Journal of Hydrology最新文献_第10页

Integrated multi-scale ecohydrogeological monitoring of spatio-temporal dynamics in karst critical zones 喀斯特关键带多尺度生态水文地质时空动态综合监测

IF 6.3 1区地球科学 Q1 ENGINEERING, CIVIL

Journal of Hydrology

Pub Date : 2026-01-30 DOI: 10.1016/j.jhydrol.2026.135027

Nataša Ravbar , Metka Petrič , Mitja Ferlan , Uroš Novak , Janez Kermavnar , Lado Kutnar , Aleksander Marinšek , Daniel Žlindra , Blaž Kogovšek , Erika Kozamernik , Cyril Mayaud , David Štefanič , Sara Skok , Janez Mulec , Stanka Šebela , Urša Vilhar

Contemporary environmental concerns highlight the vulnerability of karst environments to changing hydrometeorological patterns and vegetation disturbance, necessitating a unified, interdisciplinary strategy for comprehensive understanding. This paper critically examines the current state of research. To overcome the identified gaps, it presents an integrated multi-scale ecohydrogeological monitoring approach tailored to karst critical zones (KCZ) and its spatio-temporal variability. Forested karst aquifer in Slovenia is used as a case study to demonstrate and assess the strengths and limitations of the proposed monitoring framework. To decipher flow dynamics and propose customized data collection strategies the approach combines surface and underground sites and employs advanced methods adapted to the challenges of karst environments. The results highlight the benefits and advancements of monitoring and sampling approaches to ensure representativeness in heterogeneous environments. The focus is on the use of enhanced precipitation monitoring systems to expand sampling areas nearly fivefold and improve precipitation and throughfall measurements. Additionally, customized lysimeter techniques for karst soils and microscale adaptations for cave exploration have been developed, addressing the challenges of instrument placement in environments with significant variability. Further opportunities lie in improving instrument protection, integrating sensor networks, combining remote sensing and scaling from plot to aquifer level. However, challenges remain in achieving spatio-temporal representativeness and ensuring the operational reliability of snow monitoring, soil solution sampling and drip flow measurements. Threats include environmental pressures and hydrometeorological conditions, equipment tampering and funding stability. Nevertheless, this comprehensive approach improves monitoring of ecohydrogeological processes in the KCZ, promotes interdisciplinary collaboration and environmental resource management.

当代环境问题突出了喀斯特环境对变化的水文气象模式和植被干扰的脆弱性，需要一个统一的跨学科战略来全面理解。本文批判性地考察了研究的现状。为克服上述空白，提出了一种适合喀斯特临界带及其时空变异的多尺度综合生态水文地质监测方法。斯洛文尼亚喀斯特森林含水层作为一个案例研究，以展示和评估拟议监测框架的优势和局限性。为了破译流动动力学并提出定制的数据收集策略，该方法结合了地表和地下站点，并采用了适应喀斯特环境挑战的先进方法。结果强调了监测和采样方法的好处和进步，以确保在异构环境中的代表性。重点是使用增强型降水监测系统，将采样区域扩大近五倍，并改进降水和穿透测量。此外，针对喀斯特土壤的定制溶蚀仪技术和洞穴勘探的微尺度适应性已经开发出来，解决了仪器在显著变化的环境中放置的挑战。进一步的机会在于改善仪器保护、整合传感器网络、结合遥感和从地块到含水层的缩放。然而，在实现积雪监测、土壤溶液采样和滴流测量的时空代表性和确保操作可靠性方面仍然存在挑战。威胁包括环境压力和水文气象条件、设备篡改和资金稳定。然而，这种综合方法改善了对KCZ生态水文地质过程的监测，促进了跨学科合作和环境资源管理。

{"title":"Integrated multi-scale ecohydrogeological monitoring of spatio-temporal dynamics in karst critical zones","authors":"Nataša Ravbar , Metka Petrič , Mitja Ferlan , Uroš Novak , Janez Kermavnar , Lado Kutnar , Aleksander Marinšek , Daniel Žlindra , Blaž Kogovšek , Erika Kozamernik , Cyril Mayaud , David Štefanič , Sara Skok , Janez Mulec , Stanka Šebela , Urša Vilhar","doi":"10.1016/j.jhydrol.2026.135027","DOIUrl":"10.1016/j.jhydrol.2026.135027","url":null,"abstract":"<div><div>Contemporary environmental concerns highlight the vulnerability of karst environments to changing hydrometeorological patterns and vegetation disturbance, necessitating a unified, interdisciplinary strategy for comprehensive understanding. This paper critically examines the current state of research. To overcome the identified gaps, it presents an integrated multi-scale ecohydrogeological monitoring approach tailored to karst critical zones (KCZ) and its spatio-temporal variability. Forested karst aquifer in Slovenia is used as a case study to demonstrate and assess the strengths and limitations of the proposed monitoring framework. To decipher flow dynamics and propose customized data collection strategies the approach combines surface and underground sites and employs advanced methods adapted to the challenges of karst environments. The results highlight the benefits and advancements of monitoring and sampling approaches to ensure representativeness in heterogeneous environments. The focus is on the use of enhanced precipitation monitoring systems to expand sampling areas nearly fivefold and improve precipitation and throughfall measurements. Additionally, customized lysimeter techniques for karst soils and microscale adaptations for cave exploration have been developed, addressing the challenges of instrument placement in environments with significant variability. Further opportunities lie in improving instrument protection, integrating sensor networks, combining remote sensing and scaling from plot to aquifer level. However, challenges remain in achieving spatio-temporal representativeness and ensuring the operational reliability of snow monitoring, soil solution sampling and drip flow measurements. Threats include environmental pressures and hydrometeorological conditions, equipment tampering and funding stability. Nevertheless, this comprehensive approach improves monitoring of ecohydrogeological processes in the KCZ, promotes interdisciplinary collaboration and environmental resource management.</div></div>","PeriodicalId":362,"journal":{"name":"Journal of Hydrology","volume":"669 ","pages":"Article 135027"},"PeriodicalIF":6.3,"publicationDate":"2026-01-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146089326","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

D2Mamba: A mamba-based method for floodway obstructions segmentation from multispectral satellite imagery D2Mamba：一种基于mamba的多光谱卫星图像沟道障碍物分割方法

IF 6.3 1区地球科学 Q1 ENGINEERING, CIVIL

Journal of Hydrology

Pub Date : 2026-01-30 DOI: 10.1016/j.jhydrol.2026.135069

Shiyang Fu , Zhujun Gu , Yuebin Wang , Liqiang Zhang , Haiyan Gu , Jiasheng Wu , Guanghui Liao

Floodway obstructions are critical anthropogenic factors that exacerbate flood risks, playing a key role in flood control and ecological protection. Current obstruction extraction methods face two main limitations. First, high-resolution visible-light imagery is highly sensitive to lighting and terrain, which restricts its applicability in complex hydrological environments. Second, most methods cannot capture precise obstruction boundaries and fail to meet the fine-grained requirements of floodway management. To address these challenges, we propose D2Mamba, an automated obstruction extraction framework. The method first uses a dual-domain Mamba block (DDMB) to extract key obstruction features. It then employs an adaptive feature fusion module (AFFM) that uses attention mechanisms to fuse encoder features at the same level, enhancing representation capabilities. To balance dual-domain feature learning, we designed a dual-domain loss function (DLoss) to constrain joint optimization of spatial and frequency domain features. To evaluate the effectiveness of the proposed D2Mamba, we constructed a multispectral floodway obstructions dataset (MFOD), including 8301 images and corresponding masks. Comprehensive experiments showed that D2Mamba outperformed 15 state-of-the-art (SOTA) methods on MFOD, with F1, IoU, and OA exceeding other methods by 0.14–14.25%, 3.05–24.47%, and 0.05–14.78%, respectively. Furthermore, the proposed method also demonstrated robust adaptability in large-scale obstruction extraction within the Datengxia area.

