Pub Date : 2025-12-01Epub Date: 2025-09-28DOI: 10.1016/j.advwatres.2025.105132
Wenxin Yang, Hai Sun, Lei Zhang, Gloire Imani, Dongyan Fan, Junjie Zhong, Yongfei Yang, Jun Yao
In the geological storage of hydrogen and carbon dioxide in underground salt caverns, the water-injected salt dissolution technology is widely applied in the expansion of salt caverns. During the salt dissolution, a variety of minerals with different properties are often involved, and there are obvious competitive interactions among them. This competition is primarily manifested in differences in diffusion rates, dissolution rates, and ionic concentration equilibrium constraints. In this study, we developed a lattice Boltzmann model, taking into account the competitive dissolution mechanisms of different minerals under the constraint of ion equilibrium. The model was used to investigate permeability changes of porous structure under various injection velocities and different combinations of minerals. The results reveal that whether the physical properties of binary minerals vary greatly or not, such as reaction rate and diffusion rate, the impact of competitive dissolution cannot be ignored. And as the injection rate increases, the influence of the competitive dissolution effect on the pore structure evolution becomes greater. This research provides theoretical insights into binary minerals' competitive dissolution mechanisms and references for its applications in fields such as environmental science, resource development, and chemical engineering.
{"title":"Competitive dissolution of binary minerals in porous media: A lattice Boltzmann study","authors":"Wenxin Yang, Hai Sun, Lei Zhang, Gloire Imani, Dongyan Fan, Junjie Zhong, Yongfei Yang, Jun Yao","doi":"10.1016/j.advwatres.2025.105132","DOIUrl":"10.1016/j.advwatres.2025.105132","url":null,"abstract":"<div><div>In the geological storage of hydrogen and carbon dioxide in underground salt caverns, the water-injected salt dissolution technology is widely applied in the expansion of salt caverns. During the salt dissolution, a variety of minerals with different properties are often involved, and there are obvious competitive interactions among them. This competition is primarily manifested in differences in diffusion rates, dissolution rates, and ionic concentration equilibrium constraints. In this study, we developed a lattice Boltzmann model, taking into account the competitive dissolution mechanisms of different minerals under the constraint of ion equilibrium. The model was used to investigate permeability changes of porous structure under various injection velocities and different combinations of minerals. The results reveal that whether the physical properties of binary minerals vary greatly or not, such as reaction rate and diffusion rate, the impact of competitive dissolution cannot be ignored. And as the injection rate increases, the influence of the competitive dissolution effect on the pore structure evolution becomes greater. This research provides theoretical insights into binary minerals' competitive dissolution mechanisms and references for its applications in fields such as environmental science, resource development, and chemical engineering.</div></div>","PeriodicalId":7614,"journal":{"name":"Advances in Water Resources","volume":"206 ","pages":"Article 105132"},"PeriodicalIF":4.2,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145217975","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-01Epub Date: 2025-09-25DOI: 10.1016/j.advwatres.2025.105125
Daniel Stalder, Shangyi Cao, Daniel W. Meyer, Patrick Jenny
Flow in fractured porous media is associated with high uncertainty, particularly regarding fracture properties and their overall configuration within the domain. This is especially pronounced for disconnected fractures of smaller yet comparable size to the domain. Consequently, ensemble averages are often used to capture this statistical variability and predict the expected behavior. This leads to enormous computational costs, as flow simulations of single realizations with millions of fractures are extremely expensive; and much more so full Monte Carlo studies involving hundreds of realizations. Alternatively, a recently introduced model aims to directly estimate expected flow rates and pressure fields. The model involves few degrees of freedom, leading to low-cost computations. This is achieved by using integro-differential equations involving non-local kernel functions that encompass the statistical information of fractures. So far this statistical integro-differential fracture model (Sid-FM) considers only ensembles with identical fractures having constant aperture and lengths. In this paper Sid-FM is extended to account for arbitrary fracture aperture profiles and reservoirs with fractures following specified length distributions, which is a crucial step towards applications with realistic fractured reservoirs. In a series of numerical experiments, it is demonstrated that the Sid-FM’s predictions are in excellent agreement with Monte Carlo reference data, which are based on many fracture-resolving simulations. The applicability is demonstrated through statistically one-dimensional cases, laying crucial groundwork for 2D and 3D extensions. Future work will focus on further generalizations and extensions such as transport processes and 2D/3D applications.
