Journal of Hydrology最新文献_第4页

Physics-constrained neural network for daily pan evaporation forecasting in hyper-arid climates optimized by the Bat Algorithm 基于Bat算法优化的超干旱气候条件下蒸发皿蒸发量日预报的物理约束神经网络

IF 6.3 1区地球科学 Q1 ENGINEERING, CIVIL

Journal of Hydrology

Pub Date : 2026-04-01 Epub Date: 2026-01-12 DOI: 10.1016/j.jhydrol.2026.134936

Abdullah A. Alsumaiei

With soaring evaporation rates and shrinking freshwater resources, hyper-arid regions require accurate instruments to measure atmospheric water loss, making pan evaporation forecasting a crucial aspect of contemporary water resources management. This paper proposes a hybrid framework that integrates the Physics-Constrained Neural Network (PCNN) and the Bat Algorithm (BA) to predict daily pan evaporation in Kuwait. The proposed PCNN incorporates physical constraints into the loss function, including vapor pressure deficit, net radiation, and aerodynamic resistance based on surface energy balance theory, to ensure both predictive accuracy and physical plausibility, unlike traditional machine learning models. Daily meteorological data and Class A pan evaporation data from two different stations, Kuwait International Airport (KIA) and Abdaly, are used to train and test the model. The obtained results demonstrate a high accuracy and good generalizability with RMSE of 0.904 mm/day, 1.186 mm/day, and R² of 0.953 and 0.884 at KIA and Abdaly, respectively. The model’s consistency with thermodynamic principles is also confirmed by a new metric called physics residual RMSE (PRMSE). Tests of robustness in the presence of synthetic noise show that the model is insensitive to uncertainty in its inputs. The added value of the PCNN–BA framework is demonstrated through systematic comparison with established data-driven models, showing that the proposed approach achieves competitive predictive accuracy while explicitly enforcing physical consistency. The resulting framework is computationally efficient and scalable, making it suitable for hyper-arid environments and directly applicable to desert agriculture, irrigation scheduling, and water resources management under data-limited and water-scarce conditions.

随着蒸发率的飙升和淡水资源的萎缩，极度干旱地区需要精确的仪器来测量大气水分流失，这使得蒸发皿蒸发预测成为当代水资源管理的一个重要方面。本文提出了一个混合框架，集成了物理约束神经网络（PCNN）和蝙蝠算法（BA）来预测科威特的每日蒸发皿蒸发量。与传统的机器学习模型不同，PCNN将物理约束纳入损失函数，包括蒸汽压亏缺、净辐射和基于表面能量平衡理论的空气动力阻力，以确保预测精度和物理合理性。使用科威特国际机场（KIA）和Abdaly两个不同站点的每日气象数据和A类蒸发皿蒸发数据来训练和测试模型。在KIA和Abdaly的RMSE分别为0.904 mm/day和1.186 mm/day， R2分别为0.953和0.884，具有较高的准确度和较好的推广性。该模型与热力学原理的一致性也被一个称为物理残差RMSE （PRMSE）的新度量所证实。在合成噪声存在下的鲁棒性测试表明，该模型对输入的不确定性不敏感。通过与已建立的数据驱动模型的系统比较，证明了PCNN-BA框架的附加价值，表明所提出的方法在明确强制物理一致性的同时实现了竞争性的预测准确性。由此产生的框架具有计算效率和可扩展性，使其适用于极度干旱环境，并直接适用于数据有限和缺水条件下的沙漠农业、灌溉调度和水资源管理。

{"title":"Physics-constrained neural network for daily pan evaporation forecasting in hyper-arid climates optimized by the Bat Algorithm","authors":"Abdullah A. Alsumaiei","doi":"10.1016/j.jhydrol.2026.134936","DOIUrl":"10.1016/j.jhydrol.2026.134936","url":null,"abstract":"<div><div>With soaring evaporation rates and shrinking freshwater resources, hyper-arid regions require accurate instruments to measure atmospheric water loss, making pan evaporation forecasting a crucial aspect of contemporary water resources management. This paper proposes a hybrid framework that integrates the Physics-Constrained Neural Network (PCNN) and the Bat Algorithm (BA) to predict daily pan evaporation in Kuwait. The proposed PCNN incorporates physical constraints into the loss function, including vapor pressure deficit, net radiation, and aerodynamic resistance based on surface energy balance theory, to ensure both predictive accuracy and physical plausibility, unlike traditional machine learning models. Daily meteorological data and Class A pan evaporation data from two different stations, Kuwait International Airport (KIA) and Abdaly, are used to train and test the model. The obtained results demonstrate a high accuracy and good generalizability with RMSE of 0.904 mm/day, 1.186 mm/day, and <em>R</em><sup>2</sup> of 0.953 and 0.884 at KIA and Abdaly, respectively. The model’s consistency with thermodynamic principles is also confirmed by a new metric called physics residual RMSE (PRMSE). Tests of robustness in the presence of synthetic noise show that the model is insensitive to uncertainty in its inputs. The added value of the PCNN–BA framework is demonstrated through systematic comparison with established data-driven models, showing that the proposed approach achieves competitive predictive accuracy while explicitly enforcing physical consistency. The resulting framework is computationally efficient and scalable, making it suitable for hyper-arid environments and directly applicable to desert agriculture, irrigation scheduling, and water resources management under data-limited and water-scarce conditions.</div></div>","PeriodicalId":362,"journal":{"name":"Journal of Hydrology","volume":"668 ","pages":"Article 134936"},"PeriodicalIF":6.3,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145957200","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

Semi-analytical modeling of transient flow to a partially penetrating variable-discharge well in a complex aquifer-aquitard system 复杂含水层-含水层系统部分穿透变流量井瞬态流动半解析建模

IF 6.3 1区地球科学 Q1 ENGINEERING, CIVIL

Journal of Hydrology

Pub Date : 2026-04-01 Epub Date: 2026-01-13 DOI: 10.1016/j.jhydrol.2026.134967

Yabing Li , Zhifang Zhou , Ning Zhang

Accurate prediction of transient flow and vertical leakage in multilayer aquifer systems is critical for sustainable groundwater management and contaminant risk assessment. Conventional groundwater models often simplify subsurface conditions by assuming homogeneous aquitards, isotropic aquifers, constant pumping rates, and idealized boundary conditions, limiting their applicability in realistic field settings. This study develops semi-analytical solutions for three-dimensional transient flow toward a partially penetrating well under variable discharge in a complex aquifer-aquitard system. The model incorporates aquifer anisotropy and vertical heterogeneity in aquitard hydraulic conductivity (K) and specific storage (S_s). It considers representative upper boundary conditions: constant head, water table with delayed drainage, and no-flux. Solutions are derived in the Laplace-Hankel domain and numerically inverted to quantify drawdown and leakage response. Results show that stronger aquitard S_s decay enhances early- and mid-time drawdowns, while greater aquitard K decay limits vertical leakage and decreases aquitard and upper aquifer drawdowns. Compared with a homogeneous aquitard K, increasing dimensionless decay exponent of K from 1.5 to 2.5 reduces stable-stage leakage rates by 45% to 86%. Aquifer anisotropy intensifies vertical hydraulic gradient contrasts and promotes drawdown accumulation. Variable pumping induces nonlinear leakage responses and transient flow reversals. Boundary conditions significantly influence stable-stage leakage rates: compared to the constant-head boundary, the leakage rate at the aquifer-aquitard interface is 3% lower under the water table boundary and 10% lower under the no-flux boundary. The model offers a robust tool for evaluating leakage-driven risks to groundwater quality in complex, vertically heterogeneous hydrogeologic systems such as glacial and alluvial basins.