泄洪道障碍是加剧洪水风险的重要人为因素，在防洪和生态保护中发挥着关键作用。目前的阻塞提取方法面临两个主要的局限性。首先，高分辨率可见光图像对光照和地形高度敏感，限制了其在复杂水文环境中的适用性。其次，大多数方法不能捕获精确的障碍物边界，不能满足精细的泄洪道管理要求。为了解决这些挑战，我们提出了D2Mamba，一个自动障碍物提取框架。该方法首先使用双域曼巴块（DDMB）提取关键障碍物特征；然后，它采用自适应特征融合模块（AFFM），该模块使用注意机制在同一级别融合编码器特征，增强表示能力。为了平衡双域特征学习，我们设计了一个双域损失函数（DLoss）来约束空间和频域特征的联合优化。为了评估D2Mamba的有效性，我们构建了一个多光谱的洪道障碍物数据集（MFOD），包括8301幅图像和相应的掩模。综合实验表明，D2Mamba在MFOD上优于15种最先进（SOTA）方法，其中F1、IoU和OA分别比其他方法高出0.14-14.25%、3.05-24.47%和0.05-14.78%。此外，该方法对大腾峡地区的大面积障碍物提取也具有较强的适应性。

{"title":"D2Mamba: A mamba-based method for floodway obstructions segmentation from multispectral satellite imagery","authors":"Shiyang Fu , Zhujun Gu , Yuebin Wang , Liqiang Zhang , Haiyan Gu , Jiasheng Wu , Guanghui Liao","doi":"10.1016/j.jhydrol.2026.135069","DOIUrl":"10.1016/j.jhydrol.2026.135069","url":null,"abstract":"<div><div>Floodway obstructions are critical anthropogenic factors that exacerbate flood risks, playing a key role in flood control and ecological protection. Current obstruction extraction methods face two main limitations. First, high-resolution visible-light imagery is highly sensitive to lighting and terrain, which restricts its applicability in complex hydrological environments. Second, most methods cannot capture precise obstruction boundaries and fail to meet the fine-grained requirements of floodway management. To address these challenges, we propose D2Mamba, an automated obstruction extraction framework. The method first uses a dual-domain Mamba block (DDMB) to extract key obstruction features. It then employs an adaptive feature fusion module (AFFM) that uses attention mechanisms to fuse encoder features at the same level, enhancing representation capabilities. To balance dual-domain feature learning, we designed a dual-domain loss function (DLoss) to constrain joint optimization of spatial and frequency domain features. To evaluate the effectiveness of the proposed D2Mamba, we constructed a multispectral floodway obstructions dataset (MFOD), including 8301 images and corresponding masks. Comprehensive experiments showed that D2Mamba outperformed 15 state-of-the-art (SOTA) methods on MFOD, with F1, IoU, and OA exceeding other methods by 0.14–14.25%, 3.05–24.47%, and 0.05–14.78%, respectively. Furthermore, the proposed method also demonstrated robust adaptability in large-scale obstruction extraction within the Datengxia area.</div></div>","PeriodicalId":362,"journal":{"name":"Journal of Hydrology","volume":"669 ","pages":"Article 135069"},"PeriodicalIF":6.3,"publicationDate":"2026-01-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146089327","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

High-precision numerical simulation framework for the integrated modeling of urban “Source-Plant-Network-River” water environment 城市“源-植物-网络-河流”水环境一体化建模的高精度数值模拟框架

IF 6.3 1区地球科学 Q1 ENGINEERING, CIVIL

Journal of Hydrology

Pub Date : 2026-01-29 DOI: 10.1016/j.jhydrol.2026.135036

Guangxue Luan , Jingming Hou , Fuqiang Wang , Tian Wang , Donglai Li

Urban water environments are facing increasingly severe pollution challenges, and rigorous numerical models have become indispensable for mitigating urban water pollution. Building on the “Gridding + GPU acceleration + Dynamic Link Library (DLL)” approach, this study develops an advanced coupled model that integrates (i) two-dimensional (2D) surface-water hydrodynamics and water-quality transport, (ii) 2D non-point-source pollutant (NPSP) build-up and wash-off, and (iii) one-dimensional (1D) pipe-network drainage and pollutant discharge, thereby enabling integrated simulation of the urban “Source-Plant-Network-River” (SPNR) system. The model employs high-resolution structured grids and a spatiotemporal flux scheme for multi-component pollutants in surface runoff, allowing accurate representation of NPSP wash-off and transport driven by coupled hydrological-hydrodynamic processes. DLL-based bidirectional coupling is implemented to dynamically link 2D surface processes and 1D pipe network hydraulics and water quality processes while reducing distortions in parameter transfer across modules. GPU acceleration, together with optimized water-quality flux computations and removal of redundant operations, significantly improves computational efficiency. The model is applied to the main urban area of Changzhi City under three spatially distributed rainfall scenarios. Performance is evaluated against observations of inundation depth, zoned drainage/sewage discharge, and combined sewer overflow (CSO) flow and water quality. The results show that the Nash-Sutcliffe efficiency (NSE) exceeds 0.7 for inundation depths at four flood-prone locations and for flow and pollutant concentrations at three representative drainage-outfall zones and four CSO outfalls. On an RTX 3070 workstation, the optimized model completes a 7.22 h simulation on 8,484,785 uniform structured grids coupled with 32,982 pipe-network nodes in 8.15 h, reducing runtime by 12.6% compared with the pre-optimization model. The proposed modeling framework is robust and efficient, offering strong potential for high-precision integrated simulations of urban water environments from source to receiving waters, as well as for evaluation, forecasting, early warning, and comprehensive water-environment management from a watershed perspective.

