comparing the spatial distribution of landslides and environmental conditions. The calculated information values were then used as input features for machine learning models. Third, landslide susceptibility modeling was performed using three widely used machine learning algorithms: Random Forest (RF), Support Vector Machine (SVM), and Convolutional Neural Network (CNN). Landslide samples were compiled from two sources: 318 historical landslides recorded in the study area and 42 potential landslide hazards identified through InSAR deformation analysis, resulting in a total of 360 positive samples. A buffer zone of 1000 m was established around these landslides to avoid spatial autocorrelation, and an equal number of non-landslide points were randomly generated outside the buffer zones as negative samples. The dataset was then divided into training and testing sets with a ratio of 70% and 30%, respectively.ResultsThe landslide susceptibility maps produced by the RF, SVM, and CNN models exhibit similar spatial patterns. In general, low and very low susceptibility zones are mainly distributed in the southwestern part of the irrigation district, where terrain conditions are relatively gentle and rainfall intensity is lower. In contrast, high and very high susceptibility zones are concentrated in the western and northern mountainous areas. These regions are characterized by steep slopes, complex geological structures, and higher rainfall levels, which promote slope instability through increased infiltration, elevated pore water pressure, and reduced effective stress in rock and soil masses. The susceptibility distribution also shows strong consistency with the spatial pattern of historical landslides, indicating that the models effectively capture the key environmental controls of landslide occurrence. Statistical analysis further reveals that rainfall plays an important role in landslide development in the study area, which is consistent with the feature importance analysis obtained from the machine learning models. Comparative evaluation of model performance indicates that all three machine learning models demonstrate good predictive capability. The AUC values of all models exceed 0.85, suggesting reliable classification performance. Among the three models, the SVM model achieves the best overall performance, followed by the CNN model, while the RF model shows relatively lower accuracy. After incorporating the InSAR-derived deformation velocity as a dynamic factor, the predictive performance of the models improves to varying degrees. For the RF-InSAR and CNN-InSAR models, several evaluation metrics show moderate improvements. In contrast, the SVM-InSAR model demonstrates consistent enhancement across all evaluation metrics. Its AUC value increases by approximately 4.2% compared with RF-InSAR and by 0.4% compared with CNN-InSAR, indicating that the SVM model benefits most from the integration of dynamic deformation information. Furthermore, statistical analysis of susceptibility zones across different irrigation subregions reveals that several areas exhibit particularly high landslide risk. For example, the Shaofeng Direct Irrigation Area, Liuguangling Area, and Sandu Baotian Area have high and very high susceptibility zones covering 78.50%, 58.87%, and 58.56% of their respective areas, indicating a high likelihood of landslide occurrence. Although some other irrigation zones show lower proportions of high susceptibility areas, their absolute areas remain significant, suggesting the presence of potential landslide hazards that require continuous monitoring.ConclusionsThis study proposes a landslide susceptibility assessment framework that integrates time-series InSAR deformation monitoring with machine learning models for mountainous hydraulic engineering regions. By incorporating dynamic deformation information into the susceptibility modeling process, the proposed approach effectively enhances the ability to identify potential landslide-prone areas. The results demonstrate that the integration of InSAR-derived deformation velocity significantly improves the predictive performance of landslide susceptibility models, particularly for the SVM model. The approach not only captures the static environmental conditions controlling landslide occurrence but also reflects the dynamic deformation processes that indicate slope instability. From an engineering perspective, the susceptibility maps generated in this study provide valuable information for identifying high-risk zones within the Quanmutang Irrigation District. These results can support infrastructure planning, slope stability monitoring, and disaster prevention strategies in mountainous hydraulic engineering projects.Overall, the methodology developed in this study offers a practical and effective framework for integrating remote sensing deformation monitoring with data-driven susceptibility modeling. The findings highlight the importance of incorporating dynamic surface deformation information into landslide hazard assessment and provide technical support for landslide risk management, disaster prevention, and sustainable development in mountainous irrigation districts and other similar engineering environments.

