Steel fiber reinforced concrete has been widely used in engineering practice due to its superior mechanical properties. However， unlike other fibers， steel fibers are susceptible to corrosion in corrosive environments. This paper systematically reviews the research progress and key findings on the corrosion of steel fiber concrete at home and abroad， including the corrosion mechanism， as well as the effects of corrosion on the mechanical and durability properties of concrete. Previous studies have shown that the corrosion resistance of steel fibers is influenced by several main factors， such as the water-cement ratio， fiber type， concrete cracks， and the erosion environment. Besides， steel fibers corrode only within approximately 5 mm of the surface layer in uncracked concrete. The strength of the steel fibers gradually decreases as corrosion advances， while the corrosion products play dominant roles in the filling and compression of pores at different stages. These processes affect the bonding performance between the steel fibers and the cement matrix， thereby altering the mechanical properties of the concrete. Consequently， the mechanical properties of the concrete are closely related to the corrosion degree of steel fibers. Most existing studies have not considered the impact of the corrosion degree on the strength of steel fiber concrete， and the quantitative relationship between the corrosion degree of steel fibers and the degradation of concrete mechanical property has not been established. Finally， this paper discusses the challenges in steel fiber corrosion research and suggests future research trend.

Soil salinization is one of the key factors affecting the sustainable development of agriculture in the Hetao Irrigation District. Accurate prediction of soil salinization and sodification trend is an important means to formulate scientific prevention and control measures. Traditional studies usually use the total amount of soil salinity in the root zone as an indicator to evaluate the evolution trend of salinization， while rarely paying attention to changes in ion composition. Due to the different physical and chemical behaviors and migration capacities of different salt ions， soil salts not only changes in total amount， but also change in composition of salt ions under the influence of external forces such as irrigation leaching， evaporation， freezing and thawing. To conduct in-depth research on the distribution and transport patterns of salt ions in different land use types， this paper selects a typical area in the Yichang Irrigation Area of Inner Mongolia's Hetao Irrigation District as the research area. From 2017 to 2019， observations were conducted on the distribution processes of various ions， analyzing the dynamic processes of water， salt， and ions， as well as the characteristics of ion differentiation.The soil moisture content increases with depth due to water infiltration and water-holding capacity， while variability showes an inverse trend， decreasing with soil depth. The electrical conductivity （EC） profile of abandoned cultivated land decreases with depth， with surface accumulation being the main feature of overall salt distribution. The distribution of salt ions in soils of different land types is primarily characterized by highly differentiated Na⁺， Cl^-， and SO₄ ^2- ions. The HCO₃ ^- content is also significant but is relatively uniformly distributed in the soil profile. The contents of Ca²⁺ and Mg²⁺ are noticeably lower， especially in wasteland where both their composition and proportion are minimal. The mobility differentiation of various salt ions is evident. The migration degree and rate of cations are in the order of Na⁺ > Mg²⁺ > Ca²⁺， which correlates with ion migration rates through soil layers. Among anions， Cl^- shows the strongest differentiation， while HCO^3- exhibits weaker differentiation.

Central China has a typical temperate and subtropical monsoon climate， 60% of which is mountainous， and the spatial and temporal characteristics of atmospheric precipitation and water vapor sources are very complex. By searching the literature on precipitation hydrogen and oxygen isotopes and water vapor sources in central China （Hunan， Hubei and Henan） in recent years， the isotopic characteristics of precipitation in central China and its related aspects are sorted out and summarized， and the future research on the stable isotopes of hydrogen and oxygen of atmospheric precipitation in central China is prospected. The results show that the precipitation isotope compositions in central China show seasonal variations of low summer and high winter in time， which fully reflect the precipitation effect of the distribution of hydroxide stable isotopes； there is a spatial trend of decreasing from south to north， and the areas of high precipitation hydroxide stable isotope values are mainly located in low-latitude and low-elevation areas such as eastern Hunan and eastern Hubei， and the high-elevation mountainous areas such as western Hubei and western Henan are the areas of low values. The slopes of the local atmospheric precipitation lines are closer to the global and national atmospheric precipitation lines， but there are different degrees of deviations in the intercepts of each region， with imbalanced isotope fractionation in wet and rainy regions in Hunan and Hubei， and strong secondary evaporation under the arid and low rainfall clouds in Henan. Precipitation is the main factor affecting the isotopic composition of atmospheric precipitation in central China， and there are inverse temperature effects and elevation effects in some regions. The water vapor of atmospheric precipitation in central China is mainly influenced by the oceanic air masses carried by the southwest monsoon and southeast monsoon in the summer half of the year， while the atmospheric precipitation air masses in the winter half of the year mainly come from the interior of the Asian-European continent and local evaporative water vapor. The above results provide a theoretical basis for the study of the ecohydrological cycle process in central China， and provide scientific guidance for the rational regulation of regional water resources.

The maximum height of the upstream earth-rock cofferdam of a hydropower station is 62 m， and the height of the slope formed by the cofferdam and foundation pit is 135 m. The earth-rock cofferdam is located on deep overburden layer with a maximum thickness of 71 m. The overburden layer has the characteristics of “low bearing capacity， small permeability coefficient， low shear strength and high compressibility”. In order to ensure the safety and stability of the cofferdam and slope， vibro-replacement stone column is used in design to reinforce the deep overburden layer of the cofferdam foundation. The depth of the designed stone column of a hydropower station far exceeds the existing technology level. It is the deepest stone column applied ever in engineering practice in China on a large scale. In order to solve the problem of ultra-deep vibro-replacement stone column construction and verify the effect of stone column reinforcement and drainage， the ultra-deep vibro-replacement stone column construction equipment and technology are determined through on-site productive tests， and the physical and mechanics parameters such as shear strength， bearing capacity and compression modulus of stone column body and soil between column， permeability coefficient of stone column， etc. are obtained. A set of control methods which can be applied to large-scale construction， quality inspection of vibro-replacement stone column in cofferdam foundation are summarized. This research results can be used as a reference for the design， construction and acceptance of foundation treatment of hydropower engineering projects.

As a new type of ecological agricultural model， crayfish-rice co-culture has gradually occupied an important position in China's agricultural development. Understanding the variation characteristics of the rice growth environment is of great significance for increasing the grain yield in crayfish-rice fields and ensuring the achievement of double harvests from one water source in crayfish-rice co-culture system. Based on this， this paper reviews the current development status， industrial advantages， and still existing theoretical and technical problems in actual production of crayfish-rice co-culture system in China. Combining with existing literature and research foundations， this article summarizes the research progress on the characteristics of the crayfish-rice symbiotic system and the variation characteristics of the soil environment， field water layer control methods， and biological - ecological environment brought about by the changes in co-cultivation management methods. It also summarizes the impacts of the above characteristic changes on rice growth. Finally， it outlines future research directions from three aspects， including the coupling mechanism of the crayfish-rice co-culture system， the collaborative regulation technology for water conservation， emission reduction， and carbon reduction， and model development， aiming to provide a scientific basis for stabilizing grain production and increasing income in paddy fields under integrated rice-fishery farming and for the efficient utilization of agricultural resources.

