[Objective] Global warming has led to frequent natural disasters. The spatiotemporal evolution patterns of drought in the Mu Us Sandy Land(MUSL) are investigated. [Methods] Based on high-resolution meteorological data from 2002 to 2021, multiple method were employed, including percent of normal precipitation(PNP), empirical orthogonal function(EOF), Mann-Kendall(M-K) test, and wavelet analysis, to analyze the spatiotemporal characteristics of drought in this region across multiple time scales. [Results] The result showed that:(1) except for the monthly scale, where the drought trend was insignificant, all other time scales showed a significant alleviating trend, which was more pronounced at longer time scales.(2) Drought was spatially distributed in a central radiating pattern, with the central area experiencing more severe drought and being more sensitive to responses. Short-term droughts were characterized by intense fluctuations and frequent wet-dry transitions, while long-term droughts persisted for extended durations with weaker fluctuations.(3) PNP detected more extreme drought events at short-term time scales, exhibiting irregular periodic changes. In contrast, droughts at long-term time scales persisted for extended durations with more stable cycles, and showed significant changes around 2011. [Conclusion] These findings reveal the multi-scale characteristics of drought in the MUSL, which can provide a scientific basis for the evaluation and application of drought indicators at different time scales, regional drought monitoring, and the formulation of disaster prevention measures.

[Objective] In the context of climate change and intensified human activities, traditional hydrological drought indices based on the stationarity assumption struggle to accurately characterize the actual drought conditions. To scientifically evaluate hydrological drought under non-stationary conditions, it is necessary to develop a novel drought assessment method that can integrate both natural and human impact factors.[Methods] Based on multi-source data, the natural runoff series during the disturbance period was reconstructed, and the human activity index(HI) was quantified after comparing the performance of long short-term memory(LSTM) and random forest(RF) machine learning models. Using the generalized additive models for location, scale and shape(GAMLSS) model, stationary and non-stationary models were constructed with precipitation, temperature, and HI as covariates, and the stationary hydrological drought index(SRI) and non-stationary hydrological drought index(NSRI) were calculated. Based on the performance comparison of the two indices, the characteristics of hydrological drought under non-stationary conditions were revealed using Copula functions.[Results] The result showed that:(1) the runoff series in the study area underwent an abrupt change in 1988, exhibiting significant non-stationary characteristics, with runoff increasing by 35.5% after the change point.(2) The fitting performance of the non-stationary model was superior to that of the stationary model in all months, with the model combination including HI showing the best performance. Compared with SRI, NSRI was more accurate in identifying the process and severity of typical drought events, while the SRI showed significant underestimation.(3) From 1989 to 2020, a total of 37 hydrological drought events occurred in the river basin, with an average duration of 2.97 months and an average severity of 3.27. The overall drought trend was intensifying, with spring droughts characterized by long duration and high severity, while summer droughts exhibited short duration and high frequency.(4) Joint probability analysis based on Copula functions indicated a negative correlation between drought duration and severity. The joint return period(12.9 years) of hydrological drought was significantly shorter than the co-occurrence return period(113.4 years).[Conclusion] The constructed NSRI characterizes non-stationary drought processes more accurately than traditional method by coupling climate and human activity factors, and human activity is identified as a key driver of drought evolution. During the disturbance period, hydrological drought in the river basin has intensified. The findings can provide a scientific basis for water resource management and risk prevention in arid areas.

[Objective] Future evolution of hydrological drought in the water source area of the South-to-North Water Diversion Middle Route Project under climate change is quantified, providing a basis for the security management and scheduling optimization of water resources for the project.[Methods] Downscaled data under the SSP1-2.6, SSP2-4.5, and SSP5-8.5 scenarios from the CMCC-ESM2 model in CMIP6 were used to drive the SWAT model to simulate the runoff processes in the water source area for the future period(2026—2100). The Mann-Kendall trend test and the annual-scale standardized runoff index(SRI) were applied to identify evolution trends of runoff and hydrological drought.[Results] Compared with the baseline period(1970—2020), the future multi-year average runoff in the water source area showed an increasing trend under different climate scenarios, and the magnitude of increase expanded as radiative forcing intensified. The upper-reach Hanzhong station exhibited the most pronounced increase and trend significance, while the runoff increment was most notable at the downstream Baihe station. Significant spatial heterogeneity was observed in the future hydrological drought across the water source area. Under the SSP1-2.6 scenario, the drought frequency in the middle and lower reaches decreased to 0.33~0.45, and the drought severity was reduced compared with the baseline period. Under the SSP2-4.5 and SSP5-8.5 scenarios, the drought frequency in the upper reach increased to 0.40~0.45, and the maximum increases in drought duration and severity at the Ankang station reached 22.1% and 23.4%, respectively.[Conclusion] The water source area exhibits pronounced spatial differences in its response to climate change. Under low-to-medium emission scenarios, Hanzhong and Baihe stations are characterized by a “high-frequency, short-duration, and low-severity” pattern. Under high-emission scenarios, droughts in the middle and lower reaches tend to become more persistent and intense, imposing greater demands on water resource allocation and drought mitigation in the future for the Middle Route Project.

