Rainfall–runoff modeling is a key challenge in hydrological research. Despite the extensive application of long short-term memory (LSTM) networks in rainfall–runoff modeling, our understanding of the influence of different hyperparameter configurations on various hydrograph components, as well as the linkages between hydrological concepts and LSTM architectures, remains elusive. Here, we integrated Multi-Objective Particle Swarm Optimization (MOPSO) with LSTM hyperparameter optimization by targeting the root mean square error of the overall hydrograph (RMSEall), high-flow (RMSEhigh) and low-flow (RMSElow) dynamics, and water volume deviation (Dv). The MOPSO-LSTM framework was applied to the upstream catchments of the Miyun Reservoir in Beijing, China. At a lead time of 1d, the optimal solution achieved an NSE of 0.920 in the Chaohe River Basin, with a minimum RMSEall of 0.848 m3/s, RMSEhigh of 2.081 m3/s, RMSElow of 0.382 m3/s, and Dv of 0.002%. However, as the lead time increased to 3 and 7 days, the maximum NSE declined to 0.747 and 0.560, respectively, with process-related metrics deteriorating more substantially than water balance-related metrics. The Baihe River Basin performed better, with maximum NSE and KGE values of 0.949 and 0.970 at a lead time of 1d. Clear trade-offs among different evaluation objectives were further identified, particularly the competitive relationship between RMSEhigh and RMSElow, as well as the coupling between RMSElow and Dv. SHAP (Shapley additive explanation) and partial dependence plots (PDPs) were used to quantify and interpret the effects of hyperparameters on model performance, and the results showed that learning rate, number of units, and lookback window served as the most influential hyperparameters. Moreover, optimization preferences resulted in distinct hyperparameter configurations, where Dv-oriented solutions favored smaller learning rates, longer lookback windows, and larger batch sizes than RMSE-oriented solutions. Compared with the Chaohe River Basin, the larger Baihe River Basin favored LSTM configurations with longer lookback windows, more hidden units, higher learning rates, and lower dropout rates, which was associated with the hydrological memory of the catchment. Overall, this study provides a novel multi-objective LSTM optimization framework, improving the understanding of LSTM hyperparameters and offering practical guidance for hydrological prediction and water resource management.

Multi-objective optimization is essential for balancing water-use efficiency and ecological sustainability in urban water regulation, yet its application is hindered by the high computational cost of hydrodynamic simulations. This study proposes a knowledge-driven graph framework that integrates physically informed graph inference with NSGA-III for efficient simulation-based optimization. The urban water system is represented as a directed graph, where hydraulic mechanisms, including mass conservation, hydraulic response functions, and Muskingum routing, are explicitly embedded into local node-update operators. A breadth-first topological propagation strategy enables efficient network-wide flow evolution while preserving physical consistency. Applied to the Beijing urban water network, the proposed simulator reproduces hydrodynamic responses with Nash–Sutcliffe efficiency coefficients exceeding 0.94 and negligible systematic bias. Compared with a conventional hydrodynamic model, it achieves a computational speedup of three to four orders of magnitude while maintaining comparable accuracy. Beyond accuracy and efficiency, the graph-native architecture offers flexibility: basin-specific subgraphs are extracted and simulated without re-modeling or re-calibration, and the impacts of node failures are instantly localized and quantified. Coupling the simulator with NSGA-III enables rapid exploration of Pareto-optimal regulation strategies, reducing boundary outflow by 9.22%, increasing groundwater recharge by 14.99%, and improving water-surface area and wetted river length by approximately 20.89%. By explicitly exploiting the topological and causal structure of hydrodynamic systems rather than relying on empirical surrogate models, the proposed framework provides an efficient and physically interpretable solution for large-scale multi-objective optimization of urban water systems.