Global climate change has led to significant changes in hydrological cycle processes， resulting in an increased risk of hydrological droughts. Therefore， the study of hydrological drought changes under changing environments is of great significance for early warning of flood and drought and water resources management.Typical basins in north and south China （Xiangjiang river basin and Weihe river basin） are selected as the study areas.The period from 1982 to 2015 is selected as the base period （the historical period） and the period from 2030 to 2060 is selected as the future period. Five Global Climate Models （GCMs） provided by CMIP6 under the three scenarios of SSP126， SSP245 and SSP585 are selected to output the climate data for the future period.The Extreme Gradient Boost （XGBoost） is utilized to downscale the precipitation and temperature from GCMs to drive hydrological models to obtain streamflow in the future period， and then compare and analyze the evolution of hydrological drought characteristics in Xiangjiang river basin and Weihe river basin in different periods.The results indicate that， after downscaling by XGboost， the RMSE of simulated precipitation of Xiangjiang river basin and Weihe river basin are reduced by 57.6% and 51%， respectively， and the correlation coefficient with the measured values are improved by 20% and 32%， respectively. The RMSE of simulated temperature are reduced by 50.9% and 51%， and the correlation coefficients with the measured values are improved by 4.3% and 6.25%， respectively.The increase in precipitation in Xiangjiang river basin in the future period will increase the streamflow in this basin， thus alleviating hydrological drought， and the severity， duration， intensity and peak intensity of hydrological drought are reduced by 6.3%， 4.8%， 10.6%， and 6.5%， on average， compared with those in the historical period.In contrast to Xiangjiang basin， the increase in temperature and potential evapotranspiration decreases the streamflow in Weihe river basin.Therefore， Weihe river basin will face a greater risk of drought in the future period. Among the selected scenarios， the drought is the most severe under the SSP585 scenario. Compared with the historical period， the severity and duration of drought will be increased by 72.7% and 30.8%.

Based on in-depth research， this study systematically reviews the development history， current status， and existing problems of rural water supply and standardization. It thoroughly analyzes the demand for standardized development in rural water supply and proposes several considerations and recommendations to accelerate progress. The standardization of rural water supply should adhere to a problem-oriented and demand-driven approach， emphasize technological innovation， strengthen deep integration of industry， academia， research， and application， foster and develop new quality productive forces， and leverage the guiding role of standards to unleash innovation potential. Following the principle of balancing development and safety， a full-lifecycle technical standard system should be established， covering planning and design， construction， operation and maintenance， water quality safety， and management services. This system will consist of "2+N+M" standard system. In terms of engineering planning， it is essential to effectively align with strategic plans such as the National Water Network， urban-rural integration， rural revitalization， and rural construction to optimize the overall layout of rural water supply. For engineering construction， adherence to standards must be prioritized to ensure standardized construction of rural water supply projects， establishing and proving rural water supply system that achieves overall optimal performance. Regarding operation and management， upgrading standards is crucial to transition rural water supply toward a modernized management model characterized by specialization， equalization， and degitization， thereby promoting high-quality development in rural water supply.

Unmanned Aerial Vehicle （UAV） remote sensing technology presents a multitude of advantages， including cost-effective deployment， rapid responsiveness， versatile and user-friendly operation， superior image quality， and minimal disruption to the environment. It can be used to monitor multiple water quality indicators， such as total nitrogen， suspended matter， turbidity， total phosphorus and chlorophyll， etc.， which is the development trend of river and lake water environment monitoring technology system. To assess the applicability and reliability of this technology for monitoring river water quality， the Xiushui River—one of the five major rivers in the Poyang Lake Basin—was selected as the research subject. A total of 31 samples were collected， and key water quality parameters including suspended solids （SS）， total phosphorus （TP）， total nitrogen （TN）， turbidity （TUB）， and chlorophyll-a （Chl-a） were quantitatively retrieved using remote sensing data obtained from UAV multispectral imaging.The accuracy of monitoring was evaluated through linear fitting， coefficient of determination （R2）， normalized root mean square error （NRMSE）， and mean relative error （MRE）. Furthermore， a comparative analysis along with an economic assessment demonstrated the applicability of UAV remote sensing technology in this context. The results show that：①The remote sensing technology of UAV demonstrates a high accuracy in monitoring water quality parameters such as SS， TP， TN， TUB， and Chl-a. The coefficient of determination （R2） for the linear fitting curves between the inversion values and the measured values ranges from 0.565 to 0.89， closely aligning with a 1∶1 trend line， thereby meeting the precision requirements for water quality parameter retrieval. Among these parameters， Chl-a exhibits the highest accuracy， while TP shows the lowest； thus， they can be ranked in descending order as follows： Chl-a SS TUB TN TP. ②The normalized root mean square error （NRMSE） values for each index fall between 0.1 and 0.3， indicating a moderate preference for goodness of fit. Furthermore， the mean relative error （MRE） for each index ranges from 13% to 24%， which is generally close to the measured values with deviations remaining within a controllable range. ③Additionally， the costs and construction periods associated with UAV remote sensing monitoring are more than 50% lower than those of manual monitoring， indicating that UAV remote sensing technology is a more economical option. This study innovatively uses UAV remote sensing technology to invert river water quality， and verifies the feasibility of UAV remote sensing technology in river water quality monitoring， which provides a new method for rapid and accurate monitoring of river water quality， and has certain practical guiding significance for promoting the development of water environment monitoring technology.

The calculation of TBM tunnel segment lining structure is an important link in lining design. This paper combines an engineering example in South China to establish the lining structure calculation model using the modified usage method， beam-spring method， and three-dimensional finite element method（3D FEM）， and compares the internal forces and deformations of the lining structure calculated by each method. At the same time， the paper studies the influence of the staggered assembly angle on the lining structure forces. The results show that the maximum axial force and shear force of the structure calculated by each method are close to each other. The order of the maximum bending moment and deformation of the whole structure is as follows： modified conventional method > beam-spring method > three-dimensional finite element method. According to the results， the modified conventional method yields results with high safety reserve， simple calculation， making it suitable for preliminary design of engineering. The three-dimensional finite element method is suitable for accurate calculation and preliminary design check of segments， and can show the mechanical characteristics of the detailed structure of segments. The calculation result of beam-spring method is between other two methods， which is relatively economical and can reflect the stiffness contribution of the joint of segment more accurately than the modified conventional method， and the model is more user-friendly than the three-dimensional finite element calculation model which is suitable for engineering design. At the same time， the beam-spring method has a more obvious staggered assembly effect than the 3D FEM in calculating the maximum bending moment.

The conventional reservoir operation diagram， constrained by static regulation frameworks and linear assumptions， demonstrates limited capacity in characterizing dynamic coupling mechanisms and nonlinear constraints within cascade reservoir systems. To address this， based on the theory of system dynamics， this study built a simulation model for early flood-season storage scheduling in mega reservoirs. Taking six cascade reservoirs （Ludila， Guanyinyan， Wudongde， Baihetan， Xiluodu， and Xiangjiaba） in the middle and lower reaches of the Jinsha River as the case study， the dynamic coupling path among water level， reservoir capacity and flow rate was analyzed through system dynamics feedback loops and stock-flow architecture. The study coupled optimization scheduling techniques using LOOKUP functions and IF-THEN-ELSE conditional statements， which characterized nonlinear relationships including water-level-reservoir capacity curves and hydropower unit output constraints， achieving multi-objective dynamic simulation of coordinated reservoir impoundment operations. The Research results indicated that the system dynamics model could accurately depict the dynamic and nonlinear characteristics of the impoundment operations， and its simulation results were highly consistent with those of optimal operation schemes. Under various hydrological scenarios （wet， normal， dry， and extremely dry years）， the Nash-Sutcliffe Efficiency （NSE） values for the reservoir operation water levels exceed 0.99， and the NSE values for the reservoir discharge processes surpass 0.97. The mean absolute error （MAE） values for reservoir operation water levels have not exceeded 0.11 m， while the root mean square error （RMSE） values have not surpassed 0.28 m. For reservoir discharge simulations， the MAE values are below 54 m3/s， and the RMSE values are less than 171 m3/s. Compared with the conventional operation scheme， the system dynamics simulation scheme can increase the multi-year average hydropower output by 2.495 billion kWh/a （an improvement rate of 3.21%）， reduce the spilled water volume by 1.435 billion m3/a （an improvement rate of 15.61%）， and decrease the average computation time for the joint operation of mega reservoirs from 6 hours to 5 seconds. The results not only thoroughly demonstrate the accuracy and reliability of the simulation model， but also provide scientific and technical support for the joint operation of mega reservoirs.