{"title":"Statistical integro-differential fracture model (Sid-FM) for isolated fractures with variable apertures and lengths","authors":"Daniel Stalder, Shangyi Cao, Daniel W. Meyer, Patrick Jenny","doi":"10.1016/j.advwatres.2025.105125","DOIUrl":"10.1016/j.advwatres.2025.105125","url":null,"abstract":"<div><div>Flow in fractured porous media is associated with high uncertainty, particularly regarding fracture properties and their overall configuration within the domain. This is especially pronounced for disconnected fractures of smaller yet comparable size to the domain. Consequently, ensemble averages are often used to capture this statistical variability and predict the expected behavior. This leads to enormous computational costs, as flow simulations of single realizations with millions of fractures are extremely expensive; and much more so full Monte Carlo studies involving hundreds of realizations. Alternatively, a recently introduced model aims to directly estimate expected flow rates and pressure fields. The model involves few degrees of freedom, leading to low-cost computations. This is achieved by using integro-differential equations involving non-local kernel functions that encompass the statistical information of fractures. So far this statistical integro-differential fracture model (Sid-FM) considers only ensembles with identical fractures having constant aperture and lengths. In this paper Sid-FM is extended to account for arbitrary fracture aperture profiles and reservoirs with fractures following specified length distributions, which is a crucial step towards applications with realistic fractured reservoirs. In a series of numerical experiments, it is demonstrated that the Sid-FM’s predictions are in excellent agreement with Monte Carlo reference data, which are based on many fracture-resolving simulations. The applicability is demonstrated through statistically one-dimensional cases, laying crucial groundwork for 2D and 3D extensions. Future work will focus on further generalizations and extensions such as transport processes and 2D/3D applications.</div></div>","PeriodicalId":7614,"journal":{"name":"Advances in Water Resources","volume":"206 ","pages":"Article 105125"},"PeriodicalIF":4.2,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145217979","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-01Epub Date: 2025-11-10DOI: 10.1016/j.advwatres.2025.105172
Sabrina N. Volponi, Rajitha Senavirathna, Timothy R. Ginn
Advective–dispersive transport in natural systems is often modeled assuming steady one-dimensional flow and complete cross-sectional mixing, the latter allowing the use of a constant dispersion coefficient. In reality, however, many environments exhibit pre-asymptotic dispersion (PAD), where the effective dispersion coefficient increases as solutes progressively sample velocity heterogeneity. Time- and/or distance-dependent dispersion coefficients have historically been used to address PAD but can produce inconsistencies when solute pulses interact, particularly in nonlinear systems where superposition fails. Here we generalize our recent closed-form solution for one-dimensional advective–dispersive transport with PAD to the case of transient velocity. This new solution involves expressing dispersion as a function of both solute age in the flow (residence time), and physical time. These solutions apply to both initial and boundary value problems over an infinite domain, including cases with first-order reactions. We validate our boundary value solution by reproducing breakthrough curves from turbulent, accelerating pipe flow experiments. Our results show that a solute’s “memory” of reduced early-time dispersion continues to influence its transport even after complete mixing is achieved. More broadly, this framework fills a critical modeling gap by enabling physically consistent analysis of solute transport under PAD and transient flow—conditions common in natural systems but rarely studied together.