多层含水层系统瞬态流动和垂直泄漏的准确预测对地下水可持续管理和污染物风险评估至关重要。传统的地下水模型通常通过假设均匀的含水层、各向同性含水层、恒定的抽水速率和理想化的边界条件来简化地下条件，限制了它们在实际现场环境中的适用性。本文研究了复杂含水层-含水层系统中变流量下部分穿透井三维瞬态流动的半解析解。该模型考虑了含水层的各向异性和垂向非均质性，包括含水层导水率(K)和比库容（Ss）。考虑了具有代表性的上边界条件：恒定水头、延迟排水的地下水位和无通量。在Laplace-Hankel域中推导了解，并对解进行了数值反演，以量化下降和泄漏响应。结果表明，较强的含水层Ss衰减增强了早期和中期的降水，而较大的含水层K衰减限制了垂向渗漏，降低了含水层和上部含水层的降水。与均匀出水K相比，将K的无量纲衰减指数从1.5提高到2.5，可使稳定级泄漏率降低45%至86%。含水层各向异性加剧了垂直水力梯度对比，促进了压降积累。变量泵送引起非线性泄漏响应和瞬态流动逆转。边界条件对稳定阶段渗漏率影响显著：与等水头边界相比，地下水位边界下含水层-含水层界面渗漏率降低3%，无通量边界下渗漏率降低10%。该模型为在复杂的、垂直非均质水文地质系统（如冰川和冲积盆地）中评估泄漏驱动的地下水质量风险提供了一个强大的工具。

{"title":"Semi-analytical modeling of transient flow to a partially penetrating variable-discharge well in a complex aquifer-aquitard system","authors":"Yabing Li , Zhifang Zhou , Ning Zhang","doi":"10.1016/j.jhydrol.2026.134967","DOIUrl":"10.1016/j.jhydrol.2026.134967","url":null,"abstract":"<div><div>Accurate prediction of transient flow and vertical leakage in multilayer aquifer systems is critical for sustainable groundwater management and contaminant risk assessment. Conventional groundwater models often simplify subsurface conditions by assuming homogeneous aquitards, isotropic aquifers, constant pumping rates, and idealized boundary conditions, limiting their applicability in realistic field settings. This study develops semi-analytical solutions for three-dimensional transient flow toward a partially penetrating well under variable discharge in a complex aquifer-aquitard system. The model incorporates aquifer anisotropy and vertical heterogeneity in aquitard hydraulic conductivity (<em>K</em>) and specific storage (<em>S<sub>s</sub></em>). It considers representative upper boundary conditions: constant head, water table with delayed drainage, and no-flux. Solutions are derived in the Laplace-Hankel domain and numerically inverted to quantify drawdown and leakage response. Results show that stronger aquitard <em>S<sub>s</sub></em> decay enhances early- and mid-time drawdowns, while greater aquitard <em>K</em> decay limits vertical leakage and decreases aquitard and upper aquifer drawdowns. Compared with a homogeneous aquitard <em>K</em>, increasing dimensionless decay exponent of <em>K</em> from 1.5 to 2.5 reduces stable-stage leakage rates by 45% to 86%. Aquifer anisotropy intensifies vertical hydraulic gradient contrasts and promotes drawdown accumulation. Variable pumping induces nonlinear leakage responses and transient flow reversals. Boundary conditions significantly influence stable-stage leakage rates: compared to the constant-head boundary, the leakage rate at the aquifer-aquitard interface is 3% lower under the water table boundary and 10% lower under the no-flux boundary. The model offers a robust tool for evaluating leakage-driven risks to groundwater quality in complex, vertically heterogeneous hydrogeologic systems such as glacial and alluvial basins.</div></div>","PeriodicalId":362,"journal":{"name":"Journal of Hydrology","volume":"668 ","pages":"Article 134967"},"PeriodicalIF":6.3,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145962385","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 Discrete fractal set (DFS) method for high–accuracy reconstruction and nonlinear flow simulation in rough rock fractures 基于离散分形集（DFS）的粗糙岩体裂隙高精度重建与非线性流动模拟

IF 6.3 1区地球科学 Q1 ENGINEERING, CIVIL

Journal of Hydrology

Pub Date : 2026-04-01 Epub Date: 2026-01-22 DOI: 10.1016/j.jhydrol.2026.135001

Jinjie Liu, Long Xu, Fusheng Zha, Shan Wu, Qiao Wang, Yuan Zhang

Understanding fluid flow through rough rock fractures is essential in numerous geoscientific and engineering applications. Surface roughness introduces enhanced viscous dissipation and inertial effects, thereby amplifying nonlinear flow behavior. Accurate reconstruction of rough fracture geometries, followed by their integration into nonlinear flow models, is crucial for capturing these effects with higher accuracy. This study presents a Discrete Fractal Set (DFS) method for reconstructing rough fracture surfaces and evaluating their influence on nonlinear flow. The approach segments rough profiles into Basic Rough Cells (BRCs), which are then grouped into ordered segments based on a peak ratio criterion. The Weierstrass–Mandelbrot (W–M) function is subsequently applied to each segment for localized fractal refinement. Sensitivity analysis reveals that a peak ratio threshold of 1.0 achieves an optimal balance between reconstruction accuracy and robustness. The DFS method, validated against standard JRC profiles and natural fracture surfaces, outperforms the conventional W–M approach in reconstruction quality, achieving MSE values consistently below 0.11 compared with values often exceeding 0.7 for the W–M equation. To characterize flow behavior, the generated aperture and hydraulic aperture fields are incorporated into the Forchheimer equation, yielding the DFS–Forchheimer equation. Comparative validation against both experimental and numerical results demonstrates that the proposed model improves outlet flow rate prediction accuracy by approximately 4.15% and reduces prediction variability, confirming its enhanced reliability in nonlinear flow simulation.