Accurate and robust retrieval of chlorophyll-a (Chl-a) concentrations in optically complex lakes is critical for effective water quality monitoring, especially under frequent extreme climatic conditions. This study proposes an integrated framework combining optical water type (OWTs) classification and ensemble learning algorithms to enhance the re- mote sensing retrieval accuracy of Chl-a in Dongting Lake and Poyang Lake during 2020-2023, particularly during the extreme drought event of 2022. Firstly, K-means clustering combined with spectral feature analysis was used to classify both lakes into four distinct OWTs. Dongting Lake classifications effectively separated phytoplankton-dominated waters (OWT1, OWT2, OWT4) from those dominated by non-algal suspended matter (OWT3). For Poyang Lake, classifications identified mixed water types (OWT1), frequent algal bloom areas (OWT2), high biomass phytoplankton areas (OWT3), and areas significantly masked by non-algal particles (OWT4). Further analysis indicated that OWT2 in Dongting Lake is prone to algal blooms, while OWT3 and OWT4 are dominated by non-algal particles. Secondly, the performance of ensemble learning methods (Bagging, Boosting, Stacking, and Voting) was evaluated against traditional single models (SVR and BPNN). The results demonstrated the superior stability and predictive accuracy of the Voting strategy under low Chl-a conditions in Dongting Lake, achieving a maximum MAPE reduction of 84.76 %. Meanwhile, the Stacking method exhibited outstanding robustness in Poyang Lake's complex optical conditions, with RMSE reduced by up to 93.12 %. Additionally, hierarchical modeling based on OWT-specific ensemble methods significantly reduced errors compared to traditional global models, with decreases of 36 % in Dongting Lake and 42.19 % in Poyang Lake. Finally, this study revealed that the extreme drought event in the Yangtze River basin in 2022 significantly altered seasonal and interannual variations in the OWTs of Dongting Lake and Poyang Lake. The drought led to lake shrinkage and increased dominance of non-algal suspended particles, highlighting the vulnerability of these connected lake ecosystems under extreme climatic conditions. The research emphasizes the importance of combining a lake-specific OWTs classification and advanced ensemble learning techniques for precise water quality assessment under dynamic environmental conditions.

Vertical slot fishways represent critical ecological migration facilitation structures and have been globally implemented to restore fish passage. However, most studies to date focus primarily on fishway hydraulics and fish behavior, with limited investigation into sediment deposition effects that may compromise functionality. To address this gap, we integrated physical modeling and numerical simulations to systematically analyze sediment deposition in a vertical slot fishway and its impacts on common carp upstream migration. Results indicate that sediment deposition raised fish vertical swimming positions by an average of 5.0 cm, thereby reducing pool activity space by 5.2–20.2%, altering flow patterns, and disrupting carp bottom-migration behavior. Consequently, carp exhibited increased exploratory behavior and directional uncertainty. Moreover, sediment-induced vertical vortices elevated fish energy consumption, decreasing upstream migration success from 89% to 48%. Multiple linear regression confirmed that average sediment deposition height significantly affects both migration rate and vertical swimming positions, whereas mean deposition slope demonstrates negligible influence. This study elucidates the multifaceted impacts of sediment deposition on fishway efficacy, providing a scientific basis for optimizing designs to enhance migration success and long-term functionality.

Study region: Dongting Lake Basin (DTLB) is a sub-basin of the Yangtze River Basin, Southern China, and is an important agricultural production area in China. Study focus: This study (1) evaluated the reliability of terrestrial water storage anomaly data in the DTLB and the associated trend and seasonal characteristics using gravity satellite data and various model datasets, (2) explored the hydrological and meteorological drought characteristics and response to climatic indices in the DTLB over the past 20 years. New hydrological insights for the region: Results show that the DTLB suffered from four moderate to extreme hydrological drought events during 2002-2022, with a total duration of 22 months and a total water storage deficit of-2201 mm (similar to 578.9 billion m(3)). The hydrological drought index (water storage deficit index, WSDI) showed lower drought frequency, but a longer duration of a single drought event than the meteorological drought index (standardized precipitation index, SPI). The meteorological and hydrological drought characteristics exist certain differences in sub-basins of the DTLB, jointly affected by global warming, climate indices, topography, temperature, etc. El Nino-Southern Oscillation was the major climatic indices affecting meteorological and hydrological droughts in the DTLB. This study provides a scientific reference for dynamic monitoring and prediction of drought, analysis of drought propagation and rational water resource management in the DTLB.

Repeatedly solving flow around structures with varying parameters using computational fluid dynamics (CFD) is often essential for structural design. This study proposes a boundary-assimilation Fourier neural operator (BAFNO) method to address the challenges of manually setting initial conditions for CFD. The focus of the BAFNO is on the generalization ability to predict initial flow fields without relying on observational data. BAFNO addresses the boundary constraint requirements of the existing physics-informed neural operator models in parametric geometries. Inspired by the ghost node method, the domain boundary conditions are assimilated into the loss function instead of adding penalty terms. Meanwhile, the structure boundaries are assimilated into a damping source term using a level set function. BAFNO can flexibly handle parametric geometries with different shapes and quantities. Subsequently, a series of numerical experiments for flow-around structures are conducted to confirm the performance of the BAFNO. The results indicate that the BAFNO has strong generalization capability, and the BAFNO + CFD can obtain dynamic stable fields faster than the direct CFD.

