The Pinglu Canal construction involves extensive cutoff engineering， which creates semi-enclosed river segments with reduced flow velocity， thereby limiting hydrodynamic connectivity and adversely affecting wetland habitats and biodiversity. To optimize water exchange efficiency， this study employs three-dimensional numerical simulations to analyze the rectification scheme of gravel spur dikes in a representative cutoff section of the Pinglu Canal. Key parameters including water exchange time and spatial distribution of low-velocity zones are investigated to evaluate how channel narrowing rates and weir spacing affect water exchange efficiency. Further optimization is achieved through perforation strategies at dike roots and heads. Results demonstrate that under operating parameters of 175m dike spacing and 15% narrowing rate，the straightened section exhibits the shortest water exchange time and the smallest proportion of low-velocity zones at typical cross-sections. Water exchange time is reduced by 29.7% to 35.6% compared to pre-remediation conditions， with low-velocity zones in characteristic cross-sections decreasing to 11.3%、8.9% and 19.5%. Additionally， incorporating openings in the dike bodies further improved flow velocity distribution characteristics between dikes， with perforations at dike roots exhibiting superior performance in flow redistribution compared to head modifications. This research provides a scientifically validated approach for hydrodynamic optimization and ecological restoration in navigational channel engineering， offering practical solutions to mitigate environmental impacts of river regulation projects.

This study investigated the spatiotemporal evolution of vegetation cover and land use in seven typical counties of Chifeng City within the Upper Xiliao River Basin， and examined their effects on river sediment concentration， providing scientific support for ecological management and soil-water conservation in semi-arid regions. Using multi-source remote sensing data （MODIS NDVI and land use data） from 2000 to 2024， combined with hydrological station observation data， we employed trend analysis， spatial statistics， and correlation analysis to systematically assess the dynamics of vegetation， land use changes， and their comprehensive effects on river sediment concentration. The results indicate that： ①From 2000 to 2024， vegetation cover in the study area improved significantly （NDVI annual growth rate of 0.003 0， p<0.05）， showing a “low in the north， high in the south” spatial distribution and exhibiting clear seasonal fluctuations. ②The land use pattern changed notably， with forest area increasing from 6.9% to 10.8%， grassland area decreasing from 41.9% to 38.3%， bare land area decreasing from 7.5% to 4.9%， agricultural land fluctuating， and impervious surfaces continuously expanding. ③River sediment concentration significantly decreased， with annual change rates at Xinglongpo， Taipingzhuang， and Dianzi stations being -0.958 4、-0.467 4 and -0.387 4 kg/m3， respectively. ④River sediment concentration showed significant negative correlations with NDVI and forest area （r ≥ 0.602 and r ≥ 0.514）， while had a significant positive correlation with bare land area （r ≥ 0.699）. These findings suggested that the improvement of vegetation cover and optimization of land use in the Upper Xiliao River Basin have jointly contributed to the reduction of river sediment concentration. Future efforts should focus on ecological restoration and land use optimization， particularly in areas with lower vegetation coverage， to reduce sedimentation risk and enhance soil and water conservation capacity in the watershed.

In long-distance water conveyance channels， variations in cross-sectional shapes frequently occur to accommodate differing design flow rates or terrain conditions across sections. Such changes in the channel cross-section can lead to fluctuations in the water surface， potentially exceeding the safety height limits of the channel. Consequently， simulating water flow in channels with non-uniform cross-sections holds significant importance. Currently， the explicit finite volume method is extensively employed for studying such problems. However， its computational efficiency is relatively low in long-distance channel simulations due to the strict time-step constraints imposed by the Courant condition. The Preissmann method， known for its unconditional stability， is widely adopted for simulating unsteady channel flows. Nevertheless， there remains a lack of studies evaluating its applicability in simulating water flow in channels with gradually varying cross-sections. In this paper， we derive the one-dimensional governing equation form of open channel flow in non-uniform cross-sections suitable for the Preissmann method and introduce the discretization steps using this method. The accuracy of the Preissmann method in simulating steady and unsteady flows in gradually changing cross-sectional channels is assessed by comparing simulation results with analytical solutions from the literature or simulation outcomes obtained using the Hec-Ras model.

Most studies on the automatic control of open canal systems only focus on the distant downstream water level of each canal pool. However， for irrigation open canals， the offtakes are large in number and scattered in location. Controlling only the distant downstream water level of each canal pool may lead to uncontrollable water level before the upstream offtake gates， resulting in low distribution efficiency of irrigation water and an insufficient water supply guarantee rate during key growth periods of crops. To address this problem， starting from the simplification of Saint-Venant Equations， a linearized control-oriented model that can predict the multi-point hydraulic response alongside the canal was derived. The model has good prediction accuracy for the hydraulic response of the sections of the distant downstream end and upstream offtakes. By distinguishing the requirements of different control points for water level regulation accuracy， a kind of differential setting strategy of control targets for different control points was proposed. Then， the multi-point hydraulic control modeling method for irrigation open canals was set up based on the soft constraint strategy. The single-pool test case and multi-pool test case derived from the classic test canal with a gentle bed slope proposed by the American Society of Civil Engineers were taken as controlled members. Furthermore， the model predictive controllers were designed， and the control performance was examined. The results show that， compared with the traditional single-point hydraulic control modeling method based on the Integrator-Delay model， the multi-point hydraulic control modeling method can actively control the hydraulic response of upstream offtake sections by adjusting the penalty weight of the controller objective function. The water level control stability before the offtake gates and the water supply reliability were improved by 91.0% and 69.5% in the single-pool test case， respectively， while 40.3% and 17.4% in the multi-pool test case. The multi-point hydraulic control modeling method proposed in this paper can fully utilize the limited control resources and improve the hydraulic controllability of the offtake sections along the irrigation canal， showing strong potential in precision irrigation.