在许多地球科学和工程应用中，了解流体在粗糙岩石裂缝中的流动是必不可少的。表面粗糙度引入了增强的粘性耗散和惯性效应，从而放大了非线性流动行为。精确重建粗裂缝几何形状，然后将其整合到非线性流动模型中，对于更高精度地捕获这些影响至关重要。提出了一种离散分形集（DFS）方法，用于粗糙断裂面的重建，并评估其对非线性流动的影响。该方法将粗轮廓分割成基本粗细胞（BRCs），然后根据峰值比标准将其分组为有序段。随后将weerstrass - mandelbrot （W-M）函数应用于每个片段进行局部分形细化。灵敏度分析表明，峰值比阈值为1.0可以在重建精度和鲁棒性之间达到最佳平衡。DFS方法在标准JRC剖面和天然裂缝表面上进行了验证，在重建质量方面优于传统的W-M方法，其MSE值始终低于0.11，而W-M方程的MSE值通常超过0.7。为了描述流动特性，将生成的孔径场和水力孔径场合并到Forchheimer方程中，得到DFS-Forchheimer方程。与实验和数值结果的对比验证表明，该模型将出口流量预测精度提高了约4.15%，降低了预测变异性，验证了该模型在非线性流动模拟中的可靠性。

{"title":"A Discrete fractal set (DFS) method for high–accuracy reconstruction and nonlinear flow simulation in rough rock fractures","authors":"Jinjie Liu, Long Xu, Fusheng Zha, Shan Wu, Qiao Wang, Yuan Zhang","doi":"10.1016/j.jhydrol.2026.135001","DOIUrl":"10.1016/j.jhydrol.2026.135001","url":null,"abstract":"<div><div>Understanding fluid flow through rough rock fractures is essential in numerous geoscientific and engineering applications. Surface roughness introduces enhanced viscous dissipation and inertial effects, thereby amplifying nonlinear flow behavior. Accurate reconstruction of rough fracture geometries, followed by their integration into nonlinear flow models, is crucial for capturing these effects with higher accuracy. This study presents a Discrete Fractal Set (DFS) method for reconstructing rough fracture surfaces and evaluating their influence on nonlinear flow. The approach segments rough profiles into Basic Rough Cells (BRCs), which are then grouped into ordered segments based on a peak ratio criterion. The Weierstrass–Mandelbrot (W–M) function is subsequently applied to each segment for localized fractal refinement. Sensitivity analysis reveals that a peak ratio threshold of 1.0 achieves an optimal balance between reconstruction accuracy and robustness. The DFS method, validated against standard JRC profiles and natural fracture surfaces, outperforms the conventional W–M approach in reconstruction quality, achieving MSE values consistently below 0.11 compared with values often exceeding 0.7 for the W–M equation. To characterize flow behavior, the generated aperture and hydraulic aperture fields are incorporated into the Forchheimer equation, yielding the DFS–Forchheimer equation. Comparative validation against both experimental and numerical results demonstrates that the proposed model improves outlet flow rate prediction accuracy by approximately 4.15% and reduces prediction variability, confirming its enhanced reliability in nonlinear flow simulation.</div></div>","PeriodicalId":362,"journal":{"name":"Journal of Hydrology","volume":"668 ","pages":"Article 135001"},"PeriodicalIF":6.3,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146025917","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

Differences in organic matter sources between suspended particulates and sediments in a lake during the frozen period 冰冻期湖泊中悬浮微粒和沉积物有机质来源的差异

IF 6.3 1区地球科学 Q1 ENGINEERING, CIVIL

Journal of Hydrology

Pub Date : 2026-04-01 Epub Date: 2026-02-03 DOI: 10.1016/j.jhydrol.2026.135094

Xiaohui Ren , Ruihong Yu , Yanjie Mi

Elucidating organic matter sources in lakes during the frozen period (FP) is crucial for understanding carbon and nitrogen cycling in global cold-region lakes. However, limited information exists on the sources and differences between suspended particulate organic matter and sediment organic matter in lakes during the FP. This study examined the stable carbon (δ¹³C) and nitrogen (δ¹⁵N) isotopic compositions of organic matter in suspended particulates and sediments from Daihai Lake to investigate the sources and implications of organic matter during the FP (January). Organic carbon in suspended particulates and total nitrogen in suspended particulates ranged from 0.45 to 1.22 mg/L and 0.08 to 0.20 mg/L, respectively, while organic carbon in sediments and total nitrogen in sediments ranged from 3.57 to 14.90 g/kg and 0.44 to 1.68 g/kg, respectively. Based on organic index (OI) and organic nitrogen (ON), suspended particulates (OI: 0.10 mg/L; ON: 0.12 mg/L) were slightly contaminated, whereas sediments (OI: 13.13 g/kg; ON: 1.08 g/kg) were heavily contaminated. The suspended particulate organic matter exhibited mixed source signatures, with exogenous inputs (sewage: 27.9% and soil: 22.2%) and endogenous production (phytoplankton: 25.8%). In contrast, sediment organic matter was predominantly exogenous inputs (soil: 40.3% and sewage: 26.6%). This discrepancy highlights a key process in organic matter transport and transformation under ice-covered conditions: phytoplankton-derived organic matter underwent preferential degradation during sedimentation, whereas terrestrial organic matter was more readily deposited and preserved. Notably, significant nitrogen isotope fractionation during sedimentation indicates that preferential mineralization of organic nitrogen and denitrification played key regulatory roles in the nitrogen cycling. The findings highlight the need to prioritize controlling exogenous organic matter inputs to lakes to ensure sustainable ecosystem.

阐明冻结期湖泊有机质来源对了解全球寒区湖泊碳氮循环具有重要意义。然而，关于FP期间湖泊中悬浮颗粒有机质和沉积物有机质的来源和差异的信息有限。本研究通过对滇东南东南东南东南地区沉积物和悬浮颗粒物中有机质的稳定碳（δ13C）和稳定氮（δ15N）同位素组成的分析，探讨滇东南东南东南东南地区FP（1月）的有机质来源及其意义。悬浮颗粒物中有机碳和总氮含量分别为0.45 ~ 1.22 mg/L和0.08 ~ 0.20 mg/L，沉积物中有机碳和总氮含量分别为3.57 ~ 14.90 g/kg和0.44 ~ 1.68 g/kg。以有机指数（OI）和有机氮（on）为指标，悬浮颗粒（OI: 0.10 mg/L, on: 0.12 mg/L）为轻度污染，沉积物（OI: 13.13 g/kg, on: 1.08 g/kg）为重度污染。悬浮颗粒有机质呈现混合来源特征，外源输入（污水占27.9%，土壤占22.2%）和内源产生（浮游植物占25.8%）。相比之下，沉积物有机质主要是外源输入（土壤：40.3%，污水：26.6%）。这种差异突出了冰覆盖条件下有机物运输和转化的一个关键过程：浮游植物来源的有机物在沉积过程中优先降解，而陆源有机物更容易沉积和保存。值得注意的是，沉积过程中显著的氮同位素分馏表明有机氮的优先矿化和反硝化作用在氮循环中起关键调节作用。研究结果强调，需要优先控制外源有机物质输入到湖泊，以确保可持续的生态系统。

