The water conservancy industry is promoting digital twins in an unprecedented situation. The basis of smart water management is urgently needed to solve the problem that three-dimensional visualization of water conservancy projects cannot meet the requirements of professional visualization software. In this paper， a three-dimensional visualization platform for water conservancy is built based on technologies such as Cesium， Vue and Springboot. This platform has realized the key application technologies of thematic three-dimensional visualization such as multi-source data fusion， flood discharge and flood evolution， and laid a foundation for later digital twin visualization.

Dam deformation is the visual representation of the state of a dam under the influence of internal and external loads during its operation. Building a high-precision deformation prediction model is of significant importance for dam safety monitoring and operational assessment. Existing dam displacement models have long training times， and their prediction accuracy and generalization capabilities are often suboptimal， making them unsuitable for accurate short- to medium-term displacement predictions. This paper coupled Long Short-Term Memory （LSTM） with the Transformer framework and introduced an improved Invasive Particle Swarm Optimization （IPSO） algorithm for optimization to create the IPSO-LSTM-Transformer （ILT） dam deformation prediction model. Using data from measurement point 11-1 on the vertical axis of the Jinshuitan arch dam as an example， this paper analyzed and predicted deformation time series data from 6，150 data sets. The research results demonstrate that the model's prediction accuracy decreases to a certain extent as the prediction horizon increases， but it maintains good predictive capabilities within a prediction step of 10. Compared to traditional Particle Swarm Optimization algorithms， the ILT model significantly improves the model's optimization precision and convergence speed. When compared to the single-step and multi-step prediction results of RNN， LSTM， and IPSO-Transformer neural network models， the ILT model exhibits higher accuracy and better stability， even when trained with limited data. The research findings provide new technical means for the precise short-to-medium-term prediction of dam displacement during operation.

A multi-strategy fusion improved Golden Jackal Optimization Algorithm （MGJO） is proposed to address the shortcomings of the Golden Jackal Optimization Algorithm in solving complex or high-dimensional optimization problems， such as being prone to local optima， slow convergence speed， and low computational accuracy. Firstly， by introducing a chaotic mapping strategy to initialize the population instead of random parameters， the algorithm can generate initial solutions with good diversity in the search space and avoid the initial population distribution deviating from the optimal value. Secondly， a nonlinear dynamic inertia weight is proposed to make the search process more realistic， effectively balancing the algorithm′s global and local search capabilities. Finally， the position update strategy of Cauchy mutation is introduced to fully utilize the guiding role of the optimal individual to improve population diversity， effectively exploring unknown regions and avoiding the algorithm falling into local optima. In order to verify the optimization accuracy， convergence performance， and stability of the improved Golden Jackal Optimization Algorithm， eight benchmark test functions with different features are selected for experiments. The results show that among the 8 benchmark test functions， the improved Golden Jackal Optimization Algorithm has achieved optimal results in terms of mean， standard deviation， and optimal value. In addition， the results of Wilcoxon’s sign rank test indicate that the improved Golden Jackal Optimization Algorithm is significantly superior in statistics. Through practical applications， it shows that the Golden Jackal Optimization Algorithm based on multi-strategy fusion improvement can effectively estimate the parameters of the Muskingum Model， and the optimization effect is significantly better than the particle swarm optimization algorithm， the sine cosine optimization algorithm， and the Golden Jackal Optimization Algorithm. This further verifies the effectiveness of multi-strategy fusion improvement and the superiority of the improved algorithm in parameter optimization. This provides an effective new method for more accurate estimation of the parameters of the nonlinear Muskingum Model.

In recent years， the frequent occurrence of urban waterlogging disasters caused by extreme rainfall due to climate change has posed a threat to urban water safety and sustainable development in China. Accurately grasping the public opinion and emotions in the disaster-stricken areas is of great importance for improving the situational awareness capabilities of emergency management departments in dealing with waterlogging disasters. In today's era of intelligent networks， the increasing importance of social media as a platform for people to voice their problems and suggestions has made it a major carrier of public sentiment and societal opinion， providing a new avenue for obtaining information about natural disasters. A key technical challenge that needs to be urgently addressed is how to quickly extract urban flood disaster information from social media， and how to perform thematic categorization and sentiment analysis of natural disaster information to accurately grasp the thematic categories of regional disaster situations and public opinion trends. Taking Sina Weibo as an example， this article elaborates on the methods of collecting and pre-processing flood disaster data， and constructs a thematic classification and sentiment analysis model of urban flood disaster information based on FastText to accurately capture the thematic categories and public opinion orientations of disaster-stricken areas. The research results， using the “7.20” heavy rain and flood disaster in Zhengzhou in 2021 as an example show that the methods proposed in this article achieve intelligent extraction and analysis of urban flood disaster data on social media. The theme classification model achieves an F1 score of over 0.80 for the classification prediction of the eight predefined categories， and the sentiment analysis model is generally able to accurately predict data labelled as “negative” in sentiment， which indicates that the FastText-based urban flood disaster information theme classification and sentiment analysis model constructed in this article can meet the needs of urban emergency management departments to dynamically grasp the development of flood disasters and public emotions. It holds significant guiding importance for flood prevention and disaster mitigation planning， calming public emotions， and pinpointing rescue efforts in real time.

Utilizing satellite remote sensing technology for the retrieval of water depth is a crucial method for quickly acquiring water depth information over large areas. However， traditional optical satellite images often suffer from low inversion accuracy due to various factors， including the presence of sediments， suspended particles， and chlorophyll in the water. These constituents alter the color and turbidity of the water， making it challenging to accurately estimate water depth through remote sensing techniques. To mitigate the influence of these water color materials on the accuracy of water depth inversion， a novel approach was proposed， centered around a mathematical model and the use of Sentinel-2 remote sensing imagery. This approach involves analyzing the mathematical relationships between water depth and the surface area of different types of water channels. This method deduces the river surface depth model by generalizing and classifying the river section. It combines the Sentinel-2 remote sensing images to determine the relevant parameters in the mathematical model， and derives an expression that can directly calculate the water depth from the water surface area， which can minimize the influence of water color on water depth retrieval from remote sensing images. To validate the effectiveness of this approach， it was applied to the Xiaoqing River， and the results were compared with ground-truth data. The average absolute error in water depth inversion at two hydrographic stations， Huangtaiqiao and Shicun， was found to be 0.03 meters and 0.23 meters， respectively. These results demonstrate a significant improvement in the accuracy of water depth estimation. The method proves to be more reliable in areas with varying water color and turbidity. It can quickly obtain water body depth information based on satellite remote sensing technology and has promotion and application value.

The daily peaking capacity is one of the main performance indicators of hydropower station which is of great significance in the grid scheduling of combined water， wind and photovoltaic power. This paper defines the concept of daily peaking capacity of hydropower station from the aspects of peaking magnitude and peaking power， and calculates the daily peaking capacity of hydropower stations based on four typical daily load diagrams of fixed shapes and a generalized typical daily load diagram. For the typical daily load diagrams of fixed shapes， the calculation method of the hydropower stations’ maximum peaking magnitude under the known peaking time is given. For the generalized typical daily load diagram， the calculation method of the maximum peaking power is given. Four reservoirs of hydropower plants with different regulation performance in the midstream and downstream of the Hongshui River were used as research objects. The peaking capacity of each reservoir was calculated at different initial water levels and different average daily inflow under five different typical daily load diagrams. The results show that For hydropower stations with strong regulating capacity such as seasonal and annual regulation， their daily peaking capacity is mainly affected by the reservoir water level， while the average daily inflow has little effect on it. For daily regulated hydropower stations， when the inflow is small， the daily peaking capacity is greatly affected by the daily initial water level and the daily average inflow. When the inflow reaches a certain value， the daily initial water level has less influence on the daily peaking capacity. When the average inflow and initial water level are determined， the maximum peaking amplitude and maximum peaking power of the hydropower station as well as related operation data can be quantitatively derived， which facilitates the optimal effect of the hydropower station to participate in the grid peaking. The concept of daily peaking capacity of hydropower station proposed in this paper is objective and comprehensive. The established method of calculating daily peaking capacity has strong practicality and operability which can provides scientific basis to determine the daily peaking capacity of hydropower station.