{"title":"A closed-form solution for pre-asymptotic dispersion with transient velocity","authors":"Sabrina N. Volponi, Rajitha Senavirathna, Timothy R. Ginn","doi":"10.1016/j.advwatres.2025.105172","DOIUrl":"10.1016/j.advwatres.2025.105172","url":null,"abstract":"<div><div>Advective–dispersive transport in natural systems is often modeled assuming steady one-dimensional flow and complete cross-sectional mixing, the latter allowing the use of a constant dispersion coefficient. In reality, however, many environments exhibit pre-asymptotic dispersion (PAD), where the effective dispersion coefficient increases as solutes progressively sample velocity heterogeneity. Time- and/or distance-dependent dispersion coefficients have historically been used to address PAD but can produce inconsistencies when solute pulses interact, particularly in nonlinear systems where superposition fails. Here we generalize our recent closed-form solution for one-dimensional advective–dispersive transport with PAD to the case of transient velocity. This new solution involves expressing dispersion as a function of both solute age in the flow (residence time), and physical time. These solutions apply to both initial and boundary value problems over an infinite domain, including cases with first-order reactions. We validate our boundary value solution by reproducing breakthrough curves from turbulent, accelerating pipe flow experiments. Our results show that a solute’s “memory” of reduced early-time dispersion continues to influence its transport even after complete mixing is achieved. More broadly, this framework fills a critical modeling gap by enabling physically consistent analysis of solute transport under PAD and transient flow—conditions common in natural systems but rarely studied together.</div></div>","PeriodicalId":7614,"journal":{"name":"Advances in Water Resources","volume":"206 ","pages":"Article 105172"},"PeriodicalIF":4.2,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145485561","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Hydrological monitoring establishes the foundation for scientific studies of water resources. The introduction of image-based hydrometry methods offers a non-intrusive and highly efficient approach for collecting discharge data from rivers. However, challenges persist for image-based hydrometry methods, such as partial observation, light reflection, unknown surface-to-depth coefficients, and unphysical velocity estimates requiring post-processing. In this study we propose a new image-based functional optimization method to address these issues. This Lagrangian multipliers formulation transforms river surface velocity and discharge estimation into the minimization of a distance in the functional space. The algorithm effectively handles partial, large scale, and high reflection observations, and the velocity field obtained through this method adheres to the physical governing equations. Three experiments were conducted across rivers of different scales, lighting, and weather conditions in France and China to demonstrate the applicability and robustness of the proposed algorithm. Through optimization of an objective functional, the results reveal that the proposed method can directly estimate river surface velocity and discharge from an image sequence using numerical model without the need for post-processing outliers or missing values. The study demonstrates that the proposed method has the potential to be widely deployed in different hydrological monitoring scenarios to improve hydrologic monitoring efficiency.
{"title":"Measurement as functional minimization: An image-based method for river surface velocity and discharge estimation","authors":"Kailin Huang , Lionel Pénard , Hua Chen , Chong-Yu Xu","doi":"10.1016/j.advwatres.2025.105160","DOIUrl":"10.1016/j.advwatres.2025.105160","url":null,"abstract":"<div><div>Hydrological monitoring establishes the foundation for scientific studies of water resources. The introduction of image-based hydrometry methods offers a non-intrusive and highly efficient approach for collecting discharge data from rivers. However, challenges persist for image-based hydrometry methods, such as partial observation, light reflection, unknown surface-to-depth coefficients, and unphysical velocity estimates requiring post-processing. In this study we propose a new image-based functional optimization method to address these issues. This Lagrangian multipliers formulation transforms river surface velocity and discharge estimation into the minimization of a distance in the functional space. The algorithm effectively handles partial, large scale, and high reflection observations, and the velocity field obtained through this method adheres to the physical governing equations. Three experiments were conducted across rivers of different scales, lighting, and weather conditions in France and China to demonstrate the applicability and robustness of the proposed algorithm. Through optimization of an objective functional, the results reveal that the proposed method can directly estimate river surface velocity and discharge from an image sequence using numerical model without the need for post-processing outliers or missing values. The study demonstrates that the proposed method has the potential to be widely deployed in different hydrological monitoring scenarios to improve hydrologic monitoring efficiency.</div></div>","PeriodicalId":7614,"journal":{"name":"Advances in Water Resources","volume":"206 ","pages":"Article 105160"},"PeriodicalIF":4.2,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145383870","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-01Epub Date: 2025-11-05DOI: 10.1016/j.advwatres.2025.105168