{"title":"Differences in organic matter sources between suspended particulates and sediments in a lake during the frozen period","authors":"Xiaohui Ren , Ruihong Yu , Yanjie Mi","doi":"10.1016/j.jhydrol.2026.135094","DOIUrl":"10.1016/j.jhydrol.2026.135094","url":null,"abstract":"<div><div>Elucidating organic matter sources in lakes during the frozen period (FP) is crucial for understanding carbon and nitrogen cycling in global cold-region lakes. However, limited information exists on the sources and differences between suspended particulate organic matter and sediment organic matter in lakes during the FP. This study examined the stable carbon (δ<sup>13</sup>C) and nitrogen (δ<sup>15</sup>N) isotopic compositions of organic matter in suspended particulates and sediments from Daihai Lake to investigate the sources and implications of organic matter during the FP (January). Organic carbon in suspended particulates and total nitrogen in suspended particulates ranged from 0.45 to 1.22 mg/L and 0.08 to 0.20 mg/L, respectively, while organic carbon in sediments and total nitrogen in sediments ranged from 3.57 to 14.90 g/kg and 0.44 to 1.68 g/kg, respectively. Based on organic index (OI) and organic nitrogen (ON), suspended particulates (OI: 0.10 mg/L; ON: 0.12 mg/L) were slightly contaminated, whereas sediments (OI: 13.13 g/kg; ON: 1.08 g/kg) were heavily contaminated. The suspended particulate organic matter exhibited mixed source signatures, with exogenous inputs (sewage: 27.9% and soil: 22.2%) and endogenous production (phytoplankton: 25.8%). In contrast, sediment organic matter was predominantly exogenous inputs (soil: 40.3% and sewage: 26.6%). This discrepancy highlights a key process in organic matter transport and transformation under ice-covered conditions: phytoplankton-derived organic matter underwent preferential degradation during sedimentation, whereas terrestrial organic matter was more readily deposited and preserved. Notably, significant nitrogen isotope fractionation during sedimentation indicates that preferential mineralization of organic nitrogen and denitrification played key regulatory roles in the nitrogen cycling. The findings highlight the need to prioritize controlling exogenous organic matter inputs to lakes to ensure sustainable ecosystem.</div></div>","PeriodicalId":362,"journal":{"name":"Journal of Hydrology","volume":"669 ","pages":"Article 135094"},"PeriodicalIF":6.3,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146110449","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

Spatiotemporal variation of net anthropogenic phosphorus input and its delayed effect on lake phosphorus in a plateau lake of southwestern China 西南高原湖泊人为净磷输入的时空变化及其对湖泊磷的延迟效应

IF 6.3 1区地球科学 Q1 ENGINEERING, CIVIL

Journal of Hydrology

Pub Date : 2026-04-01 Epub Date: 2026-01-17 DOI: 10.1016/j.jhydrol.2026.134969

Huanyao Liu , Cen Meng , Shuaibing Wang , Yuyuan Li , Hui Fu , Wei Ouyang

Plateau lakes are highly sensitive to anthropogenic phosphorus (P) enrichment. Despite recent management strategies reducing watershed P inputs, lake eutrophication persists, indicating a delayed recovery of water P level in response to external P reduction. This study focuses on Xingyun Lake, a typical shallow plateau lake in central Yunnan Province (Yuxi City), China, and applies the net anthropogenic P input (NAPI) model to evaluate the influences of different P sources in the watershed of Xingyun Lake from 1989 to 2020. The convergent cross mapping (CCM) approach was utilized to detect nonlinear and time-lagged causal relationships between NAPI and total phosphorus (TP) dynamics. Segmented regression analysis was further used to identify thresholds of P inputs leading to abrupt changes in lake TP concentrations. NAPI ranged from 1,148 ± 304 to 6,984 ± 2,206 kg km⁻² yr⁻¹ during the study period and was primarily driven by food/feed net P input (P_im), ore mining P input (P_ore), and fertilizer P input (P_fer). From 2000 to 2020, P_ore was the dominant contributor to NAPI changes in the northern and eastern region, particularly in Jiangcheng and Luju towns. CCM analysis identified a significant nonlinear and unidirectional causal link from NAPI to TP with a 4-year time lag. Threshold analysis demonstrated that lake TP concentrations exhibited a nonlinear upward trend when P_fer and P_ore exceeded 806 and 1,305 kg km⁻² yr⁻¹, respectively. This study provides the first quantitative evidence for the time-lagged response of lake P dynamics to anthropogenic P inputs in plateau lake systems, offering theoretical support and technical guidance for setting watershed water quality improvement goals and optimizing policy intervention timing.

高原湖泊对人为磷富集高度敏感。尽管最近的管理策略减少了流域磷的输入，但湖泊富营养化仍然存在，表明水体磷水平的恢复是对外部磷减少的延迟响应。以云南中部典型的高原浅水湖泊星云湖为研究对象，应用净人为磷输入（NAPI）模型，评价1989 - 2020年不同磷源对星云湖流域的影响。采用收敛交叉映射（CCM）方法检测NAPI与总磷（TP）动态之间的非线性时滞因果关系。通过分段回归分析，确定了磷输入导致湖泊总磷浓度突变的阈值。在研究期间，NAPI范围为1,148±304至6,984±2,206 kg km−2 yr−1，主要由食物/饲料净磷输入（Pim）、采矿磷输入（Pore）和肥料磷输入（Pfer）驱动。2000 - 2020年，北部和东部以江城和鹿居镇为主要影响因子；CCM分析发现，NAPI与TP之间存在显著的非线性单向因果关系，存在4年的时滞。阈值分析表明，当Pfer和Pore分别超过806和1305 kg km−2 yr−1时，湖泊总磷浓度呈非线性上升趋势。本研究首次为高原湖泊系统湖泊磷动态对人为磷输入的滞后响应提供了定量证据，为制定流域水质改善目标和优化政策干预时机提供了理论支持和技术指导。