Subsurface infiltrating irrigation is a kind of underground micro-irrigation method. The irrigation water is diverted to the soil at a certain depth under the ground through the buried pipe for subsurface infiltrating irrigation，and then the soil capillary action is used to achieve direct water supply to the root zone of crops. This irrigation method is essentially the same as the irrigation process of “Subsurface Drip Irrigation”. Subsurface infiltrating irrigation can effectively improve soil environmental quality and facilitate crop growth， and it is a highly efficient and water-saving irrigation technology with broad development prospects. It is of great significance to research on subsurface drip irrigation for promoting the development of efficient water-saving agriculture. In this paper， the relevant research progress and existing problems are reviewed from the aspects of soil water transport rule， irrigation technology parameters， irrigation system， influence and regulation of infiltration irrigation pipe clogging etc.， and the future research directions are proposed，which may be referred to by scholars in the field of water-saving irrigation research. It is concluded from this review that， compared with the production practice of subsurface infiltrating irrigation， the research on its mechanism is relatively lagging behind， which limits the deep popularization and application of this technology. The main manifestations are as follows： the law of soil water transport under subsurface infiltration irrigation is not clear enough， clogging is still the limiting factor that hinders the application and development of subsurface infiltration irrigation technology， the optimization design of subsurface infiltration irrigation water distribution system needs further research. This paper suggests carrying out a numerical simulation analysis of water infiltration under different irrigation conditions in the future， selecting the appropriate treatment level， studying the main factors affecting the effect of infiltration irrigation， and seeking the optimal combination of irrigation technology parameters. Research on crop irrigation system under the condition of subsurface infiltration irrigation is carried out to form a set of reasonable and perfect irrigation system and test method. Further research should be carried out on the rule of subsurface infiltration irrigation pipe outflow， clogging mechanism and the effective control threshold of key parameters of irrigation water quality when entering the subsurface infiltration irrigation system.

Wheat drip irrigation technology is the main project of high-standard farmland construction in North China， and it is also an important technical means to deal with the shortage of water resources in the region and ensure national food security. Based on the latest research progresses of drip irrigation technology on wheat at home and abroad and the research results of our group for many years， this paper elaborates the influences of drip irrigation factors such as drip flow， dripping element spacing， drip belt laying distance on soil water distribution， the ways of drip irrigation scheduling on wheat， and the effects of drip irrigation on wheat water consumption characteristics， dry matter accumulations， yield compositions and so on. At the same time， the development trends of wheat drip irrigation in the future are put forward： smart irrigation decision-making and water-fertilizer integration technology that integrates multiple factors such as meteorology， soil and crops， It is hoped these contents would provide theoretical and technical support for accelerating the application and popularization of drip irrigation technology in the main wheat-producing areas in China..

The optimal regulation of multi-energy complementary systems helps to take advantages of different power sources. The growing installed capacity of wind and photovoltaic power has brought about the problems of clean energy consumption and grid peak shaving. A bi-level optimal scheduling model of wind， photovoltaic（PV）， hydropower， thermal power and energy storage is proposed to solve the problems. Firstly， considering that there are many variables and constraints involved in the optimal scheduling of multi-energy complementary， the model is processed hierarchically based on the output characteristics and complementary relationship of multiple energy sources， to reduce the complexity of model solving. Then， according to the demand of source-load coordination， clean energy consumption and power generation economy， the model is divided into the upper-level model of Wind-PV-Hydro-Storage joint optimal scheduling and the lower-level model of thermal power unit optimal scheduling. The optimization objectives of upper-level model are the minimum net load variance and the maximum clean energy generation， and the optimization objective of lower-level model is the minimum operation cost of thermal power units. The net load curve achieved by the upper-level model is used as the constraint of the lower-level model. Finally， taking the IEEE30-bus system as an example， the CPLEX solver of MATLAB platform is used to calculate three pre-defined scheduling scenarios. The results show that the anti-peak regulation of wind power and photovoltaic output will expand the peak-valley difference and variance of net load， increase the pressure of peak shavingand reserve of thermal power， and increase the power generation cost and unit loss. The participation of hydropower in multi-energy complementary systems has a significant improvement on the net load fluctuation of the power grid， which can effectively reduce the peaking pressure of the power system. Energy storage can reduce the conflict between smoothing net load fluctuation and improving clean energy generation to a certain extent. The optimal scheduling model proposed can effectively reduce the fluctuation of source and load， promote the consumption of clean energy and reduce the operation cost of the system， which contribute to the low-carbon and safe operation goals of the grid.

Evapotranspiration （ET） is a crucial link in water circulation. It plays an important role in the global water cycle and surface energy balance， and has significant impacts on climate， ecosystems， and water resources management. Therefore， the quality of evapotranspiration data is crucial for the precise management of global water resources. This study conducted accuracy validation and spatiotemporal comparison of three ET products in the Northern Hemisphere， selecting the ET products that is more suitable for the Northern Hemisphere， providing suggestions for strengthening the combination of remote sensing and ground observation research. Using the monthly average measured data from FluxNet2015 flux sites to verify three ET products， it was found that PML_ V2 product has the highest accuracy in the Northern Hemisphere， followed by GLDAS， and finally MOD16A2， with correlation coefficients R of 0.66， 0.57， and 0.56， respectively； The root mean square error （RMSE） is 2.46， 5.68， and 12.42 mm/month， respectively； The average biases are 14.36%， 16.86%， and 35.02%， respectively. The GLDAS ET product has the ability to monitor daily scale ET， and the consistency between the daily average estimated value and the measured value at the flux tower site is high. The correlation coefficient R is 0.74， and the RMSE and Bias are 1.62mm/day and 27.90%， respectively. Overall， on the time scale， all three ET products can simulate the seasonal changes in the Northern Hemisphere， with higher summer evapotranspiration and lower winter evapotranspiration. The three ET products in summer all have overestimation phenomena on different land cover types， and the simulation results in other seasons are better than the ground observation values. Among them， MOD16A2 performs the worst， and the overestimation phenomenon is the most obvious. In addition， during the period from 2001 to 2020， except for arid areas， the spatial distribution of the three ET products was relatively consistent in most regions， with a correlation coefficient R greater than 0.6. This study provides scientific recommendations for selecting suitable ET data sources for conducting evapotranspiration studies in the Northern Hemisphere by evaluating the uncertainty and product quality of different ET products.