[Objective] Flood risk assessment is crucial for developing regional disaster prevention and mitigation systems. However, current research focuses mainly on large-scale areas such as counties and cities, with limited attention to small-scale urban districts. The aim is to effectively assess urban flood risk and enhance the refinement level of risk assessment.[Methods] A flood risk assessment method suitable for urban areas was established by integrating the MIKEFLOOD hydrodynamic modeling technique with a game theory-based combination weighting model. Taking the urban area of Dexing City as an example, multi-source data were collected, including rainfall and runoff process from 12:00 on June 18 to 12:00 on June 23, 2022, a high-resolution digital elevation model(DEM) of the urban area obtained from UAV aerial photography, and measured river cross-sections. A MIKEFLOOD hydrodynamic model for the Dexing urban area was then constructed for flood simulation. After validating the model, four hazard indicators were selected based on the flood simulation result: maximum inundation depth, maximum inundation duration, maximum inundation area, and maximum inundation flow velocity. These were combined with indicators of environmental vulnerability and disaster-bearing body exposure to establish an urban flood disaster risk assessment system. To address the indicator weighting problem within the assessment system, game theory was innovatively introduced to effectively integrate subjective experience with objective data information, thereby overcoming the limitations of single weighting method. This enabled systematic analysis of the flood disaster risk indicators and effective assessment of flood disaster risk in the Dexing urban area.[Results] The result showed that there were 22 communities in the Dexing urban area, among which 13 were classified as high-risk, including Xinyingyi Village(high-risk area of 0.115 km²), Shangzhou Community(0.067 km²), Tianmenshan Community(0.034 km²), Suihanshan Community(0.024 km²), Jishui Community(0.020 km²), Shuilvqiao Community(0.019 km²), Yinshan Community(0.013 km²), Nanmen Community(0.012 km²), Kentangshan Community(0.011 km²), Yinquan Community(0.011 km²), Wuyuan Community(0.009 km²), Shengliting Community(0.004 km²), and Wuanlu Community(0.002 km²).[Conclusion] The result indicate that the MIKEFLOOD model demonstrates good accuracy and reliability in urban flood simulation, effectively revealing the spatial distribution characteristics of flood in urban areas. The high-precision DEM obtained via UAV aerial photography is crucial for urban flood risk assessment, as it not only provides accurate topographic information for the MIKEFLOOD model but also offers a reliable basis for environmental vulnerability analysis. For small-scale urban areas, UAV aerial surveying should be used to obtain elevation data. Significant differences are observed between the weights calculated by the entropy weight method and the analytic hierarchy process, making it difficult to achieve effective balance through subjective judgment alone. In contrast, the game theory-based combination weighting method, based on mathematical principles, integrates the result of these two weighting method to achieve a more scientific and rational weight allocation, providing important reference value for weight determination in comprehensive assessment. The flood risk assessment result for the Dexing urban area provide a reliable decision-making basis for the precise prevention of flood disasters and for risk classification and control in the future. Additionally, they can serve as a theoretical and method ological reference for flood prevention and mitigation efforts in similar regions.

[Objective] To identify the major pollution sources in typical river sections within highly urbanized watersheds and enhance the scientific and precise management of urban river water quality, the Qinhuai River Basin in Nanjing was selected as a case study. By integrating the fluorescence characteristics of dissolved organic matter(DOM) with hydrochemical parameters, an evaluation framework is developed for pollution source analysis and water quality correlation.[Methods] Water samples from typical cross-sections of the main and tributary channels of the Qinhuai River were analyzed using three-dimensional excitation-emission matrix(EEM) fluorescence spectroscopy combined with parallel factor analysis(PARAFAC) to extract DOM fluorescent components. Fluorescence indices such as SUVA254 and spectral slope(S_R) were used to further interpret their source characteristics. Meanwhile, cations(K⁺, Ca²⁺, Na⁺, Mg²⁺) and other water quality indicators were measured, and the correlation analysis with the water quality index(WQI) was conducted to explore the evolution patterns of river water quality under multi-source pollution.[Results] The result showed that the fluorescent components C1—C4 were associated with surface runoff, sewage discharge, and sediment release, respectively. Among them, C3 was significantly elevated near sewage discharge points, C4 indicated endogenous release from sediments, and C2 showed composite characteristics of multiple sources. No significant correlations were observed among DOM spectral indices, suggesting strong interference from endogenous DOM on fluorescence signals, which limited their ability to be used independently for pollution source identification. Hydrochemical analysis revealed a negative correlation between cation concentrations and water quality conditions. The concentrations of the four ions synchronously increased in regions with low WQI, indicating that sewage concentration effects dominated their distribution. A significant positive correlation was observed between Na⁺ and NO^-₃-N. Combined with effluent monitoring data from sewage treatment plants, this indicated that the tailwater of sewage treatment plants was the primary source of total nitrogen in the watershed.[Conclusion] The combined analysis of DOM spectral characteristics and hydrochemical indicators helps identify key pollution types and driving factors in urban rivers, providing technical support for refined pollution control and zoned management strategies in urban river systems.