Study region: Groundwater overexploitation area in Beijing, China. Study focus: Balancing mechanistic representation, simulation accuracy, and computational efficiency remains a critical challenge in groundwater modeling. This study proposes a novel knowledge-driven framework for groundwater regime simulation. The core is an interpretable knowledge graph (KG) that embeds fundamental hydrogeological laws directly into its reasoning engine. The KG integrates stratigraphic structures, topology, and hydrogeological parameters. Darcy's law and the water balance principle are embedded to drive the reasoning process, while the Harris Hawks Optimization algorithm is used for global parameter calibration. The framework enables automatic flow direction identification, inter-unit flux estimation, and groundwater level (GWL) simulation. New hydrological insights for the region: The GWL simulations demonstrate high accuracy, achieving mean absolute error, root mean square error, and Nash-Sutcliffe efficiency of 0.67 m, 0.86 m, and 0.94 for shallow aquifers, and 1.46 m, 1.79 m, and 0.96 for deep aquifers, respectively. Scenario predictions reveal that, in the future six years, average GWL declines by 0.3 m/ year in dry years and rises by over 1.6 m/year in wet years; while +/- 25% changes in groundwater abstraction result in decline of 0.4 m/year or rise of 1.2 m/year, respectively. By unifying mechanistic knowledge and reasoning rules, this framework features physical consistency, computational efficiency, and structural flexibility, providing a scalable theoretical and methodological foundation for refined groundwater management.

Rainwater harvesting systems (RHS) are extensively executed to manage stormwater control and water shortage issues in cities. However, the influences of rainfall characteristics on the performances of RHS are still not deeply explored. In this research, a methodology framework is developed to explore the influences of rainfall characteristics on stormwater control and water saving performances of RHS, by using daily precipitation data during 1968-2017 at 30 stations across the Beijing region as a testbed. The proposed methodology framework applies a water balance model, which requires daily rainfall and water demands, catchment area, tank size, runoff losses and first flush as input data, to assess water saving efficiency (WSE), reliability (R), and stormwater capture efficiency (SCE) of RHS, and a clustering approach to determine if same rainfall patterns produce similar/ equivalent outputs of RHS. Rainfall characteristics are quantified using 10 rainfall indices, and their influences on the performances of RHS are investigated. Results demonstrate that WSE, R, and SCE of RHS emerge less scattered when clustering becomes more accurate. Equivalent outputs of RHS for combined water demand can be obtained by dividing 30 stations into eight clusters. Furthermore, 6 rainfall indices (mean annual rainfall, standard precipitation index, rainfall concentration degree, ratio of dry and rainy days, consecutive number of dry days per year, average duration of dry days) are found significantly influencing the stormwater control and water saving performances of RHS. Therefore, they are recommended to be carefully considered to quantify rainfall characteristics and support the performance evaluation and design of RHS.

In view of the subjectivity and limitations of scenario setting, as well as the time-consuming nature of computations and difficulty of obtaining global optimal solutions, this paper proposes an ecological water replenishment optimisation model by coupling a hydrodynamic model and an optimisation algorithm. The proposed model aims to obtain the most optimal water replenishment plan to meet specific objectives, by using the hydrodynamic model to obtain relationship curves between various hydrodynamic indicators (e.g. water level, discharge, flow velocity) and different ecological water replenishment flows, and combining these curves with the optimisation algorithm. The model was applied to the Beijing section of the Yongding River. The minimum water replenishment flow at the cross section of Guanting Reservoir is 30 m3/s, to meet the goal of full-channel waterflow connectivity in the Beijing section of Yongding River. A comprehensive optimal ecological water replenishment plan is proposed under the consideration of three objectives in the best state possible, i.e. the maximum guaranteed rate of suitable ecological flow, the highest monthly average ecological water replenishment efficiency and the maximum guaranteed rate of full-channel waterflow connectivity. This study provides novel insights and methodologies for the formulation of ecological water replenishment plans for water-deficient rivers.