As a unit hydrograph method widely used in engineering practice， the synthetic unit hydrograph of Guangdong is obtained from the analysis and synthesis of rain and flood data before 1983， which reflects the production-confluence relationship before urbanization in Guangdong Province. However， there are few studies on the evaluation and modification of its applicability in urbanized areas. Taking Jingtian Watershed of Shenzhen City， Guangdong Province as an example， based on the rainstorm and flood records of Jingtian Hydrology Station from 2011 to 2019， this paper evaluates the applicability of the comprehensive unit line in Guangdong Province by using evaluation indexes such as Nash efficiency coefficient， identification index and correlation coefficient， and conducts a sensitivity analysis of important parameters by Spearman rank correlation coefficient method. Based on Latin hypercube sampling method and constrained least square method， the comprehensive unit line method of Guangdong Province is modified. The results show that the Jingtian Basin has experienced rapid urban expansion since the 1980s， and the proportion of impervious surface has exceeded 90%. The comprehensive unit line method in Guangdong Province has obvious shortcomings in describing the storm-flood process in this basin. The two schemes proposed in this paper can improve the simulation effect of the model by determining important parameters and modifying the unit line. Among them， the cross-validation parameter optimization method based on grid search and the unit line correction method based on constrained least square method proposed for small samples show a good simulation performance， and the indexes such as Nash efficiency coefficient， identification index and correlation coefficient are significantly improved， which better reflects the characteristics of production and convergence in Jingtian Basin. Compared with the recommended unit line of the integrated unit line method in Guangdong Province， the modified unit hydrograph method shows the characteristics of larger peak discharge， earlier peak time and smaller total duration， which is more in line with the law of production and confluence of urbanization.

This paper analyzed the pre-stressing mechanism of the steel wire and the damage process of the pipeline， and proposes a method of PCCP wire broken warning and alarm threshold assessment based on mechanical simulation analysis. The PCCP finite element model was established by ABAQUS program， and the complete construction process including laying bedding， installing pipelines， backfilling in steps was simulated， and the stress state of pipeline prestressing steel wires， steel cylinders and pipe core concrete was calculated under nearly 100 working conditions with different depths of burial and different design working pressures. It was proposed to take the maximum permissible number of broken wires of the pipe core concrete cracked under the action of water hammer as the early warning threshold for the bursting of the pipeline， and the maximum permissible number of broken wires in the limit state of the buried bearing capacity of the pipeline as the alarm threshold for pipe bursting， and analyzed the relationship between the PCCP wire broken warning and the alarm threshold under different working pressures and burial depths.

As an important non-engineering measure， the optimal operation of reservoir flood control can reduce the loss caused by flood disasters with little investment and plays a key role in flood control and disaster reduction. In this paper， considering the safety of the reservoirs themselves and the safety of the downstream flood control section， a multi-objective optimal model for real-time flood control operation of reservoirs is established with the objective function of minimizing the maximum flow of the downstream flood control section and minimizing the maximum water level of each reservoir. Based on the “filter operator”， an improved multi-objective Lichtenberg algorithm （MOLA） is proposed， which is used to solve the multi-objective model， and the real-time flood control operation schemes of the reservoirs are obtained， to enhance the practicability of operation solutions. Finally， a comprehensive screening method based on hierarchical clustering and the physical meaning of Pareto front is proposed to screen the scheduling schemes on the Pareto front and select limited ones for the scheduler， to increase the focus of decision-making. Taking the Shiguanhe flood control system of the Huaihe River Basin as an example， this paper applies the multi-objective optimal model for real-time flood control operation of reservoirs. The result shows that the proposed model based on the improved multi-objective Lichtenberg algorithm has high computational efficiency and strong applicability. The gradient analysis method is used to quantitatively analyze the mutual feedback relationship between the maximum water level of Nianyushan Reservoir， the maximum water level of Meishan Reservoir， and the maximum flow of Jiangjiaji flood control section. The results show that the water level change of Meishan Reservoir has a more significant impact on the flow of Jiangjiaji section， and the Meishan Reservoir is a priority in flood risk control of Shiguanhe River Basin. The research results can provide decision-making support for real-time flood control operation of reservoirs.

Irrigation is one of the types of land management that has the greatest impact on regional climate， and its impact has been increasing in recent decades. A large number of papers have demonstrated the cooling effect of irrigation， particularly in arid and semi-arid regions. However， the cooling effect in humid regions has been found to be not significant， and the investigation of the cooling effect pertaining to extreme high temperatures has received comparatively less attention.In this study， a sliding window search algorithm is used to investigate the impact of augmenting the proportion of irrigated areas on the occurrence of extreme high temperatures in the Yangtze River Basin based on a global gridded temperature dataset and a historical irrigation distribution dataset. To isolate the climatic effects of various factors in the observed data， the multiple linear regression techniques are employed. Furthermore， a control test is set up by using the WRF model of the coupled irrigation module to perform numerical simulations for verification.The results show that irrigation in the Yangtze River Basin has a cooling effect on extreme high temperatures， and the intensity of the cooling effect depends on the proportion of irrigated area， which becomes more obvious with the increase in irrigated area， but when the proportion of irrigated area exceeds the threshold （0.25~0.30）， the cooling effect gradually diminishes. Numerical simulation experiments verified the results of the analysis based on observed data， but the model would overestimate the cooling effect of irrigation to some extent. Meanwhile， it was found that irrigation could increase soil moisture and latent heat flux and decrease sensible heat flux and soil heat flux， and the cooling effect would be weakened as the proportion of irrigated area increased， the surface moisture increased and the albedo decreased. The results of this research can provide a theoretical basis for the scientific management of agricultural irrigation in the Yangtze River Basin， which is of great scientific significance.

Sound and vibration are the direct carriers of running data of rotating machinery equipment. According to the principle of voiceprint monitoring， through data collection， pre-processing and extraction of eigen values， a voiceprint sample and model library of running data and voiceprint monitoring system of vibration data of large pumping station units is established by adopting convolution neural network deep learning recognition model so as to realize the real-time monitoring of the unit running data of the large pumping station.

Tetracycline hydrochloride （TCH） is widely used in medical， livestock breeding， and agricultural industries. However， because it is not fully absorbed by humans and animals while used， it enters environmental water bodies with feces and urine. TCH can damage the ecological environment， harm human health， and produce drug-resistant bacteria， posing threats to the environment and human health. Therefore， finding an economical and effective removal method is crucial for water quality safety. Currently， the removal methods for TCH in water mainly include chemical， biological， and physical methods. Compared with chemical and biological treatments， the adsorption method in physical treatments for removing TCH from polluted water bodies has the advantages of low cost， environmental friendliness， high efficiency， stability， and easy operation. In choosing adsorption materials， carbon-based and mineral materials often appear powdery， are not easily recoverable， and carry the risk of secondary pollution. In contrast， gel materials are suitable for removing organic pollutants. Thus， finding an inexpensive adsorbent with good adsorption performance is currently a hot subject. Montmorillonite （MMT） is a clay mineral with a stable crystal structure and high mechanical strength. It has a surface with many permanent negative charges. Due to its low cost， easy availability， and superior adsorption properties， it is commonly used in wastewater treatment to adsorb heavy metal ions， dyes， phenolic compounds， polycyclic aromatic hydrocarbons， etc.， with good removal effects. However， there is currently no research specifically on using MMT for TCH adsorption. Against this background， this paper aims to study the adsorption performance and effects of MMT adsorbing TCH. Through research， it is found that the Freundlich isothermal adsorption model and the pseudo-second-order kinetic model fit well for MMT adsorbing TCH， indicating that its adsorption is multilayered and can reach 87% of the total adsorption amount within 60 minutes. The Langmuir saturation adsorption amount can reach 118.6mg/g. The Gibbs free energy ΔG0 is always negative， indicating that the adsorption of TCH by MMT is a spontaneous process. The calculated enthalpy and entropy changes are both positive， indicating that the adsorption process is endothermic and occurs randomly on the solid-liquid surface. Increasing the temperature promotes adsorption， which is consistent with the conclusions drawn from temperature effect experiments， further indicating that the adsorption process is an endothermic reaction and increasing the temperature is conducive to adsorption. The solution pH significantly affects adsorption performance. As the pH increases from 3 to 10， the adsorption amount first increases and then decreases， with the optimal pH range being 6-8. This is because， under strongly acidic conditions， H+ competes with TCH for adsorption， resulting in a lower adsorption amount of TCH by MMT. As the pH value increases and the H+ concentration decreases， the competitive adsorption with TCH is reduced， so the electrostatic adsorption of TCH by MMT increases. But as the pH continues to rise， TCH gradually transforms into H2TC- and HTC2- forms， creating electrostatic repulsion with negatively charged MMT and causing a decrease in adsorption. Based on the above experiment results and characterization analysis， the mechanism suggests that the adsorption process of MMT on TCH includes both physical adsorption and chemical actions， with the latter being dominant. The Al-O-H group in MMT can form hydrogen bonds with the tricarbamoyl amide and hydroxyl groups on the TCH molecule. At pH < 5， there is also a certain electrostatic attraction. In summary， due to the low cost of MMT， its fast adsorption rate， and high adsorption capacity， it has unique advantages in the adsorption performance of TCH， thus promising a bright application prospect in the actual TCH wastewater treatment.