The wave run-up is a crucial parameter in the design of river dikes. It is usually determined through model tests and empirical formulas. For a long time， there has been a lack of on-site measurement data to support this. To reasonably determine the height of the dike，it is very necessary to carry out on-site observations and analyses of the wave run-up along river dikes. From May to October 2024， a comprehensive all-weather riverine wave observation system was employed to perform synchronous monitoring of wave parameters， wind speed， water levels， and wave run-up at three representative river sections in Zhejiang Province： Xiasha section at Qiantang River， Zhashang section at Cao'e River， and Xinkaihe section at Changxing Port of Tiaoxi River Basin. Based on the field-measured data， the relationship between wave run-up and wave height was analyzed and compared with the existing formula of code about dike engineering. The results indicate that under no-flow conditions， for a sloped dike with a gradient of 1∶2， the ratio of significant wave run-up with significant wave height（R _s /H _s） is approximately 1.75 under normal wave action and approximately 1.5 under oblique wave action with the angle of 45°to dike. When oblique waves with the angle of 45° to dike act on vertical dike， the ratio of run-up with wave height （R _s /H _s） is 0.85， which is marginally lower than the run-up estimated basing on the formula of specifications for dike. According to the observed relationship between wave period and wave height， the calculation formula for run-up in code of dike can be optimized. This formula can be simplified from a dual-factor relationship incorporating both wave height and wavelength to a single-factor relationship that considers only wave height. The outcomes of this simplified run-up calculation correspond closely with the measured values. In Xiasha section of the Qiantang River， flood and ebb tidal currents reduce the run-up of normally incident waves. As a result， the ratio of wave run-up with wave height （R _s /H _s） for slope of 1：2 gradient is approximately 0.94， which is significantly lower than the ratio observed under no-flow conditions.

Gabion weirs are eco-friendly river training structures. Flow over gabion weirs can induce local scour， which threatens structural stability. Current studies on scour at gabion weirs ignore the effect of upstream sediment supply， leaving the characteristics of scour under such conditions unclear. Based on laboratory flume tests， this study investigated the scour patterns and the temporal development of scour dimensions of gabion weirs under the impact of upstream sediment supply， and analyzed the response of the equilibrium scour dimensions to sediment supply intensity. The results indicate that scour occurs immediately upstream and downstream of the gabion weir， with the contours of the scour holes being interconnected through the weir. When the sediment supply rate approaches the sediment transport capacity of the flow， the upstream and downstream scour dimensions initially increase rapidly， followed by significant oscillations around the equilibrium value （quasi-live-bed scour）. When the sediment supply rate approaches zero， the upstream and downstream scour dimensions oscillate initially， and then increase monotonically with time towards an equilibrium value without obvious oscillation （quasi-clear-water scour）. The equilibrium upstream and downstream scour depth and scour length at the gabion weir increase approximately linearly with decreasing sediment supply intensity or increasing flow intensity. Based on experimental data and theoretical analysis， predictors for the equilibrium upstream and downstream scour dimensions at the gabion weir were proposed， providing scientific support for the scour design of gabion weirs.

In order to explore the characteristics of flood sediment in the loess hilly and gully area of the Yellow River， this paper took the Dali River as the research object， based on the observation data of 292 flood events at the Suide Hydrological Station at the outlet of the basin from 1960 to 2019， and used the K-medoid cluster analysis method to select key hydrological parameters such as runoff depth （H）， flood duration （T） and peak flood flow （Q_p ） to divide four flood types. The runoff characteristics and sediment transport laws of four types of floods， namely small， medium， large and extreme， were studied by using the runoff depth-sediment transport modulus （H-SY） curve and the sediment content-flow （SSC-Q） hysteresis curve. The results show that the linear fitting of the runoff depth-sediment transport modulus （H-SY） of different floods in the Dali River Basin shows that the sediment transport modulus increases significantly with the increase of flood scale. Among them， the sediment transport modulus of extreme floods is 3 to 26 times that of other floods. Medium and large-scale flood events account for only 25% of the multi-year flood events in the basin， but the runoff and sediment production account for about 60% of the multi-year runoff and sediment production in the basin. In the sediment content-flow （SSC-Q） hysteresis curve of the Dali River Basin， the composite curve accounts for the largest proportion （73%）， and increases significantly with the increase of flood scale， reflecting the complex process of dispersed sediment sources and multiple batches of sediment entering the river under heavy rainfall over a long period of time. The implementation of various measures for soil and water conservation has reduced the sediment content of various floods by 20%–40% on average. However， the proportion of large-scale and extreme floods in the runoff and sediment production of floods in the basin is still increasing， which is closely related to the increase in the frequency of extreme rainstorms under climate change. This provides a scientific basis for the subsequent simulation of water and sediment processes， erosion prevention and control， and response to extreme climate events in the Loess Plateau basin.