[Objective] Propidium monoazide(PMA) staining technique, a molecular tool that selectively inhibits DNA amplification from nonviable microorganisms, has recently shown unique advantages in monitoring viable microorganisms in aquatic ecosystems. Conventional molecular detection method fail to distinguish between viable and nonviable bacteria, leading to result that cannot reflect the viability of microorganisms. PMA can permeate the compromised membranes of dead cells, bind to intracellular DNA, and block its amplification, thereby enabling targeted analysis of viable microbial communities.[Methods] A systematic review of Chinese and English literature related to PMA staining technique is conducted to comparatively analyze its advantages and disadvantages in monitoring viable microorganisms in aquatic ecosystems.[Results] By adjusting PMA concentration, optimizing photoactivation conditions, and integrating with qPCR or flow cytometry, the detection sensitivity and specificity are significantly improved. Currently, this technology has been successfully applied across diverse fields such as the detection of viable pathogens in drinking water, wastewater treatment, functional microorganism tracking, and the assessment of viable microbial communities in algal blooms.[Conclusion] PMA technology provides an effective tool for observing the “vital status” of aquatic microorganisms. However, its application requires optimization tailored to specific scenarios to provide more precise technical support for drinking water safety, urban water supply quality assurance, aquatic ecosystem restoration, and cyanobacterial bloom management.

[Objective] The denitrification performance and microbial mechanisms of polyurethane(PU) media with different pore sizes in denitrification biofilters(DNBFs) are investigated to clarify the relationship between medium pore size and operating parameters, to reveal the regulatory effects of pore size on biofilm characteristics and nitrogen metabolism functional genes, thereby providing technical support for groundwater nitrate pollution control.[Methods] Three DNBFs were constructed using 15 ppi(large pore), 30 ppi(medium pore), and 40 ppi(small pore) PU media. The empty bed hydraulic retention time(EBHRT) was set to 3, 2, 1.5, and 1 h. The nitrate nitrogen(NO^-₃-N) removal rate and volumetric denitrification rate(R_vd) were monitored. The biomass on the medium surface and the content and composition of extracellular polymeric substances(EPS) were determined. The microbial community structure was analyzed via high-throughput sequencing, and the abundance of nitrogen metabolism functional genes(nirKS, norBC, nosZ) was predicted using PICRUSt2.[Results] At EBHRT=1.5 h, the DNBF with 30 ppi PU medium achieved a higher NO^-₃-N removal rate(93.15%) and R_vd(0.89 kg N·m^-3·d^-1), outperforming those with the 40 ppi(0.83 kg N·m^-3·d^-1) and 15 ppi(0.75 kg N·m^-3·d^-1) media. The biomass on this medium surface was 93.6 mg·cm^-3, and the extracellular polymeric substances(EPS) production was 49.61 mg·g^-1, both of which were higher than those of the other two media. Microbial community analysis showed that the 30 ppi medium enriched a higher relative abundance of the phylum Proteobacteria(63.93%) and denitrifying bacterial groups(e.g., Simplicispira: 15.87%; Cloacibacterium: 14.24%). PICRUSt2-based prediction of nitrogen function further revealed that the abundance of denitrification genes(nirKS, norBC, nosZ) in the 30 ppi medium was 1.6 times and 1.5 times higher than that of the 15 ppi and 40 ppi media, respectively.[Conclusion] The 30 ppi PU significantly improves the denitrification efficiency of DNBFs by increasing biomass and EPS production, enriching denitrifying functional bacterial groups, and enhancing the abundance of key genes. These findings provide reliable support for constructing efficient and stable groundwater nitrate removal systems, and also provide a theoretical basis and technical reference for promoting economically feasible denitrification processes in industrial and municipal wastewater treatment.