By analyzing the spatial and temporal distribution, evolution law and driving factors of land subsidence in Tongzhou area, we can provide scientific basis for regional disaster prediction, prevention and urban planning. This study utilized the IPTA-based SBAS-InSAR method, incorporating multiple data sources to obtain surface deformation results for Tongzhou District from 2003 to 2018, and the distribution characteristics and spatiotemporal evolution patterns of land subsidence were studied in Beijing's Tongzhou District. Additionally, using the MIC method, combined with spatial analysis techniques and statistical methods, the study analyzed the impact of factors such as groundwater levels in different aquifers and the thickness of compressible layers on land subsidence. The results revealed that land subsidence in Beijing's Tongzhou District is unevenly distributed, with major subsidence areas concentrated in BMAC and Taihu. During the study period, surface deformation rates ranged from -147 to 24 mm/yr, showing an initial acceleration followed by a deceleration trend. The third and fourth aquifers were identified as the dominant factors influencing changes in land subsidence, and interactions between different aquifers may also play a role. For varying compressible thicknesses, the second and third layers were the primary contributors to land subsidence. This study provides valuable insights into the mechanisms of land subsidence in Tongzhou District, offering a foundation for further research and informing decision-making for sustainable urban development.

Maintaining machinery wells, key to groundwater sustainability, is vital for managing these precious resources. In keeping with the need for effective groundwater management, this study introduces a screening process leveraging the Density-Based Spatial Clustering of Applications with Noise (DBSCAN) clustering algorithm and incorporating point of interest (POI) and spatial correlation assessment. Specifically, this approach was utilized in the Fangshan District, Beijing, to detect illegal water extraction wells and bolster water resource management. By integrating POI and land use categorization, we performed a water demand overlay analysis, considering public water provision and well distribution. ArcGIS spatial analysis helped us pinpoint POIs with significant water demand. The DBSCAN clustering algorithm was then employed to scrutinize potential problematic wells. The credibility and practicality of this methodology were confirmed through field studies, meeting current governance standards. The study identified 14 potential unregistered wells, of which 5 were confirmed to be unauthorized, primarily located in Shidu and Zhangfang Towns. Field checks confirmed these findings, highlighting the need for improved groundwater management and validating our method's reliability. Based on these findings, we propose additional steps to enhance groundwater extraction management. This research shows how the use of POI data and the DBSCAN algorithm can aid groundwater resource management in Fangshan District and potentially serve as a model for other regions.

Accurate and prompt flood forecasting is essential for effective decision making in flood control to help minimize or prevent flood damage. We propose a new custom deep learning model, IF-CNN-GRU, for multi-step-ahead flood forecasting that incorporates the flood index (IF) to improve the prediction accuracy. The model integrates convolutional neural networks (CNNs) and gated recurrent neural networks (GRUs) to analyze the spatiotemporal characteristics of hydrological data, while using a custom recursive neural network that adjusts the neural unit output at each moment based on the flood index. The IF-CNN-GRU model was applied to forecast floods with a lead time of 1–5 d at the Baihe hydrological station in the middle reaches of the Han River, China, accompanied by an in-depth investigation of model uncertainty. The results showed that incorporating the flood index IF improved the forecast precision by up to 20%. The analysis of uncertainty revealed that the contributions of modeling factors, such as the datasets, model structures, and their interactions, varied across the forecast periods. The interaction factors contributed 17–36% of the uncertainty, while the contribution of the datasets increased with the forecast period (32–53%) and that of the model structure decreased (32–28%). The experiment also demonstrated that data samples play a critical role in improving the flood forecasting accuracy, offering actionable insights to reduce the predictive uncertainty and providing a scientific basis for flood early warning systems and water resource management.

As coal seams are mined at greater depths, the threat of high water pressure from the confined aquifer in the floor that mining operations face has become increasingly prominent. Taking the Madaotou mine field in the Datong Coalfield as the research object, in the context of mining under pressure, for the main coal seams in the mining area, first of all, an improved evaluation method for the vulnerability of floor water inrush is adopted for hazard prediction. Secondly, numerical simulation is used to conduct a simulation analysis on the fault zones in high-risk areas. By using the fuzzy C-means clustering method (FCCM) to improve the classification method for the normalized indicators in the original variable-weight vulnerability evaluation, the risk zoning for water inrush from the coal seam floor is determined. Then, through the numerical simulation method, a simulation analysis is carried out on high-risk areas to simulate the disturbance changes of different mining methods on the fault zones so as to put forward reasonable mining methods. The results show that the classification of the variable-weight intervals of water inrush from the coal seam floor is more suitable to be classified by using fuzzy clustering, thus improving the prediction accuracy. Based on the time effect of the delayed water inrush of faults, different mining methods determine the duration of the disturbance on the fault zones. Therefore, by reducing the disturbance time on the fault zones, the risk of karst water inrush from the floor of the fault zones can be reduced. Through prediction evaluation and simulation analysis, the evaluation of the risk of water inrush in coal mines has been greatly improved, which is of great significance for ensuring the safe and efficient mining of mines.