{"title":"Spatiotemporal variation of net anthropogenic phosphorus input and its delayed effect on lake phosphorus in a plateau lake of southwestern China","authors":"Huanyao Liu , Cen Meng , Shuaibing Wang , Yuyuan Li , Hui Fu , Wei Ouyang","doi":"10.1016/j.jhydrol.2026.134969","DOIUrl":"10.1016/j.jhydrol.2026.134969","url":null,"abstract":"<div><div>Plateau lakes are highly sensitive to anthropogenic phosphorus (P) enrichment. Despite recent management strategies reducing watershed P inputs, lake eutrophication persists, indicating a delayed recovery of water P level in response to external P reduction. This study focuses on Xingyun Lake, a typical shallow plateau lake in central Yunnan Province (Yuxi City), China, and applies the net anthropogenic P input (NAPI) model to evaluate the influences of different P sources in the watershed of Xingyun Lake from 1989 to 2020. The convergent cross mapping (CCM) approach was utilized to detect nonlinear and time-lagged causal relationships between NAPI and total phosphorus (TP) dynamics. Segmented regression analysis was further used to identify thresholds of P inputs leading to abrupt changes in lake TP concentrations. NAPI ranged from 1,148 ± 304 to 6,984 ± 2,206 kg km<sup>−2</sup> yr<sup>−1</sup> during the study period and was primarily driven by food/feed net P input (P<sub>im</sub>), ore mining P input (P<sub>ore</sub>), and fertilizer P input (P<sub>fer</sub>). From 2000 to 2020, P<sub>ore</sub> was the dominant contributor to NAPI changes in the northern and eastern region, particularly in Jiangcheng and Luju towns. CCM analysis identified a significant nonlinear and unidirectional causal link from NAPI to TP with a 4-year time lag. Threshold analysis demonstrated that lake TP concentrations exhibited a nonlinear upward trend when P<sub>fer</sub> and P<sub>ore</sub> exceeded 806 and 1,305 kg km<sup>−2</sup> yr<sup>−1</sup>, respectively. This study provides the first quantitative evidence for the time-lagged response of lake P dynamics to anthropogenic P inputs in plateau lake systems, offering theoretical support and technical guidance for setting watershed water quality improvement goals and optimizing policy intervention timing.</div></div>","PeriodicalId":362,"journal":{"name":"Journal of Hydrology","volume":"668 ","pages":"Article 134969"},"PeriodicalIF":6.3,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145995705","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

Uncovering the dynamic role of bedrock-stored water in ecosystem evapotranspiration 揭示基岩储水在生态系统蒸散中的动态作用

IF 6.3 1区地球科学 Q1 ENGINEERING, CIVIL

Journal of Hydrology

Pub Date : 2026-04-01 Epub Date: 2026-01-19 DOI: 10.1016/j.jhydrol.2026.134987

Yang Qiu , Fawang Zhang , Lin Gao , Zekang He , Hanxiang Xiong , Ling Peng , Zhiye Wang , Hao Cui , Cheng Su , Defang Zhang , Shizheng Zhou , Chuanming Ma , Aiguo Zhou

Water stored in bedrock is crucial for ecosystem water supply, sustaining plant function, and drought resilience. However, conventional ecohydrological frameworks focus on soil moisture, often neglecting bedrock water and lacking adequate methodologies for its quantification. Here, we introduce a dynamic bedrock water evapotranspiration model and estimate the daily bedrock-derived evapotranspiration across China using multi-year meteorological and remote sensing datasets. Temporal and spatial patterns of evapotranspiration from bedrock-stored water reveal that over half of shallow bedrock ecosystems use bedrock water, contributing 10.84% of total evapotranspiration and 7.99% of vegetation transpiration. In arid/semi‑arid and seasonally dry regions, seasonal water-carbon decoupling increases dependence on bedrock water, particularly in southwestern and northwestern China. This occurs when transpiration demand rises before the peaks of precipitation and/or photosynthesis, necessitating deeper water sources. Causal machine learning shows that in water‑stressed climates, bedrock water enhances ecosystem productivity and carbon stocks by sustaining transpiration (FDR < 0.05). These results challenge the prevailing soil-centric paradigm and underscore the critical role of bedrock-stored water in enhancing ecosystem productivity and drought resilience. By linking hydrology, plant physiology, and carbon cycling across climatic gradients, our study supports improvements to land surface models and the prediction of ecosystem responses to intensifying drought.

储存在基岩中的水对生态系统供水、维持植物功能和抗旱能力至关重要。然而，传统的生态水文框架侧重于土壤水分，往往忽视基岩水，缺乏适当的量化方法。本文引入了一个动态基岩水蒸散模型，并利用多年气象和遥感资料估算了中国基岩日蒸散量。基岩储水蒸散的时空格局表明，一半以上的浅层基岩生态系统利用基岩水，贡献了10.84%的总蒸散和7.99%的植被蒸腾。在干旱/半干旱和季节性干旱地区，季节性水碳脱钩增加了对基岩水的依赖，特别是在中国西南和西北地区。当蒸腾需求在降水和/或光合作用高峰之前上升时，就会发生这种情况，因此需要更深层的水源。因果机器学习表明，在缺水气候下，基岩水通过维持蒸腾作用提高生态系统生产力和碳储量（FDR < 0.05）。这些结果挑战了主流的以土壤为中心的范式，强调了基岩储水在提高生态系统生产力和抗旱能力方面的关键作用。通过将水文、植物生理学和跨气候梯度的碳循环联系起来，我们的研究支持改进陆地表面模型和预测生态系统对干旱加剧的响应。

{"title":"Uncovering the dynamic role of bedrock-stored water in ecosystem evapotranspiration","authors":"Yang Qiu , Fawang Zhang , Lin Gao , Zekang He , Hanxiang Xiong , Ling Peng , Zhiye Wang , Hao Cui , Cheng Su , Defang Zhang , Shizheng Zhou , Chuanming Ma , Aiguo Zhou","doi":"10.1016/j.jhydrol.2026.134987","DOIUrl":"10.1016/j.jhydrol.2026.134987","url":null,"abstract":"<div><div>Water stored in bedrock is crucial for ecosystem water supply, sustaining plant function, and drought resilience. However, conventional ecohydrological frameworks focus on soil moisture, often neglecting bedrock water and lacking adequate methodologies for its quantification. Here, we introduce a dynamic bedrock water evapotranspiration model and estimate the daily bedrock-derived evapotranspiration across China using multi-year meteorological and remote sensing datasets. Temporal and spatial patterns of evapotranspiration from bedrock-stored water reveal that over half of shallow bedrock ecosystems use bedrock water, contributing 10.84% of total evapotranspiration and 7.99% of vegetation transpiration. In arid/semi‑arid and seasonally dry regions, seasonal water-carbon decoupling increases dependence on bedrock water, particularly in southwestern and northwestern China. This occurs when transpiration demand rises before the peaks of precipitation and/or photosynthesis, necessitating deeper water sources. Causal machine learning shows that in water‑stressed climates, bedrock water enhances ecosystem productivity and carbon stocks by sustaining transpiration (FDR < 0.05). These results challenge the prevailing soil-centric paradigm and underscore the critical role of bedrock-stored water in enhancing ecosystem productivity and drought resilience. By linking hydrology, plant physiology, and carbon cycling across climatic gradients, our study supports improvements to land surface models and the prediction of ecosystem responses to intensifying drought.</div></div>","PeriodicalId":362,"journal":{"name":"Journal of Hydrology","volume":"668 ","pages":"Article 134987"},"PeriodicalIF":6.3,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146001382","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

Hydro-meteorological factors and human activities contribute comparably to weekly water quality dynamics in the Yangtze River Basin 水文气象因子与人类活动对长江流域周水质动态的贡献相当