Yalin Song , Xiaoqing Shi , André Revil , Qilin Wang , Xinqiang Du , Jichun Wu
Toluene is a common groundwater contaminant originating from leaks of petroleum products and industrial effluents. Accurately assessing the in-situ biodegradation rate of such contaminants is crucial for evaluating the effectiveness of bioremediation strategies. However, traditional drilling and sampling methods are costly and incapable of in-situ biodegradation rate assessment. In recent years, Spectral induced polarization (SIP) has demonstrated to be an effective tool for real-time monitoring of microbial activity. However, to date, limited researches have investigated its mechanism to real-time monitoring of toluene biodegradation. To address this gap, nine soil column experiments were conducted to monitor the biodegradation of dissolved-phase toluene using the SIP method. Biodegradation was qualitatively confirmed through changes in dissolved oxygen (DO), nitrate (NO3-), and carbon isotope ratios (δ13C). The results indicate that the observed increase in quadrature conductivity primarily reflects bacterial growth during biodegradation rather than variations in dissolved toluene concentration. The maximum specific growth rate () of the inoculated bacteria was estimated to be 0.035 d-1 and the corresponding toluene biodegradation rate constant was 0.018 d-1. These findings demonstrate that SIP offers strong potentials as a quantitative, non-invasive, technique for tracking microbial degradation processes in porous media, providing theoretical and methodological support for future field-scale bioremediation applications.
{"title":"Real-time monitoring of dissolved toluene biodegradation in column experiments using spectral induced polarization","authors":"Yalin Song , Xiaoqing Shi , André Revil , Qilin Wang , Xinqiang Du , Jichun Wu","doi":"10.1016/j.advwatres.2025.105168","DOIUrl":"10.1016/j.advwatres.2025.105168","url":null,"abstract":"<div><div>Toluene is a common groundwater contaminant originating from leaks of petroleum products and industrial effluents. Accurately assessing the in-situ biodegradation rate of such contaminants is crucial for evaluating the effectiveness of bioremediation strategies. However, traditional drilling and sampling methods are costly and incapable of in-situ biodegradation rate assessment. In recent years, Spectral induced polarization (SIP) has demonstrated to be an effective tool for real-time monitoring of microbial activity. However, to date, limited researches have investigated its mechanism to real-time monitoring of toluene biodegradation. To address this gap, nine soil column experiments were conducted to monitor the biodegradation of dissolved-phase toluene using the SIP method. Biodegradation was qualitatively confirmed through changes in dissolved oxygen (DO), nitrate (NO<sub>3</sub><sup>-</sup>), and carbon isotope ratios (δ<sup>13</sup>C). The results indicate that the observed increase in quadrature conductivity primarily reflects bacterial growth during biodegradation rather than variations in dissolved toluene concentration. The maximum specific growth rate (<span><math><msub><mi>μ</mi><mi>m</mi></msub></math></span>) of the inoculated bacteria was estimated to be 0.035 <span>d</span><sup>-1</sup> and the corresponding toluene biodegradation rate constant was 0.018 <span>d</span><sup>-1</sup>. These findings demonstrate that SIP offers strong potentials as a quantitative, non-invasive, technique for tracking microbial degradation processes in porous media, providing theoretical and methodological support for future field-scale bioremediation applications.</div></div>","PeriodicalId":7614,"journal":{"name":"Advances in Water Resources","volume":"206 ","pages":"Article 105168"},"PeriodicalIF":4.2,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145441707","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The primary factors causing road damage in high-latitude areas are frost heave and thaw settlement, which are governed by hydrothermal changes. An indoor freeze-thaw cycle test based on similarity theory was conducted to analyze the hydrothermal evolution of high-latitude frozen soil in Northeast China. The results were verified by field data. A hydro-thermal coupling Eq. was established by using finite element software. The experimental and simulation results were compared to validate the finite element model of the soil column. The freeze-thaw cycles exhibited three distinct phases: freezing initiation, peak freezing intensity, and thawing. The soil moisture at the end of each freeze-thaw cycle increased, decreased, and increased. The proposed hydrothermal coupling model was used to describe the long-term freeze-thaw behavior of typical subgrade sections in high-latitude permafrost regions of China. The model based on similar particle gradation proved accurate.