Accurate and timely soil moisture inversion is a crucial foundation for implementing precision irrigation management in farmlands， significantly enhancing irrigation water resource utilization efficiency. This study developed enhanced Thermal Optical Trapezoid Model （TOTRAM） and Optical Trapezoid Model （OPTRAM） by incorporating nonlinear dry/wet boundaries and phenological characteristics， using multi-source remote sensing data from Landsat-8/9 and Sentinel-2A/B. The OPTRAM model further accounts for the influence of phenological stages.The accuracy of the enhanced models in soil moisture inversion for arid farmland was then evaluated using measured soil moisture data.The findings of the study demonstrated that the original TOTRAM model exhibited superior performance in comparison to the basic OPTRAM model， with root mean square errors （RMSE） of 0.183 cm3/cm3 and 0.141 cm3/cm3， respectively. The implementation of non-linear dry/wet boundaries resulted in a substantial enhancement in the performance of both models， achieving comparable reductions in RMSE to 0.145 cm3/cm3 （TOTRAM） and 0.126 cm3/cm3 （OPTRAM）. The phenology-integrated OPTRAM model demonstrated optimal accuracy， with a root mean square error （RMSE） of 0.096 cm3/cm3. In summary， the enhanced OPTRAM model exhibited superior precision in the retrieval of soil moisture in arid agricultural soil compared to TOTRAM， while concurrently maintaining the higher spatiotemporal resolution advantages afforded by optical satellite data. This advancement provides critical technical support for precision irrigation management systems.

Pumping station technology upgrading is an important part of building a national water network， realizing the goal of “dual carbon”， and integrating water resources， ecology and energy security. Overseas pumping stations have made significant progress in hydraulic design， material application， informatization and energy synergy， which is manifested in the enhancement of adaptability to extreme working conditions， improvement of the level of intelligent operation and maintenance， improvement of the ecological mode of operation and the development of green energy supply system. However， under complex conditions such as high sand content and strong corrosion， there are still problems such as insufficient design accuracy， imperfect failure prediction， unclear ecological response mechanism and limited multi-energy synergy optimization. In this paper， the latest foreign research results and engineering practices are sorted out， and key technologies such as inverse problem and machine learning fusion design， digital twin technology， fish-friendly flow channel and ecological protection measures， and wind-solar storage and complementary evaluation are summarized. It also proposes technical paths for the national water network and the follow-up project of South-to-North Water Diversion， which will provide reference for the modernization of water conservancy infrastructure and the construction of intelligent water conservancy during the 15th Five-Year Plan period.

In order to reveal the co-evolution mechanism of water-energy-food-ecology （WEFE） system security and economic resilience in the Beijing-Tianjin-Hebei urban agglomeration， this study used the panel data of 13 prefecture-level cities in the Beijing-Tianjin-Hebei region from 2007 to 2022 to construct a three-dimensional evaluation framework of pressure-state-response， and systematically analyzed the characteristics and influencing factors of regional coupling spatial network from 2007 to 2022 by comprehensively using the coupling coordination degree model， modified gravity model， social network analysis and secondary assignment procedure. By quantifying the nonlinear correlation between resource constraints and economic resilience， this paper aims to break the limitations of traditional single-factor research and propose a differentiated networked governance path， so as to provide theoretical support and decision-making basis for urban agglomerations to break through the path dependence of "resource consumption for growth" and realize sustainable development of ecological priority. The results show that from 2007 to 2022， the average coupling coordination degree of the WEFE system security and economic resilience in Beijing-Tianjin-Hebei increased from 0.38 to 0.48， but regional differences expanded significantly， and the spatial pattern evolved from a “core-periphery” fragmented structure to a “multipolar network” form. While the number and density of network associations are increasing， the correlation and stability still need to be optimized： Beijing and Tianjin form technology capital spillover through the “siphon-radiation” effect， and intermediary cities such as Shijiazhuang and Baoding play the role of “structural holes” to promote peripheral integration； Geographical proximity significantly strengthens network association， while population quality， government efficiency， technological innovation， openness level and climate risk differences have inhibiting effects on network formation. Based on this， this study proposes to promote the multi-directional circulation of elements by constructing a dynamic monitoring and three-level governance structure， implementing a core node feedback mechanism and reward fund to promote network collaboration， and establishing a cross-domain collaboration， talent sharing and climate adaptation mechanism， so as to realize the differentiated networked governance of the coupling of WEFE system security and economic resilience in Beijing-Tianjin-Hebei.

The Sixth Coupled Model Comparison Program （CMIP6） data is often employed to evaluate potential global-scale changes in climate variables. However， it is less efficacious in projecting changes at the regional level. The Ili River Basin in Xinjiang Uygur Autonomous Region of China （IRBC）， characterized by significant elevation variation and complex terrain， gives rise to considerable discrepancies in data analyses based on limited number of meteorological stations. Consequently， reliable meteorological data are essential for climatic and hydrological research. This study develops CMIP6 data into a new multi-model ensemble dataset by four approaches： reliability ensemble average （REA）， convolutional neural network （CNN）， random forest （RF）， and Bayesian model average （BMA）. Additionally， we employed performance indicators such as the Taylor Diagram， Taylor Skill Score （TSS）， and Kling-Gupta Efficiency （KGE） to identify the method with the best simulation effect on precipitation and temperature in the study area. The results demonstrate that the BMA dataset exhibits the most effective simulation of precipitation patterns across a range of extreme precipitation indices. In calibration period（1961-1999）， the annual total precipitation when daily wet day amount 95th percentile （R95PTOT） is the highest. In the validation period（2000-2014）， BMA's rank sum ratio （RSR） is 1.1， ranking first， indicating that its comprehensive simulation effect is the best. The KGE results of annual maximum consecutive 1-day （Rx1day） and 5 day precipitation （Rx5day）， and R95PTOT are 0.032， 0.39 and 0.52， respectively， which are all higher than the values obtained from other datasets. In the average temperature simulation， the RF dataset performed the best. The standardization deviation （SD）， the center root mean square error （CRMSE）， the correlation coefficient （r） and TSS in the calibration period are 1.005， 0.088， 0.996 and 0.49， respectively. These values represent the most favorable outcomes compared to other datasets. Similarly， the RF also outperformed others during the validation period， with a RSR value of 1.18. The findings of this study assess the efficacy of various approaches in simulating meteorological data within the study area. This can provide a usable method for the analysis of meteorological data in future scenarios and provide a scientific foundation for the management of meteorological disasters and water resources in the IRBC.

To address the the complexity of wind resource simulation for mountainous wind farms， this study takes a complex mountainous wind farm as an example. Based on existing anemometer tower data near the wind farm， and considering the relative relationship between the terrain shape and the prevailing wind direction， the computational fluid dynamics （CFD） software Windsim 12.0 was employed to establish two digital terrain models with different boundary ranges of the wind farm. Numerical simulations of the flow fields in the wind farm area were simulated， and the power generation results from the two models were compared with the actual power output of the wind farm. The results indicate that， to accurately assess wind resources in complex mountainous wind farm areas， it is essential to avoid insufficient development of numerical terrain flow field simulations and the adverse effects of boundary conditions on internal flow fields during modeling. This study highlights the critical role of digital terrain model construction in wind resource assessment for complex mountainous wind farms， providing a reference for wind resource evaluation and rational utilization in such terrains， as well as technical support and empirical insights for wind energy development in similar environments.