Study region: Beijing Region Study focus: Limited studies have yet been done on regionalization of multi-performances of rainwater harvesting systems (RHS). In this study, the daily rainfall records from 77 stations are used, and a hydro-economic model is adopted to examine regional water saving, stormwater control and economic performances of RHS by proposing a regionalization approach. New hydrological insights for the region: Higher water saving efficiency (WSE) and reliability (R) of RHS are linked with lower water demand, larger tank size and greater rainfall conditions. Differently, higher stormwater capture efficiency (SCE) is related to higher water demand, larger tank size and lower rainfall. The WSE and R of RHS demonstrate substantial regional differences; their maximized obtainable values closely depend on water demand scenarios and range from 9% to 99%, with smaller values in western mountain area but larger values in northeast suburban and central urban areas. The maximum obtainable values of SCE range from 61% to 100%, with higher values in western mountain area, but lower values in northeast suburban and central urban areas. A 10 m3 tank size can provide the highest benefit-cost ratio of RHS across the Beijing region. The regionalization approach proposed is a useful guidance for the implementation of RHS and sustainable urban water management, and thus it has the potential for wide uses in other regions with high rainfall gradients.

Abstract In view of the subjectivity and limitations of traditional scenario setting in devising ecological water replenishment scheme for water-deficient rivers, as well as problems such as time-consuming computations and the difficulty of obtaining global optimal solutions, this paper proposes an ecological water replenishment optimization model coupling a hydrodynamic model and an optimization algorithm. By leveraging the hydrodynamic model to obtain relationship curves between various hydrodynamic indicators (e.g., water level, discharge, flow velocity) under different ecological water replenishment scenarios, and combining these curves with the optimization algorithm, the model aims to obtain the most optimal water replenishment scheme to meet specific objectives. Using the ecological water replenishment project in the Beijing section of the Yongding River as a case study, the model calculates the minimum thresholds of the water replenishment flow of the Guanting Reservoir are 3 m3/s and 30 m3/s, respectively, when the Mountain Gorge section and the Beijing section of Yongding River are in the state of full-channel waterflow connectivity. On the premise of the maximum guarantee rate of suitable ecological flow in Sanjiadian section, the highest monthly average ecological water replenishment efficiency from Guanting to Sanjiadian section and the maximum guarantee rate of full-channel waterflow connectivity, this study proposes a comprehensive optimal ecological water replenishment scheme to make each goal in a better state as much as possible, which not only provides novel insights and methodologies for the formulation of ecological water replenishment scheme for the Yongding River, but also holds important reference value for the development of ecological water replenishment scheme of other water-deficient rivers.

Evapotranspiration (ET) plays an important role in restoring stormwater retention capacity of bioretention systems and improving microclimate conditions in urban areas. However, there is still a lack of applicable methods to accurately predict ET from bioretention systems. In this study, the applicability of five commonly used models (i.e., ASCE Penman-Monteith, Priestley-Taylor, Penman, Hargreaves, and Blaney-Criddle) in predicting ET from bioretention systems was assessed based on lysimeter measurement data from a field-scale bioretention system located in Beijing. The measured average daily ET from the bioretention system using a lysimeter is 2.24 mm center dot d(-1). ET from the bioretention system is significantly related to solar radiation, vapor pressure deficit, and air temperature (p < 0.001). The ASCE Penman-Monteith, Priestly-Taylor, Penman, and Blaney-Criddle models underestimate average daily ET by 23.31%, 18.50%, 18.93%, and 29.60%, respectively. However, the Hargreaves model overestimates ET by 35.13%. Incorporating various crop coefficients in different plant growth stages can significantly improve the accuracy of these five models. The Penman model is demonstrated to be the most accurate in predicting ET from the bioretention system, followed by the Priestly-Taylor, and ASCE Penman-Monteith models. These three models have relative errors <= 2.68%, RMSE <= 0.66 mm center dot d(-1), R-2 >= 0.92, and NSE >= 0.95. The Hargreaves or Blaney-Criddle models could not accurately predict ET from the bioretention system because solar radiation is not considered in these two models. The results provide references for selecting and implementing models to predict ET from bioretention systems.