[Objective] Under the “dual carbon” goals and increasingly stringent ecological constraints, tapping the potential of existing hydropower stations and enhancing renewable energy integration capacity are crucial for building a new power system. A coordinated planning method for secondary hydropower development and multi-energy complementarity suitable for river basins with mature hydropower development is proposed.[Methods] Taking the Danjiangkou-Wangfuzhou cascade reservoirs in the Han River as the study object, a short-term optimal dispatch model was first constructed with dual objective of maximizing cascade power generation and generation benefits, incorporating downstream ecological flow constraints. After the feasible domain for hydropower capacity expansion was determined through life-cycle economic evaluation, a regional hydro-photovoltaic(PV) complementary coordinated operation model was further established with the objective of maximizing renewable energy integration and minimizing the variance of the system's residual load.[Results] The Danjiangkou Hydropower Station was found to possess the technical potential for capacity expansion to 1500 MW. Life-cycle economic comparison identified the 1200 MW expansion scheme as the optimal solution. Implementation of a 10~20 cm over-storage operation strategy for the Wangfuzhou Reservoir during the non-flood season effectively compensated for the peak shaving capacity loss caused by fixed ecological flow release. When complemented by an installed photovoltaic capacity of 750 MW, the system's total annual power generation reached 4 billion kWh, representing a 10.7% increase compared to the baseline scheme. Furthermore, the variance of the system's residual load was smaller, and the power deficit rate decreased from 2.48% to 2.43%.[Conclusion] The proposed system solution of “expansion potential assessment-ecological and peak-shaving coordination-multi-energy complementarity optimization” can provide a theoretical basis and practical reference for basin-level secondary hydropower development and capacity configuration of hydro-PV complementary systems. The configuration of Danjiangkou expansion to 1200 MW coupled with 750 MW PV is the optimal scheme, demonstrating significant advantages in improving power generation benefits and renewable energy integration.

[Objective] The high-precision ultra-short-term wind speed prediction technology is investigated to address the key bottleneck constraining the efficient utilization of wind power: the challenge of improving prediction accuracy caused by the non-stationary nature of wind speed and the inherent limitations of deep learning models.[Methods] Taking two wind farms in Jiangxi as the research objects, a dual-frequency encoder-decoder model based on an attention mechanism and long short-term memory(LSTM) was established. The model applied convolution theory to perform multi-scale decomposition of wind speed sequences, effectively mitigating the impact of non-stationarity. A novel frequency-time attention mechanism was proposed to adaptively capture high-frequency fluctuation characteristics through the interaction of features in the frequency and time domains. A two-layer LSTM encoder-decoder architecture was adopted to effectively extract low-frequency trend features through hierarchical nonlinear mapping and suppress error accumulation in multistep predictions.[Results] Comparative experiments using data from the two wind farms showed that the proposed model demonstrated excellent performance in 15-step, 30-step, and 60-step predictions. For 30-step prediction, the mean absolute error(MAE), mean squared error(MSE), and root mean squared error(RMSE) were 0.340 1, 0.281 6, and 0.530 6, respectively. For 60-step prediction, the MAE, MSE, and RMSE were 0.471 3, 0.504 7, and 0.710 2, respectively. This demonstrated that the proposed model showed a significant improvement in prediction accuracy compared to baseline models, along with outstanding generalization ability.[Conclusion] By leveraging collaborative modeling in the frequency and time domains and hierarchical feature extraction, the proposed model effectively addresses the non-stationarity of wind speed and the limitations in model accuracy. It demonstrates high precision and excellent generalization in multistep prediction, providing a reliable technical solution for ultra-short-term wind speed prediction in wind power generation scenarios.

[Objective] The Jinghe River Basin in a semi-arid region is selected as the study area to analyze the spatiotemporal variation characteristics of runoff and to investigate the response relationship of runoff to precipitation changes in the river basin.[Methods] Spatiotemporal variations in runoff were analyzed using runoff data from hydrological stations and precipitation data within the river basin from 2022 to 2024. The river basin was divided into four regions using the K-Means clustering algorithm based on the contribution of station precipitation to runoff, and the precipitation-runoff lag relationship in different regions was further analyzed. Finally, short-term runoff prediction models for the Jinghe River with lead times of 1 h, 3 h, and 6 h were developed based on a long short-term memory(LSTM) neural network.[Results] The result showed that:(1) the daily average runoff of the Jinghe River exhibited pronounced seasonal variation, with the flood season being the high-discharge period. Extreme peaks were concentrated in mid-to-late July, with three consecutive years showing a single extreme peak with multiple fluctuations.(2) Analysis of three typical flood peaks in the Jinghe River showed that when antecedent soil moisture content was similar, the maximum hourly precipitation and precipitation duration within the river basin were the key factors influencing the flood peaks. Specifically, longer precipitation duration, higher intensity, and larger total precipitation led to longer flood duration and higher peak discharges. When the antecedent soil moisture content in the 0~28 cm soil layer was below 0.2 m³/m³, the basin-wide average lag time reached 30 h, nearly twice that when the soil moisture content exceeded 0.4 m³/m³. Regional differences were observed in the precipitation-runoff transformation within the river basin. Under higher antecedent soil moisture content conditions, the optimal average lag times for regions 1~4 were 11 h, 12 h, 14 h, and 28 h, respectively. Under lower soil moisture content conditions, the lag times were 23 h, 30 h, 34 h, and 35 h, respectively.(3) A short-term runoff prediction model for the Jinghe River was established using the LSTM machine learning algorithm. The effects of different influencing factors on the model were tested, revealing that antecedent discharge at Jingcun and Yuluoping stations and basin average areal precipitation were identified as key factors. For lead times of 1 h, 3 h, and 6 h, the optimal models achieved Nash-Sutcliffe efficiency coefficients of 0.902, 0.774, and 0.676, respectively, and root mean square errors of 2.788 m³/s, 4.230 m³/s, and 5.070 m³/s, respectively. Prediction accuracy decreased with increasing lead time. The model performed well in simulating runoff during large floods and stable-discharge periods.[Conclusion] The identified lag effect of flood peaks on basin precipitation and the developed runoff prediction models can provide technical support for water resource regulation and flood prediction in the Jinghe River Basin and offer new insights for studying the precipitation-runoff relationships in small and medium-sized river basins in semi-arid regions.