The wind-photovoltaic-hydro complementary energy is an effective strategy to enhance the consumption of renewable energy. In the context of short-term operational scheduling， the multi-energy complementary dispatch of cascade reservoir systems necessitates the coordination of power generation and load balancing across various reservoirs and diverse renewable energy sources. Furthermore， the scheduling process must consider the time delay associated with water flow through the cascade reservoirs， resulting in a lag effect on power discharge during the scheduling period. This scenario presents a multi-objective optimization challenge， requiring a balance between the benefits of power generation both within and outside the scheduling period， as well as the benefits of peak-shaving during the scheduling period. This study develops a short-term multi-objective optimization model for the wind-photovoltaic-hydro complementary energy in cascade reservoir systems. The model generates a Pareto-optimal set of solutions， encompassing power generation benefits， peak-shaving benefits， and potential power outside the scheduling period. Additionally， it investigates the conflicting and synergistic relationships among these multiple objectives and quantitatively assesses the comprehensive benefits of complementary dispatch through comparative analysis. Using the cascade hydropower station group in the Yuanshui River Basin as a case study， model simulations for typical dry and wet seasons yield the following key findings： ① There exists a conflict between long-term and short-term scheduling benefits. As the energy storage capacity and lagged power within the cascade reservoirs increase， the power generation benefits within the scheduling period diminish. ② A trade-off exists between power generation and peak-shaving objectives， with this trade-off becoming less significant as the potential power decreases. ③ Hydro power compensation for wind and photovoltaic generation effectively enhances the total system output， with additional generation during the dry and wet periods amounting to 8.08 million and 5.2 million kWh， respectively. In peak-shaving mode， the maximum residual load is reduced by 880，000 and 60，000 kW during the dry and wet seasons， respectively. The multi-objective optimization scheduling model for wind-photovoltaic-hydro complementary energy in cascade reservoirs provides valuable insights for advancing the integration and consumption of renewable energy.

Piping is a typical form of seepage failure in earth-rock dams， and its evolution process is accompanied by internal physical field changes within the earth-rock dam structure. Investigating the relationship between the variations in the internal physical field of earth-rock dams and the evolution of piping is of significant importance for identifying the state of piping progression. In view of this， a multi-field information fusion analysis framework for piping evolution is first constructed based on the multiple physical field information obtained during the operation of earth-rock dams. Building on this foundation， the seepage pressure， temperature， electrical potential， and flow velocity fields during the piping evolution process in earth-rock dams are integrated. By applying normalization and dimensionality reduction techniques， a matrix decomposition fusion method and principal component analysis （PCA） are employed to perform fusion analysis of the multi-field fitting functions. Subsequently， a quantitative characterization function with established correlations is constructed to analyze the logical symbiotic relationships among seepage pressure， temperature， electrical potential， and flow velocity fields. Finally， by analyzing the characteristics of different stages in the evolution of piping， the normalized timeline of piping evolution in earth-rock dams is divided into three stages： （0， 0.481）， （0.481， 0.663）， and （0.663， 1）. These stages correspond to the initial phase， expansion phase， and critical phase of piping evolution， respectively. The proposed method for multi-field information fusion and state identification in the piping evolution process of earth-rock dams offers a valuable reference for seepage safety monitoring and early warning in such structures.

While reservoir dams have made significant contributions to flood control， power generation， navigation， and other areas， they also pose potential safety hazards， such as the risk of dam failure. With the increasing frequency of extreme weather events， the risk of dam failure has grown， making dam breach flood analysis especially important. The simulation of flood evolution in the event of a dam breach is a critical method for risk analysis. Previous studies have generally used the broad-crested weir formula for simulating breach flow， which does not account for the influence of factors such as downstream water depth and the development process of the breach. To fully consider these factors， this study proposes a two-dimensional dam breach flood evolution coupled breach model based on the summary of prior research results. This model conceptualizes the development process of the breach shape and integrates reservoir water balance analysis with two-dimensional hydrodynamic computations to simulate both the breach formation process and the downstream flood evolution. It enables the automatic calculation of breach flow and real-time exchange of data with the downstream flood evolution. The model is validated using empirical formulas and experimental data， with the simulation results showing good agreement with both the empirical formulas and experimental observations， confirming the reliability of the model. The model is then applied to simulate the dam break flood of a specific reservoir. Based on the simulation results， the processes of flood flow variation， reservoir water level changes， flood inundation， and the evolution of the flood at key points and typical cross-sections are analyzed. The results demonstrate that the simulated outcomes of the proposed model are reasonable， consistent with the actual physical processes， and suitable for application in dam break flood studies.

To achieve the safe and stable operation of reservoirs and the efficient utilization of water resources in the basin， the key lies in clarifying the operation rules and dispatching strategies of reservoirs. To this end， aiming at the problems that traditional neural networks have when simulating reservoir dispatching operation strategies， such as vanishing gradients， getting stuck in local optimal solutions， and difficulty in determining hyperparameters， an improved deep neural network model （TSD） combining the Stochastic Gradient Descent with Momentum （SGDM） and the Tree-structured Parzen Estimator （TPE） is proposed. It has enhanced the accuracy and applicability of the deep neural network model by improving the training method of network parameters and automatically optimizing hyperparameters. Meanwhile， considering the numerous factors affecting reservoir dispatching decisions， the F-test method （FR） and the Mutual Information method （MIR） have been adopted respectively to achieve the optimization of input factors. Further， six simulation methods for reservoir dispatching strategies were constructed based on the Decision Tree model （DT）， Support Vector Machine model （SVM）， and improved deep neural network model： FR-DT， MIR-DT， FR-SVM， MIR-SVM， FR-TSD and MIR-TSD. And case studies were carried out with four reservoirs， namely Ahai， Jin’anqiao， Ertan and Pubugou. The research results show that compared with other methods， the FR-TSD and MIR-TSD methods have larger deterministic coefficients and smaller root mean square errors and mean absolute errors when simulating the dispatching strategies on various reservoirs. This indicates that the simulation effects of the FR-TSD and MIR-TSD methods are better and the accuracy is higher， and they can simulate the dispatching operation strategies of reservoirs more accurately， thus providing decision support for the actual operation of reservoir dispatching. Furthermore， an analysis of the simulation results of different reservoirs under the same method reveals that the simulation effect of the model is influenced by the regulating capacity of the reservoir and the allowable range of water level variation. Reservoirs with smaller regulating capacity and allowable range of water level variation have poorer simulation effects， while those with larger ones have better effects.

The change of temperature stress in high altitude area has a significant influence on the overall bearing capacity of concrete double-curvature arch dam. In order to analyze the overall bearing capacity of the concrete double-curvature arch dam of ZY hydropower station under temperature effects， and ensure the safety design， construction and operation of the project， a three-dimensional nonlinear finite element analysis method of TFINE program was employed to establish a numerical model of the integrated arch dam-foundation system. Firstly， under normal water level conditions， the displacement， stress and yield state of arch dam under temperature decrease and increase scenarios are analyzed， and unfavorable conditions are obtained. Then， under this working condition， the bearing capacity of arch dam is analyzed by overload method， such as arch dam displacement， unbalanced force， complementary energy norm and yield zone volume. The results show that under the action of temperature stress， the safety of the arch dam meets the requirements under the normal water level condition， and the temperature drop condition is a more unfavorable condition. When the overload ratio is 1.2~1.5， the upstream dam heel cracks； when the overload ratio is 4.0~5.0， the dam body is in nonlinear deformation； when the overload ratio is 7.5~8.5， the dam body loses its bearing capacity.