城市水环境面临着日益严峻的污染挑战，严谨的数值模拟已成为缓解城市水污染不可缺少的手段。基于“网格+ GPU加速+动态链接库（DLL）”方法，本研究开发了一种先进的耦合模型，该模型集成了(i)二维（2D）地表水流体动力学和水质运输，（ii）二维非点源污染物（NPSP）的积累和冲刷，以及（iii）一维（1D）管网排水和污染物排放，从而实现了城市“源-植物-网络-河流”（SPNR）系统的综合模拟。该模型采用高分辨率结构网格和地表径流中多组分污染物的时空通量方案，可以准确表示由水文-水动力耦合过程驱动的NPSP冲刷和输送。基于dll的双向耦合实现了二维表面过程与一维管网水力和水质过程的动态连接，同时减少了模块间参数传递的失真。GPU加速，加上优化的水质通量计算和去除冗余操作，显著提高了计算效率。将该模型应用于长治市主城区三种降雨空间分布情景。性能是根据淹没深度、分区排水/污水排放、综合下水道溢流（CSO）流量和水质的观察来评估的。结果表明，在4个洪水易发区的淹没深度、3个代表性排水口和4个CSO排水口的流量和污染物浓度方面，Nash-Sutcliffe效率（NSE）超过0.7。在RTX 3070工作站上，优化后的模型在8.15 h内完成了8,484,785个均匀结构网格和32,982个管网节点的7.22 h的仿真，与优化前模型相比，运行时间减少了12.6%。该模型框架稳健高效，为实现城市水环境从源头到受水的高精度综合模拟，以及流域评价、预报、预警和综合水环境管理提供了强大的潜力。

{"title":"High-precision numerical simulation framework for the integrated modeling of urban “Source-Plant-Network-River” water environment","authors":"Guangxue Luan , Jingming Hou , Fuqiang Wang , Tian Wang , Donglai Li","doi":"10.1016/j.jhydrol.2026.135036","DOIUrl":"10.1016/j.jhydrol.2026.135036","url":null,"abstract":"<div><div>Urban water environments are facing increasingly severe pollution challenges, and rigorous numerical models have become indispensable for mitigating urban water pollution. Building on the “Gridding + GPU acceleration + Dynamic Link Library (DLL)” approach, this study develops an advanced coupled model that integrates (i) two-dimensional (2D) surface-water hydrodynamics and water-quality transport, (ii) 2D non-point-source pollutant (NPSP) build-up and wash-off, and (iii) one-dimensional (1D) pipe-network drainage and pollutant discharge, thereby enabling integrated simulation of the urban “Source-Plant-Network-River” (SPNR) system. The model employs high-resolution structured grids and a spatiotemporal flux scheme for multi-component pollutants in surface runoff, allowing accurate representation of NPSP wash-off and transport driven by coupled hydrological-hydrodynamic processes. DLL-based bidirectional coupling is implemented to dynamically link 2D surface processes and 1D pipe network hydraulics and water quality processes while reducing distortions in parameter transfer across modules. GPU acceleration, together with optimized water-quality flux computations and removal of redundant operations, significantly improves computational efficiency. The model is applied to the main urban area of Changzhi City under three spatially distributed rainfall scenarios. Performance is evaluated against observations of inundation depth, zoned drainage/sewage discharge, and combined sewer overflow (CSO) flow and water quality. The results show that the Nash-Sutcliffe efficiency (<em>NSE</em>) exceeds 0.7 for inundation depths at four flood-prone locations and for flow and pollutant concentrations at three representative drainage-outfall zones and four CSO outfalls. On an RTX 3070 workstation, the optimized model completes a 7.22 h simulation on 8,484,785 uniform structured grids coupled with 32,982 pipe-network nodes in 8.15 h, reducing runtime by 12.6% compared with the pre-optimization model. The proposed modeling framework is robust and efficient, offering strong potential for high-precision integrated simulations of urban water environments from source to receiving waters, as well as for evaluation, forecasting, early warning, and comprehensive water-environment management from a watershed perspective.</div></div>","PeriodicalId":362,"journal":{"name":"Journal of Hydrology","volume":"668 ","pages":"Article 135036"},"PeriodicalIF":6.3,"publicationDate":"2026-01-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146071824","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Impacts of spatial-temporal rainfall structures and antecedent wetness on flood variability at the catchment scale 时空降水结构和前期湿度对流域尺度洪水变率的影响

IF 6.3 1区地球科学 Q1 ENGINEERING, CIVIL

Journal of Hydrology

Pub Date : 2026-01-29 DOI: 10.1016/j.jhydrol.2026.135019

Wencong Yang , Changming Li , Hanbo Yang

Rainfall amounts largely determine flood magnitudes, but the influence of spatial–temporal rainfall structures and antecedent wetness remains unclear. This study presents a new approach to derive flood variability using different combinations of “event indicators”, which describe spatial–temporal rainfall dynamics and antecedent wetness at the catchment scale, while holding rainfall intensity fixed. Specifically, a multivariate distribution generates synthetic event indicators conditioned on rainfall intensity for historical high-rainfall events, and a data-driven runoff model predicts flood peaks based on these indicators. Applied to daily flood events in 140 catchments (304–86,475 km²) across the Eastern Monsoon Region of China, the approach uncovers the effects of event indicators on event flood variability and flood quantile uncertainty. Single-peak temporal rainfall and spatial uniform rainfall are prevalent across events. Antecedent wetness has the greatest impact on flood peak variability, with an attributable coefficient of variance (CV) of 0.23, followed by temporal (CV = 0.16) and spatial rainfall structures (CV = 0.05). Spatial-temporal rainfall correlation has minimal effects. Antecedent wetness and spatial rainfall structures exhibit greater impact in drier catchments, while temporal structures have less impact in larger, elongated catchments. For flood quantiles, variability in event indicators results in a mean CV of 0.15 across catchments for 50-year flood estimates, while uncertainty associated with unobserved hydrologic conditions accounts for 79% of total sampling uncertainty. These findings emphasize the need to incorporate spatial–temporal rainfall variability and antecedent wetness into flood risk estimation and process understanding.

降雨量在很大程度上决定了洪水的大小，但时空降雨结构和前期湿度的影响尚不清楚。该研究提出了一种利用“事件指标”的不同组合来推导洪水变率的新方法，这些“事件指标”描述了流域尺度上的时空降雨动态和先前的湿度，同时保持降雨强度固定。具体而言，多变量分布生成以历史高降雨事件的降雨强度为条件的综合事件指标，数据驱动的径流模型根据这些指标预测洪峰。将该方法应用于中国东部季风区140个流域（304-86,475 km2）的日洪水事件，揭示了事件指标对事件洪水变率和洪水分位数不确定性的影响。单峰时间降水和空间均匀降水在各事件中普遍存在。前缘湿度对洪峰变率的影响最大，其归因方差系数（CV）为0.23，其次是时间（CV = 0.16）和空间降雨结构（CV = 0.05）。时空降水相关性影响最小。先前的湿度和空间降雨结构对干旱流域的影响更大，而时间结构对较大的狭长流域的影响较小。对于洪水分位数，事件指标的可变性导致各流域50年洪水估计的平均CV为0.15，而与未观测到的水文条件相关的不确定性占总采样不确定性的79%。这些发现强调了在洪水风险评估和过程理解中纳入降雨时空变率和前期湿度的必要性。