Based on the knowledge and understanding of modernization of the physical irrigation district and digitalization of the analog irrigation district， this paper analyzes and summarizes the “five major functions” of the construction of modernized digital irrigation district including “Identification system of irrigation district”， “Stereoscopic Perception System”， “Precise Control System”， “Information Exchange System” and “Water Allocation System”. In order to meet the requirements of intelligent water projects and efficient operation and management of irrigation district， to efficiently simulate water flow in canal system， the generalized map of irrigation district with “joint flow process” as input and output is systematically sorted out， and the key application structure system of modernized digital irrigation district is studied and proposed. That is to build an interactive overall structure with “database” as the carrier， “mathematical model” as the support layer， “nine professional applications” as the interaction layer including “organization management”， “project management”， “security management”， “pumping station management”， “water saving and water supply management”， “economic management”， “information management” and “public services” on the basis of “one digital map of the irrigation district”. It is expected to effectively improve the safety， equity， reliability and flexibility of irrigation water supply services.

In our present stage， agricultural water right trading is dispersed and the scale is smaller， and the market activity of agricultural water right trading is not in full swing. With the contradiction between supply and demand of water resources becoming increasingly tense， agriculture has necessity to stimulate the decisive role of market in water resources allocation， and further create greater trading space for agricultural water rights. Agricultural water rights involve multi-level management systems， but few studies have analyzed the factors affecting the development of agricultural water rights trading from different dimensions in detail， so it is impossible to systematically promote the development of agricultural water rights trading. Therefore， this paper divides the key factors affecting the development of agricultural water rights trading into three dimensions： macro social economy， middle market players and their mutual relations， and micro trading elements. A qualitative analysis is made of the substantive attributes of each factor index in different dimensions and its impact on the development of agricultural water rights trading， and the macro-support for the development of agricultural water rights trading is found. This paper clarifies the development ideas of agricultural water rights trading market， and then focus on the role of micro-trading elements in the process of water rights trading. From three dimensions， it puts forward suggestions on how to strengthen the agricultural water rights trading policy and promote the construction of rural water system connectivity in multiple ways， clarify the role positioning of government and enterprises， and encourage the synchronous transfer of agricultural water rights trading and land to activate the agricultural water rights trading market so as to improve the agricultural water rights trading market to provide the construction foundation and ideas.

The plain river network area has a flat terrain and slow flow velocity， resulting in a limited ability to resist pollution. Therefore， enhancing the connectivity of the water system is crucial for improving the quality of the water environment. This article focuses on the Yuejin polders which has poor water system connectivity in Shengze Town as the subject of research. By conducting water diversion experiments and constructing a hydrodynamic model of the river network in the polders area， the study explores the scheduling and operation rules of water conservancy work， methods of connecting the river network， and the corresponding changes in the water environment. The study utilizes water age as a time scale parameter to measure the connectivity of the water system. It establishes a river network model for hydrodynamics， water quality， and water age in the Yuejin polders. The simulation evaluates the connectivity and water quality improvement in the polders area under five water diversion schemes. Additionally， an evaluation system of pressure， state and response is constructed to quantitatively analyze changes in water system connectivity and the level of water environmental improvement resulting from water diversion. The results reveal that from the perspective of the length of water exchange， the exchange time of water bodies in the main river channels of the research area is within 5 days. The greater the connectivity of these water bodies， the shorter the water age becomes. From the perspective of changes in water quality indicators， during the water diversion experiment， the most noticeable change in the water environment inside the polders is the level of ammonia nitrogen. It shows a strong response to water diversion， and simulation results indicate that water age generally aligns with the concentration of ammonia nitrogen. Enhanced connectivity of water bodies leads to a shorter water age， which aids in reducing ammonia nitrogen levels. From the perspective of water diversion scheme， the uni-source water diversion and uni-region drainage scheme has limited effectiveness in improving the connectivity of the water system within polders， especially in areas with weak water flow in the river network. This scheme also increases variations in water flow within the polders， resulting in spatial differences in connectivity and water quality in the river network. Consequently， local stagnant water areas or dead water areas may be formed. Therefore， the combination of multiple sources of water diversion and multi-regional drainage scheme can enhance the connectivity of the water system and improve water environmental quality.

The Sihu Basin is located in the hinterland of the Jianghan Plains， with low-lying terrain and prominent flood disasters. In response to the construction of drainage projects in the Sihu Basin， this paper couples natural runoff model， controlled runoff model， one-dimensional hydrodynamic model of river network with lake regulation and storage calculation model to construct a hydrological and hydrodynamic simulation model of the Sihu Basin. By taking 2016 and 2020 as typical years， this paper studies the impact of drainage project construction and different scheduling strategies on flood control safety in the Sihu Basin. The results show that the construction of drainage projects can significantly reduce the flood control pressure in the middle and lower regions of the Sihu Basin. Under the rainstorm conditions in 2016 and 2020， the flood peak water level of Honghu Lake dropped by 1.19 m and 1.41 m respectively， and it dropped from above the guaranteed water level to below the warning water level. The implementation of the joint regulation and storage of Changhu Lake and Honghu Lake can effectively improve the overall drainage benefit of the Sihu Basin. Under the conditions of rainstorm in 2016 and 2020， the flood peak water level of Honghu Lake only increased by 0.05 m and 0.16 m， and did not exceed the warning water level， the flood peak water level of Changhu Lake decreased by 0.73 m and 0.40 m， significantly improving the flood control safety of the upper region of the Sihu Basin. However， the water level of Changhu Lake still exceeded the warning water level due to the flow capacity of sluice and channel. The construction of the drainage projects has overall improved the flood control safety of the rivers and canals in the Sihu Basin. Except for the Pailao River （West）， Shaluo canal， and Luoshan Main Canal， which have experienced a certain increase in the average highest water level after participating in the drainage of the basin， the average highest water level of other major rivers and canals has decreased. The construction of the drainage projects can reduce the dependence of the middle and lower regions of the Sihu Basin on the Xintankou Pump Station and Gaotankou Pump Station. Under the rainstorm conditions in 2016 and 2020， the proportion of the total drainage of the two pump stations in the total drainage of the basin decreased by 13.23%~19.35%， and the flood control safety of Honghu Lake was improved. The implementation of the joint regulation and storage of Changhu Lake and Honghu Lake can improve the safety of flood control in the upper region of the Sihu Basin and reduce the pressure of drainage of the Tianguan pump station. After the implementation of the joint regulation and storage， the total drainage of the Tianguan pump station decreased by 60.97% and 57.23% respectively under the rainstorm conditions in 2016 and 2020. This research can provide a reference for improving the application level of drainage projects and enhancing flood control and security capabilities in the Sihu Basin of the Jianghan Plains.