{"title":"Hydrothermal evolution of high-latitude frozen soil during freeze-thaw cycles","authors":"Jiao Huang , Xiabing Yue , Xueying Wang , Hongwei Zhang","doi":"10.1016/j.advwatres.2025.105121","DOIUrl":"10.1016/j.advwatres.2025.105121","url":null,"abstract":"<div><div>The primary factors causing road damage in high-latitude areas are frost heave and thaw settlement, which are governed by hydrothermal changes. An indoor freeze-thaw cycle test based on similarity theory was conducted to analyze the hydrothermal evolution of high-latitude frozen soil in Northeast China. The results were verified by field data. A hydro-thermal coupling Eq. was established by using finite element software. The experimental and simulation results were compared to validate the finite element model of the soil column. The freeze-thaw cycles exhibited three distinct phases: freezing initiation, peak freezing intensity, and thawing. The soil moisture at the end of each freeze-thaw cycle increased, decreased, and increased. The proposed hydrothermal coupling model was used to describe the long-term freeze-thaw behavior of typical subgrade sections in high-latitude permafrost regions of China. The model based on similar particle gradation proved accurate.</div></div>","PeriodicalId":7614,"journal":{"name":"Advances in Water Resources","volume":"206 ","pages":"Article 105121"},"PeriodicalIF":4.2,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145093972","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-01Epub Date: 2025-11-01DOI: 10.1016/j.advwatres.2025.105163
Ching-Min Chang , Chuen-Fa Ni , Chi-Ping Lin , I-Hsian Lee , Wei-Cheng Lo
It is expected that climate change will lead to an increase in the extreme rainfall time series with nonstationary characteristics worldwide. This means that the assumption of stationarity in temporal rainfall fluctuations, used in the traditional stochastic analysis of temporal variation of rainfall events, may underestimate the temporal rainfall variability. Quantifying the fluctuations in soil moisture in response to variations in rainfall events is crucial for understanding the effects of climate change on regional water availability. Therefore, the purpose of this article is to generalize the previous results presented in the literature on quantifying temporal variability of soil moisture, which only apply to temporal stationarity in the random rainfall perturbation field. In this article, a general solution for the theoretical soil moisture semivariogram is derived to quantify the temporal variability of the nonstationary soil moisture perturbation field driven by the temporal nonstationary rainfall perturbation field. The use of the Fourier-Stieltjes spectral representation and the representation theorem enables the development of the theoretical soil moisture semivariogram expressed in the Fourier frequency domain. In addition, an approximate solution of the soil moisture semivariogram is given for the case that the temporal fluctuations in the rainfall field can be characterized by long-range power-law correlations. The influence of the coefficients in the rainfall power-law semivariogram and in the diffusion-injection model on the temporal variability of the nonstationary soil moisture perturbation field is analyzed.