IF 6.3 1区地球科学 Q1 ENGINEERING, CIVIL

Journal of Hydrology

Pub Date : 2026-04-01 Epub Date: 2026-01-26 DOI: 10.1016/j.jhydrol.2026.135025

Leifang Li , Taihua Wang , Ge Li , Han Cheng , Shulei Zhang , Dawen Yang

Monitoring and managing water quality is challenged by pronounced short-term fluctuations in key indicators, driven by both hydro-meteorological variability and human activities. Although substantial research has focused on water-quality modeling, effectively characterizing week-scale dynamics using weekly data remains limited. To address this gap, this study develops weekly water-quality prediction models for ten sub-basins of the Yangtze River using seven machine-learning algorithms. The models demonstrate robust performance, with Nash–Sutcliffe efficiency (NSE) values ranging from 0.41 to 0.74, and are used to reconstruct weekly time series of four key indicators, pH, DO, CODMn, and NH₃-N, from 2012 to 2018. Among the evaluated algorithms, Random Forest Regression performs best, achieving average NSE values of 0.56–0.81 despite a 45% missing-data rate. Shapley Additive Explanations (SHAP) analysis reveals that hydro-meteorological and anthropogenic drivers contribute comparably to weekly water-quality variability, with hydro-meteorological and agricultural sectors emerging as dominant controls in water quality. About half of the sub-basins show water quality exceedances mainly associated with hydro-meteorological conditions, whereas the remainder are influenced primarily by anthropogenic pressures, underscoring the need for differentiated and sector-targeted management strategies. By reconstructing missing records and quantifying driver importance, this study provides a practical framework for interpreting short-term water quality fluctuations and diagnosing weekly exceedance events. The proposed approach supports risk-informed water quality management in large river basins and is transferable to other monsoonal and temperate catchments.

监测和管理水质受到水文气象变化和人类活动共同推动的关键指标的短期显著波动的挑战。尽管大量的研究集中在水质建模上，但利用每周数据有效地表征周尺度动态仍然有限。为了解决这一差距，本研究使用7种机器学习算法开发了长江10个子流域的每周水质预测模型。该模型表现出稳健的性能，NSE值在0.41 ~ 0.74之间，并用于重建2012 - 2018年pH、DO、CODMn和NH3-N四个关键指标的周时间序列。在评估的算法中，随机森林回归表现最好，尽管丢失数据率为45%，但平均NSE值为0.56-0.81。Shapley加性解释（SHAP）分析表明，水文气象和人为驱动因素对每周水质变化的贡献相当，其中水文气象和农业部门正在成为水质的主要控制因素。大约一半的子流域的水质超标主要与水文气象条件有关，而其余的则主要受到人为压力的影响，强调需要有区别的和针对部门的管理战略。通过重建缺失记录和量化驱动因素的重要性，本研究为解释短期水质波动和诊断每周超标事件提供了实用的框架。拟议的方法支持在大型河流流域进行风险知情的水质管理，并可转移到其他季风和温带流域。

{"title":"Hydro-meteorological factors and human activities contribute comparably to weekly water quality dynamics in the Yangtze River Basin","authors":"Leifang Li , Taihua Wang , Ge Li , Han Cheng , Shulei Zhang , Dawen Yang","doi":"10.1016/j.jhydrol.2026.135025","DOIUrl":"10.1016/j.jhydrol.2026.135025","url":null,"abstract":"<div><div>Monitoring and managing water quality is challenged by pronounced short-term fluctuations in key indicators, driven by both hydro-meteorological variability and human activities. Although substantial research has focused on water-quality modeling, effectively characterizing week-scale dynamics using weekly data remains limited. To address this gap, this study develops weekly water-quality prediction models for ten sub-basins of the Yangtze River using seven machine-learning algorithms. The models demonstrate robust performance, with Nash–Sutcliffe efficiency (NSE) values ranging from 0.41 to 0.74, and are used to reconstruct weekly time series of four key indicators, pH, DO, CODMn, and NH<sub>3</sub>-N, from 2012 to 2018. Among the evaluated algorithms, Random Forest Regression performs best, achieving average NSE values of 0.56–0.81 despite a 45% missing-data rate. Shapley Additive Explanations (SHAP) analysis reveals that hydro-meteorological and anthropogenic drivers contribute comparably to weekly water-quality variability, with hydro-meteorological and agricultural sectors emerging as dominant controls in water quality. About half of the sub-basins show water quality exceedances mainly associated with hydro-meteorological conditions, whereas the remainder are influenced primarily by anthropogenic pressures, underscoring the need for differentiated and sector-targeted management strategies. By reconstructing missing records and quantifying driver importance, this study provides a practical framework for interpreting short-term water quality fluctuations and diagnosing weekly exceedance events. The proposed approach supports risk-informed water quality management in large river basins and is transferable to other monsoonal and temperate catchments.</div></div>","PeriodicalId":362,"journal":{"name":"Journal of Hydrology","volume":"668 ","pages":"Article 135025"},"PeriodicalIF":6.3,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146048582","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

Multi attribute refined identification of flood-affected bodies based on multi-source data fusion 基于多源数据融合的洪灾体多属性精细化识别

IF 6.3 1区地球科学 Q1 ENGINEERING, CIVIL

Journal of Hydrology

Pub Date : 2026-04-01 Epub Date: 2026-02-08 DOI: 10.1016/j.jhydrol.2026.135104

Yutie Jiao , Zongkun Li , Wei Ge , Meimei Wu , Bo Wang , Yadong Zhang , Pieter van Gelder

Lands and populations are the most direct and core disaster-bearing bodies in floods. Accurate and comprehensive identification their attributes is critical for differentiated flood prevention and mitigation strategies. However, two key challenges persist in current practices. First, the accuracy of urban land function (ULF) identification based on machine learning is constrained by data and grid scale, yet research on their impacts remains insufficient. Second, location-based service (LBS) data has sampling bias and nighttime distortion in characterizing dynamic population distribution, and its spatial resolution is insufficient for high-precision flood simulations. For ULF identification, abundant comparison schemes are generated through data traversal and multi-scale fusion, and an ensemble learning model is constructed to select the optimal ULF identification scheme. This avoids the accuracy uncertainty caused by subjective selection, and the results provide reliable data support for economic loss assessment and subsequent population spatial interpolation. For dynamic population distribution, a human-land relationship matching method based on spatiotemporal behavioral laws is proposed to reduce the impact of data bias. Meanwhile, spatial downscaling is achieved through regional division, land type and area weight calculation, generating dynamic population distribution maps with high spatiotemporal resolution. The results support the analysis of population mobility’s impact on flood risk. Hydraulic simulation is coupled with GIS (geographic information system) analysis to construct a grid-based diagnostic framework for multi-attributes of disaster-bearing bodies, including land function, population size, water depth, and spatial location. Case studies show that this framework provides reliable support for the accurate and comprehensive identification of flood disaster-bearing bodies.