[Objective] The persistent decline in the water level of Fuxian Lake has significantly impacted the ecological environment within its basin and the production and livelihoods in surrounding areas. Investigating the causes of this decline is crucial for protecting Fuxian Lake's ecosystem and realizing its socio-economic value.[Methods] A distributed hydrological model, Water and Energy Transfer Processes in Large River Basin(WEP-L), was applied to simulate the water resource volume of the Fuxian Lake Basin from 1956 to 2023. Based on the long-term average net water balance under natural conditions, the contributions of meteorological drought, the diversion of outflow from Xingyun Lake, water consumption, and external drainage to the reduction of Fuxian Lake's water volume in 2023 were quantified using observational data, contribution rate analysis, and the water balance equation.[Results] The result showed that:(1) the water level of Fuxian Lake generally showed a slow declining trend at a rate of 0.12 m·a^-1, with a drop of 1.6 m below the multi-year average water level.(2) From 1956 to 2023, the inflow into Fuxian Lake from the entire basin showed a declining trend. In 2023, the net water balance of Fuxian Lake was-111 million m³, a decrease of 204 million m³ compared with the 93 million m³ under the multi-year average natural conditions.(3) According to the calculation of the water balance equation, the contribution rate of reduced precipitation and increased lake evaporation caused by rising temperatures to the net water balance decline of Fuxian Lake was 65.7%, the contribution rate of the Xingyun Lake diversion project was 19.6%, and the contribution rate of water consumption and external drainage was 14.7%.[Conclusion] The decline in Fuxian Lake's water level is influenced by multiple factors. The study highlights the dominant role of climate change, providing a scientific basis for watershed managers to formulate water resource allocation strategies and holding great significance for ensuring the ecological security of Fuxian Lake.

[Objective] Due to unclear physical mechanisms and numerous influencing factors, modeling analysis of complex water resource coupling systems is one of the key challenges in the current research on hydrology and water resources. It is of great significance for the scientific management of water resources to clarify the practicality and accuracy of method such as data-driven method and system dynamics(SD).[Methods] Based on the previous research achievements of the research team, the water supply-power generation-environment(WPE) coupling system in the upper reaches of the Yangtze River Basin was taken as the research object. Using data from 1998 to 2020, 11 typical data-driven method were compared with the SD model to systematically evaluate their performance in the modeling of complex water resource coupling systems.[Results] The result showed that:(1) except for the SD model, all the other models exhibited different degrees of overfitting. The SD model demonstrated the best performance, with an average simulated R² of approximately 0.81 for all variables, showing strong explanatory power for system changes, and an average MAPE of approximately 11%, indicating high prediction accuracy.(2) The multiple linear regression(MLR) model performed relatively well, with an average simulated R² of approximately 0.57 for all variables and an average MAPE of about 19%.[Conclusion] The SD model can effectively solve the problems of complex water resource coupling systems, demonstrating strong stability and reliability. However, it requires modelers to have a high level of proficiency in complex system theories and method. Although the MLR model cannot comprehensively characterize the WPE system, it can meet the basic analysis requirements with simple operation and high practical applicability. The relevant research findings can provide a reference for the modeling and analysis of complex water resource systems.