The aim of this paper is to explore the spatial distribution of ecosystem services， the trade-offs and synergistic relationships among ecosystem services， and the key factors driving the changes of these services in the Han River Basin. Five ecosystem services， namely， habitat quality （HQ）， water yield （WY）， carbon storage （CF）， soil conservation （SC）， and net primary productivity （NPP）， were selected for analysis， and the synergies and trade-offs among them were assessed using Spearman’s method， and the XGBoost-SHAP model was adopted for the analysis of driving factors. The results showed： ① Each ecosystem service showed significant imbalance in spatial and temporal scales. Temporally， water production and soil retention showed a trend of decreasing and then increasing， habitat quality showed a trend of increasing and then decreasing and then increasing， and net primary productivity and carbon storage showed a trend of continuous increase. Spatially， water production was more affected by rainfall presenting a larger yield in the eastern region， while the other four ecosystem services generally showed smaller values in the east. ② Water yield showed obvious trade-offs with the other four services， especially with carbon storage， while the other four services mainly showed synergistic effects. ③ Rainfall is the main natural factor influencing water yield， while elevation has a significant effect on net primary productivity， habitat quality and carbon storage， and slope is a key determinant of soil conservation.

The increasing nitrogen input load in the Pearl River Delta urban agglomeration has led to a rising trend in total nitrogen （TN） levels in the Pearl River Estuary. To reveal the spatiotemporal variations of different nitrogen forms and identify the sources of nitrate nitrogen （NO??-N） ， the dominant inorganic nitrogen species， surface water sampling was conducted during both flood and ebb tides in the wet and dry seasons of 2023 across the complex river network of the estuary. Based on the analysis of the spatiotemporal distribution patterns of various nitrogen forms， nitrate sources were quantitatively apportioned using nitrate isotopes （δ1?N-NO?? and δ1?O-NO??） and the SIAR model. Results indicated that nitrate was the predominant form of inorganic nitrogen， with a mean concentration of 1.85 mg/L. Except for a few sampling points affected by upstream cross-border inflows， ammonia nitrogen （NH??-N） and nitrite nitrogen （NO??-N） concentrations were generally low at most monitoring sites， typically below 0.50 mg/L and 0.10 mg/L， respectively. Although higher nitrogen concentrations were observed at some locations during the dry season， their contribution to the annual total load was substantially lower than that during the wet season. The isotopic compositions ranged from +3.08‰ to +10.03‰ for δ1?N-NO?? and from +0.51‰ to +5.12‰ for δ1?O-NO??. Quantitative source apportionment revealed that the primary contributors to nitrate pollution were agricultural sources （44.7%）， followed by domestic sewage （36.4%） and industrial wastewater （18.9%）. No significant differences in contribution proportions were observed between flood and ebb tides； during flood tide， the contributions from agricultural， domestic， and industrial sources were 45.6%， 36.2%， and 18.2%， respectively， while during ebb tide， they were 43.4%， 36.5%， and 20.1%. The significant spatial heterogeneity in nitrogen concentrations and fluxes， along with the identified source characteristics， provides a scientific basis for developing targeted total nitrogen pollution control strategies in the Pearl River Estuary.

The Jinsha River serves as the main stem of the upper Yangtze River， contributing approximately 50% to the total runoff discharge and 60% to the total sediment load in the upper Yangtze River Basin. Its extensive tributary network and complex sediment sources render its hydrological dynamics highly complex. In recent years， the operation of key cascade reservoirs has significantly altered flow and sediment transport characteristics， necessitating an in-depth analysis of these changes. In this study， hydrological data （1977-2020） from the Xiangjiaba（Pingshan） Hydrological Station in the lower reaches of the Jinsha River were used to investigate sediment transport dynamics during flood events before and after the impoundment of the Xiangjiaba Reservoir and Xiluodu Reservoir. Flood events were extracted from the time series data， and analytical methods， including Mann?Kendall （M?K） analysis， sediment rating curve analysis， and hysteresis pattern analysis methods， were systematically applied. The key findings indicated that a 99% reduction in sediment load occurred during the postimpoundment period， which was primarily attributable to reservoir sediment trapping. The discharge?sediment concentration relationship exhibited fundamental regime shifts. The original robust power-law relationship between water discharge and sediment load was disrupted after reservoir operation. Under moderate and high discharge levels （10 000~13 000 m3/s）， the sediment transport capacity was no longer governed primarily by hydrodynamic forces but was instead constrained by sediment supply limitations. However， this flow range remained the dominant sediment transport range. The sediment rating curves significantly flattened （the coefficient decreased by 87%）， reflecting reservoir-induced flood peak attenuation and changes in the discharge?sediment relationship. Asynchrony analysis of flood events revealed that under natural conditions， sediment source depletion during flood events led to insufficient subsequent supply， causing sand peaks to exhibit predominantly leading asynchrony. After reservoir operation， a notable change in asynchrony characteristics between flood and concentration peaks occurred， with the sediment peak lag pattern becoming dominant， driven by a substantial weakening of hydrodynamic conditions due to elevated reservoir water levels. This study elucidated the mechanistic shifts in water?sediment transport in the lower Jinsha River， thereby providing theoretical foundations for the formulation of reservoir sediment management strategies.

The morphology of river channels is crucial in the management and conservation of river systems. A thorough understanding of the evolutionary processes of river morphology significantly enhances the ability to manage and protect these vital watercourses. This study focuses on the morphological characteristics of the channel between Yichang and Datong in the middle and lower reaches of the Yangtze River， using Landsat remote sensing imagery from 1988 to 2023. Key morphological indicators， including the area of river islands， average water surface width， meander ratio， and bifurcation coefficient， were assessed. By incorporating coefficients of variation， change rates， and recovery rates， the study analyzed the sensitivity and recoverability of the river's morphological adjustments in response to changing environmental conditions. The results indicate several important trends in the morphological evolution of the Yangtze River： ① From 1988 to 2002， the water and sediment conditions in the middle and lower reaches of the Yangtze River were relatively balanced， with a notable 11% increase in river island area and an 8% decrease in average water surface width. However， following the operation of the Three Gorges Reservoir and upstream cascade reservoirs after 2003， substantial changes in water and sediment conditions occurred. As a result， the area of river islands decreased by 9%， while the average river width increased by 6%， with minor changes in sinuosity and bifurcation coefficient. ② The river island area was the most sensitive morphological indicator during the study period. Significant differences were found in the sensitivity of morphological indicators before and after the construction of the Three Gorges Reservoir， indicating its substantial impact on river morphology. ③ Under the regulation of the Three Gorges Reservoir， the average annual rate of morphological change decreased from 1% before the construction of the reservoir to 0.33% afterward. The rate of channel morphological evolution decelerated， reflecting the long-term impact of the reservoir on the river system. From 2003 to 2005， the rate of morphological change was 12%， largely due to the continuous erosion of the river channel. However， as erosion stabilized， the recovery rate declined from 52% in 2005 to 34% by 2023. In summary， the evolution of river morphology in the middle and lower reaches of the Yangtze River has been profoundly influenced by human activities， including the construction of the Three Gorges Reservoir and upstream cascade reservoirs， riverbank protection works， and sand mining.

Yellow River Pump Stations Face Severe Sand Erosion and Cavitation Issues： Application of Epoxy Carborundum and Polymer Ceramic Composite Coatings for Pump Protection and Repair.The pumps at Yellow River water-lifting stations are severely affected by sand erosion and cavitation. Sand particles in the water and the collapse of cavitation bubbles cause pitting， material loss， and gradual degradation of the pump's flow components. Without protective measures， this can lead to reduced structural strength， decreased water output， lower efficiency， increased energy consumption， and even equipment failures， impacting water supply reliability. This paper analyzes the application of epoxy carborundum coatings and polymer ceramic composite coatings in addressing pump erosion issues， based on the practical experience of pump erosion control at the Hongsibu Pump Station in Ningxia. It explores the characteristics of these coatings in erosion protection， proposes technical approaches for applying epoxy carborundum or polymer ceramic composite coatings to pump bodies， and provides details on application timing， methods， and expected outcomes. Additionally， it suggests applying polymer ceramic composite coatings to pump impellers， outlines relevant technique requirements， and describes the impeller coating effects. Field inspections have shown that the application of epoxy carborundum and polymer ceramic composite coatings in Yellow River water-lifting pump stations can extend the major overhaul cycle of pumps by approximately 1 000 hours and increase their effective service life by more than 100%. The coating protection model introduced in this paper provides technical support for erosion protection in Yellow River pumps.