{"title":"Impacts of spatial-temporal rainfall structures and antecedent wetness on flood variability at the catchment scale","authors":"Wencong Yang , Changming Li , Hanbo Yang","doi":"10.1016/j.jhydrol.2026.135019","DOIUrl":"10.1016/j.jhydrol.2026.135019","url":null,"abstract":"<div><div>Rainfall amounts largely determine flood magnitudes, but the influence of spatial–temporal rainfall structures and antecedent wetness remains unclear. This study presents a new approach to derive flood variability using different combinations of “event indicators”, which describe spatial–temporal rainfall dynamics and antecedent wetness at the catchment scale, while holding rainfall intensity fixed. Specifically, a multivariate distribution generates synthetic event indicators conditioned on rainfall intensity for historical high-rainfall events, and a data-driven runoff model predicts flood peaks based on these indicators. Applied to daily flood events in 140 catchments (304–86,475 km<sup>2</sup>) across the Eastern Monsoon Region of China, the approach uncovers the effects of event indicators on event flood variability and flood quantile uncertainty. Single-peak temporal rainfall and spatial uniform rainfall are prevalent across events. Antecedent wetness has the greatest impact on flood peak variability, with an attributable coefficient of variance (CV) of 0.23, followed by temporal (CV = 0.16) and spatial rainfall structures (CV = 0.05). Spatial-temporal rainfall correlation has minimal effects. Antecedent wetness and spatial rainfall structures exhibit greater impact in drier catchments, while temporal structures have less impact in larger, elongated catchments. For flood quantiles, variability in event indicators results in a mean CV of 0.15 across catchments for 50-year flood estimates, while uncertainty associated with unobserved hydrologic conditions accounts for 79% of total sampling uncertainty. These findings emphasize the need to incorporate spatial–temporal rainfall variability and antecedent wetness into flood risk estimation and process understanding.</div></div>","PeriodicalId":362,"journal":{"name":"Journal of Hydrology","volume":"669 ","pages":"Article 135019"},"PeriodicalIF":6.3,"publicationDate":"2026-01-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146071823","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Application of a coupled mechanistic and data-driven model for water level prediction considering the temporal and spatial effects of runoff evolution in cascade hydropower stations 考虑径流演化时空效应的耦合机制与数据驱动模型在梯级水电站水位预测中的应用

IF 6.3 1区地球科学 Q1 ENGINEERING, CIVIL

Journal of Hydrology

Pub Date : 2026-01-29 DOI: 10.1016/j.jhydrol.2026.135057

Yubin Zhang , Xiaoqun Wang , Jijian Lian , Pingping Luo

Accurate water level prediction in cascade hydropower systems remains challenging due to complex runoff evolution and dynamic regulation. This study proposes a coupled model that integrates a hydrodynamic model (HD), a water balance model (WB), and a backpropagation neural network (BP) combining mechanistic and data-driven approaches to address these challenges. The HD model, established between Zhentou dam (ZTB) and Shaping II Hydropower Station (SP) based on the Saint-Venant equations, explicitly accounts for dynamic storage capacity and runoff evolution effects that significantly influence short-term forecasting accuracy. The WB describes water level changes through physical balance laws, while the BP captures nonlinear fluctuation patterns from historical data. To combine these strengths, a particle swarm optimization algorithm is introduced to dynamically weight model outputs, allowing the framework to adaptively correct prediction errors in real time. Application results show that the coupled HD–WB–BP model (New method) markedly improves predictive performance. Compared with the individual WB and BP models, the New method achieves approximately 30% lower Root Mean Square Error (RMSE), with Nash–Sutcliffe efficiency (NSE) consistently above 0.90, a correlation coefficient (r) of 0.96, and a Mean Absolute Percentage Error (MAPE) of 3.8% for both 3-hour and 24-hour horizons. The model also effectively reduces prediction lag and provides more accurate peak estimation under conditions influenced by dynamic storage and runoff evolution. In addition, the new method keeps P95 latency at = 10174.68 ms, safely within the 5-minute cycle. This demonstrates a favorable balance between accuracy and efficiency for near real-time operation. Overall, the proposed approach provides a robust and scalable framework for water level forecasting in cascade hydropower systems, underscoring the advantages of combining physical–mechanistic and data-driven models to enhance reliability and operational decision-making.

由于梯级水电系统的径流演化和动态调节复杂，对其水位的准确预测具有一定的挑战性。本研究提出了一个耦合模型，该模型集成了水动力学模型（HD）、水平衡模型（WB）和反向传播神经网络（BP），结合了机制和数据驱动的方法来解决这些挑战。基于Saint-Venant方程建立了镇头大坝与沙坪二水电站之间的HD模型，该模型明确考虑了动态库容和径流演化效应，对短期预报精度有显著影响。WB通过物理平衡定律描述水位变化，而BP从历史数据中捕获非线性波动模式。为了结合这些优点，引入粒子群优化算法对模型输出进行动态加权，使框架能够实时自适应地纠正预测误差。应用结果表明，新方法的HD-WB-BP耦合模型显著提高了预测性能。与单独的WB和BP模型相比，新方法的均方根误差（RMSE）降低了约30%，Nash-Sutcliffe效率（NSE）始终高于0.90，相关系数(r)为0.96，平均绝对百分比误差（MAPE）为3.8%。该模型还有效地减小了预测滞后，在受动态蓄水量和径流演变影响的条件下提供了更准确的峰值估计。此外，新方法将P95延迟保持在= 10174.68 ms，安全地保持在5分钟的周期内。这证明了近实时操作的准确性和效率之间的良好平衡。总体而言，该方法为梯级水电系统的水位预测提供了一个强大且可扩展的框架，强调了将物理机制模型和数据驱动模型相结合的优势，以提高可靠性和运营决策。

{"title":"Application of a coupled mechanistic and data-driven model for water level prediction considering the temporal and spatial effects of runoff evolution in cascade hydropower stations","authors":"Yubin Zhang , Xiaoqun Wang , Jijian Lian , Pingping Luo","doi":"10.1016/j.jhydrol.2026.135057","DOIUrl":"10.1016/j.jhydrol.2026.135057","url":null,"abstract":"<div><div>Accurate water level prediction in cascade hydropower systems remains challenging due to complex runoff evolution and dynamic regulation. This study proposes a coupled model that integrates a hydrodynamic model (HD), a water balance model (WB), and a backpropagation neural network (BP) combining mechanistic and data-driven approaches to address these challenges. The HD model, established between Zhentou dam (ZTB) and Shaping II Hydropower Station (SP) based on the Saint-Venant equations, explicitly accounts for dynamic storage capacity and runoff evolution effects that significantly influence short-term forecasting accuracy. The WB describes water level changes through physical balance laws, while the BP captures nonlinear fluctuation patterns from historical data. To combine these strengths, a particle swarm optimization algorithm is introduced to dynamically weight model outputs, allowing the framework to adaptively correct prediction errors in real time. Application results show that the coupled HD–WB–BP model (New method) markedly improves predictive performance. Compared with the individual WB and BP models, the New method achieves approximately 30% lower Root Mean Square Error (RMSE), with Nash–Sutcliffe efficiency (NSE) consistently above 0.90, a correlation coefficient (r) of 0.96, and a Mean Absolute Percentage Error (MAPE) of 3.8% for both 3-hour and 24-hour horizons. The model also effectively reduces prediction lag and provides more accurate peak estimation under conditions influenced by dynamic storage and runoff evolution. In addition, the new method keeps P95 latency at = 10174.68 ms, safely within the 5-minute cycle. This demonstrates a favorable balance between accuracy and efficiency for near real-time operation. Overall, the proposed approach provides a robust and scalable framework for water level forecasting in cascade hydropower systems, underscoring the advantages of combining physical–mechanistic and data-driven models to enhance reliability and operational decision-making.</div></div>","PeriodicalId":362,"journal":{"name":"Journal of Hydrology","volume":"668 ","pages":"Article 135057"},"PeriodicalIF":6.3,"publicationDate":"2026-01-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146072664","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Comparative analysis of GAMLSS modeling approaches for nonstationary runoff dynamics in the Yellow River Basin of China 黄河流域非平稳径流动态的GAMLSS模拟方法比较分析