In view of the traditional agricultural irrigation in hills and mountains， which usually uses artificial control solenoid valve for irrigation， the problems of water waste and high labor cost has come into being. A set of automatic irrigation control system based on fuzzy control and PID control is designed. The stability of PID control and adaptability of fuzzy control are combined to control irrigation time. The irrigation system of hills and mountains built by Simulink is controlled by conventional PID and fuzzy PID respectively. The simulation experiments show that compared with the conventional PID control， the fuzzy PID control system has significantly reduced overshoot and adjustment time， and has higher stability， adaptability and adjustment performance. It can be used in the automatic irrigation system of hilly and mountainous agriculture.

With the proposal of the carbon peaking and carbon neutrality goals， China’s energy structure is gradually shifting from relying on fossil fuels to relying mainly on renewable and clean energy systems. However， clean energy represented by wind and photovoltaic power generation is difficult to operate and consume separately due to output fluctuations and uncertainties. Hydropower stations have good regulation capabilities and can operate in conjunction with wind farms and photovoltaic power stations to alleviate the instability of new energy generation. The southwestern region of China is rich in hydropower resources， and a large number of cascade hydropower stations with regulatory reservoirs have been built. The complementary operation of cascade hydropower with wind power and photovoltaic resources around the watershed， and the full utilization of the regulating capacity of cascade hydropower to bundle and deliver high-quality electricity， is an important means of achieving energy transformation. To quantify the carrying capacity of cascade hydropower stations on wind power and photovoltaic resources， this paper proposes a nested calculation model for medium to long-term optimal scheduling of hydropower to short-term optimal allocation of hydro-wind-photovoltaic power. The short-term scheduling calculation boundary is determined based on the mid-term and long-term optimization scheduling calculation results of hydropower， the scale of wind and photovoltaic power during short-term scheduling is adjusted， and scheduling with the goal of maximizing source load matching， maximizing power generation， and minimizing abandoned power are optimized. Decoupling is carried out by using a hierarchical nested strategy to calculate the scale of wind and photovoltaic resources that can be carried by cascade hydropower stations. Applying this model to the cascade hydropower station group in the Yalong River Basin， the scale of wind and photovoltaic resources carried by the “three reservoirs and seven cascade” cascade hydropower stations in the Yalong River Basin is calculated. The research results show that considering the 2025 planning level year and 5% curtailment rate constraint， the cascade hydropower in the Yalong River basin can carry 1.396 million kW of wind power resources and 10.74 million kW of photovoltaic power resources. The research results have a certain reference value for the optimization of water solar energy complementary power generation systems in the watershed.

Pump storage is of great significance to the development of renewable energy and the construction of a new energy system， and help to achieve the “dual carbon” goal. Fully understand the functions and functions of pumping storage， sort out the policy evolution and development process in the process of modernization of China's pumping storage， analyze the current status of Chinese pumping storage， and look forward to its future. The inherent relationship between the modernization of the entire country is conducive to promoting the high-quality development of the new stages of pumping in the new stage， and it is expected to accelerate and promote the formation of a modern industry with strong international competitiveness in my country.

In response to the current problems in water quality prediction models， such as the complexity of the data itself， noise interference in signal processing， and insufficient decomposition depth， which make it difficult for a single decomposition to fully capture the nonlinear features of the signal， this paper proposes a water quality prediction model based on secondary decomposition. This innovative method firstly uses the Complete Ensemble Empirical Mode Decomposition with Adaptive Noise （CEEMDAN） to decompose the raw data. Then， Variational Mode Decomposition （VMD） is adopted to perform a secondary decomposition of the Intrinsic Mode Function （IMF） with the highest entropy value. Finally， the processed time series are put into the TCN-light GBM multi-feature prediction model. At the same time， the Sparrow Search Algorithm （SSA） is used to optimize the prediction model. By taking the water quality of Yufu River in Shandong Province as an example， the Root Mean Square Error （RMSE） of this model is 0.105 3， the Mean Absolute Error （MAE） is 0.081 5， and the coefficient of determination （R ²） is 0.947 1.The predictive metrics of the model are compared with those from popular contemporary deep learning and neural network algorithms， such as the Gated Recurrent Unit （GRU）， Long Short-Term Memory （LSTM）， Light Gradient Boosting Machine （Light GBM）， and Temporal Convolutional Networks （TCN）， among others. The results show that， in terms of R ²， the model achieved improvements of 53.04%， 70.41%， 66.07%， and 65.20% respectively. In terms of RMSE， the model represented reductions of 62.76%， 65.50%， 64.93%， and 64.80% respectively. And in terms of MAE， the model witnessed decreases of 62.76%， 66.24%， 63.80%， and 65.24% respectively. Therefore， it is evident that the model based on CEEMDAN-VMD-TCN-light GBM exhibits superior predictive performance， which can reduce the fluctuation of water quality sequences more effectively and improve the ability to capture nonlinear features of signals.

In order to improve the spatio-temporal accuracy of rainfall and evaporation distribution in the Grid-XAJ model， strengthen the hydrological cycle， and provide a foundation for further coupling of the WRF model with the Grid-XAJ model， we constructed the WRF-driven Grid-XAJ model. Firstly， we used the successive correction method to obtain the WRF-merged rainfall by merging WRF forecast rainfall with gauged rainfall. Then， the WRF forecast meteorological data were input into the Grid-XAJ model， and the hourly evaporation capacity of each grid was calculated by Penman-Monteith method （PE_PM ）. Finally， the Grid-XAJ model was driven by WRF-merged rainfall and PE_PM to simulate floods in the humid Tunxi catchment. The results showed that： ① The WRF-merged rainfall exhibited high accuracy and a fine spatial distribution. Compared with WRF forecast rainfall， the WRF-merged rainfall had a higher correlation （RR≥0.99） and a better fitting degree（NSE≥0.98） to actual rainfall， and errors of rainfall peaks （-8.1%~3.5%） and cumulative rainfall （-2.0%~6.7%） were significantly reduced. In terms of spatial distribution， the WRF-merged rainfall displayed more complex spatial information compared to rainfall interpolated from gauged rainfall， as indicated by Shannon Entropy （SE） significantly increased （30.4%~48.2%）. Additionally， the WRF-merged rainfall can incorporate rainfall centers from both the WRF forecast rainfall and the rainfall interpolated by gauged rainfall. ② The PE_PM exhibited hourly variations and showed a close correlation with rainfall. In terms of spatial distribution， PE_PM was closely related to elevation. Regions with higher elevations displayed a relatively lower mean PE_PM values， while areas with moderate elevations exhibited the highest mean PE_PM values， and slightly lower values were observed in lower elevation areas. ③ The WRF-driven Grid-XAJ model held significant potential in flood forecasting. The flood forecasting accuracy of the Grid-XAJ model， when driven by WRF-merged rainfall and PE_PM， was significantly enhanced compared to the model driven by WRF forecast rainfall and PE_PM . The NSE values of flood forecasts were all above 0.90， and the qualified rates of flood volumes， flood peaks and peak occurrence times were 100%.

