Christa Brelsford, Ethan T. Coon, Mark Wang, Nathanael Rosenheim, Nicholas Brake, Liv Haselbach, Paola Passalacqua
In coupled human and natural systems, developing an observation strategy which maximizes insight into both the natural system and the human system is a challenging multi-objective optimization problem. In this article, we describe the expansion of a flood risk observation system in Southeast Texas designed to improve our understanding of both physical and socioeconomic exposure to hydrological hazards at fine spatial scales, in the context of a structured hazard-exposure-vulnerability risk framework. We describe a new approach for assessing the spatial extent through which a flood sensor's observations can be assumed to be relevant, and estimate the population served within each sensor's area of information using downscaled socio-demographic data. As hydrological observations and modeling move to ever finer scale, assessing the information they contain in the context of both social and natural systems becomes increasingly important for developing actionable scientific insights.
{"title":"Multi-Objective Urban Observational Strategies: A Risk-Based Framework for Expanding Flood Sensor Networks","authors":"Christa Brelsford, Ethan T. Coon, Mark Wang, Nathanael Rosenheim, Nicholas Brake, Liv Haselbach, Paola Passalacqua","doi":"10.1029/2025wr041135","DOIUrl":"https://doi.org/10.1029/2025wr041135","url":null,"abstract":"In coupled human and natural systems, developing an observation strategy which maximizes insight into both the natural system and the human system is a challenging multi-objective optimization problem. In this article, we describe the expansion of a flood risk observation system in Southeast Texas designed to improve our understanding of both physical and socioeconomic exposure to hydrological hazards at fine spatial scales, in the context of a structured hazard-exposure-vulnerability risk framework. We describe a new approach for assessing the spatial extent through which a flood sensor's observations can be assumed to be relevant, and estimate the population served within each sensor's area of information using downscaled socio-demographic data. As hydrological observations and modeling move to ever finer scale, assessing the information they contain in the context of both social and natural systems becomes increasingly important for developing actionable scientific insights.","PeriodicalId":23799,"journal":{"name":"Water Resources Research","volume":"54 1","pages":""},"PeriodicalIF":5.4,"publicationDate":"2026-01-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146042628","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}
Peiyuan Sun, Chao Deng, Yicong Dai, Xin Yin, Hongbin Li
With the advancement of deep neural networks and physics-data-driven coupling methods, significant new opportunities have emerged for improving watershed hydrological simulations. Attention mechanisms (AM) and physical process-wrapped recurrent neural networks (PRNN) enhance model performance by dynamically highlighting key hydrological features and integrating physical concepts to learn complex processes. However, in PRNN models where runoff relies on storage-driven hydrological processes, the potential of employing AM to quantify and interpret respective contributions of aquifer-delineated storage layers remains underexplored. This study addresses this gap by integrating an AM into a PRNN based on the Hydrologiska Byråns Vattenbalansavdelning, yielding a diagnostic framework to assess whether the model's internal representations are consistent with established hydrological principles. Validated on 451 catchment attributes and meteorology for large-sample studies basins, the proposed PRNN-θaf-A variant yields a median Nash-Sutcliffe efficiency of 0.72. The diagnostic analysis reveals two key findings: (a) the soil moisture layer (S3) is the dominant contributor to runoff generation, accounting for a mean annual attention weight of 0.53; (b) the model's seasonal behavior is systematically linked to basin climate, demonstrated by a strong correlation (R2 = 0.68) between the seasonal attention shift for snowpack (S1) and the basin snow fraction. These results illustrate the potential of the framework to produce internal representations that are physically plausible and consistent with established hydrological principles, thereby bridging the gap between process-based understanding and data-driven modeling.