IF 6.3 1区地球科学 Q1 ENGINEERING, CIVIL

Journal of Hydrology

Pub Date : 2026-01-29 DOI: 10.1016/j.jhydrol.2026.135048

Ben Niu , Yi Li , Yurui Fan , Lei Gong , Lei Wang , Taishan Wang

Quantifying the driving effects of climate change and human activities on nonstationary runoff dynamics is essential. However, the systematic assessments of nonstationary characteristics and their multiple driving mechanisms at the basin scale remain insufficient. This study compared two Generalized Additive Model for Location, Scale, and Shape (GAMLSS) modeling approaches–Continuous-series modeling (Mode1) and monthly-segmented modeling (Mode2)–to analyze the hydrological nonstationarity characteristics of the Yellow River Basin in China and elucidate the driving mechanisms of runoff processes by incorporating covariates of time, circulation, climate, and human factors. The results indicated that: (1) Taking time as a covariate, Mode2 significantly enhanced model robustness by isolating seasonal dynamics. Mode2 raised the average correlation of the location parameter (μ) with the monthly runoff series to R = 0.82 (range 0.70–0.90), a 30–50% increase over Mode1. Furthermore, the nonstationary standardized runoff index (NSRI) aligned more accurately with actual hydrological fluctuations. (2) Taking circulation indices as covariates, circulation indices (AMO, PDO and NINO3) predominantly governed large-scale hydrological trends, with PDO exerting a more pronounced regulatory effect on extreme events in the downstream region under Mode2. (3) The climate-human composite-driven model had the best fitting effect on runoff, particularly at lower timescales (1-, 3-, and 6-month scales) in the middle and lower reaches, where the interactions among total precipitation (TP), snowmelt (SMT), and soil water storage capacity (SWC) explained over 80% of runoff variations (This model yielded the lowest AICc—15% lower than the single climate model—and the highest explanatory power, with R² = 0.85 at 1-month, and 0.70 at 3-month scales). This study suggests that Mode2, with its precise characterization of seasonal differentiation and human dynamics, is more suitable for refined water resource management and extreme drought-flood prediction, whereas Mode1 remains efficient for analyzing interdecadal circulation effects. By addressing three key challenges—capturing monthly runoff nonstationarity, integrating multi-factor drivers, and validating runoff simulations—this study greatly improves runoff modeling and drought detection accuracy, laying a scientific foundation for adaptive management under the combined pressures of climate change and human activities.

量化气候变化和人类活动对非平稳径流动态的驱动效应是必要的。然而，对流域尺度上的非平稳特征及其多重驱动机制的系统评价仍然不足。本文通过对连续序列模型（model1）和月分割模型（model2）两种广义加性模型（GAMLSS）建模方法的比较，分析了黄河流域水文非平稳性特征，并结合时间、环流、气候和人为因素等协变量阐明了径流过程的驱动机制。结果表明：(1)以时间为协变量，模型2通过隔离季节动态显著增强了模型的稳健性。模型2将位置参数（μ）与月径流序列的平均相关性提高到R = 0.82（区间0.70 ~ 0.90），比模型1提高了30 ~ 50%。此外，非平稳标准化径流指数（NSRI）与实际水文波动更为吻合。(2)以环流指数为协变量，环流指数（AMO、PDO和NINO3）主导大尺度水文趋势，其中在模式2下，PDO对下游地区极端事件的调节作用更为明显。(3)气候-人类复合驱动模式对径流的拟合效果最好，特别是在较低的时间尺度（1、3、6个月尺度）中下游地区，总降水量、融雪量和土壤蓄水能力的相互作用解释了80%以上的径流变化（该模式的aicc最低，比单一气候模式低15%，解释能力最高，1个月和3个月的R2分别为0.85和0.70）。研究表明，模型2更适合于精细水资源管理和极端旱涝预测，而模型1更适合于分析年代际环流效应。通过解决月度径流非平稳性捕获、多因素驱动整合和径流模拟验证三个关键挑战，本研究极大地提高了径流建模和干旱探测的精度，为气候变化和人类活动联合压力下的适应性管理奠定了科学基础。

{"title":"Comparative analysis of GAMLSS modeling approaches for nonstationary runoff dynamics in the Yellow River Basin of China","authors":"Ben Niu , Yi Li , Yurui Fan , Lei Gong , Lei Wang , Taishan Wang","doi":"10.1016/j.jhydrol.2026.135048","DOIUrl":"10.1016/j.jhydrol.2026.135048","url":null,"abstract":"<div><div>Quantifying the driving effects of climate change and human activities on nonstationary runoff dynamics is essential. However, the systematic assessments of nonstationary characteristics and their multiple driving mechanisms at the basin scale remain insufficient. This study compared two Generalized Additive Model for Location, Scale, and Shape (GAMLSS) modeling approaches–Continuous-series modeling (Mode1) and monthly-segmented modeling (Mode2)–to analyze the hydrological nonstationarity characteristics of the Yellow River Basin <em>in China</em> and elucidate the driving mechanisms of runoff processes by incorporating covariates of time, circulation, climate, and human factors. The results indicated that: (1) Taking time as a covariate, Mode2 significantly enhanced model robustness by isolating seasonal dynamics. Mode2 raised the average correlation of the location parameter (μ) with the monthly runoff series to R = 0.82 (range 0.70–0.90), a 30–50% increase over Mode1. Furthermore, the nonstationary standardized runoff index (NSRI) aligned more accurately with actual hydrological fluctuations. (2) Taking circulation indices as covariates, circulation indices (AMO, PDO and NINO3) predominantly governed large-scale hydrological trends, with PDO exerting a more pronounced regulatory effect on extreme events in the downstream region under Mode2. (3) The climate-human composite-driven model had the best fitting effect on runoff, particularly at lower timescales (1-, 3-, and 6-month scales) in the middle and lower reaches, where the interactions among total precipitation (TP), snowmelt (SMT), and soil water storage capacity (SWC) explained over 80% of runoff variations (This model yielded the lowest AICc—15% lower than the single climate model—and the highest explanatory power, with R<sup>2</sup> = 0.85 at 1-month, and 0.70 at 3-month scales). This study suggests that Mode2, with its precise characterization of seasonal differentiation and human dynamics, is more suitable for refined water resource management and extreme drought-flood prediction, whereas Mode1 remains efficient for analyzing interdecadal circulation effects. By addressing three key challenges—capturing monthly runoff nonstationarity, integrating multi-factor drivers, and validating runoff simulations—this study greatly improves runoff modeling and drought detection accuracy, laying a scientific foundation for adaptive management under the combined pressures of climate change and human activities.</div></div>","PeriodicalId":362,"journal":{"name":"Journal of Hydrology","volume":"669 ","pages":"Article 135048"},"PeriodicalIF":6.3,"publicationDate":"2026-01-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146071822","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Groundwater responses to decadal rainfall variability in semi-arid South Africa 半干旱南非地下水对年代际降水变化的响应