[Objective] Global climate change, population growth, and intensified social activities have led to variations in the spatiotemporal distribution of water resources, posing challenges to the development and utilization of water resources. Optimizing the water scheduling of cascade reservoir groups is crucial for water resource allocation. However, current research lacks sufficient consideration of decision-makers' preferences for water shortage patterns and time.[Methods] To address the challenges in water resource allocation, an optimal water volume scheduling model for cascade reservoir groups was established with the primary objective of minimizing water supply deficits. Three water shortage pattern strategies—total water supply deficit control, centralized distribution, and wide-shallow distribution—were proposed, along with water shortage time strategies that adjusted weights across different periods. Then, an analysis was conducted using the Toutun River Basin as a case study.[Results] In the case study of water shortage pattern strategies, when the inflow frequency was p=50%, the wide-shallow distribution pattern result ed in the smallest supply deficit(493 700 m³), the centralized distribution pattern result ed in the largest deficit(1 797 400 m³), and the total water supply deficit control pattern fell in between(973 300 m³). When the inflow frequencies were p=95% and p=75%, although the total supply deficits under the three patterns were similar, the water shortage was more evenly distributed under the wide-shallow distribution pattern. In the analysis of the water shortage time strategy, using p=95% as an example, after applying the strategy, the water supply deficits in March to May decreased from 58 400 m³, 5 735 800 m³, and 6 246 600 m³ to 49 000 m³, 134 000 m³, and 124 600 m³, respectively. However, the deficits in other months increased from 32 496 200 m³ to 44 223 200 m³, and the total water supply deficit remained essentially unchanged.[Conclusion] In summary, the proposed strategies are proven effective. The wide-shallow distribution pattern demonstrates clear advantages in ensuring supply stability. The water shortage time strategy can reduce water supply deficits in specific periods, but it does not reduce the total deficit—it only redistributes the time of the shortage. The findings can serve as a reference for water allocation of cascade reservoir groups.

[Objective] Accurate water level prediction is essential for flood control, navigation safety, and water resource management. Traditional hydrological models often fail to capture the complex spatiotemporal dependencies and nonlinear characteristics in water level fluctuations. [Methods] To address these limitations, the Hydro Fusion Net deep learning framework was proposed, integrating CNN-LSTM module and self-attention mechanism in a parallel architecture to improve prediction accuracy. The CNN-LSTM module extracted local spatiotemporal dependencies, while the self-attention mechanism captured global contextual relationships. The outputs of both modules were fused for final prediction. This framework was optimized for deployment on the Ascend CANN platform, leveraging high-performance computing to achieve efficient training and prediction.[Results] Experiments on the Xijiang River dataset showed that Hydro Fusion Net outperformed conventional models, achieving an RMSE of 0. 394 and an R² of 0. 894. Furthermore, with optimization on Ascend CANN, the training speed increased by 2. 3 times, and energy consumption reduced by 17%, indicating significant computational advantages. [Conclusion] The proposed framework effectively captures complex hydrological dynamics, providing real-time water level prediction support for intelligent hydrological management and disaster prevention.

[Objective] The CSU-HIDRO and WSR88D algorithms are the primary dual-polarization radar quantitative precipitation estimation(QPE) algorithms used in operational meteorology. However, their applicability varies under different climatic conditions, necessitating a detailed evaluation of the suitability of these two algorithms in the peri-Shanghai region.[Methods] Data from the S-band dual-polarization radar in Qingpu, Shanghai, and minute-level rainfall data from automatic stations within the radar coverage were utilized. Based on 46 typical precipitation events in the peri-Shanghai region from 2023 to 2024, the CSU-HIDRO and WSR88D algorithms were evaluated and verified using four method: non-stratified assessment, stratification by observed rainfall intensity(rain rate stratification), stratification by station-to-radar distance(distance stratification), and classification of influencing weather systems.[Results] The result showed that:(1) the accuracy of the CSU-HIDRO algorithm was higher than that of the WSR88D algorithm in unstratified, rain rate-stratified, and distance-stratified evaluation. Moreover, the CSU-HIDRO algorithm achieved the best precipitation estimation performance within the range of hourly rainfall intensity between 35 and 50 mm and within distances of 10 to 75 km from the radar.(2) Verification based on weather system types revealed that both algorithms demonstrated comparable precipitation estimation performance for weather systems forced by upper-level cold advection. However, for samples of weather systems forced by low-level warm advection, quasi-barotropic systems, and baroclinic frontogenesis systems, the CSU-HIDRO algorithm outperformed the WSR88D algorithm in quantitative precipitation estimation.[Conclusion] Overall, the CSU-HIDRO algorithm demonstrates stronger applicability in the peri-Shanghai region. It provides higher estimation accuracy, especially for moderate rainfall intensity and medium distances, thereby providing a better option for radar-based precipitation retrieval in this region.