{"title":"Integrating Physical Parameterization and Attention Mechanisms in Recurrent Neural Networks for Hydrological Modeling: Quantification of Storage Layers Dynamics and Meteorological Responses Within the PRNN Model Framework","authors":"Peiyuan Sun, Chao Deng, Yicong Dai, Xin Yin, Hongbin Li","doi":"10.1029/2024wr039800","DOIUrl":"https://doi.org/10.1029/2024wr039800","url":null,"abstract":"With the advancement of deep neural networks and physics-data-driven coupling methods, significant new opportunities have emerged for improving watershed hydrological simulations. Attention mechanisms (AM) and physical process-wrapped recurrent neural networks (PRNN) enhance model performance by dynamically highlighting key hydrological features and integrating physical concepts to learn complex processes. However, in PRNN models where runoff relies on storage-driven hydrological processes, the potential of employing AM to quantify and interpret respective contributions of aquifer-delineated storage layers remains underexplored. This study addresses this gap by integrating an AM into a PRNN based on the Hydrologiska Byråns Vattenbalansavdelning, yielding a diagnostic framework to assess whether the model's internal representations are consistent with established hydrological principles. Validated on 451 catchment attributes and meteorology for large-sample studies basins, the proposed PRNN-θaf-A variant yields a median Nash-Sutcliffe efficiency of 0.72. The diagnostic analysis reveals two key findings: (a) the soil moisture layer (S3) is the dominant contributor to runoff generation, accounting for a mean annual attention weight of 0.53; (b) the model's seasonal behavior is systematically linked to basin climate, demonstrated by a strong correlation (<i>R</i><sup>2</sup> = 0.68) between the seasonal attention shift for snowpack (S1) and the basin snow fraction. These results illustrate the potential of the framework to produce internal representations that are physically plausible and consistent with established hydrological principles, thereby bridging the gap between process-based understanding and data-driven modeling.","PeriodicalId":23799,"journal":{"name":"Water Resources Research","volume":"13 1","pages":""},"PeriodicalIF":5.4,"publicationDate":"2026-01-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146034167","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}
Rainwater resources play a vital role in water resources balance, particularly in water-scarce regions. Regulating rainwater resources can help mitigate water supply-demand seasonal mismatch and support sustainable revegetation. However, the influence of revegetation on the rainwater resources use potential (RWUP)—comprising effective soil moisture and surface runoff—through land-atmosphere feedbacks remains uncertain. Here, we quantify the variations in the mismatch between water supply and demand as represented by precipitation (<i>P</i>) and the climatically appropriate for existing conditions P (<span data-altimg="/cms/asset/174f6d49-545c-4804-be31-e07951c954c3/wrcr70661-math-0001.png"></span><mjx-container ctxtmenu_counter="41" ctxtmenu_oldtabindex="1" jax="CHTML" role="application" sre-explorer- style="font-size: 103%; position: relative;" tabindex="0"><mjx-math aria-hidden="true" location="graphic/wrcr70661-math-0001.png"><mjx-semantics><mjx-mrow><mjx-mover data-semantic-children="0,1" data-semantic- data-semantic-role="latinletter" data-semantic-speech="ModifyingAbove upper P With ˆ" data-semantic-type="overscore"><mjx-over style="padding-bottom: 0.105em; padding-left: 0.486em; margin-bottom: -0.551em;"><mjx-mo data-semantic- data-semantic-parent="2" data-semantic-role="overaccent" data-semantic-type="operator" style="width: 0px; margin-left: -0.278em;"><mjx-c></mjx-c></mjx-mo></mjx-over><mjx-base><mjx-mi data-semantic-annotation="clearspeak:simple" data-semantic-font="italic" data-semantic- data-semantic-parent="2" data-semantic-role="latinletter" data-semantic-type="identifier"><mjx-c></mjx-c></mjx-mi></mjx-base></mjx-mover></mjx-mrow></mjx-semantics></mjx-math><mjx-assistive-mml