IF 6.3 1区地球科学 Q1 ENGINEERING, CIVIL

Journal of Hydrology

Pub Date : 2026-01-29 DOI: 10.1016/j.jhydrol.2026.135022

Elisa Bjerre , Søren Jessen , Raphael Schneider , Karen G. Villholth , Matthys A. Dippenaar , Trine Enemark , Rena Meyer , Jason Hallowes , Torben O. Sonnenborg , Thokozani Kanyerere , Karsten H. Jensen

This study examines groundwater responses to climate trends and variability in semi-arid South Africa (SA). Groundwater is a vital water resource in the semi-arid regions of Africa, which are characterized by low, erratic rainfall and intermittent streamflow. In these regions, water scarcity is intensifying due to rising irrigation demands, urbanization, and climate change. Semi-arid SA experiences significant rainfall variability at inter-annual and decadal timescales. However, groundwater responses to SA’s climate remains underexplored, partly due to limited observational data. This study analyses long-term trends and variability in rainfall, temperature, and groundwater levels in the Hout/Sand River catchment from 1940 to 2022 using climate and groundwater indices. Groundwater response to rainfall variability is assessed through correlations between the Standardized Precipitation Index (SPI) and the Standardized Groundwater Index (SGI) across annual to decadal timescales. Results indicate no significant trends in total annual rainfall, but rainfall patterns intensified manifested as higher daily rainfall intensity, longer dry periods, and shorter wet periods. The highest SGI-SPI correlation occurs at 7- and 8-year accumulation periods (r = 0.80), which indicates a dependence of groundwater levels on antecedent rainfall consistent with the region’s decadal climate variability of dry and wet epochs. The study adds to our conceptual understanding of groundwater responses to large-scale climatic patterns, which is essential for assessing future water availability under climate change.

本研究考察了半干旱南非（SA）地下水对气候趋势和变率的响应。地下水是非洲半干旱地区的重要水资源，这些地区的特点是降雨量少、不稳定和水流断断续续。在这些地区，由于灌溉需求的增加、城市化和气候变化，水资源短缺正在加剧。半干旱南亚在年际和年代际时间尺度上具有显著的降水变率。然而，地下水对南亚气候的响应仍未得到充分研究，部分原因是观测数据有限。本研究利用气候和地下水指数分析了1940年至2022年Hout/Sand河流域降雨、温度和地下水水位的长期趋势和变化。通过标准化降水指数（SPI）和标准化地下水指数（SGI）在年至年代际时间尺度上的相关性来评估地下水对降雨变率的响应。结果表明，年降水量变化趋势不明显，但降水模式的增强表现为日降水强度增大、干期延长、湿期缩短。SGI-SPI相关性在7年和8年积累期最高（r = 0.80），表明地下水水位对前期降水的依赖与该地区干湿期的年代际气候变率一致。这项研究增加了我们对地下水对大规模气候模式响应的概念理解，这对于评估气候变化下未来的水资源供应至关重要。

{"title":"Groundwater responses to decadal rainfall variability in semi-arid South Africa","authors":"Elisa Bjerre , Søren Jessen , Raphael Schneider , Karen G. Villholth , Matthys A. Dippenaar , Trine Enemark , Rena Meyer , Jason Hallowes , Torben O. Sonnenborg , Thokozani Kanyerere , Karsten H. Jensen","doi":"10.1016/j.jhydrol.2026.135022","DOIUrl":"10.1016/j.jhydrol.2026.135022","url":null,"abstract":"<div><div>This study examines groundwater responses to climate trends and variability in semi-arid South Africa (SA). Groundwater is a vital water resource in the semi-arid regions of Africa, which are characterized by low, erratic rainfall and intermittent streamflow. In these regions, water scarcity is intensifying due to rising irrigation demands, urbanization, and climate change. Semi-arid SA experiences significant rainfall variability at inter-annual and decadal timescales. However, groundwater responses to SA’s climate remains underexplored, partly due to limited observational data. This study analyses long-term trends and variability in rainfall, temperature, and groundwater levels in the Hout/Sand River catchment from 1940 to 2022 using climate and groundwater indices. Groundwater response to rainfall variability is assessed through correlations between the Standardized Precipitation Index (SPI) and the Standardized Groundwater Index (SGI) across annual to decadal timescales. Results indicate no significant trends in total annual rainfall, but rainfall patterns intensified manifested as higher daily rainfall intensity, longer dry periods, and shorter wet periods. The highest SGI-SPI correlation occurs at 7- and 8-year accumulation periods (r = 0.80), which indicates a dependence of groundwater levels on antecedent rainfall consistent with the region’s decadal climate variability of dry and wet epochs. The study adds to our conceptual understanding of groundwater responses to large-scale climatic patterns, which is essential for assessing future water availability under climate change.</div></div>","PeriodicalId":362,"journal":{"name":"Journal of Hydrology","volume":"669 ","pages":"Article 135022"},"PeriodicalIF":6.3,"publicationDate":"2026-01-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146072670","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A dynamic multi-objective inversion framework for seepage parameters based on monitoring data: case study of an earth-rockfill dam 基于监测数据的渗流参数动态多目标反演框架——以某土石坝为例

IF 6.3 1区地球科学 Q1 ENGINEERING, CIVIL

Journal of Hydrology

Pub Date : 2026-01-29 DOI: 10.1016/j.jhydrol.2026.135064

Dian-Qing Li , Qing Kang , Kang Yan , Jin-Ping He , Yong Liu

Accurate inversion of seepage parameters is fundamental to conducting seepage safety analysis for earth-rockfill dams. Traditional parameter inversion methods often rely on the static analysis of hydraulic head data, ignoring the time-dependent effects during dam operation. This study innovatively presents a dynamic multi-objective inversion framework for seepage parameters, containing surrogate model establishment, multi-objective function updating, and multi-source data decision. Four surrogate models, namely OED-SVM, OED-XGBoost, LHS-SVM, LHS-XGBoost are developed to substitute computationally intensive finite element simulation based on Latin hypercube sampling (LHS), orthogonal experimental design (OED), support vector machine (SVM) and extreme gradient boosting (XGBoost). Multi-objective functions can be updated in real-time by continuously incorporating new static hydraulic head monitoring data and updating its time series features, enabling the dynamic inversion of parameters. The non-dominated sorting genetic algorithm (NSGA-Ⅱ) method is employed to explore the solution space, and optimal solutions are obtained by integrating multi-source data. The effectiveness of the proposed inversion framework is validated through a case study of an earth-rockfill dam. The established OED-SVM and OED-XGBoost models demonstrate high predictive accuracy. The multi-source decision index developed can yield optimal solutions for seepage parameters. The seepage parameters obtained from the model inversion are input into the finite element model for forward modeling. The total error between the hydraulic head obtained from finite element simulation and the observed hydraulic head is within 5%. These findings provide an effective and innovative framework for the inversion of seepage parameters, and offer strong support for the seepage safety analysis of earth-rockfill dams.