Soil moisture content（SMC） and temperature are important parameters affecting the hydrological cycle and climate change. In order to monitor SMC and temperature timely and accurately， this paper proposes a method to detect based on hyperspectral technology. The dataset which contains a lot of noise and redundant information is obtained during five days of field measurements. After being denoised by Savitzky-Golay filter， the successive projection algorithm（SPA） is first used to extract the characteristic wavelength of hyperspectral database. Then， the hyperparametric weight and bias value of the support vector machine regression（SVR） are optimized by the genetic algorithm（GA）.At the same time， the SPA-GA-SVR hybrid algorithm model proposed in this paper is constructed to predict the soil moisture content（SMC） and temperature（T）.The prediction performance is compared with BP neural network， SPA-optimized BP neural network， SVR， SPA-optimized SVR and GA-optimized SVR.Results show that there is no significant difference in the prediction ability of each model when the soil moisture is less than 30%.However， hybrid models has obvious advantages over the single neural network or machine learning model on the whole dataset. Finally， the SPA-GA-SVR algorithm is superior to other models in all indicators， in which R ² is 0.981 and RMSE is 0.473% in soil moisture prediction and R ² is 0.963 and RMSE is 0.883℃ in soil temperature prediction. Research indicates that the SVR model optimized by SPA and GA can accurately predict soil moisture and temperature based on hyperspectral data.This method， which has certain application value and practical significance， can be applied to portable hyperspectral and unmanned aerial vehicle to realize real-time monitoring of soil moisture and temperature， and provide a theoretical reference for sowing and irrigation in the future.

To effectively solve the high-dimensional problem of water resources carrying capacity classification and evaluation， and reveal the dynamic changes in water resources carrying capacity， an improved projection pursuit clustering model is proposed. For the improved projection pursuit clustering model and the projection pursuit clustering model based on the maximum information entropy principle， the reasonable range and optimal value of density window width were derived by analyzing the negative entropy change law of projection values； When the density window width is the optimal value， the improved projection pursuit clustering model performs better in classification evaluation than the projection pursuit clustering model based on the maximum information entropy principle. This article uses an improved projection pursuit clustering model to dynamically evaluate the water resources carrying capacity status in Jiangsu Province. The water resources carrying capacity level from 2009 to 2020 and 2022 was level III， and the water resources carrying capacity level in 2021 was level II. A grey GM （1，1） model was established to predict the water resources carrying capacity status of the province as level II from 2023 to 2030. The improved projection pursuit clustering model more effectively extracts the structural feature information of high-dimensional data for water resources carrying capacity evaluation indicators， further enhancing the accuracy of the water resources carrying capacity classification evaluation model and making the evaluation results more objective and reasonable. Through comparative analysis of the contribution rates of indicators in 2022 and 2009， the measures taken by the province to save water in industry and agriculture， accelerate socio-economic development， and protect water resources have promoted the continuous improvement of water resources carrying capacity. Based on the contribution rate of evaluation indicators to the water resources carrying capacity status and evaluation standards in Jiangsu Province， this paper deeply analyzes the shortcomings of the water resources carrying capacity in the province， and puts forward relevant suggestions to improve the water resources carrying capacity status， ensuring that the water resources system carrying capacity status in Jiangsu Province reaches level I as soon as possible and achieving sustainable utilization of water resources.

To meet the requirements of developing high head and huge capacity hydropower plants， this paper proposes a new Pelton turbine that joints three runners by one vertical shaft to have a rated head of 1 000 m and a single unit capacity close to 800 MW. The design concept of increasing capacity by increasing the number of runners and the structural components of the new turbine-generator unit are explained， the theoretical design parameters of the runner are calculated and presented， the performance characteristics and the flow patterns in runner and casing are predicted and optimized by CFD simulation. The results show that the rated operating efficiency of the turbine can reach 87%， and the operating characteristic curves are smooth for different heads and outputs， the water splashing interference in the casing is an important factor affecting the efficiency of the turbine. By optimizing the layout of the runners and nozzles in the casing to suppress the splashing interference， the efficiency level， torque oscillation， and operation stability of the Pelton turbine can be improved. This preliminary attempt has demonstrated the feasibility of the concept and also found the importance of optimizing the water splashing in the casing， which has reference values for further researches.

The seepage failure caused by the change of reservoir water level is one of the main reasons for the instability of the earth-rock dam. Moreover， the landslide would become geological disasters easily， which poses a great threat to human life and property. Aiming at the influence of reservoir water level rise on the landslide of earth-rock dam， this paper uses a homogeneous earth dam as the research background. The plastic zone and horizontal displacement changes of the dam slope under the critical instability state are studied by COMSOL Multiphysics. Based on the FEM strength reduction method， the stability of the dam slope under three conditions of normal water level， design flood level and check flood level are analyzed. The results show that the phreatic line of earth-rock dam increases with the rise in reservoir water level. The maximum plastic strain and horizontal displacement increase linearly， and the maximum values of the strength reduction factor （SRF）， the deformation behavior is mainly a shear slip. The factor of safety （FOS） under three conditions are 1.894，1.855 and 1.831， respectively. The stability of the dam body decreases continuously， but it is higher than the critical minimum factor of safety.

In order to improve the accuracy of dam displacement prediction， this paper proposes a novel comprehensive prediction method based on deep learning. Initially， a multi-level data denoising technology based on Complete Ensemble Empirical Mode Decomposition with Adaptive Noise（CEEMDAN） and Singular Spectrum Analysis （SSA） is introduced. This effectively eliminates the noise and outliers in the monitoring data， improving data quality， and providing more reasonable dam deformation data for subsequent predictions. Subsequently， a deep learning model based on Convolutional Neural Networks （CNN） and Gated Recurrent Units （GRU） is constructed. The CNN is used to extract rich features from the monitoring data， and the GRU is utilized to capture and process the long-term dependencies in the time series data. To enhance the model’s performance， a self-attention mechanism is introduced to help the model better handle and recognize complex patterns in the data. Compared with other prediction methods， experimental results show that this method significantly improves the accuracy and stability of dam displacement predictions， providing a new approach for the field of dam deformation monitoring.

As the core equipment of hydroelectric power plants， hydroelectric generator units inevitably generate vibration and pressure pulsation problems during long-term operation. Long term vibration can accelerate the wear and fatigue of mechanical parts， reduce the lifespan and reliability of the unit， and may also lead to unstable operation of the unit， and even cause serious accidents， posing a threat to the safety of the power plant. These vibration and pulsation problems stem from multiple factors， including the dynamics of water flow， the design and manufacturing quality of impellers， and the complex interactions between the unit and the hydraulic system. These vibration and pressure pulsation issues have a negative impact on the performance， reliability， and safety of hydroelectric generators. Therefore， it is particularly important to conduct in-depth research on the vibration and pressure pulsation problems of hydroelectric generator units during operation and find effective solutions. This article analyzes the experimental data of Unit 3 of Ansha Hydropower Station and delves into the operational stability issues caused by pressure pulsation and vibration of hydroelectric generator units. By using the Wilcoxon rank-sum test and Pearson correlation coefficient to analyze the pressure pulsation， vibration， and swing signals during the operation of water turbine units， it was found that there is a strong correlation between the overall signals of hydraulic turbine units， especially in the swing signal. Finally， methods and strategies to reduce pressure pulsation and vibration issues generated during the operation of hydroelectric generator units were explored for the design， operation， and maintenance stages. By taking these measures， we can effectively reduce the negative impact of vibration and pressure pulsation， ensure the safe and stable operation of hydroelectric generator units， and contribute to the development of sustainable energy.