{"title":"Characterization of temporal nonstationary random soil moisture perturbation fields","authors":"Ching-Min Chang , Chuen-Fa Ni , Chi-Ping Lin , I-Hsian Lee , Wei-Cheng Lo","doi":"10.1016/j.advwatres.2025.105163","DOIUrl":"10.1016/j.advwatres.2025.105163","url":null,"abstract":"<div><div>It is expected that climate change will lead to an increase in the extreme rainfall time series with nonstationary characteristics worldwide. This means that the assumption of stationarity in temporal rainfall fluctuations, used in the traditional stochastic analysis of temporal variation of rainfall events, may underestimate the temporal rainfall variability. Quantifying the fluctuations in soil moisture in response to variations in rainfall events is crucial for understanding the effects of climate change on regional water availability. Therefore, the purpose of this article is to generalize the previous results presented in the literature on quantifying temporal variability of soil moisture, which only apply to temporal stationarity in the random rainfall perturbation field. In this article, a general solution for the theoretical soil moisture semivariogram is derived to quantify the temporal variability of the nonstationary soil moisture perturbation field driven by the temporal nonstationary rainfall perturbation field. The use of the Fourier-Stieltjes spectral representation and the representation theorem enables the development of the theoretical soil moisture semivariogram expressed in the Fourier frequency domain. In addition, an approximate solution of the soil moisture semivariogram is given for the case that the temporal fluctuations in the rainfall field can be characterized by long-range power-law correlations. The influence of the coefficients in the rainfall power-law semivariogram and in the diffusion-injection model on the temporal variability of the nonstationary soil moisture perturbation field is analyzed.</div></div>","PeriodicalId":7614,"journal":{"name":"Advances in Water Resources","volume":"206 ","pages":"Article 105163"},"PeriodicalIF":4.2,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145424064","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-01Epub Date: 2025-10-08DOI: 10.1016/j.advwatres.2025.105140
A. Chiofalo , V. Ciriello , D.M. Tartakovsky
Computationally inexpensive surrogates of process-based models, such as deep neural networks, enable ensemble-based computations used in risk assessment, data assimilation, etc. However, generation of large datasets required to train a neural network can be as expensive as the ensemble simulations themselves. We ameliorate this challenge by using data from multifidelity (MF) groundwater simulations and transfer learning (TL) to reduce data generation costs while maintaining model accuracy. As a computational example, we train a deep convolutional neural network (CNN) to reconstruct permeability fields from saturation maps derived from a multiphase flow model. Starting with very low- and low-fidelity data generated on increasingly coarse meshes, we pretrain the CNN, followed by output-layer training and fine-tuning using only a limited number of high-fidelity samples. We demonstrate the surrogate’s robustness when interpreting low-quality inputs — such as interpolated maps or data affected by noise — which has strong implications for the applicability in practical hydrogeological scenarios. This multilevel MF-TL strategy achieves a favorable trade-off between computational efficiency and predictive accuracy, significantly outperforming high-fidelity-only approaches under the same computational budget.
{"title":"Transfer learning of neural surrogates on multifidelity groundwater simulations","authors":"A. Chiofalo , V. Ciriello , D.M. Tartakovsky","doi":"10.1016/j.advwatres.2025.105140","DOIUrl":"10.1016/j.advwatres.2025.105140","url":null,"abstract":"<div><div>Computationally inexpensive surrogates of process-based models, such as deep neural networks, enable ensemble-based computations used in risk assessment, data assimilation, etc. However, generation of large datasets required to train a neural network can be as expensive as the ensemble simulations themselves. We ameliorate this challenge by using data from multifidelity (MF) groundwater simulations and transfer learning (TL) to reduce data generation costs while maintaining model accuracy. As a computational example, we train a deep convolutional neural network (CNN) to reconstruct permeability fields from saturation maps derived from a multiphase flow model. Starting with very low- and low-fidelity data generated on increasingly coarse meshes, we pretrain the CNN, followed by output-layer training and fine-tuning using only a limited number of high-fidelity samples. We demonstrate the surrogate’s robustness when interpreting low-quality inputs — such as interpolated maps or data affected by noise — which has strong implications for the applicability in practical hydrogeological scenarios. This multilevel MF-TL strategy achieves a favorable trade-off between computational efficiency and predictive accuracy, significantly outperforming high-fidelity-only approaches under the same computational budget.</div></div>","PeriodicalId":7614,"journal":{"name":"Advances in Water Resources","volume":"206 ","pages":"Article 105140"},"PeriodicalIF":4.2,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145261759","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-01Epub Date: 2025-10-06DOI: 10.1016/j.advwatres.2025.105136