To investigate the impacts of freeze-thaw cycles on the mechanical characteristics of riverbank soils and bank stability， freeze-thaw cycling tests were conducted on bank soil specimens with varying clay contents and moisture contents. Deformation measurements and direct shear tests were performed on specimens subjected to different numbers of freeze-thaw cycles. Subsequently， numerical simulations of typical riverbank cross-sections were implemented to analyze freeze-thaw effects on bank stability. The results indicate that during freeze-thaw cycling， low-moisture-content soils exhibit “freeze-shrinkage and thaw-expansion” volumetric change pattern， while high-moisture-content soils display “freeze-expansion and thaw-shrinkage” pattern， with volumetric change rates ranging from approximately 1.9% to 3.6%. Soil cohesion initially increases then decreases with rising moisture content， whereas the internal friction angle shows a monotonic decreasing trend. Freeze-thaw cycles significantly degrade soil shear strength of riverbank soils. Both cohesion and internal friction angle decrease progressively with increasing cycles， showing respective reductions of 10–47% and 9–19% after 7 cycles， with cohesion rather than internal friction angle being more susceptible to freeze-thaw effects. Unlike conventional riverbank behavior， freeze-thaw action induces accelerated bank destabilization， substantially increasing collapse risks during thaw periods and advancing failure occurrence timelines. Specifically， freeze-thaw affected banks collapse earlier than non-affected counterparts under identical conditions. These findings provide valuable insights for understanding riverbank collapse mechanisms and fluvial evolution processes in cold regions.

Drought-flood abrupt alternation， a compound extreme climate phenomenon， has caused serious negative impacts on the ecological environment， agriculture production， and social economy. In order to deeply understand the regional characteristics and overall evolutionary patterns of drought-flood abrupt alternation events in summer in China， this study uses observational data and employs two indices， the Long-cycle Drought-Flood Abrupt Alternation Index （LDFAI） and the Short-cycle Drought-Flood Abrupt Alternation Index （SDFAI）， to identify the drought-flood abrupt alternation events in summer across the eight sub-regions in China. It further reveals the spatial and temporal distribution， trends and future persistence of the frequency and intensity of these events， and finally focuses on the characteristics of the short-cycle drought-flood events with different alternation timings and their influencing factors. The results show that： ① Drought-flood abrupt alternation events are primarily concentrated in eastern China， particularly in South China （SC）， Southwest China （SW）， Northeast China （NE）， and Eastern Northwest China （ENW）. The frequency of both long-cycle and short-cycle flood-to-drought events shows a significant increasing trend. Additionally， the future trends in the frequency and intensity of these events across all the sub-regions are consistent with those observed over the past 58 years. ② The short-cycle drought-flood abrupt alternation events with alternation timing between May and June generally exhibit higher intensity and show the characteristics of gradual enhancement. ③ The short-cycle drought-flood abrupt alternation events with different alternation timings primarily concentrate in the mid-latitude regions （20°N~40°N）. In particular， the alternation timing of drought-flood abrupt alternation events corresponded well to the advance of summer rain belt in eastern China. Meanwhile， the occurrence frequency of severe flood-to-drought events with different alternation timings remains relatively stable and consistently high. ④ Comparing to the earlier period （1961-1990）， influenced by the earlier onset of the South China Sea summer monsoon， the drought-to-flood abrupt alternation events with alternation timing between May and June （June and July） increased （decreased） in eastern China during the later period （1991-2018）. Similarly， flood-to-drought events with alternation timing between June and July （May and June， July and August） increased （decreased）.

Against the backdrop of climate change and high-intensity human activities， large-scale coal development in arid regions poses severe challenges to water resource sustainability， exacerbating the conflict between coal mining and water scarcity. This study selects the Nalinxili coalfield—a typical arid area rich in coal resources with important surface river and lake systems—as the research area， to quantitatively assess the impact of under-river and lake mining activities on the dynamic interaction processes between surface water and groundwater， aiming to provide a scientific basis for the sustainable utilization of regional water resources and ecological environmental protection. This paper innovatively constructs a multi-process dynamic model based on SWAT-MODFLOW coupling， effectively integrating surface hydrological and groundwater flow simulation processes， comprehensively considering the interactions of mining subsidence， the development of water-conducting fractured zones， and hydrogeological processes. Relying on subsidence prediction theory and a water-conducting fractured zone height calculation method， the model achieves real-time dynamic coupling between mining disturbance parameters and the hydrogeological model， thereby more accurately depicting the complex interaction mechanisms between surface water and groundwater under mining disturbance. The model validation results show that it has high simulation accuracy （R ²=0.785 1）. Simulation predictions indicate that continuous coal mining in the Nalinxili coalfield will lead to a significant decline in groundwater levels over the next 20 years， with localized maximum drawdowns expected to reach 15~20 m. Maximum surface subsidence is estimated to reach 5 400 mm. Simultaneously， the hydrological function of the regional lake Hadatu Naor will transform from a recharge source in wet seasons to a discharge area in dry seasons， further exacerbating the risk of regional water scarcity. In addition， the study reveals that the development height range of the water-conducting fractured zone is 28.71~116.35 m， which has not yet affected the deep main aquifer， providing critical boundary information for mine area deep groundwater pollution risk prevention and control. The research results can provide scientific support for similar mine areas to formulate adaptive water resources management strategies that take into account both energy development and water resources protection.

This article aims to assess the impact of climate change on the spatial extent and depth classes of flood disasters in the Yongding River floodplain. A coupled model was generated by integrating the SWAT model with the 2DHECRAS model. The coupled model includes 245 computational cross-sections and nine bridges， all calibrated using measured data. The measured data from the Lugouqiao Hydrological Station of the Yongding River and the simulated flow hydrograph are used as the basis for calibration. The correlation coefficients vary between 0.996 and 0.999， and the total squared errors range from 0.134 to 0.330. Three greenhouse gas concentration scenarios （RCP2.6； RCP4.5 and RCP8.5） were considered to forecast the extreme rainstorm and flood series. Different return periods were selected to evaluate the relative extreme rainstorm， flood grades， flood detention， flooded area， and inundated depths under the impact of climate change. According to the DEM data of the floodplain， flood inundation maps with different recurrence intervals in the floodplain are drawn. The results show that under the future climate scenario models， the flood detention in the floodplain （P=1%） are 201、 210 and 216 million m3 respectively. The relevant inundated areas are 129.4、 135.5 and 139.9 km2 respectively， accounting for 24.7%、 25.9% and 26.8% of the total area of the floodplain respectively. For P=0.1%， the flood detention is 335、 352 and 378 million m3 respectively. The relevant inundated areas are 295.8、309.8 and 318.6 km2 respectively， accounting for 56.6%、 59.3% and 61.0% of the total area of the floodplain respectively. It shows that under the influence of future climate change， the flood inundation area and intensity will increase significantly. Particularly， under the RCP8.5 climate scenario， the storm rainfall， total flood volume， and inundation range in the floodplain are larger than those under the other two climate scenarios. The results prove that it is a new way to combine the greenhouse gas effect with the hydrological model to conduct research on flood disaster projection in a flood plain.