[Objective] To address the low efficiency and high risks associated with manual inspections of long-distance hydraulic tunnels, an intelligent tunnel inspection system is developed that integrates an autonomous inspection robot with deep learning-based crack identification and quantification, enabling accurate and efficient detection of tunnel defects.[Methods] Through multi-sensor fusion(the robot integrates laser SLAM, depth vision, encoders, and IMU), the system achieves autonomous obstacle avoidance and path planning in complex tunnel environments. Combined with passive RFID tag-assisted positioning and calibration, it ensures robotic movement accuracy within ±2 cm. The robot body is equipped with an intelligent fill-light system and a ring-shaped imaging device, enabling image capture in low-light and confined spaces. Through multiple rounds of testing and optimization of key parameters—including travel speed(0.6~1.0 m/s), exposure time(1.5~4 ms), and camera gain(3~9 dB)—the system achieves stable vertical acquisition of high-definition tunnel wall imagery. Captured images feed into CrackARNet, a crack detection model based on an enhanced U-Net architecture. This model incorporates residual connections and channel attention mechanisms, outperforming mainstream models like U-Net, TernausNet, and Mask R-CNN on public crack datasets(with approximately 4% improvement in IoU). The system automatically outputs crack prediction masks, visualizes result, and extracts morphological parameters such as crack length and width.[Results] Experiments demonstrate that the system achieves approximately threefold improvement in single-task detection efficiency compared to traditional manual method. The CNN algorithm can identify cracks as narrow as 2 pixels, with millimeter-level measurement accuracy.[Conclusion] By integrating multi-sensor fusion and AI algorithms, this system provides a feasible integrated solution for precise navigation, efficient imaging, and intelligent diagnostics in long-distance hydraulic tunnels, offering significant engineering application value.

[Objective] With the increasing number of water transfer projects in China, establishing an efficient intelligent inspection system is crucial for the long-term safe operation and maintenance of water transfer projects. Precise positioning within hydraulic tunnels is key and challenging for ROVs to achieve autonomous inspection, while existing ROV underwater positioning method primarily target marine environments, making conventional positioning approaches difficult to apply effectively in hydraulic tunnels.[Methods] An underwater positioning technology based on enhanced visual-inertial SLAM is proposed. Aiming at the problems of limited feature regions and blurred features in tunnel images, an underwater image enhancement network based on DE-MFET is constructed. This network fuses depth information and multi-scale feature enhancement modules to enhance the channel response of important image features, highlight the local edge details of underwater images, and improve the quantity and efficiency of feature matching between SLAM keyframes. To counteract frequent visual odometry failures caused by weak features and strong feature homogeneity in hydraulic tunnels, enhanced visual odometry is fused with inertial navigation data to further correct the ROVs instantaneous pose, improving positioning accuracy in hydraulic tunnels.[Results] Validation using a self-developed hydraulic tunnel dataset, the LSUI and the UIEB public underwater dataset demonstrates that the proposed image enhancement method significantly improves underwater image quality, outperforming other enhancement method in UCIQE, UIQM scores, and ORB feature matching quantity. ROV underwater positioning experiments in Hubei's Ebei Water Resources Allocation Project confirm that the enhanced visual-inertial SLAM method effectively improves positioning accuracy within hydraulic tunnels.[Conclusion] This positioning method exhibits notable superiority and robustness in hydraulic tunnel environments, providing a critical research foundation for intelligent inspection solutions in water transfer projects.

[Objective] The aim is to investigate the hydraulic noise characteristics and control strategies of large mixed-flow turbines, specifically targeting the hydraulic whistling noise generated during the operation of a specific large mixed-flow turbine. [Methods] The Reynolds-averaged Navier-Stokes(RANS) method was used to perform numerical simulations of the large mixed-flow turbine under hydraulic whistling conditions, both before and after modifying trailing edges of the runner blades. Pressure pulsations were monitored at internal measurement points of key flow-passing components. Using fluid-acoustic coupling, the distribution of external sound pressure levels and internal sound power levels in flow-passing components such as the runner and draft tube were analyzed. The flow field distribution patterns and mechanisms of hydraulic noise reduction before and after the modification of the runner blade trailing edges were analyzed. [Results] The result showed that abnormal pressure pulsation frequencies in the range of 300 to 400 Hz were detected in flow-passing components including the volute, runner, and draft tube. Areas with high sound power levels were mainly located near the trailing edges of the runner blades. [Conclusion] The findings indicate that complex vortex structure is one of the causes of hydraulic noise in turbines. Modifying the trailing edges of the runner blades can achieve hydraulic noise reduction in turbines.