Vibration monitoring of hydropower units plays a crucial role in evaluating unit health status and providing early warnings of deterioration. Accurate prediction of the trend of vibration deflection of unit components can detect the abnormal operation of the unit in advance， and effectively enhance the safety and stability of the unit operation. However， vibration oscillation signals of hydropower units monitored on-site contain noise， and the current methods for predicting the vibration oscillation of hydropower units mostly predict the single position of the hydropower units， leading to incomplete evaluation of the operational stability of the hydropower units， and the interaction among the vibration oscillation of multiple measurement points is not fully considered. To address these challenges， this paper proposes using Cross Graph Neural Networks （CrossGNN） to predict the vibration oscillation signal of multiple positions of hydropower units. CrossGNN is a graph neural network model with linear complexity. The multi-component vibration signals of hydropower units can be interoperable across scales and variables， thus implicitly revealing the spatio-temporal relationship of vibration signals at multiple positions of hydropower units. In order to effectively deal with the noise in time dimension， the adaptive multi-scale identifier （AMSI） is used to construct a multi-scale time series with noise reduction effect. Then a cross-scale neural network is proposed to accurately extract scales with clear trend and weak noise. At the same time， cross-variable GNN is constructed based on the homogeneity and heterogeneity among different variables. By focusing synchronously on edges with higher significance scores and restricting edges with lower scores， the time and space complexity of CorssGNN can be linearly related to the length L of the input sequence， thus effectively improving the prediction accuracy and generality. Finally， taking a hydropower station unit in China as an example， the effectiveness of the proposed method is verified by predicting the vibration oscillation signal of six positions of the unit， which provides a new idea for the vibration prediction of the hydropower station.

Dams obstruct fish migration upstream and pose risks for fish moving downstream. To enhance the ecological connectivity of rivers， extensive and in-depth research has been conducted abroad on the risks to fish navigating dams， as well as upstream and downstream passage technologies. Compared with international studies， domestic research in China has largely focused on upstream fish passage technologies， while studies on the evaluation， design of downstream passage channels， and protective measures for fish are quite lacking. With China's growing emphasis on natural ecology， the restoration and reconstruction on river habitats and fish migration routes gradually commence， and the design， construction， and research of downstream fish passage routes become imperative. Conducting a comprehensive review research about the current state of safety research for fish navigating downstream through dams abroad， this paper highlights the main risks posed by dams to fish moving downstream， and details fish-friendly downstream passage channels or methods， including their principles， characteristics， and operational requirements. It discusses safety barrier types and layout strategies designed to be placed upstream of the diversion channel to prevent fish from entering the turbine and to guide them towards the entrance of the bypass， and compares the usage scenarios and limitations of different behavioral barriers （electric， acoustic， visual， and bubble barriers）. Finally， it summarizes the existing issues and challenges in current research on downstream fish passage through dams， and offers the following recommendations for future domestic research： ① Develop geometric and hydraulic design standards for downstream channels； ② Conduct long-term monitoring of fish navigating downstream through dams； ③ Focus on the performance and applicable conditions of downstream fish passage technologies and facilities； ④ Adopt a cross-disciplinary collaborative approach.

The high energy efficiency of hydraulic machinery has always been the direction pursued by many scholars. Centrifugal pump is a kind of widely used hydraulic machinery， and its efficiency is generally between 70% and 80%. The method of optimizing centrifugal pump and improving its efficiency is worth discussing. In this paper， a vertical single-stage single-suction volute centrifugal pump is used to calculate and analyze the simulated fluid. Based on RNG turbulence model， the fluid simulation calculation of centrifugal pump under seven different flow inlet conditions of 0.5Q_d ~1.3Q_d was carried out， and the hydraulic performance of centrifugal pump under large and small flow conditions was obtained. The main performance was low efficiency and high head under low flow conditions. The opposite was true for large large flow conditions， with high efficiency and low head. After that， the impeller of centrifugal pump was optimized by Bezier spline curve， which mainly shows the change of bending degree of edge profile of blade inlet. Five schemes including the initial model were established. The simulation numerical calculation of different schemes was carried out under rated working conditions， and the changes of efficiency and head with the bending of blade inlet edge profile were analyzed. The difference of blade pressure， velocity vector and outlet pipe flow pattern under different schemes was compared. Finally， the optimal scheme was obtained， with the efficiency slightly increased to 91.2% and the head became closer to the design head.

Under the combined influence of climate change and human activities， urban flooding disasters have become increasingly frequent， posing severe impacts on socio-economic development. To effectively address urban flood risks， conducting scientific flood simulation and early warning studies in urban watersheds has become a top priority. This study takes the Wusha River Basin in Nanchang City， Jiangxi Province， as the research area. It constructs the WRF （Weather Research and Forecasting） model and the MIKE SHE/MIKE 11 coupled model to simulate precipitation， river water levels， and runoff processes in the urban watershed， respectively. Firstly， a four-layer one-way nested grid WRF model was developed to simulate a precipitation event in the basin in July 2020. The results show that the WRF model effectively captures the precipitation characteristics of the basin， with a simulation bias （Bias） of -0.3 mm. Secondly， the MIKE SHE/MIKE 11 coupled model was constructed to simulate the rainfall-runoff process in the Wusha River Basin from January 1 to August 26， 2022. Within this framework， the MIKE 11 model was primarily responsible for simulating river water level changes， providing hydrodynamic boundary conditions for the MIKE SHE model. Taking the river water level simulation near the Wanli Station of the Wusha River as an example， the MIKE 11 model achieved a coefficient of determination （R2） of 0.86， indicating its ability to accurately reflect the dynamic trends of river water levels. Meanwhile， the MIKE SHE/MIKE 11 coupled model integrated surface and subsurface hydrological processes in the basin， demonstrating strong performance in simulating the measured runoff at Wanli Station. Over the entire simulation period， the Nash-Sutcliffe Efficiency （NSE） coefficient of the model reached 0.7， with simulated flow rates closely matching observed values. The WRF model and the MIKE coupled model established in this study provide effective technical support for urban flood early warning and disaster risk assessment， holding significant importance for the formulation of urban flood prevention and mitigation strategies.

Under the drive of the “dual carbon” goals， hydro-wind-solar multi-energy complementary systems have become an important solution for decarbonization China's power system. However， most existing research focused on regulation reservoirs， neglecting the complementary operation of hydro-wind-solar systems with daily regulating reservoirs. This paper used the Duobu Hydropower Station and surrounding wind and solar power stations as research cases， constructed a short-term optimal scheduling model for hydro-wind-solar complementarity， and used dynamic programming to solve it. The typical daily water level changes， unit start-stop status， and load distribution of the hydropower station had been analyzed. The simulation results were compared with the actual operation results to verify the model effectiveness. The results showed that in typical days of non-flood season， the fore-bay water level decreased firstly and then increased， with only unit 1 and 2 operating. In typical days of flood season， the water level basically remained at the normal water level， with all units operating. The average power generation benefit by optimized scheduling in March 2023 was 3.052×10⁵ kWh， compared to 3.013×10⁵ kWh by historical actual operation， increasing the efficiency of power generation by 1.28%. The research results have important practical value for guiding the complementary operation of water-wind-solar power in the Duobu Hydropower Station.