The water imbalance between inflow and outflow of cascade reservoirs will significantly affect both power generation efficiency and navigation scheduling accuracy. To thoroughly investigate and address this issue， the cascade sections in the lower reaches of the Jinsha River were taken as the case study. Based on the basic geographic information， meteorology， runoff， and reservoir operation data of the basin， numerical analysis methods were used to analyze the interval and time distribution characteristics of water flow differences in the cascade sections of the Jinsha River. The SWAT model was utilized to simulate inter-reservoir inflow processes， and the key factors contributing to water imbalance were identified. Finally， the correction formulas for inflow-outflow discrepancies were constructed. Results reveal that： ① The most pronounced water balance discrepancy in the lower Jinsha River cascade sections occurs between the Xiluodu and Xiangjiaba reservoirs. Analysis of daily water volume data from 2013 to 2018 shows that 67% of the time periods exhibited inflow at Xiangjiaba being less than outflow from Xiluodu， mainly concentrated in non-flood seasons. ②The simulation results of SWAT model indicate that the mean inter-reservoir flow between Xiluodu and Xiangjiaba reaches 142 m3/s in flood season and 26 m3/s in non-flood seasons respectively， which should not be neglected. ③Through systematic evaluation of five potential mechanisms causing water imbalance，calculation errors in Xiangjiaba’s inflow-outflow measurements are identified as the primary contributor for the water imbalance. ④ After incorporating inter-reservoir inflow and applying the correction formula， the mean water difference between the inflow of Xiangjiaba and the outflow of Xiluodu in the flood season and non-flood season changes from 127.1 and -29.0 m³/s before correction to 7.4 and 4.8 m³/s after correction， respectively， and the negative time periods of the difference between the inflow of Xiangjiaba and the outflow of Xiluodu is significantly decreased. This study provides a simple， scientific and effective method for reviewing and correcting the water balance of the lower Jinsha River cascade reservoirs， offering significant theoretical and practical value. By correcting the water imbalance， the operational efficiency of the cascade reservoirs can be improved， and technical support can be provided to ensure the rational use of water resources and ecological balance in the Yangtze River basin.

Following the operation of the Three Gorges Reservoir （TGR）， the water level at Zhicheng Station in the middle Yangtze River has been significantly influenced by reservoir regulation and channel bed scouring. To investigate the impacts of TGR regulation and water level-discharge relationship alterations on Zhicheng Station′s water level， this study reconstructs the discharge and water level processes unaffected by TGR regulation by integrating the GR4J hydrological model with the Muskingum channel routing algorithm. Three simulation scenarios were established using the controlled variable method to quantitatively attribute water level variations. The results indicate that after TGR operation， trends between discharge and water level variations at Zhicheng Station exhibited certain asynchrony. Summer， autumn， and annual mean water levels exhibited more pronounced declining trends compared to corresponding discharge patterns， whereas the rising trends of spring and winter water levels were notably weaker than those of discharge. Channel bed incision has modified the water level-discharge relationship， resulting in varying degrees of water level decline under identical discharge conditions. From 2003 to 2019， Zhicheng Station experienced a total water level decline of 0.36 m， with 99.7% attributed to discharge-water level relationship changes induced by channel bed erosion， while TGR regulation only contributed 0.01 m. The maximum seasonal decline in water level， reaching 1.01 m， was observed during autumn. This decline was attributed to changes in the water level-discharge relationship， contributing 45% to the decline， and the regulation operations of the Three Gorges Reservoir， contributing the remaining 55%. Although TGR water replenishment elevated water levels by 0.24~0.38 m during spring and winter， these increases were completely offset by water level declines caused by water level-discharge relationship modifications. TGR regulation exerted distinct seasonal impacts： during the initial impoundment phase （2003-2008）， its effects on spring， autumn， and winter water levels were -0.21 m， 0 m， and -0.20 m， respectively， shifting to 0.24 m， 0.38 m， and -0.60 m during the normal operation period （2009-2019）. This study enhances comprehensive understanding of TGR regulation’s hydrological impacts on Zhicheng Station and elucidates the mechanistic drivers of water level evolution in the Zhicheng reach.