[Objective] The stability and failure modes of bedding rock slopes are closely related to the shear mechanical behavior of layered rock masses. To thoroughly reveal the underlying mechanisms, the effects of different shear angles and water-bearing states on the shear mechanical behavior of bedded limestone are investigated. [Methods] Variable-angle shear tests were conducted, combining with digital image correlation(DIC) and microseismic monitoring techniques, on bedded limestone specimens under both natural and saturated water-bearing states at different shear angles(30°~60°). Their mechanical properties, strain field evolution, and microseismic signal characteristics were systematically studied. [Results] The result demonstrated that:(1) with increasing shear angle, the peak shear stress, normal stress, and shear stress of the specimens gradually decreased, while the failure mode transitioned from shear fracture of the rock matrix to slip along the bedding planes.(2) At lower shear angles, the specimens tended to develop shear-tensile composite fractures with relatively fragmented failure patterns. As the shear angle increased, the effects of normal stress weakened, and the failure mode gradually transformed into a single slip mode dominated by shear along the bedding planes.(3) Under the natural state, microseismic activity exhibited a clear sequence of “active-quiet-precursor-mainshock” with high signal peaks, indicating typical brittle failure. Under the saturated state, the lubricating effects of water caused microseismic activity to remain continuously active, but the signal peaks were significantly lower, demonstrating microseismic signal characteristics of a transition from brittle to plastic failure.(4) The shear angle dominated the macroscopic deformation and failure mode of bedded limestone specimens. At lower angles, microseismic activity was dispersed with high cumulative counts, while at higher angles, microseismic precursor signals became concentrated with a significant quiet period, [Results]ing in more sudden failure. [Conclusion] The research findings provide theoretical and data support for stability evaluation and disaster early warning of bedding rock slopes under rainfall conditions by revealing the variable-angle shear mechanical properties and microseismic signal characteristics of bedded limestone under different water-bearing states.

[Objective] To address the deficiencies in spatial feature capturing and nonlinear fitting in current regional karst collapse susceptibility studies, a modeling framework based on graph convolutional networks(GCN) is proposed. [Methods] The fractal characteristics of karst development were integrated to quantitatively evaluate the degree of karst development. Shichang Township in Bijie City, Guizhou Province, was selected as the study area. 16 disaster-inducing factors were selected to establish a GCN-based collapse susceptibility evaluation model, and the regional karst collapse susceptibility was investigated by comparing the GCN model with support vector machine(SVM) and random forest(RF) models. [Results] The result showed that different strata in the same karst area had varying impacts on karst collapse. Converting complex karst morphologies into fractal dimensions enabled the quantitative assessment of karst development degree, effectively avoiding subjective judgment bias. All three evaluation models mitigated the issue of insufficient nonlinear fitting, and the orientation of very high/high susceptibility zones was roughly consistent with the trend of karst troughs in the study area, and the models exhibited both similarities and differences. Among them, the accuracy, sensitivity, specificity, precision, and AUC of the GCN model were significantly higher than those of the other two models. The very high susceptibility zones identified by the GCN model displayed a continuous strip-like distribution, which aligned with the orientation of the karst troughs in the study area, with an overlap rate of 92%. Moreover, the GCN model concentrated 33 hazard points(accounting for 75%) in these very high susceptibility zones, while the area of these zones only accounted for 18.18% of the total area, demonstrating the model's strong capability to identify the clustering of hazard points in high susceptibility zones. [Conclusion] By integrating the fractal characteristics of karst development, the GCN model accurately identifies the very high susceptibility zones in the central and western parts of the study area, and the spatial distribution is fully consistent with the strong karst development zones. Compared with traditional single-indicator evaluation, the model captures the spatial heterogeneity of collapse risks more effectively. The findings enhance the understanding of the karst development degree in Shichang Township, Jinsha County, Bijie City. Additionally, a relatively suitable evaluation system for karst collapse susceptibility is developed. The GCN model demonstrates more prominent performance in regional karst collapse susceptibility evaluation.

[Objective] Water resource management in the Yangtze River Basin faces many challenges, especially low agricultural water use efficiency and incomplete data measurement. These problems not only affect the rational allocation of water resources but also restrict the effective management of droughts in the river basin. To address these problems, an accounting method for the total water leakage and discharge in irrigation areas under the condition of data scarcity is developed. [Methods] An irrigation area in the middle and lower reaches of the Yangtze River was taken as an example. Multi-source remote sensing data and an improved SCS-CN model considering irrigation-induced surface runoff were utilized. Key factors such as rainfall, evaporation, irrigation water volume, surface runoff, and water supply from waterworks were integrated to estimate water consumption and water loss from the reservoir to the irrigation area, and the gap between potential and actual evapotranspiration was calculated. [Results] The result showed that the proportion of deep leakage and water discharge loss in the water supply of the irrigation area reservoir was significant, and the ratio of unmeasured water volume to the reservoir outflow water volume reached as high as 1.11. High water-saving potential could be realized in the future through measures such as precise irrigation regulation and canal seepage control renovation. [Conclusion] The results show that through the calculation and analysis of the monthly water supply and demand of the reservoir, it is found that during the summer irrigation period from June to October, the reservoir outflow matches well with the irrigation water demand, and the loss ratio is relatively lower, which verifies the feasibility of the method. Based on reliable data sources and rigorous methodological logic, this method can provide an effective reference for reservoir water management in irrigation areas with data scarcity. It offers a scientific basis for prioritizing daily water supply during dry seasons, optimizing reservoir outflow scheduling to reduce losses, and improving water resource utilization efficiency, thereby alleviating drought and water scarcity issues.