The existing deformation monitoring models of concrete dams rarely consider the dynamic spatiotemporal correlation of multiple measurement points， thus limiting the prediction accuracy of deformation. To overcome this shortcoming， this paper constructs a dynamic prediction model of spatial deformation of concrete dam based on cluster analysis， long short-term memory （LSTM） network and attention mechanism， which can improve the prediction accuracy of deformation of target measurement point by introducing dynamic correlation factors reasonably. Taking the horizontal displacement monitoring data of several measurement points of a gravity dam as an example， the effectiveness of the model and the importance of spatial dynamic correlation factors are verified by comparing with three other LSTM-based models. Furthermore， the contribution degree of multiple factors to the model output is quantified by two interpretable machine learning algorithms for different measurement points， and the ranking results are consistent with the prior knowledge of dam engineering. The research results can provide reference for promoting the engineering application of intelligent dam deformation monitoring model.

Reservoirs primarily designed for flood control frequently employ either a single fixed discharge operation or compensation-based operation in actual scheduling practices. However， The lack of interconnected， coordinated， and unified flood control operations among reservoir groups within river basins hinders the full utilization of hydrological compensation effects and storage capacity compensation effects between reservoirs.Domestic and international scholars have conducted extensive research on joint flood control operation for cascade reservoirs， focusing on methodologies such as Linear Programming （LP）， Nonlinear Programming （NLP）， Dynamic Programming （DP）， and various intelligent algorithms. Nevertheless， due to inherent limitations in mathematical optimization algorithms—particularly the "curse of dimensionality" and poor convergence stability—these approaches have rarely been integrated into practical reservoir operation plans or real-time flood control scheduling systems.This study addresses this gap by deeply analyzing the inherent hydraulic linkages within cascade reservoir systems. We propose a novel， relatively simple and implementable method： the Cascade Reservoir Flood Control with Complementary Storage （CRFC-CS） method， achieved by coupling physical principles with numerical algorithms. Based on graded water level partitioning for the downstream reservoir（s）， a compensation-based control release model for cascade reservoirs is established. This model achieves dynamic inter-reservoir linkage and complementary utilization of flood control storage capacity， effectively addressing flood control scheduling challenges in basins characterized by complex flood regional composition. The CRFC-CS model is computationally tractable with clear and straightforward operational procedures， ensuring strong engineering practicality. It successfully achieves seamless coordination and unification across the flood control planning phase （e.g.， storage allocation， rule formulation） and the entire real-time flood operation phase， effectively resolving the disconnect between planning and execution.To validate its efficacy， the method was applied to flood control operations for upstream reservoirs in the Qingjiang River Basin （a primary tributary of the Yangtze River）. Compared to the conventional fixed discharge operation method， implementing CRFC-CS yielded a 52.5% reduction in the system's maximum flood retention volume while maintaining equivalent flood safety standards. This conclusively validates the practicality and superior performance of CRFC-CS in enhancing storage utilization efficiency and optimizing flood control operations. The method provides an efficient and reliable new solution for coordinated flood control of reservoir groups in complex river basins.

The characteristics of water and sediment transport in rivers include hydrodynamic and sediment transport features， which can help analyze the water and sediment movement of rivers and the changes in river geomorphology. The calculation and detection of water and sediment transport characteristics in river channels have important guiding significance for the safety monitoring and dredging of hydropower stations and reservoirs. In the calculation of water and sediment transport characteristics in rivers， empirical analysis of scenarios and energy balance methods require strict information on river parameters. In the calculation of diverse river channel shapes， obtaining hydraulic parameters often proves challenging. Therefore， in order to accurately determine the sedimentation state of the reservoir and evaluate the erosion and sedimentation morphology of the hydropower station channel， the effective shear method was used to calculate the water and sediment transport characteristics of the hydropower station， and the optimized calculation formula was embedded into the Delft 3D model to improve the sediment transport module of the model. Drawing on the optimization scheme of domestic shear force calculation methods， this study removed the difficult to obtain hydraulic radius and slope data of the river channel， and simplified the calculation of effective shear force. And the optimized shear calculation method was integrated into the sediment transport module of Delft 3D， while combining the hydrodynamic module of the model to obtain accurate river sediment carrying capacity data. In the actual measurement and verification of typical rivers in China， the reliability of the Delft 3D model and effective shear calculation method was verified through channel information and indoor calculations. Experiments have shown that compared with other methods， the mean square error of the improved effective shear method is the smallest， at 0.56， and the correlation coefficient is the highest， at 0.83， indicating that the improved effective shear method performs the best in the algorithm. Meanwhile， in the simulation experiments of hydropower stations and river channels， the average absolute error of flow velocity calculation decreased by 0.04， and the relative error decreased by 16.5% after model optimization. In the calculation of sediment concentration， the average absolute error of sediment concentration after model optimization decreased by 0.1kg·m^-3， and the average relative error decreased by 38.2%. The experiment shows that the model constructed in the study has significantly improved the predictive performance of river flow velocity and sediment concentration， exhibiting higher accuracy and reliability. After optimizing the Delft 3D model through effective shear calculation methods， the model exhibits better compatibility. By constructing a calculation method for the sediment carrying capacity of rivers， it is possible to simulate the sedimentation amount of reservoirs in the next 40 years， providing scientific support for the judgment of sedimentation status in hydropower stations and providing data basis for the regulation of reservoirs under hydropower stations. Therefore， the simulation method of water and sediment transport characteristics based on Delft 3D model and effective shear force has practical value in the operation and management of hydropower stations.

The linkage effect of diversion water level during the construction diversion of large-scale water conservancy projects has a significant impact on flood control and safety. Existing research mostly focuses on the analysis of single project construction diversion， and the study of hydraulic characteristics of multi-regional collaborative construction diversion is not sufficient. Based on a two-dimensional mathematical model of water flow， taking the section from Qishi navigation junction to Qingnian navigation junction of Pinglu Canal as an example， the variation law of water surface profile in different diversion types of river sections was analyzed， and the linkage effect of water level along the river during multi-regional collaborative construction was explored. The results show that the water level along the river is significantly affected by the diversion method and terrain conditions. At low flow rates， the water level mainly changes with the longitudinal shape of the riverbed. As flow rate increases， the longitudinal control ability of the riverbed weakens， and the river's plane shape and local bends gradually become dominant factors. There are significant spatial differences in the linkage between collaborative construction and water level in different diversion types of areas. Cutting and connecting bends can effectively reduce the water level in the upstream and local river sections， and combined with river channel expansion can further increase the water level drop. However， local bend checkpoints weaken the impact of excavation on water level， leading to a weakened regional linkage effect. The impact of return water caused by changes in tailwater level is jointly determined by collaborative construction and the size of incoming flow. At low flow rates， tailwater changes can be well transmitted upstream and have a wider range of effects， while at high flow rates， they are mainly limited by the cut and curved river sections that are not connected at the end. Based on the above findings， in future similar engineering construction， priority should be given to connecting the curved river section， improving the water level control capability through section excavation， and arranging construction projects reasonably to optimize the flow terrain and strengthen regional synergy. The research results can provide reference for multi-regional collaborative construction diversion optimization and water level regulation decision-making in large-scale water conservancy projects.

Branching channels are common in nature. After the construction of the upstream reservoir hub， the downstream clear water scouring causes the riverbed armoring of the branching channel to become normal. Therefore， it is of great significance to study the laws related to the riverbed armoring of the branching channel under the action of clear water. Through the flume test， it is found that in the dynamic process of the damage and formation of the armoring layer in the branching river bed caused by clear water scouring， the bed-load transport rate presents a fluctuating change of first increasing and then decreasing. And with the increase of flow， the peak value of the transport rate increases， the occurrence time is advanced， and the formation time of the armoring layer is shortened. The armoring degree of the main branch is higher than that of the branch， and the increase of discharge intensifies the damage of armoring layer， resulting in the overall increase of sediment diameter and the reduction of particle size distribution. In addition， the erosion and deposition of riverbed topography are significant. The main branch erodes violently and the branch deposits， and there is a dynamic balance between the flow and the riverbed topography.