River ice recognition serves as a technical prerequisite for ice-flood monitoring and timely emergency response. Currently， optical satellite remote sensing is an important means of river ice detection， yet it often depends on shortwave infrared （SWIR） bands and struggles to simultaneously meet the demands of large-scale and high-precision monitoring. The wide field-of-view of the Chinese GF-6 satellite offers new technological possibilities in this regard. Targeting the Ningxia–Inner Mongolia reach of the upper Yellow River， which is characterized by complex ice-jam conditions and extensive geographical span， this study utilizes the newly added visible bands from GF-6 WFV data. By analyzing the spectral differences between river ice， water bodies， and other land-cover types during the ice-jam season， a normalized river ice-water classification index （NRIWI） model is proposed using the violet band （0.40~0.45 μm） and the yellow band （0.59~0.63 μm）， enabling accurate identification of river ice distribution within the channel. To evaluate the performance of NRIWI， several other visible-spectrum-based indices applicable to river ice recognition were compared， including the Normalized Difference Water Index （NDWI）， Normalized Difference Unfrozen Water Index （NDUWI）， and Normalized Difference Glacier Index （NDGI）. Six typical areas with frequent historical ice-jam events along the Ningxia–Inner Mongolia reach were selected， including river sections controlled by hydraulic engineering structures， meandering sections prone to ice jams， and canyon-type sections affected by mountain shadows. River ice recognition experiments were carried out in these areas to analyze and compare the overall accuracy， Kappa coefficient， and threshold sensitivity of each index. The results indicate that the overall accuracy of NRIWI in all typical areas exceeds 96%， with Kappa coefficients above 0.9. Compared with other indices models， NRIWI provides higher recognition accuracy， lower regional variability， and reduced sensitivity to threshold changes. It effectively suppresses interference from small ice features， easily confused pixels， and mountain shadows， accurately recognizes river ice extent， and proves suitable for various ice-jam-prone sections within the Ningxia–Inner Mongolia reach of the Yellow River.

Reservoir regulation rules are important tools for guiding reservoir operations and promoting the efficient use of water resources. Current research on derivation methods for reservoir regulation rules rarely considers sediment-related factors. During flood seasons， the incoming sediment load to the Three Gorges Reservoir （TGR） accounts for 90.5% of its annual sediment influx. If existing methods are directly applied to extract flood season regulation rules for the TGR without considering the impacts of incoming and outgoing sediment on reservoir operations， the resulting rules may fail to comprehensively reflect the reservoir′s operational response process to water and sediment influxes from tributaries and main streams during flood periods. Therefore， this paper introduces sediment factors into the extraction of flood season regulation rules for the TGR. Using historical operation data and incoming and outgoing sediment data of the reservoir as samples， we propose a deep learning-based method for reservoir regulation rule extraction. This approach employs both the dam-front water level at the end of operational period and the period-averaged outlet sediment transport rate as joint decision variables， while incorporating the hydrological-sedimentary dynamics of upstream tributaries and main streams in the Yangtze River basin， along with their time-lag effects on reservoir regulations. Firstly， a RF-LSTM model is established， which uses the Random Forest （RF） algorithm to screen the decision factors that have a significant impact on decision variables. Then， to achieve the extraction and simulation of the reservoir regulation rules， we take these decision factors as input variables for the long short-term memory （LSTM） neural network to further recognize the mapping relationship between the decision factors and the decision variables. Subsequently， an autoregressive approach is employed to correct errors in the results with poor simulation accuracy. Finally， this method is applied to extract water and sediment regulation rules for the TGR during the flood season. The results show that compared with the LSTM model， the RF-LSTM model effectively screens key decision factors influencing reservoir operation decision variables and exhibits higher simulation accuracy and stability. The NSE values of dam-front water level at the end-of-period and time-averaged outflow sediment transport rate in validation period are 0.995 and 0.891， respectively， improved by 1.7% and 1.0%. After error correction， the NSE value of the outflow sediment transport rate increases to 0.944， an improvement of 5.6%， indicating a significant increase in simulation accuracy. The research results can provide a valuable reference for the extraction of flood season regulation rules for the TGR.

The acquisition of high-resolution precipitation data is fundamental to the accurate modeling of terrestrial water cycle processes. Although satellite precipitation products have significant advantages in terms of spatial representativeness and continuity， their relatively low spatial resolution and substantial uncertainty still limit their effectiveness in practical applications. A framework integrating satellite precipitation and gauge-based precipitation based on Bayesian combination method was proposed in this paper. The applicability of three satellite precipitation products （CMORPH， GPM， and TRMM） and their merged precipitation products （MCMORPH， MGPM， and MTRMM） were evaluated according to measured precipitation in the Zishui basin from 2001 to 2019. The results showed that ① For satellite precipitation products， the daily-scale accuracy is relatively low （CC≤0.583， RMSE≥7.57 mm）， and the correlation with the observed precipitation in the spatial scale is weak and the error is large（CC<0.44， RMSE>10.21 mm）. In contrast， the CC of the merged precipitation products with the observed precipitation reaches 0.988 and effectively reflects the spatial distribution of precipitation in the Zishui basin， but overestimates the observed precipitation （median RB of 1.72%~11.22%）. ② In terms of hydrological utility assessment， the SWAT model using observed precipitation data performs better in the Zishui basin， with simulation results reaching the “good” level during the calibration and the validation periods. The streamflow simulation results of the satellite precipitation products during the calibration and the validation periods are “satisfactory” with NSE of 0.59 and 0.54， respectively. The MGPM of the fusion precipitation product also performs “good” with NSE of 0.66 and 0.68 during the calibration and the validation periods， respectively， and MCMORPH和MTRMM reach the “satisfactory” level. Compared with the original satellite precipitation products， the evaluation index values of the merged precipitation products are significantly improved and have better applicability in the study area. The research results can provide reference for precipitation data fusion and hydrological simulation in data-scare regions.