准确的渗流参数反演是进行土石坝渗流安全分析的基础。传统的参数反演方法往往依赖于水头数据的静力分析，忽略了大坝运行过程中的时间依赖效应。创新性地提出了一种包含代理模型建立、多目标函数更新和多源数据决策的渗流参数动态多目标反演框架。建立了OED-SVM、OED-XGBoost、LHS-SVM、LHS-XGBoost四个替代模型，替代基于拉丁超立方采样（LHS）、正交实验设计（OED）、支持向量机（SVM）和极限梯度提升（XGBoost）的计算密集型有限元仿真。通过不断整合新的静压水头监测数据并更新其时间序列特征，实现多目标函数的实时更新，实现参数的动态反演。采用非支配排序遗传算法（NSGA-Ⅱ）方法探索解空间，通过整合多源数据得到最优解。通过对某土石坝的实例分析，验证了该反演框架的有效性。建立的OED-SVM和OED-XGBoost模型具有较高的预测精度。所建立的多源决策指标可以得到渗流参数的最优解。将模型反演得到的渗流参数输入到有限元模型中进行正演模拟。有限元模拟得到的水头与实测水头的总误差在5%以内。这些研究成果为渗流参数反演提供了有效和创新的框架，为土石坝渗流安全分析提供了有力支持。

{"title":"A dynamic multi-objective inversion framework for seepage parameters based on monitoring data: case study of an earth-rockfill dam","authors":"Dian-Qing Li , Qing Kang , Kang Yan , Jin-Ping He , Yong Liu","doi":"10.1016/j.jhydrol.2026.135064","DOIUrl":"10.1016/j.jhydrol.2026.135064","url":null,"abstract":"<div><div>Accurate inversion of seepage parameters is fundamental to conducting seepage safety analysis for earth-rockfill dams. Traditional parameter inversion methods often rely on the static analysis of hydraulic head data, ignoring the time-dependent effects during dam operation. This study innovatively presents a dynamic multi-objective inversion framework for seepage parameters, containing surrogate model establishment, multi-objective function updating, and multi-source data decision. Four surrogate models, namely OED-SVM, OED-XGBoost, LHS-SVM, LHS-XGBoost are developed to substitute computationally intensive finite element simulation based on Latin hypercube sampling (LHS), orthogonal experimental design (OED), support vector machine (SVM) and extreme gradient boosting (XGBoost). Multi-objective functions can be updated in real-time by continuously incorporating new static hydraulic head monitoring data and updating its time series features, enabling the dynamic inversion of parameters. The non-dominated sorting genetic algorithm (NSGA-Ⅱ) method is employed to explore the solution space, and optimal solutions are obtained by integrating multi-source data. The effectiveness of the proposed inversion framework is validated through a case study of an earth-rockfill dam. The established OED-SVM and OED-XGBoost models demonstrate high predictive accuracy. The multi-source decision index developed can yield optimal solutions for seepage parameters. The seepage parameters obtained from the model inversion are input into the finite element model for forward modeling. The total error between the hydraulic head obtained from finite element simulation and the observed hydraulic head is within 5%. These findings provide an effective and innovative framework for the inversion of seepage parameters, and offer strong support for the seepage safety analysis of earth-rockfill dams.</div></div>","PeriodicalId":362,"journal":{"name":"Journal of Hydrology","volume":"669 ","pages":"Article 135064"},"PeriodicalIF":6.3,"publicationDate":"2026-01-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146072662","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Enhanced saturated hydraulic conductivity estimation in fine-grained soils: a voting regressor ensemble framework 增强的饱和水力电导率估计在细粒土壤：投票回归集合框架

IF 6.3 1区地球科学 Q1 ENGINEERING, CIVIL

Journal of Hydrology

Pub Date : 2026-01-29 DOI: 10.1016/j.jhydrol.2026.135041

Hong Zhang , You Gao , De’an Sun

Accurate prediction of saturated hydraulic conductivity (k_s) is crucial for hydrology, agriculture, and contaminant modeling. Traditional lab methods are costly and time-consuming, while theoretical models lack adaptability and require complex calibration. To address these limitations, this study compared the predictive performance of six machine learning algorithms for estimating k_s on an augmented dataset using the synthetic minority over-sampling technique: eXtreme Gradient Boosting (XGBoost), Categorical Boosting (CatBoost), Random Forest (RF), Gene Expression Programming (GEP), Support Vector Machine (SVM), Multilayer Perceptron (MLP). The results revealed that XGBoost, RF, and CatBoost suffered from significant overfitting and poor extrapolation. In contrast, SVM, MLP, and GEP demonstrated superior generalization capabilities, indicating their robust ability to capture physical patterns. To further enhance prediction robustness, a hybrid ensemble method based on the Voting Regressor (VR) framework is proposed, which integrates the optimal base models from SVM, MLP, and GEP; this approach significantly improved accuracy in both interpolation and extrapolation scenarios. The proposed VR model accurately simulates k_s across diverse soil types using only basic soil physical parameters. It demonstrates superior performance (R² = 0.992 and RMSLE = 0.049 cm/s) compared to traditional theoretical models, as validated against measured data from the literature. This research provides a reliable computational tool for precision irrigation management in agriculture and groundwater pollution risk assessment.

饱和水力电导率（ks）的准确预测对水文学、农业和污染物建模至关重要。传统的实验室方法成本高，耗时长，而理论模型缺乏适应性，需要复杂的校准。为了解决这些限制，本研究比较了六种机器学习算法的预测性能，用于使用合成少数过采样技术在增强数据集上估计ks：极端梯度增强（XGBoost），分类增强（CatBoost），随机森林（RF），基因表达编程（GEP），支持向量机（SVM），多层感知器（MLP）。结果显示，XGBoost、RF和CatBoost存在明显的过拟合和较差的外推。相比之下，SVM、MLP和GEP表现出优越的泛化能力，表明它们具有捕获物理模式的强大能力。为了进一步提高预测的鲁棒性，提出了一种基于投票回归（VR）框架的混合集成方法，该方法将SVM、MLP和GEP的最优基础模型集成在一起；这种方法在插值和外推场景中都显著提高了精度。所提出的VR模型仅使用基本的土壤物理参数就能准确地模拟不同土壤类型的ks。与传统的理论模型相比，该模型表现出优越的性能（R2 = 0.992, RMSLE = 0.049 cm/s），并通过文献中的实测数据进行了验证。该研究为农业精准灌溉管理和地下水污染风险评估提供了可靠的计算工具。