土地和人口是洪水最直接、最核心的受灾体。准确和全面地识别它们的属性对于制定差异化的防洪和减灾战略至关重要。然而，在当前的实践中仍然存在两个关键挑战。首先，基于机器学习的城市土地功能（ULF）识别精度受到数据和网格尺度的限制，对其影响的研究不足。其次，基于位置服务（LBS）的数据在刻画人口动态分布时存在采样偏差和夜间失真，其空间分辨率不足以实现高精度的洪水模拟。对于ULF识别，通过数据遍历和多尺度融合生成丰富的比较方案，并构建集成学习模型来选择最优的ULF识别方案。避免了主观选择带来的精度不确定性，为经济损失评估和后续人口空间插值提供了可靠的数据支持。针对人口动态分布，提出了一种基于时空行为规律的人地关系匹配方法，以减少数据偏差的影响。同时，通过区域划分、土地类型和面积权重计算实现空间降尺度，生成高时空分辨率的动态人口分布图。研究结果支持了人口流动对洪水风险影响的分析。将水力学模拟与地理信息系统（GIS）分析相结合，构建基于网格的承灾体多属性诊断框架，包括土地功能、人口规模、水深、空间位置等。实例研究表明，该框架为准确、全面地识别洪水承灾体提供了可靠的支持。

{"title":"Multi attribute refined identification of flood-affected bodies based on multi-source data fusion","authors":"Yutie Jiao , Zongkun Li , Wei Ge , Meimei Wu , Bo Wang , Yadong Zhang , Pieter van Gelder","doi":"10.1016/j.jhydrol.2026.135104","DOIUrl":"10.1016/j.jhydrol.2026.135104","url":null,"abstract":"<div><div>Lands and populations are the most direct and core disaster-bearing bodies in floods. Accurate and comprehensive identification their attributes is critical for differentiated flood prevention and mitigation strategies. However, two key challenges persist in current practices. First, the accuracy of urban land function (ULF) identification based on machine learning is constrained by data and grid scale, yet research on their impacts remains insufficient. Second, location-based service (LBS) data has sampling bias and nighttime distortion in characterizing dynamic population distribution, and its spatial resolution is insufficient for high-precision flood simulations. For ULF identification, abundant comparison schemes are generated through data traversal and multi-scale fusion, and an ensemble learning model is constructed to select the optimal ULF identification scheme. This avoids the accuracy uncertainty caused by subjective selection, and the results provide reliable data support for economic loss assessment and subsequent population spatial interpolation. For dynamic population distribution, a human-land relationship matching method based on spatiotemporal behavioral laws is proposed to reduce the impact of data bias. Meanwhile, spatial downscaling is achieved through regional division, land type and area weight calculation, generating dynamic population distribution maps with high spatiotemporal resolution. The results support the analysis of population mobility’s impact on flood risk. Hydraulic simulation is coupled with GIS (geographic information system) analysis to construct a grid-based diagnostic framework for multi-attributes of disaster-bearing bodies, including land function, population size, water depth, and spatial location. Case studies show that this framework provides reliable support for the accurate and comprehensive identification of flood disaster-bearing bodies.</div></div>","PeriodicalId":362,"journal":{"name":"Journal of Hydrology","volume":"669 ","pages":"Article 135104"},"PeriodicalIF":6.3,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146138394","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 causality-informed and Copula-based framework for predictive assessment of downstream water-quality risks under extreme scenarios 基于因果关系和copula的极端情景下下游水质风险预测评估框架

IF 6.3 1区地球科学 Q1 ENGINEERING, CIVIL

Journal of Hydrology

Pub Date : 2026-04-01 Epub Date: 2026-02-12 DOI: 10.1016/j.jhydrol.2026.135123

Quan Zhou , Jiajia Zhang , Tian Lan , Yongqin David Chen , Siye Wei , Xiuwen Ren , Wenjing Wang , Chunju Zheng , Renren Wu

In view of heightened hydroclimatic extremes and increasing human activities, aquatic ecosystems have been increasingly exposed to extreme water quality and hydrometeorological disturbances, with downstream dissolved oxygen (DO) dynamics affected by multiple water quality and meteorological drivers. To evaluate these risks, a downstream water-quality risk assessment framework is established to quantify the probability of future hypoxic events and to examine the transmission pathways of multi-source stressors. A CA-LSTM water quality prediction model is developed, integrating features derived from causal discovery and time–frequency analysis into a gated-memory attention architecture, thereby enhancing interpretability and predictive performance of DO dynamics. Joint variations of water-quality and meteorological factors that exert the strongest influence on predicted downstream DO are incorporated into the analytical framework. Vine Copula structure is employed to characterize the dependency structure between selected indicators and predicted downstream DO, which allowed the quantification of tail risks under extreme conditions. Application to the Shatiansisheng Station demonstrated that the CA-LSTM model achieved DO forecasts with an RMSE of 0.52 mg/L and an R² of 0.91, accurately reproducing both long-term patterns and short-term fluctuations. Single-factor analysis indicated that downstream water temperature (WT) exerted the strongest influence on downstream DO. Under combined extreme scenarios, when upstream DO fell below 5 mg/L and downstream WT exceeded the 90th percentile, the probability of downstream DO declining below the threshold within one day approached certainty. The proposed framework provides a quantitative basis for early warning of hypoxic events and offers methodological support for catchment-scale water-quality management.

由于极端水文气候的加剧和人类活动的增加，水生生态系统越来越多地暴露于极端水质和水文气象干扰中，下游溶解氧（DO）动态受到多种水质和气象驱动因素的影响。为了评估这些风险，建立了下游水质风险评估框架，以量化未来缺氧事件的概率，并检查多源应激源的传递途径。建立了CA-LSTM水质预测模型，将因果发现和时频分析的特征整合到门控记忆注意力架构中，从而提高了DO动力学的可解释性和预测性能。将对预测下游DO影响最大的水质和气象因子的联合变化纳入分析框架。采用Vine Copula结构表征所选指标与预测下游DO之间的依赖关系结构，从而可以量化极端条件下的尾部风险。在沙天寺生站的应用表明，CA-LSTM模型对DO的预测RMSE为0.52 mg/L， R2为0.91，既能准确地再现长期模式，也能准确再现短期波动。单因素分析表明，下游水温对DO的影响最大。在综合极端情景下，当上游DO低于5 mg/L，下游WT超过第90个百分位数时，一天内下游DO低于阈值的概率接近确定。提出的框架为缺氧事件的早期预警提供了定量基础，并为流域尺度的水质管理提供了方法学支持。