The shallow groundwater aquifer in Yanghuang Irrigation District of Hongsibu， Ningxia shows obvious spatial heterogeneity， which makes the accurate description of aquifer system face great challenges and may affect the reliability and accuracy of groundwater numerical simulation. This study focuses on a representative area within the irrigation district mentioned above. Based on the geological exploration data and laboratory test data in the study area， 40 groups of three-dimensional heterogeneous numerical models of groundwater seepage are randomly established by finite difference method， and the random simulation and uncertainty analysis of three-dimensional heterogeneity of groundwater are carried out. The drawdown of shallow groundwater level in the next 10 years is simulated and predicted， and the influence of formation heterogeneity on numerical simulation of groundwater is analyzed. The results show that the maximum variation of groundwater storage in 40 groups of heterogeneity is 0.993 m³/d， the average value is 0.253 m/d， and the maximum error is 0.016%， which meets the requirements of water balance. The maximum difference and minimum difference of each “realized” water level reduction value are 1.48m and 0.002 m respectively， which are located near the drainage well and at the northern edge of the study area. The models with a high permeability coefficient （20.09~43.20 m/d） accounted for a large proportion （such as the 10th “realization”）， resulted in a wide range of water level decline and fast rate， with a maximum decline of 1.45 m. The fluctuation and standard deviation of water level in Baozitan village in the middle of the study area are large， and the uncertainty is obvious. Under 95% confidence， the average upper and lower limits of water level fluctuation in the whole region are only 0.11m. In the model with the same lithologic content ratio， the spatial distribution difference of permeability coefficient （such as the 20th and 30th “realization”） leads to significant changes in the range and rate of water level decline. Building the model many times can make the simulation and prediction of groundwater more rigorous.

The establishment of a multi-field coupled model has become an important means of predicting the evolution of concrete properties， and the values of the parameters in the model have a great influence on the calculation results. Aiming at the problem of insufficient simulation accuracy due to the difficulty in determining parameters in the finite element simulation of early-age concrete， this paper proposes a multi-field coupled simulation method of concrete chemistry-heat-humidity （CTH） that integrates physical experiments and BP neural network parameter inversion. Firstly， an optimized chemical affinity function and a modified evaporable water equation are introduced to achieve an accurate simulation of the multi-stage evolution of relative humidity. subsequently， a parameter inversion method based on BP neural network is used to invert and optimize the key parameters of the CTH multi-field coupled model by combining physical experimental data， to construct a high-precision numerical model， which is then applied to the simulation calculations of concrete chunk casting. The training results of the BP neural network model of this method have good stability， and the predicted value of the humidity decline period fits the experimental data by more than 90%. The simulation results show that the constructed CTH multi-field coupling model can accurately simulate the spatial and temporal evolution of the temperature and humidity fields inside the concrete. This research result provides theoretical and methodological support for the parameter calibration and engineering prediction of multi-field coupled model for early-age concrete.

To investigate the influence of filler particle gradation on the dewatering and consolidation performance of geotubes， this study focused on four common particle size fractions in engineering applications： silt， fine sand， medium sand， and coarse sand. Laboratory hanging bag simulation tests were conducted to investigate the temporal variations in dewatering rate， soil loss （seepage volume）， and internal solid-liquid mixture pressure as the content of each particle size fraction was progressively increased. A mechanistic analysis of soil loss from the bag was performed based on the particle size distribution of the discharged seepage soil. The results indicate that the dewatering process of the bag can be divided into three distinct phases based on dewatering rate： rapid dewatering phase， clogging phase， and stable phase. Additionally， based on soil retention performance， the process can be divided into two stages： the soil loss phase and the filter formation phase. By analyzing the effects of silt content， coefficient of uniformity （Cu）， and coefficient of curvature （Cc） of the filler on the dewatering rate， this study established a composite parameter incorporating these three factors. A functional relationship was then derived between this composite parameter and the geotube dewatering rate. This relationship provides a valuable reference for estimating the dewatering rate of locally sourced soils in practical engineering applications.

In roller compaction construction， the continuation of the compaction route on the construction surface requires the use of a turning and cross-vehicle zone. Traditional turning path planning methods （Method I） suffer from issues such as excessive turning area and uneven compaction. Rational path planning for steering is critical to enhancing compaction intelligence and ensuring quality. Building upon the traditional steering method （Method I）， this study proposes a novel approach （Method II） that employs continuous steering to reduce the turning and cross-vehicle area. The reliability of Method II has been validated through theoretical analysis and engineering applications. Results indicate that， compared to Method I， Method II theoretically reduces the length of the cross-vehicle zone and the driving path by 27.6% and 27.7%， respectively. After adjacent compaction trajectories were completed， Method II reduced the areas with fewer than 2 compactions and more than 4 compactions by 16.63% compared to Method I， achieving a more uniform distribution of compaction passes. In engineering practice， the integration of unmanned roller compaction with Method II reduced the length of the cross-vehicle zone by 54.4% compared to the combination of manual compaction and Method I. This research provides technical support for optimizing driving paths and improving compaction efficiency in turning and cross-vehicle zones during unmanned roller operations.

The rational setting of shield tail grease pressure is a key factor in ensuring the safety of shield tail sealing structures in shield tunnel construction. This study proposes a tail grease pressure prediction model based on the XGBoost algorithm， using operational data collected from a shield tunneling section of the Beijing Metro. Twelve original features were selected from the tunneling data， and Spearman correlation analysis was employed to identify eight features with strong relevance to tail grease pressure as input variables for the XGBoost model. Bayesian optimization was applied to fine-tune the model’s hyperparameters. The optimized model was evaluated on an independent test dataset and compared with the prediction performance of Random Forest （RF）， Support Vector Machine （SVM）， and Artificial Neural Network （ANN） models. The results show that： ① The optimized XGBoost model achieved a root mean square error （RMSE） of 1.48， a coefficient of determination （R2） of 0.91， and a mean absolute percentage error （MAPE） of 8.86% in predicting tail grease pressure. ② Compared with RF、SVR and ANN models， the XGBoost model not only significantly improved prediction accuracy but also demonstrated strong generalization capability， providing an effective solution for solving the modeling problems of such complex systems. The research results provide a scientific basis for achieving high-precision and verifiable tail-seal grease pressure settings in subsequent engineering projects， helping to promote the transformation of grease pressure determination from experience-based judgment to a data- and model-driven scientific decision-making system.