改善城市河湖生态环境是提升区域生态环境质量的基础.传统的引调水工程对城市河湖水环境问题抑制成效有限,结合城市河湖自身特点与补水需求,在生态补水的基础上,丰富了生态补水的科学内涵与目标,确定影响城市河湖生态环境的关键指标体系,构建生态补水技术框架,阐述补水过程中涉及到的工程措施、评估指标体系、运行维护管理及优化调配机制.充分考虑生态系统与水文要素变化的响应过程,并对其关键技术问题进行详细阐述,用以辅助城市河湖恢复自身生态系统调节功能,抑制水环境问题,以期为我国城市河湖水生态保护提供参考.

Considering the two key indicators of per capita GDP and per capita energy consumption, this study comprehensively analyzed the environmental Kuznets curve（EKC） of the total water consumption and different industrial water consumptions in China from 1997 to 2020, and the multiple tests, including augmented Dickey-Fuller test, t-test, and Breusch-Godfrey test, were used to optimize the parameters of the EKC model. Then, the supply side and demand side of water resources were further analyzed. Results indicate that EKCs of total water consumption, industrial water consumption, and agricultural water consumption show inverted “N” shape, and the total water consumption, industrial water consumption, and agricultural water consumption exhibit declining trends; EKC of tertiary industrial water consumption exhibits a linear growth trend, and the tertiary industrial water consumption has not reached the declining stage yet; tertiary industrial water consumption is the main growing point of water consumption in future, and reduction of groundwater extraction should be emphasized.

暴雨山洪灾害预警是中小流域山洪灾害防控体系的薄弱环节,也是决定山洪灾害防控成败的关键.论文围绕山洪灾害预警的核心问题,从中国山洪灾害区域差异特征、山洪灾害预警技术方法、山洪灾害概率预警现状3个方面进行了综述.中国山洪灾害分布存在明显的时空差异,因此有必要根据山洪灾害的区域差异发展有针对性的预警方法.以临界雨量为指标的雨量预警是目前中国中小流域暴雨山洪灾害预警的主要技术手段,但常规方法仅给出一个(组)确定的临界雨量阈值,导致预警结果存在突出的不确定性问题.概率预警可以定量评估诸多不确定性,给出山洪灾害概率预警结果,因此具备很好的理论优势与潜在应用价值.论文展望了山洪灾害概率预警未来的研究重点与方向:①充分挖掘暴雨洪水样本信息,开展山洪灾害概率预警基础方法与技术集成研究;②加强非平稳性条件下的临界雨量阈值估算与山洪灾害概率预警研究;③综合考虑预警阈值发生概率及其致灾概率,优化"多级预警、多级响应"技术方法,推进山洪灾害综合预警业务系统建设与应用.

考虑前期影响雨量的径流模拟可量化水资源形成过程,核算降雨水资源量,提升水资源精细化管控水平.以张家坟流域为例,构建新安江单区间和三区间径流模拟模型,并基于前期影响雨量进行频率分析,分析2021年"7·12"场次降水前期影响雨量值,调整模型参数模拟径流过程,实现水资源量核算目的.通过P-III生成张家坟流域前期影响雨量值的频率曲线,建立土壤含水与前期影响雨量关系,分析不同前期影响雨量下径流影响,发现前期影响雨量与径流存在正相关关系.建立张家坟流域新安江单区间和三区间径流模型,根据前期影响雨量值调整模型K值和WM进行径流模拟,通过模拟结果进行张家坟降雨量、蒸发量、场次径流总量、河道汇入组成径流量及土壤入渗量核算.本研究提出基于前期影响雨量调整新安江模型参数模拟径流方法,实现对水资源量的分类、分区及分场次核算,为智慧水务下水资源精细化管控提供技术支持.