{"title":"Enhanced saturated hydraulic conductivity estimation in fine-grained soils: a voting regressor ensemble framework","authors":"Hong Zhang , You Gao , De’an Sun","doi":"10.1016/j.jhydrol.2026.135041","DOIUrl":"10.1016/j.jhydrol.2026.135041","url":null,"abstract":"<div><div>Accurate prediction of saturated hydraulic conductivity (<em>k</em><sub>s</sub>) is crucial for hydrology, agriculture, and contaminant modeling. Traditional lab methods are costly and time-consuming, while theoretical models lack adaptability and require complex calibration. To address these limitations, this study compared the predictive performance of six machine learning algorithms for estimating <em>k</em><sub>s</sub> on an augmented dataset using the synthetic minority over-sampling technique: eXtreme Gradient Boosting (XGBoost), Categorical Boosting (CatBoost), Random Forest (RF), Gene Expression Programming (GEP), Support Vector Machine (SVM), Multilayer Perceptron (MLP). The results revealed that XGBoost, RF, and CatBoost suffered from significant overfitting and poor extrapolation. In contrast, SVM, MLP, and GEP demonstrated superior generalization capabilities, indicating their robust ability to capture physical patterns. To further enhance prediction robustness, a hybrid ensemble method based on the Voting Regressor (VR) framework is proposed, which integrates the optimal base models from SVM, MLP, and GEP; this approach significantly improved accuracy in both interpolation and extrapolation scenarios. The proposed VR model accurately simulates <em>k</em><sub>s</sub> across diverse soil types using only basic soil physical parameters. It demonstrates superior performance (<em>R</em><sup>2</sup> = 0.992 and RMSLE = 0.049 cm/s) compared to traditional theoretical models, as validated against measured data from the literature. This research provides a reliable computational tool for precision irrigation management in agriculture and groundwater pollution risk assessment.</div></div>","PeriodicalId":362,"journal":{"name":"Journal of Hydrology","volume":"669 ","pages":"Article 135041"},"PeriodicalIF":6.3,"publicationDate":"2026-01-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146072663","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Evaluating a data space inversion surrogate model for predictive uncertainty quantification in a coupled SWAT + gwflow model 评估SWAT + gwflow耦合模型中预测不确定性量化的数据空间反演代理模型

IF 6.3 1区地球科学 Q1 ENGINEERING, CIVIL

Journal of Hydrology

Pub Date : 2026-01-29 DOI: 10.1016/j.jhydrol.2026.135028

Ehsan Qasemipour , Wesley Kitlasten , Ryan T. Bailey , Markus Pahlow , Thomas A. Cochrane

Incorporating measured data into environmental simulation models through calibration helps to improve predictive performance and reduce uncertainty. Traditionally, this process involves transferring information from observations to parameters, requiring many model runs and parameter updates. The relationships between observations (simulated equivalents) and parameters are often non-linear, which can compromise predictions. Data space inversion (DSI) explores the posterior predictive distribution by building a surrogate model based on the covariance between model outputs that correspond to 1) field measurements, and 2) predictions of interest. DSI avoids updating physical model parameters by conditioning predictions on measurements of system behaviour. DSI is applied to the Soil and Water Assessment Tool (SWAT + ) coupled with a modified groundwater flow module (gwflow) for the Winnebago watershed (U.S.) to evaluate its robustness and efficiency in predicting streamflow and groundwater and to quantify associated uncertainty. The coupling with gwflow enables spatially distributed simulation of groundwater heads using cell-based aquifer properties, allowing increased parameterisation complexity compared to SWAT + alone and providing a rigorous test case for DSI. The DSI-based model predicted streamflow and groundwater head comparably to the physical model, based on acceptable model performance metrics. The DSI-based model enables computationally efficient analysis based on relationships between measurements and predictions, making it a practical tool for uncertainty assessment. Unlike the uncertainty bounds derived from the posterior ensemble of the physically-based model (quantified using iterative ensemble method), the DSI-based model’s uncertainty bounds captured observed groundwater head values during both calibration and prediction periods, highlighting its potential for decision-support modelling.

通过校准将测量数据纳入环境模拟模型有助于提高预测性能并减少不确定性。传统上，这个过程包括将信息从观测值传递到参数，需要多次模型运行和参数更新。观测值（模拟当量）和参数之间的关系通常是非线性的，这可能会影响预测。数据空间反演（Data space inversion， DSI）通过建立一个代理模型来探索后验预测分布，该模型基于对应于1)现场测量和2)感兴趣预测的模型输出之间的协方差。DSI通过调节对系统行为测量的预测来避免更新物理模型参数。DSI应用于美国温尼贝戈流域的水土评估工具（SWAT +）和改进的地下水流量模块（gwflow），以评估其在预测河流流量和地下水方面的鲁棒性和效率，并量化相关的不确定性。与gwflow的耦合可以使用基于单元的含水层属性对地下水水头进行空间分布模拟，与单独使用SWAT +相比，增加了参数化的复杂性，并为DSI提供了严格的测试用例。基于dsi的模型基于可接受的模型性能指标，与物理模型相比，预测了水流和地下水头。基于dsi的模型能够基于测量和预测之间的关系进行计算效率分析，使其成为不确定性评估的实用工具。与基于物理的模型的后检集合（使用迭代集合方法量化）得出的不确定性边界不同，基于dsi的模型的不确定性边界捕获了校准和预测期间观测到的地下水水头值，突出了其决策支持建模的潜力。

{"title":"Evaluating a data space inversion surrogate model for predictive uncertainty quantification in a coupled SWAT + gwflow model","authors":"Ehsan Qasemipour , Wesley Kitlasten , Ryan T. Bailey , Markus Pahlow , Thomas A. Cochrane","doi":"10.1016/j.jhydrol.2026.135028","DOIUrl":"10.1016/j.jhydrol.2026.135028","url":null,"abstract":"<div><div>Incorporating measured data into environmental simulation models through calibration helps to improve predictive performance and reduce uncertainty. Traditionally, this process involves transferring information from observations to parameters, requiring many model runs and parameter updates. The relationships between observations (simulated equivalents) and parameters are often non-linear, which can compromise predictions. Data space inversion (DSI) explores the posterior predictive distribution by building a surrogate model based on the covariance between model outputs that correspond to 1) field measurements, and 2) predictions of interest. DSI avoids updating physical model parameters by conditioning predictions on measurements of system behaviour. DSI is applied to the Soil and Water Assessment Tool (<em>SWAT +</em> ) coupled with a modified groundwater flow module (<em>gwflow</em>) for the Winnebago watershed (U.S.) to evaluate its robustness and efficiency in predicting streamflow and groundwater and to quantify associated uncertainty. The coupling with <em>gwflow</em> enables spatially distributed simulation of groundwater heads using cell-based aquifer properties, allowing increased parameterisation complexity compared to SWAT + alone and providing a rigorous test case for DSI. The DSI-based model predicted streamflow and groundwater head comparably to the physical model, based on acceptable model performance metrics. The DSI-based model enables computationally efficient analysis based on relationships between measurements and predictions, making it a practical tool for uncertainty assessment. Unlike the uncertainty bounds derived from the posterior ensemble of the physically-based model (quantified using iterative ensemble method), the DSI-based model’s uncertainty bounds captured observed groundwater head values during both calibration and prediction periods, highlighting its potential for decision-support modelling.</div></div>","PeriodicalId":362,"journal":{"name":"Journal of Hydrology","volume":"668 ","pages":"Article 135028"},"PeriodicalIF":6.3,"publicationDate":"2026-01-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146072669","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0