Floods are among the most common natural disasters worldwide， and accurate flood forecasting is essential for disaster prevention and emergency management. Traditional flood forecasting models often face limitations in capturing complex rainfall patterns and peak flow characteristics. To address these challenges， this study proposes a Long Short-Term Memory （LSTM） model enhanced with a multi-head attention mechanism and a customized Floss loss function. Taking the Yongjiang River Basin in Nanning City as a case study， we collected rainfall and flood peak data from 15 historical flood events between 2008 and 2024. Using a sliding window approach， we expanded the dataset into 320 training samples. To improve the model′s generalization and convergence performance， K-fold cross-validation was applied during training. In addition， Particle Swarm Optimization （PSO） was used to automatically tune key hyperparameters such as network structure and learning rate. To mitigate the risk of underestimating flood peak levels， we designed the Floss loss function， incorporating a penalty term for underestimation and a water-level-based weighting scheme to enhance sensitivity to high water levels. In the testing phase， we compared LSTM models with and without multi-head attention mechanism under the Floss loss setting， and further evaluated the impact of different loss functions—including Huber、 MAE、 MSE and the proposed Floss—on predictive performance. The results indicate that： ①The attention mechanism significantly improves predictive accuracy， reducing the test RMSE by 41.6% from 1.9642 to 1.1462 compared to the baseline model. ②The Floss loss function， through its underestimation penalty （β=1.206 9） and water-level weighting （α=1.0）， effectively reduces underestimation errors， achieving a lower RMSE（1.1462） than Huber（1.1834）， MAE （1.1864） and MSE （1.2313）. ③The attention-based LSTM model using Floss shows no underestimation in three independent flood events， with maximum errors within 1.15 meters. These findings demonstrate that incorporating attention mechanisms and a tailored loss function can significantly enhance model accuracy and robustness， offering new methodological and technical support for intelligent flood forecasting.

The anisotropy of karst media leads to the complex and variable dynamics of karst water， and studying the response mechanisms of karst water systems is of great significance for the protection of springs. This paper takes the karst water system in Jinan as an example， and uses methods such as time series， correlation analysis， and sliding window analysis to study the dynamic characteristics of karst water， clarify the grading and development patterns of the groundwater flow system in the karst water system， and reveal the response of groundwater levels in different hierarchical groundwater flow systems of karst spring domains to atmospheric precipitation. The results show： ① The groundwater level response within the spring domain has a high degree of seasonal and cyclical variation characteristics， with different groundwater dynamics in different hierarchical flow systems， where local flow systems exhibit drastic and large-amplitude changes in groundwater dynamics， while regional flow systems show stable and small-amplitude changes. ② The groundwater dynamic response lag time of different hierarchical flow systems is different， which is characterized by regional features. It is mainly controlled by topography， geological structure， lithology， and the intensity of fracture karst development. Fault structures and fracture karst shorten the response lag time of groundwater dynamics， and local flow systems transform groundwater into regional flow systems， extending the gushing time of karst springs. ③ There is a lag in the initial response time of groundwater levels in different hierarchical flow systems to subsequent precipitation， with local flow systems having a response lag time of 0.75 to 2.32 days， and regional flow systems having a response lag time of 30.79 to 33.35 days. The degree of karst development， lithology of strata， geological structure， and the extent of fracture karst development are important factors for the differences in response lag times. The research results are expected to enrich the theoretical research of multi-level groundwater flow systems in the Jinan spring domain and provide technical support for the protection of springs， water supply， and ecological replenishment in the Jinan spring domain.

Combined supporting structures consisting of steel set and shotcrete are widely used in underground tunnels when the geological conditions are poor. However， there exist many difficulties in conducting nonlinear finite element analyses for the steel set due to its small size and dense arrangement. Therefore， according to the mechanical properties of the steel set， a four-node isoparametric element model of the steel set structure was proposed， and its mechanical action was simulated by stiffness superposition. Based on the Mises yield criterion and the secondary plastic flow model， an elastoplastic constitutive model was established to simulate the yielding and hardening behavior of steel set. The elastic damage constitutive model was used to simulate the damage softening behavior of concrete， and finally， a nonlinear numerical simulation method for the combined supporting structure of steel set and concrete was formed， which is verified to be reasonable by a planar circular tunnel example. This method was used to investigate the force characteristics of the combined supporting structure， and the results show that： when the surrounding rock pressure is small， the concrete is in an elastic state so it has a high bearing capacity and is the main bearing structure； when the surrounding rock pressure is large， the concrete undergoes compression damage， and the excess stress is transferred to the steel set， resulting in a significant increase in the stress of the steel set， and the steel set evolves into the main bearing structure； reducing the spacing between steel sets can to some extent reduce the steel set’s stress， slow down the development of concrete damage， and enhance the supporting effect of the combined supporting structure.

Groundwater level is a critical indicator for assessing the sustainable use of groundwater resources and serves as a key basis for groundwater management. Accurate prediction of groundwater levels is essential. Traditional numerical models， such as MODFLOW， are complex to construct and require numerous parameters， while data-driven neural network models often lack interpretability and generalization capabilities. In this papar， a Groundwater Physics-Informed Neural Network （GWPINN） model is developed to solve the groundwater flow equation by integrating Physics-Informed Neural Networks （PINNs） with the governing equation of groundwater flow. The model employs fully connected neural networks， using spatial-temporal coordinates as inputs and groundwater head as outputs. It incorporates the groundwater flow equation， initial conditions， and boundary conditions as physical constraints， and combines the measurement data to construct a loss function to solve the groundwater flow equation. To evaluate the effectiveness and accuracy of the algorithm， we applies the GWPINN model to single pumping well of the confined aquifer. The simulation results are compared with those of the traditional numerical model， MODFLOW. The findings show that when relying solely on physical constraints without any data， the GWPINN solver achieves a Mean Absolute Error （MAE） of 0.063 m and a Relative Root Mean Square Error （RRMSE） of 0.083%. When combining physical constraints with data-driven approaches， the MAE is 0.069 m， and the RRMSE is 0.079%. These results demonstrate that the algorithm can effectively solve groundwater flow equations， especially for those with source of wells. Additionally， the impact of spatial and temporal sampling frequency on the solution results is investigated. The study reveals that a denser spatial sampling interval of 100 meters near the well source is required to capture the rapid changes in groundwater head. On a temporal scale， an increased sampling frequency with an interval of 0.3 days is necessary at the start of pumping to capture the rapid changes in water head near the well. However， in regions farther from the pumping well or after 1 day of pumping， the spatial and temporal sampling frequency can be appropriately reduced to intervals of 200 meters and 5 days， respectively.

The joint optimal operation for sewage pumping stations is of great significance to improving the quality and efficiency of urban sewage system operation and management. Based on the model predictive control （MPC） method and the mechanism model of drainage system coupling heuristic optimization algorithm， a joint optimal scheduling decision-making system for sewage pumping stations was developed to meet the diversified needs of sewage overflow control， energy conservation in pumping station operation and convenient maintenance. The urban drainage network model－storm water management model（SWMM） was used to construct the hydrodynamic and water quality model of sewage system， and the optimal scheduling model was established based on particle swarm algorithm （PSO）. The two-stage screening mechanism was used to ensure that the global optimal feasible solution was found. The total energy consumption of pumping stations and the complexity of the control scheme before and after optimal scheduling was compared， providing a scientific basis for the integrated scheduling of sewage system plant network.