Aiming at the inconsistency between the results of precipitation water resources assessment and annual scale water resources assessment, a new evaluation method of precipitation water resource is proposed in this paper by splitting the runoff components such as antecedent precipitation recession and annual scale base flow, in order to realize the refined evaluation of precipitation water resources. The analysis results in the typical basin of Chaobai River show that the method has good adaptability, significantly improve the pertinence and accuracy of the assessment results of precipitation water resources, and fulfills the needs of water resources assessment with multiple temporal and spatial scale coordination.

城市景观缓滞水体水华控制及水力条件改善是当前城市湖泊、河流等景观水体生态健康恢复的关键问题.永定河生态补水措施用于缓解莲石湖8号湖等景观缓滞水体面临的水华易发问题,然而该补水过程对景观水体的水华控制成效尚不明确.基于2019-2021年水质、水文、藻类实测数据,通过构建莲石湖8号湖水动力-水质二维耦合数值模型,结合现场监测数据对耦合数值模型进行率定与验证,探讨了在不同补水条件下8号湖叶绿素a含量和缓流区域的变化特征,定量分析了不同外调补水条件下缓滞区空间分布、叶绿素a浓度与典型景观缓滞水体水文条件的响应关系.结果表明:不同外调补水流量能够有效控制缓滞水体内叶绿素a含量,当外调补水流量为0.23～0.96 m3/s时,莲石湖8号湖表层水体中叶绿素a浓度较低且缓流区域面积较小.该研究结果可为城市景观水体水华控制及外调补水优化方案的制定提供技术支撑.

Urban flooding is a global water disaster resulting from the expansion of urban impervious surfaces and the strengthening of extreme precipitation events, especially in China. Nonetheless, few studies have focused on the spatial distributions of urban flooding characteristics and their variations in the context of climate change. In this study, eight critical metrics (i.e., maximum flooding volume, total overloaded manholes with different flooding volumes or durations, total flooding volume, mean and maximum flooding durations, maximum inundation area, and depth) are adopted to characterize the urban flood events. The impacts of climate change on these metrics are assessed for two periods, the 2030s (2020–2049) and 2070s (2060–2089), and compared with those in the baseline period (1976–2005). The Future Science City Park in Beijing, China, is selected as our study area. The results show that all four flood events are well simulated, with both efficiency coefficients and correlation coefficients being over 0.8. The number of overloaded manholes and the total flooding volume are projected to increase 19.3%–44.8% and 171%–716% under 20‐year rainfall events due to climate change in the two future periods. The spatial distribution of overloaded manholes with different increased flooding volumes is projected to expand to almost the whole area from the region with lowland and limited drainage capacity. Furthermore, the maximum inundation area and depth are projected to increase obviously. This study will be helpful for designing and improving the drainage system, controlling urban flooding, and adapting to climate change.

In this study, a mathematical expression describing the relationship between infiltration rate and soil moisture content was deduced based on the Horton equation in order to study the relationship between infiltration rate and soil moisture content during the rainfall infiltration process. Specific experimental data on infiltration processes were used to verify the validity of the equations and to analyze the main factors affecting the infiltration rate. The results indicate the following: (1) The experimental data demonstrated a high degree of accuracy. The volumetric error of soil moisture increase and cumulative infiltration was 3.5% and the coefficient of determination (R2) was 0.87 in the 42 tests. (2) The equation obtained in this paper can well describe the relationship between infiltration rate and soil moisture content; the R2 of the fitted results was greater than 0.80 in more than 80% of the experiments. (3) The relationship between infiltration rate and soil moisture content is mainly influenced by initial soil moisture content; the higher the initial soil moisture content, the lower the initial infiltration rate, the faster the infiltration rate decreases with soil moisture content, and the lower the “relative stable infiltration rate” in the process of infiltration.