Anoop Pandey, Richa Ojha
Quantifying flow through macropores is challenging due to their discrete and heterogeneous distribution in soil. Many flow theories exist for modelling flow through macropores based on Darcian and non-Darcian approaches. However, not much is known about their applicability and performance under various geometrical characteristics of macropores. This study numerically examines the ability of four theoretically relevant models: the single-porosity model (SPM), dual-permeability model (DPM), coupled Richards equation with laminar flow in macropores (CRL), and coupled Richards equation with thin-film flow along macropores (CRTF), in capturing water flux, pressure distribution, and related hydrological responses across diverse macropore geometries under different boundary conditions. Five representative scenarios (S-1 to S-5) were formulated based on field-observed macropore characteristics, considering variations in density, size, distribution, shape, and connectivity. The details related to geometry, parameter values, initial and boundary conditions were obtained from various sources in the literature. Two-dimensional numerical analyses were performed within the COMSOL Multiphysics® environment, leveraging the Richards equation interface. Findings indicate that model selection is critically dependent on the specific hydrological variable of interest. The CRL and CRTF models reliably capture soil-moisture distribution and lateral mass exchange, whereas the DPM adequately estimates the total outflux. Notably, the CRTF model consistently yields the most accurate predictions for bottom outflux and velocity, values ranging from 0.1 to 1 mm/s, which closely align with observed field data. Although its performance reduces in scenarios characterized by poorly connected macropores that impede mass exchange. The SPM exhibits low performance in S-4 (related to shape and curvature) with an average deviation of 75–80 % between CRL and SPM. This study highlights the critical need for careful model selection based on the specific structural features of macropore networks.
{"title":"Evaluating existing flow theories for modelling macropore flow through unsaturated soils: A numerical study","authors":"Anoop Pandey, Richa Ojha","doi":"10.1016/j.advwatres.2025.105136","DOIUrl":"10.1016/j.advwatres.2025.105136","url":null,"abstract":"<div><div>Quantifying flow through macropores is challenging due to their discrete and heterogeneous distribution in soil. Many flow theories exist for modelling flow through macropores based on Darcian and non-Darcian approaches. However, not much is known about their applicability and performance under various geometrical characteristics of macropores. This study numerically examines the ability of four theoretically relevant models: the single-porosity model (SPM), dual-permeability model (DPM), coupled Richards equation with laminar flow in macropores (CRL), and coupled Richards equation with thin-film flow along macropores (CRTF), in capturing water flux, pressure distribution, and related hydrological responses across diverse macropore geometries under different boundary conditions. Five representative scenarios (S-1 to S-5) were formulated based on field-observed macropore characteristics, considering variations in density, size, distribution, shape, and connectivity. The details related to geometry, parameter values, initial and boundary conditions were obtained from various sources in the literature. Two-dimensional numerical analyses were performed within the COMSOL Multiphysics® environment, leveraging the Richards equation interface. Findings indicate that model selection is critically dependent on the specific hydrological variable of interest. The CRL and CRTF models reliably capture soil-moisture distribution and lateral mass exchange, whereas the DPM adequately estimates the total outflux. Notably, the CRTF model consistently yields the most accurate predictions for bottom outflux and velocity, values ranging from 0.1 to 1 mm/s, which closely align with observed field data. Although its performance reduces in scenarios characterized by poorly connected macropores that impede mass exchange. The SPM exhibits low performance in S-4 (related to shape and curvature) with an average deviation of 75–80 % between CRL and SPM. This study highlights the critical need for careful model selection based on the specific structural features of macropore networks.</div></div>","PeriodicalId":7614,"journal":{"name":"Advances in Water Resources","volume":"206 ","pages":"Article 105136"},"PeriodicalIF":4.2,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145324349","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-12-01Epub Date: 2025-11-25DOI: 10.1016/j.advwatres.2025.105178
Raimund Bürger , Enrique D. Fernández-Nieto , Jose Garres-Díaz , Jorge Moya