{"title":"A causality-informed and Copula-based framework for predictive assessment of downstream water-quality risks under extreme scenarios","authors":"Quan Zhou , Jiajia Zhang , Tian Lan , Yongqin David Chen , Siye Wei , Xiuwen Ren , Wenjing Wang , Chunju Zheng , Renren Wu","doi":"10.1016/j.jhydrol.2026.135123","DOIUrl":"10.1016/j.jhydrol.2026.135123","url":null,"abstract":"<div><div>In view of heightened hydroclimatic extremes and increasing human activities, aquatic ecosystems have been increasingly exposed to extreme water quality and hydrometeorological disturbances, with downstream dissolved oxygen (DO) dynamics affected by multiple water quality and meteorological drivers. To evaluate these risks, a downstream water-quality risk assessment framework is established to quantify the probability of future hypoxic events and to examine the transmission pathways of multi-source stressors. A CA-LSTM water quality prediction model is developed, integrating features derived from causal discovery and time–frequency analysis into a gated-memory attention architecture, thereby enhancing interpretability and predictive performance of DO dynamics. Joint variations of water-quality and meteorological factors that exert the strongest influence on predicted downstream DO are incorporated into the analytical framework. Vine Copula structure is employed to characterize the dependency structure between selected indicators and predicted downstream DO, which allowed the quantification of tail risks under extreme conditions. Application to the Shatiansisheng Station demonstrated that the CA-LSTM model achieved DO forecasts with an RMSE of 0.52 mg/L and an R<sup>2</sup> of 0.91, accurately reproducing both long-term patterns and short-term fluctuations. Single-factor analysis indicated that downstream water temperature (WT) exerted the strongest influence on downstream DO. Under combined extreme scenarios, when upstream DO fell below 5 mg/L and downstream WT exceeded the 90th percentile, the probability of downstream DO declining below the threshold within one day approached certainty. The proposed framework provides a quantitative basis for early warning of hypoxic events and offers methodological support for catchment-scale water-quality management.</div></div>","PeriodicalId":362,"journal":{"name":"Journal of Hydrology","volume":"669 ","pages":"Article 135123"},"PeriodicalIF":6.3,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146186390","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

Deep learning for groundwater level simulation in unconfined aquifers across the contiguous United States: Analyzing simulations at multiple lead times and integrating groundwater signatures 美国连续无约束含水层地下水位模拟的深度学习：在多个提前期分析模拟并整合地下水特征

IF 6.3 1区地球科学 Q1 ENGINEERING, CIVIL

Journal of Hydrology

Pub Date : 2026-04-01 Epub Date: 2026-01-12 DOI: 10.1016/j.jhydrol.2026.134949

Kenneth Beng Wee Boo , Ming Fai Chow , Wai Peng Wong , Ali Najah Ahmed , Ahmed El-Shafie

Simulating groundwater level accurately ahead of time is critical for various practical reasons. In this study, we simulate daily groundwater levels using meteorological data at 1-day to 7-day lead times across 249 unconfined wells in the contiguous United States. Our objectives include: 1) comparing LSTM-based Seq2Seq and Seq2One models for our simulation tasks, 2) investigating the efficacy of addressing failed models (where Nash Sutcliffe-Efficiency (NSE) < 0) by using alternative training-validation splits, and 3) integrating groundwater signatures for a more thorough model evaluation process. Our results demonstrate satisfactory to good performance with our best performing models achieving median NSE scores of 0.744 and 0.603 for 1-day and 7-day lead simulations, respectively. Notably, we show that LSTM-based Seq2One model outperforms the Seq2Seq model at 1-day lead simulation, however the difference in their performance is not statistically significant for 4-day and 7-day lead simulation scenarios. We found that 14% of our models exhibit subpar performance, which may be attributed to modeling complex underlying groundwater systems without key anthropogenic forcing variables. And we show that simply changing the training-validation splits is generally not sufficient to address these failed models. Our analysis with groundwater signatures, which are statistical aggregates of groundwater level hydrographs, reveals that most groundwater dynamics are well captured by the model, and notable correlations between model NSE and groundwater signatures are found. These findings position us towards better interpreting the capabilities and limitations of our groundwater models.

由于各种实际原因，提前准确模拟地下水位至关重要。在这项研究中，我们使用气象数据在1天到7天的提前时间模拟了美国249口无密封井的每日地下水位。我们的目标包括：1)比较基于lstm的Seq2Seq和Seq2One模型用于我们的模拟任务，2)通过使用替代训练验证分割来研究解决失败模型（其中纳什苏特克利夫效率（NSE） <； 0）的有效性，以及3)整合地下水特征以进行更彻底的模型评估过程。我们的结果显示出令人满意的良好性能，我们表现最好的模型在1天和7天的领先模拟中分别获得了0.744和0.603的中位数NSE分数。值得注意的是，我们发现基于lstm的Seq2One模型在提前1天的模拟中优于Seq2Seq模型，但是在提前4天和7天的模拟场景中，它们的性能差异没有统计学意义。我们发现14%的模型表现不佳，这可能是由于在模拟复杂的地下地下水系统时没有考虑关键的人为强迫变量。我们表明，简单地改变训练-验证分割通常不足以解决这些失败的模型。我们对地下水特征（地下水水位线的统计集合）的分析表明，该模型很好地捕捉了大多数地下水动态，并且发现模型NSE与地下水特征之间存在显著的相关性。这些发现使我们能够更好地解释我们的地下水模型的能力和局限性

{"title":"Deep learning for groundwater level simulation in unconfined aquifers across the contiguous United States: Analyzing simulations at multiple lead times and integrating groundwater signatures","authors":"Kenneth Beng Wee Boo , Ming Fai Chow , Wai Peng Wong , Ali Najah Ahmed , Ahmed El-Shafie","doi":"10.1016/j.jhydrol.2026.134949","DOIUrl":"10.1016/j.jhydrol.2026.134949","url":null,"abstract":"<div><div>Simulating groundwater level accurately ahead of time is critical for various practical reasons. In this study, we simulate daily groundwater levels using meteorological data at 1-day to 7-day lead times across 249 unconfined wells in the contiguous United States. Our objectives include: 1) comparing LSTM-based Seq2Seq and Seq2One models for our simulation tasks, 2) investigating the efficacy of addressing failed models (where Nash Sutcliffe-Efficiency (NSE) < 0) by using alternative training-validation splits, and 3) integrating groundwater signatures for a more thorough model evaluation process. Our results demonstrate satisfactory to good performance with our best performing models achieving median NSE scores of 0.744 and 0.603 for 1-day and 7-day lead simulations, respectively. Notably, we show that LSTM-based Seq2One model outperforms the Seq2Seq model at 1-day lead simulation, however the difference in their performance is not statistically significant for 4-day and 7-day lead simulation scenarios. We found that 14% of our models exhibit subpar performance, which may be attributed to modeling complex underlying groundwater systems without key anthropogenic forcing variables. And we show that simply changing the training-validation splits is generally not sufficient to address these failed models. Our analysis with groundwater signatures, which are statistical aggregates of groundwater level hydrographs, reveals that most groundwater dynamics are well captured by the model, and notable correlations between model NSE and groundwater signatures are found. These findings position us towards better interpreting the capabilities and limitations of our groundwater models.</div></div>","PeriodicalId":362,"journal":{"name":"Journal of Hydrology","volume":"668 ","pages":"Article 134949"},"PeriodicalIF":6.3,"publicationDate":"2026-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145957161","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