Making full use of modern technological means to improve the accuracy of runoff forecasting plays an important guiding role in basin flood and drought disaster defense and joint scheduling of reservoir groups. However， existing deep learning models have problems such as lack of model transparency and poor physical interpretability. To address the above problems， in this study， a conceptual hydrological model EXP-Hydro is embedded into the P-RNN layer of recurrent neural network， and a deep learning hybrid model Hybrid-DL coupled with physical mechanism is modeled. The hybrid model adopts a differential framework to realize the deep bidirectional fusion of conceptual model and neural network， which is able to train the parameters of conceptual model and neural network at the same time. And an application study is carried out in the upper reaches of Qingjiang River as an example. The results show that compared with RNN， EXP-Hydro， BP and SVM models， the Nash efficiency coefficient （NSE） of the Hybrid-DL model increases by 6.08%， 21.01%， 37.09% and 73.92%， the root-mean-square error （RMSE） decreases by 10.82%、33.73%、54.70% and 95.57%， the KGE efficiency coefficient increases by 4.78%、12.68%、26.79% and 55.74%， and the peak error TPE decreases by 4.96%、13.12%、252.84% and 297.81%. The Hybrid-DL model has good robustness and adaptability， and can provide a reliable theoretical tool for runoff forecasting in the upper reaches of Qingjiang River and even in other basins.

Optimal allocation of water resources can solve the problems of regional water resources shortage and unreasonable utilization， and in recent years optimization algorithms have become an important tool for obtaining the results of optimal allocation of water resources. To effectively alleviate the problems of slow convergence， local optimization and prematureness in existing optimal allocation algorithms， this paper introduces a new metaheuristic algorithm： the Cheetah Optimizer （CO）， with the advantages of fast convergence， robustness， and superiority-seeking ability， which is applied to the optimal allocation of water resources. Taking Taiyuan as the study area， this paper establishes a water resource optimization model with the objectives of economic， social， and ecological benefits， and CO is used to solve the model， then compares and analyzes the results with the Genetic Algorithm （GA）. The results show that the total pollutant discharge of CO is within the control target of the planning year （2030）， the regional satisfaction reaches 0.98， which can meet the demand for basic and development water； compared with the current year （2020）， the proportion of domestic water consumption decreases by 22.58%， and the proportion of water consumption in production increases by 10.53%， 25.81% and 12.5%， respectively. The water supply ratio of surface water and groundwater is reduced by 21.70% and 11.14%， respectively， the water supply ratio of reuse water and yellow diversion water is increased by 5.43% and 27.40%. It accords with the idea of water resources allocation in Taiyuan， and verifies the effectiveness of CO. Compared with GA， CO has increased regional satisfaction， integrated water value and water distribution equity by 7.10%， 3.57% and 56.52%， respectively， reduced pollutant emissions and water shortage by 3.11% and 3.85%， at the same time， it has better coordination， which verifies the superiority of the CO. The results show that CO is reasonable and feasible to solve the optimal allocation model of water resources， which can provide a new method for solving the optimal allocation model of water resources.

Dike foundation piping is one of the most important reasons for the dangerous situation and even the breach of dikes in Poyang Lake Area. It is of great significance to study and explain the disaster-causing mechanism of dike foundation piping in Poyang Lake Area from the mechanism so as to deal with the dangerous situation of piping and carry out emergency rescue scientifically.In this paper， an organic glass model tank is designed and manufactured， which integrates the functions of pore water pressure signal acquisition， seepage flow observation， erosion soil particle collection， graded constant head control， internal water circulation and so on. Based on the typical dike foundation structure of a dike in Poyang Lake Area， the physical model test of piping is carried out by using the prototype soil of dike foundation， and the whole process of piping occurrence， development and collapse of prototype multi-layer complex dike foundation is completely simulated. Combined with the experimental phenomena， the hydraulic conditions and soil particle erosion rate of the six stages （ surface crack development stage， latent layer failure stage， overlying layer failure stage， piping channel upstream stage， embankment foundation failure stage， embankment foundation collapse stage ） in the process of piping occurrence and development are analyzed， and the special embankment foundation failure mechanism of the dike is explained by analyzing the effect of vertical seepage force.The results show that the piping failure in the dike foundation structure in Poyang Lake Area is concentrated in the underlying sand layer. Under the action of osmotic pressure， the overlying weak permeable layer is penetrated by seepage， and the piping channel is formed on the contact surface between the sand layer and the overlying layer， which accelerates the erosion of soil particles and eventually leads to the dike foundation collapse. The loam layer sandwiched in the clay layer in the embankment foundation of a typical dike in the Poyang Lake Area cannot play the role of the transition layer between the strong and weak permeable layers. The law of soil particle erosion is similar to that of the embankment foundation without the transition layer. The growth rate of the cumulative erosion of the soil particles forming the surface piping outlet and the embankment foundation cracking is less than 0.1 g/s； the average overall critical horizontal gradient of the dike foundation where piping occurs is about 0.1.

In order to explore the theoretical technology and method system for the formulation of ecological water replenishment schemes for seasonal rivers， this paper proposes an ecological water replenishment mutual feedback method combining ecological flow calculation and hydrodynamic model.Based on the diagnosis of hydrological sequence variation， this method lays a foundation for the formulation of ecological water replenishment scheme by calculating the ecological flow process in the river channel.Then， according to the actual process of ecological water replenishment， a hydrodynamic model of river ecological water replenishment is established and solved， which provides a model support for the study of ecological water replenishment evolution. Finally， the hydrodynamic model is used to analyze the water replenishment evolution of different water replenishment schemes， so as to obtain the best water replenishment scheme under the given target. By taking a seasonal river DY River in the north as an example， this paper carries out an application analysis. The results show that： ① The variation point of TG reservoir inflow runoff sequence is 1980. On this basis， the runoff sequence can be divided into two parts： natural stage and human disturbance stage. ② The ecological water replenishment scheme obtained by the ecological water replenishment mutual feedback method can not only meet the ecological flow demand in the river channel， but also fully consider the characteristics of the annual variation of the river runoff process， which is of important scientific significance for understanding the ecological flow restoration process of seasonal rivers in China； ③ The ecological water replenishment mutual feedback method is applied to the ecological water replenishment project of DY River， and the effect is good. According to the requirements of the ecological water replenishment target， the ecological water replenishment scheme that meets the water supply of the whole line in DY River is obtained， that is， when the ecological water replenishment of DY River is in spring， the optimal ecological water replenishment flow from March to May is 49.82， 27.17 and 8.67 m³/s. This method can be used for the subsequent DY River to formulate corresponding ecological water replenishment schemes for other water replenishment targets， and can also provide new methods and new ideas for the formulation of ecological water replenishment schemes for seasonal rivers in other basins.