Rock mass fractures have the characteristic of random distribution in nature. After excavation and unloading， tunnels and chambers in fractured rock masses are highly susceptible to deformation and damage. Therefore， investigating the deformation behavior and its intrinsic characteristics in surrounding rock masses with stochastic fractures is of great significance. In this paper， taking a complex， highly fractured headrace tunnel in Huazhou City， Guangdong Province as the engineering background， the geometric characteristics and physical mechanical parameters of fractures are obtained， and a function probability model of parameters for fracture occurrence， traces length and spacing is established， capable of simulating the random distribution feature of nature fractures. Based on this function probability model， we use the discrete element 3DEC software to establish a DFN random fracture rock mass tunnel excavation model. The influence of geometric characteristics and mechanical parameters of fractures on the deformation of surrounding rock is analyzed. The results show that the displacement of tunnel surrounding rock generally increases exponentially with the increase of the fracture volume density. As the number of fractures increases， the fracture surfaces inside the tunnel are cut into unfavorable combinations with each other. The displacement of the surrounding rock of the arch crown decreases linearly with the increase of the internal friction angle within the fracture surface， and there is a local bulging phenomenon at the inverted arch of the entrance to the cave. As the friction angle within the fracture increases， the plastic zone within the fracture zone rapidly increases， especially above the tunnel arch top.

The significant difference in material properties between the new and old dike bodies is one of the important reasons for the numerous risks during the flood season in the Poyang Lake dikes. Based on the prototype Jiaofeng Dike， laboratory model experiments were conducted to simulate the seepage process of the dike body under different water levels. The seepage characteristics near the interface between strong and weak permeable layers and the development process of seepage failure were analyzed， and the mechanism of dike breach was discussed. The research results show that due to the significant difference in permeability coefficients at the junction of the new and old dike bodies， a large hydraulic gradient and concentrated seepage are prone to form near the interface， leading to the erosion and removal of fine particles， which eventually develops into contact scouring or piping failure. Contact erosion primarily develops progressively. When the critical hydraulic gradient is exceeded， contact erosion begins to occur， which can further trigger piping. Seepage-induced softening mainly manifests as the weakening of the soil at the slope toe due to water saturation， which can subsequently lead to slope instability. The experiment reveals four stages of dike breach： initial seepage stage， seepage expansion stage， seepage failure stage， and dike collapse stage. The research results can provide theoretical basis and technical support for the design， construction， and reinforcement of Poyang Lake dikes.

With the rapid development of China’s energy infrastructure， the scale of pipelines crossing river embankments has been increasing year by year， while seepage damage to embankment foundations caused by construction disturbances has become an increasingly prominent issue. Taking the sudden piping hazard in the West-East Natural Gas Pipeline crossing section of the Yangtze River main embankment during the flood season in 2012 as a typical case， this paper systematically reveals the disaster-causing mechanism of the hazard through geological surveys， construction record tracing， and three-dimensional seepage modeling， aiming to address the challenge of traditional anti-seepage measures failing. In response to the composite working conditions of deep seepage， dynamic water erosion， and pipeline vibration in this project， a “rigid blocking + flexible buffering” composite anti-seepage structure was developed based on the governance concept of “graded energy dissipation and rigid-flexible synergy.” The structure employs a bagged gravel layer at its end to reduce dynamic water adsorption pressure， combined with multi-layer materials （polyurethane-asphalt-tar felt and jute） to block seepage paths， and strengthens shallow protection through local weighted pressure grouting and cone probing grouting. After implementation in 2013， the structure withstood floods exceeding the warning water level in the Yangtze River in 2016 and 2020， with no new hazards occurring behind the embankment. Research findings confirm that the new anti-seepage structure effectively blocks deep seepage channels through multi-layer synergistic effects， overcoming the limitations of traditional reinforcement measures. It also exhibits advantages in earthquake resistance， durability， and cost-effectiveness. This case provides technical support and an engineering paradigm for seepage disaster prevention and control in similar pipeline-crossing embankment projects.

The assembly methods of shield segments include through-joint assembly and staggered-joint assembly. Different assembly methods cause differences in the lining structure system. Based on a water resources allocation project in South China， this paper explores the mechanical properties of the through-joint and staggered-joint segment structures of shield tunnels under internal water pressure. The following conclusions are obtained： under the action of internal water pressure， there is obvious displacement dislocation between the segment ring and the ring of the staggered-joint assembly structure. Under the combined action of the shear resistance of the longitudinal bolt and the friction and shear resistance between the segment ring， the stiffness of the staggered-joint assembly segment is obviously greater than that of the through-joint assembly， and it has good performance in terms of structural deformation and waterproof performance. In addition， the type of surrounding rock is an important boundary condition that affects the staggered joint effect. As the rock type transitions from Class II to Class IV， the staggered joint effect increases. The research results can provide scientific basis for the structural design and operational characteristics of shield tunnels.

Hydraulic gates operate under diverse in-service conditions， exhibiting complex nonlinear dynamic characteristics that hinder operational defect identification and heighten accident risks. To address this challenge within hydraulic and hydropower project contexts， a Bayesian network-based identification model is established for gate operational defects. This model enables rapid defect recognition and correlation impact analysis. Furthermore， through Bayesian network significance analysis， critical influencing factors and correlation probabilities of operational defects are clarified. The targeted reinforcement of safety measures provides key technical support for hazard-mitigation operations.