Frequent typhoons in the Beibu Gulf pose a significant threat to marine and coastal infrastructure. To address this issue, we applied a fully integrated tide-surge-wave model using the Holland typhoon model and Delft3D-FLOW-WAVE. The model simulates storm surge and typhoon waves generated by Super Typhoon Rammasun (2014). We generated five idealized typhoon tracks by systematically shifting Rammasun’s track to evaluate their potential impacts on storm surge and waves. The results indicate that nearshore storm surges in the Beibu Gulf exhibit a distinct rise-then-fall pattern, with the maximum surge occurring on the right-hand side of the typhoon track. Surge magnitude diminishes as the track shifts eastward. Significant wave heights undergo marked spatial redistribution upon the typhoon’s entry into the Gulf, transitioning from left- to right-biased asymmetry during passage. Coastal ports in the Beibu Gulf and the eastern Leizhou Peninsula experience pronounced positive surges, while the highest surge and wave intensities occur along northeastern Hainan Island and both sides of the Qiongzhou Strait. Although based on idealized tracks, this study offers critical insights for optimizing coastal disaster mitigation against extreme typhoons in the region.

The impact of the South China Sea (SCS) to the super typhoon Mangkhut (2018) and the air-sea interaction are evaluated through COAWST model in this study which fully coupled with the atmosphere model (WRF), the ocean model (ROMS), and the wave model (SWAN). A comparison of our modeled results with several buoys and tide stations revealed remarkable consistency, the minimum root mean square error (RMSE) for wind speed, significant wave height, and water level are 3.013 m/s, 0.641m, 0.007m. During the development and dissipation of super typhoon Mangkhut, the Coriolis force caused the typhoon wind field to exhibit a pronounced rightward deflection. The characteristics of the significant wave height field are generally similar to those of the wind field, although there is a temporal lag of several hours. Moreover, we researched on the spatiotemporal variations of sea surface temperature (SST) impact by the Mangkhut’s passage, and found that it exhibits two main characteristics: spatial asymmetry and temporal lag. The spatial asymmetry is primarily governed by typhoon-generated wind fields, while the temporal lag is mainly controlled by upwelling and vertical mixing processes during the typhoon’s passage, with Ekman pumping playing a pivotal role in these dynamics. This study mainly concentrates on investigating the dynamic and thermodynamic responses of the ocean during extreme weather conditions by using COAWST model.

Extreme wave events induced by marine meteorological extremes such as storm surges and tsunamis have become a critical focus in contemporary coral reef hydrodynamics research. The synergistic forcing of strong winds and extreme wave regimes poses significant challenges to the structural safety of reef engineering systems. A high-resolution numerical model was established based on OpenFOAM. The reliability and accuracy of the model were verified by comparing with experimental data. Building upon this foundation, the effects of wind on the interaction processes between solitary waves and vertical walls on reef flats were investigated. The findings demonstrate that wind enhances vortex motions during solitary wave interactions with vertical walls while attenuating wave reflection effects. At low wind speeds, the maximum horizontal force experiences minor reductions compared to windless scenarios, whereas high wind speeds induce significant increases. The maximum vertical force at the wall crest exhibits monotonic growth with increasing wind intensity. Wind action also advances the initiation of overtopping processes relative to windless. Crucially, the maximum cumulative overtopping volume amplifies progressively with wind speed, reaching approximately 1.23-4.11 times the windless at U* = 4. This study provides theoretical references for the design of protective engineering structures on coral reefs.

The angle of repose is a fundamental parameter for assessing the stability of coral reefs. However, predictive models for this angle are currently lacking. In this study, a series of laboratory experiments were undertaken to investigate the angle of repose by varying moisture content, particle shape, and particle size. Based on our experimental data, variation in the angle of repose with moisture content is classified into five distinct zones. It is demonstrated that the range of moisture content for each zone varies with particle size. Coral sands of dendrite, flake, rod, and block particles have a descending order of angle of repose, as demonstrated for a sieve size of 4.5 mm. The angle of repose for dry, submerged, and steady coral sands exhibits a correlation with the nominal diameter of particle size. Finally, extended models are proposed for predicting the angle of repose of coral sands (R2 = 0.8, Dn50 = 0.317−5.470). To facilitate use of these models, a linear relationship between sieve particle size diameter, nominal particle size diameter, and Corey shape factor, allowing for conversion among these parameters, is established. This study thereby helps to enhance our understanding of how moisture content affects angle of repose and improve our ability to predict the angle for coral grains with intricate geometries.