To study the rainfall runoff characteristics of green roofs in the northern semi humid area， this papar explores the impact of structural parameter changes on their rainwater storage effect， and provide data support and guidance for the green New Infrastructure of Sponge city. Based on measured rainfall runoff， we established comparative experiments on three types of roofs： green roofs （gradient of 0°）， ordinary roofs （gradient of 0°）， and sloping roofs （gradient of 5°）， as well as five sets of green roof experiments based on adjustment of matrix layer thickness， matrix layer， drainage layer， and drainage method structural parameters. Research shows that during the monitoring period， the average runoff coefficients of comprehensive rainfall on green roofs， ordinary roofs， and sloping roofs were 0.27， 0.50， and 0.77， respectively. The reduction effect of rainfall runoff on green roofs was significant. As the intensity of rainfall increases， the probability of green roof runoff generation gradually increases， and the total runoff control rate gradually decreases. Green roofs have a significant effect on the storage of rainfall at light to heavy rain levels. During the monitoring period， the runoff coefficient of green roof rainfall is significantly positively correlated with the total rainfall， but weakly negatively correlated with the number of days between previous rainfall intervals and previous rainfall； The green roof significantly reduces the efficiency of rainfall runoff control， corresponding to a critical value of 30 mm of rainfall intensity. The significant factors affecting the rainwater storage effect of green roofs in descending order are： substrate layer （ultra light substrate layer）， drainage method （5 cm drainage from the bottom）， substrate layer thickness （20 cm）， and drainage layer （ceramic layer）. The rainfall runoff characteristics and rainwater storage effect of green roofs are influenced by multiple internal and external factors. Based on long-term rainfall monitoring periods， further research should be conducted on the ecological effects of sponge cities with different structures and scales of green roofs.

Clarifying the pollution status and quantitatively identifying the pollution sources are an important foundation for effective control of surface water pollution. This paper proposes a data preprocessing method by comprehensively uses mathematical statistics， GIS spatial analysis， and principal component analysis-absolute principal component score-multiple linear regression （PCA-APCS-MLR） model and other methods. Using this data preprocessing method， a quantitative analysis of pollution sources and their contribution rates was conducted on 23 key water quality indicators at 49 sampling points in the surface water system of Yan'an City from 2018 to 2020. The results indicate： ① The main sources of surface water pollution are nutrient pollution （54.02%） and heavy metal pollution （17.97%）. ② The average contribution rate of nutrient pollution to various water quality indicators （52.70%） is greater than that of heavy metal pollution （47.30%）. ③ Nutrient pollution has a greater impact on the northeastern part of Yan'an City， and heavy metal pollution has a greater impact on the upper reaches of Yan River and Xiuyan River. Compared with the traditional data preprocessing method， the analysis results of the data preprocessing method proposed in this paper are more regionally representative which can provide a data support for the treatment of surface water pollution in Yan′an City and a scientific reference for the designation of surface water quality protection policies.

Understanding the characteristics of extreme climate change is an important part of groundwater resources management. To investigate the impact of extreme climate on groundwater resources， this paper takes the Dagu River groundwater resources area as an example and based on the meteorological data for over 40 years， 14 extreme climate indices are selected to analyze the trend， mutation and periodicity of meteorological changes in the study area by using methods such as linear trend method， Mann-Kendall mutation test and wavelet analysis. And based on extreme climate index and hydrological frequency analysis， the representative extreme climate years of the research area are determined， and the groundwater balance under extreme climate year is estimated by using water balance analysis method. Meanwhile， the influence of extreme climate index on groundwater depth is evaluated by correlation coefficient and periodicity analysis. The results are listed as follows： ① From 1979 to 2021， the extreme climate indexes representing high temperature and precipitation show an increasing trend， while the extreme climate indexes representing low temperature show a decreasing trend. The representative years of extreme wet years are 2007 and 2020， and the representative years of extreme dry years are 1981 and 1986. ② The extreme climate indexes mutate in the 1990s and the early 21st century， and then the trend of high temperature becomes more significant. ③The mean period corresponding to the first main period of the extreme temperature index have three main variation ranges： 20~24， 12~14 and 5~6 a. The average cycles of most extreme rainfall indices are 11~16 a. ④ Extreme precipitation index inhibits the increase in groundwater level depth. Extreme precipitation （R95p） and simple daily precipitation intensity index （SDII） are the main precipitation indexes affecting the groundwater level depth in Dagu River water source area. The period corresponding to the first main cycle of the groundwater level depth is 13~15 a， which is basically consistent with the cycle characteristics of the extreme precipitation index. This means that the groundwater level depth is significantly affected by the extreme precipitation index. ⑤ In the extreme wet year （2007）， the total groundwater recharge and discharge are about 319.4 mm and 204.8 mm， respectively， with an equilibrium difference of about 114.6 mm. In the extreme dry year （1981）， the total groundwater recharge and discharge are about 104.2 mm and 141.1 mm， respectively， with an equilibrium difference of about -36.9 mm.

Based on fractal geometry， the description of soil pore size distribution is established， and a suction model of unsaturated soil matrix considering the fractal characteristic of pores is proposed in combination with Young-Laplace theory. For the loess in Shaanxi region， the soil-water characteristic curves of different dry densities were collected， and the effects of dry density on pore distribution characteristics and matrix suction were analyzed. The results show that the loess pores have obvious fractal properties， which can be well described by the Menger Sponge fractal model. The pore fractal model parameter analysis shows that with the increase of dry density， the fractal dimension D increases approximately linearly， while the maximum aperture L decreases as a power function. Through function fitting， the characteristic relationship between loess pore size distribution and dry density is further established， and the suction capacity of loess matrix suction under different dry densities is quickly predicted.

Soil water content is one of the important factors affecting the growth of crops， and plays an important role in crop yield estimation and drought monitoring. In soil water content calculation， multiple spectral variables are generally extracted for inversion， but the spectral information contained between the variables may have redundancy and overlap. In order to extract effective feature variables and make them independent of each other， the thesis selects the feature variable screening method and verifies the applicability in soil water content inversion. Based on the UAV multispectral images， 12 types of vegetation indices such as Normalized Difference Vegetation Index （NDVI） are calculated， combined with UAV thermal infrared（TIR）data to calculate the Land Surface Temperature （LST） and the corresponding Temperature Vegetation Dryness Index （TVDI）， as well as four backscattering coefficients obtained from miniSAR data processing. XGBoost feature variables and the Best Subset Selection （BSS） algorithm were used to screen the optimal variable combinations， and then Partial Least Squares Regression （PLSR） and Random Forest Regression （RFR） algorithm was used to invert， the soil water content at the tasseling stage of winter wheat in the experimental area. The research results show that： ① The inversion results of 0~20 cm depth are better than those of 0~10 cm depth； ②Comparing the four soil moisture inversion models of XGBoost-PLSR， XGBoost-RFR， BSS-PLSR and BSS-RFR， the inversion accuracy of the RFR model at different depths is the highest； ③The inversion accuracy of the XGBoost-PLSR model is better than that of XGBoost-RFR at a soil depth of 0~10 cm， but the inversion accuracy is the opposite at a depth of 0~20 cm， where the inversion accuracy of the BSS-RFR model is higher than that of BSS-PLSR. The research results can provide theoretical and technical support for UAV multispectral remote sensing inversion of soil water content， and provide test basis for satellite remote sensing large-scale soil moisture monitoring.