As the core component of water conservancy infrastructure， large and medium-sized irrigation and drainage pumping stations play an irreplaceable role in ensuring food security， flood control and drainage， water resources allocation and ecological protection. However， the existing large and medium-sized irrigation and drainage pumping stations in China are generally faced with problems such as equipment aging， high energy consumption， low automation level and backward management means， which make them difficult to adapt to the needs of modern water conservancy development. Based on the current situation and renovation practice of irrigation and drainage pumping stations in China， this paper systematically analyzes the main problems， and expounds the necessity of modernization from six dimensions of national food security， rural revitalization， water security， ecological civilization construction， realization of the "dual carbon" goals and improvement of management ability. On this basis， this paper puts forward a renovation path targeting "facilities in good condition， engineering safety， energy saving， intelligent control"， covering core contents including mechanical and electrical equipment update， building reinforcement， energy-saving technology application， informatization and intelligent construction， and digital twin pilots. Meanwhile， it emphasizes that the implementation should be advanced through safeguard measures such as policy support， planning guidance and management system reform measures to promote the implementation. The research aims to provide theoretical basis and practical reference for speeding up the modernization process of irrigation and drainage pumping stations and supporting the high-quality development of agriculture and rural areas.

Integrating urban and rural water supply systems is crucial for achieving balanced urban-rural development. Currently， China faces significant disparities in infrastructure and management between urban and rural water supply， necessitating scientific guidance and precise management. Taking Chongqing as a case study， this paper proposes four development models based on the concept of urban-rural water supply integration and regional differences： the urban pipeline extension model， the regional interconnected pipeline model， the integrated regional water supply unit model， and the single-village upgrading model. Building upon this， an evaluation index system for construction and management effectiveness of urban rural water supply integration is formulated， covering engineering construction and operational management. The Analytic Hierarchy Process （AHP） is used to assign differentiated weights to indicators according to each model’s specific attributes， enabling a comprehensive assessment of 34 Chongqing districts and counties. Results show the urban pipeline model has the highest scores （0.60~0.82）， the single-village model the lowest （0.37~0.73）， and the regional interconnection （0.54~0.74） and integrated unit models （0.61~0.69） have intermediate levels. The findings provide valuable scientific decision-making support and theoretical guidance for advancing the construction and management of urban-rural water supply integration initiatives.

As a major production area of chili peppers in China， Guizhou′s chili cultivation holds significant importance for its local agricultural economy. Current nitrogen monitoring methods for chili plants suffer from limitations such as poor timeliness and limited coverage. Hyperspectral information offers a promising solution to address these drawbacks in traditional monitoring. This study conducted a field plot experiment using unmanned aerial vehicle （UAV） hyperspectral remote sensing to monitor chili nitrogen levels. Random Forest （RF） was employed to identify nitrogen-sensitive spectral bands， and multiple machine learning algorithms were utilized to develop dynamic inversion models for chili nitrogen estimation， identifying the optimal model. For total nitrogen inversion， the Random Forest model—using vegetation indices and selected spectral bands （36 bands） as independent variables and total chili nitrogen as the dependent variable—demonstrated superior performance， achieving a test set R2=0.879， RMSE=0.004 and MAPE=15.135%. This research provides a theoretical foundation and technical support for precision fertilizer management in Guizhou′s chili cultivation， contributing to improved nitrogen fertilizer utilization efficiency and the advancement of high-quality agricultural productivity.

In order to investigate the change rule of cumulative reference crop emergence （ET ₀） forecast accuracy under different foresight periods， and optimize the ET ₀ forecasting method in agricultural irrigation management， this study took the Jalaid Banner Station of Cholle Irrigation District in Inner Mongolia as the object of study. Based on the day-by-day weather prediction data and the measured meteorological data of the same period， the Hargreaves-Samani （HS） model was used together with the FAO Penman-Monteith （PM） formula to systematically analyzed the difference in accuracy between the single-day ET ₀ forecast and the cumulative ET ₀ forecast during the 1~7 d forecast period， and its dynamic changes with the extension of the forecast period. Using the results of PM formula as the reference value， the study clarifies the advantages of cumulative forecasting and the optimal forecasting period by comparing the goodness-of-fit （R2）， mean absolute error （MAE）， and error distribution characteristics of the HS model under different forecasting periods. The results show that： ① The mean value of R2is 0.89 and the mean value of MAE is 0.36 mm/d for the single-day ET₀ forecast， and the maximum single-day error reaches 0.40 mm/d， which indicates that there is a significant uncertainty in the single-day forecast； ② the accuracy of the cumulative ET ₀ forecasts is significantly better than that of the single-day forecasts， and the R2 for all the foresight periods （1~7 d） is improved to more than 0.92 and the MAE decreases steadily with the increase in the number of cumulative days. Specifically， the mean values of MAE for 1~7 d cumulative forecasts are 0.32、0.30、0.40、0.29、0.21、0.24、0.23 mm/d， respectively， among which the MAE for 7 d cumulative forecast is the lowest （0.23 mm/d）， which reduces by 36.1% compared with that of single-day forecast， and the range of error fluctuation converges significantly； ③ When the cumulative forecast period is extended to 7 d， the cumulative effect is effectively smoothed， the negative impact of short-term fluctuations of meteorological factors is weakened， the systematic bias of the model is suppressed， and the stability of the forecast is significantly enhanced. In conclusion， this study reveals that the cumulative ET ₀ forecast accuracy improves with the extension of the foresight period， and verifies that 7 d is the optimal cumulative foresight period. The results of this study can provide a theoretical basis for the development of dynamic irrigation plans in irrigation districts， especially in data-driven agricultural water management， and the reliability and practicability of irrigation forecasts can be significantly improved by using 7 d cumulative ET ₀ forecasts. In the future， the coupled optimization of multi-scale ET₀ forecasts can be further explored by combining with numerical weather prediction models.