The Saint-Venant–Exner (SVE) model is widely used for the description of sediment transport including bedload, erosion, and deposition processes. A modified version of the SVE model, which includes sediment concentration, incorporates exchange of sediment between the fluid and an erodible bed and a non-hydrostatic pressure for the fluid along with non-equilibrium entrainment and deposition velocities, is introduced. Gravitational effects on erosion are described by an effective shear stress formulation. This modified SVE model is derived from a general approach with density variations. It preserves the mass of both the sediment and the fluid, and satisfies a dissipative energy balance. On the other hand, well-balanced finite volume schemes adapted for SVE models are derived since standard well-balanced schemes for the Saint-Venant system with fixed bottom are in general no more well-balanced when applied to the SVE model. The latter property is due to the uncontrolled numerical diffusion associated with the bed evolution equation. Two novel techniques to achieve the well-balanced property for the modified SVE model are proposed. The first is a new polynomial-viscosity-matrix-based (PVM) scheme, denoted “PVM-2I”, that modifies the numerical approximation of the bed evolution equation according to its related characteristic speed. The second is a physically motivated correction of the numerical diffusion term for the Rusanov and Harten–Lax–van Leer (HLL) schemes. The proposed schemes are positivity-preserving for the water height. Numerical solutions are compared with exact solutions with gravitational effects, with a novel exact solution in non-equilibrium conditions, and with experimental data. It is illustrated how the use of standard non-well-balanced schemes leads to a large artificial (unphysical) erosion and completely degraded solutions. This undesirable behaviour is avoided by the proposed well-balanced schemes. Moreover, it is demonstrated that for dam-break flows the inclusion of non-hydrostatic pressure improves the prediction of the water surface and sediment evolution, while for overtopping flow erosion tests, accounting for erosion–deposition exchanges between the bedload and suspended sediment layers leads to better agreement with experimental data.
{"title":"Well-balanced physics-based finite volume schemes for Saint-Venant–Exner-type models of sediment transport","authors":"Raimund Bürger , Enrique D. Fernández-Nieto , Jose Garres-Díaz , Jorge Moya","doi":"10.1016/j.advwatres.2025.105178","DOIUrl":"10.1016/j.advwatres.2025.105178","url":null,"abstract":"<div><div>The Saint-Venant–Exner (SVE) model is widely used for the description of sediment transport including bedload, erosion, and deposition processes. A modified version of the SVE model, which includes sediment concentration, incorporates exchange of sediment between the fluid and an erodible bed and a non-hydrostatic pressure for the fluid along with non-equilibrium entrainment and deposition velocities, is introduced. Gravitational effects on erosion are described by an effective shear stress formulation. This modified SVE model is derived from a general approach with density variations. It preserves the mass of both the sediment and the fluid, and satisfies a dissipative energy balance. On the other hand, well-balanced finite volume schemes adapted for SVE models are derived since standard well-balanced schemes for the Saint-Venant system with fixed bottom are in general no more well-balanced when applied to the SVE model. The latter property is due to the uncontrolled numerical diffusion associated with the bed evolution equation. Two novel techniques to achieve the well-balanced property for the modified SVE model are proposed. The first is a new polynomial-viscosity-matrix-based (PVM) scheme, denoted “PVM-2I”, that modifies the numerical approximation of the bed evolution equation according to its related characteristic speed. The second is a physically motivated correction of the numerical diffusion term for the Rusanov and Harten–Lax–van Leer (HLL) schemes. The proposed schemes are positivity-preserving for the water height. Numerical solutions are compared with exact solutions with gravitational effects, with a novel exact solution in non-equilibrium conditions, and with experimental data. It is illustrated how the use of standard non-well-balanced schemes leads to a large artificial (unphysical) erosion and completely degraded solutions. This undesirable behaviour is avoided by the proposed well-balanced schemes. Moreover, it is demonstrated that for dam-break flows the inclusion of non-hydrostatic pressure improves the prediction of the water surface and sediment evolution, while for overtopping flow erosion tests, accounting for erosion–deposition exchanges between the bedload and suspended sediment layers leads to better agreement with experimental data.</div></div>","PeriodicalId":7614,"journal":{"name":"Advances in Water Resources","volume":"206 ","pages":"Article 105178"},"PeriodicalIF":4.2,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145598733","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号