display="inline" unselectable="on"><math altimg="urn:x-wiley:00431397:media:wrcr70661:wrcr70661-math-0001" display="inline" location="graphic/wrcr70661-math-0001.png" xmlns="http://www.w3.org/1998/Math/MathML"><semantics><mrow><mover accent="true" data-semantic-="" data-semantic-children="0,1" data-semantic-role="latinletter" data-semantic-speech="ModifyingAbove upper P With ˆ" data-semantic-type="overscore"><mi data-semantic-="" data-semantic-annotation="clearspeak:simple" data-semantic-font="italic" data-semantic-parent="2" data-semantic-role="latinletter" data-semantic-type="identifier">P</mi><mo data-semantic-="" data-semantic-parent="2" data-semantic-role="overaccent" data-semantic-type="operator">ˆ</mo></mover></mrow>$widehat{P}$</annotation></semantics></math></mjx-assistive-mml></mjx-container>) in the Loess Plateau from 1999 to 2018. We examine the spatiotemporal patterns of RWUP by employing a coupled land-atmosphere model with scenario experiments, and reveal the mechanisms through which revegetation influences water resources balance. We quantify the maximum vegetation carrying capacity (MVC) under optimal RWUP regulation. Our findings show that revegetation could increase both <i>P</i> and <span data-altimg="/cms/asset/43a43458-c21d-49ed-9ec6-77a9e
{"title":"Revegetation Rebalances Water Resources by Enhancing Rainwater to Increase Vegetation Carrying Capacity in China's Loess Plateau","authors":"Jihui Zhang, Baoqing Zhang, Xuejin Wang, Furong Yang, Xining Zhao, Yanyan Cheng","doi":"10.1029/2025wr040307","DOIUrl":"https://doi.org/10.1029/2025wr040307","url":null,"abstract":"Rainwater resources play a vital role in water resources balance, particularly in water-scarce regions. Regulating rainwater resources can help mitigate water supply-demand seasonal mismatch and support sustainable revegetation. However, the influence of revegetation on the rainwater resources use potential (RWUP)—comprising effective soil moisture and surface runoff—through land-atmosphere feedbacks remains uncertain. Here, we quantify the variations in the mismatch between water supply and demand as represented by precipitation (<i>P</i>) and the climatically appropriate for existing conditions P (<span data-altimg=\"/cms/asset/174f6d49-545c-4804-be31-e07951c954c3/wrcr70661-math-0001.png\"></span><mjx-container ctxtmenu_counter=\"41\" ctxtmenu_oldtabindex=\"1\" jax=\"CHTML\" role=\"application\" sre-explorer- style=\"font-size: 103%; position: relative;\" tabindex=\"0\"><mjx-math aria-hidden=\"true\" location=\"graphic/wrcr70661-math-0001.png\"><mjx-semantics><mjx-mrow><mjx-mover data-semantic-children=\"0,1\" data-semantic- data-semantic-role=\"latinletter\" data-semantic-speech=\"ModifyingAbove upper P With ˆ\" data-semantic-type=\"overscore\"><mjx-over style=\"padding-bottom: 0.105em; padding-left: 0.486em; margin-bottom: -0.551em;\"><mjx-mo data-semantic- data-semantic-parent=\"2\" data-semantic-role=\"overaccent\" data-semantic-type=\"operator\" style=\"width: 0px; margin-left: -0.278em;\"><mjx-c></mjx-c></mjx-mo></mjx-over><mjx-base><mjx-mi data-semantic-annotation=\"clearspeak:simple\" data-semantic-font=\"italic\" data-semantic- data-semantic-parent=\"2\" data-semantic-role=\"latinletter\" data-semantic-type=\"identifier\"><mjx-c></mjx-c></mjx-mi></mjx-base></mjx-mover></mjx-mrow></mjx-semantics></mjx-math><mjx-assistive-mml display=\"inline\" unselectable=\"on\"><math altimg=\"urn:x-wiley:00431397:media:wrcr70661:wrcr70661-math-0001\" display=\"inline\" location=\"graphic/wrcr70661-math-0001.png\" xmlns=\"http://www.w3.org/1998/Math/MathML\"><semantics><mrow><mover accent=\"true\" data-semantic-=\"\" data-semantic-children=\"0,1\" data-semantic-role=\"latinletter\" data-semantic-speech=\"ModifyingAbove upper P With ˆ\" data-semantic-type=\"overscore\"><mi data-semantic-=\"\" data-semantic-annotation=\"clearspeak:simple\" data-semantic-font=\"italic\" data-semantic-parent=\"2\" data-semantic-role=\"latinletter\" data-semantic-type=\"identifier\">P</mi><mo data-semantic-=\"\" data-semantic-parent=\"2\" data-semantic-role=\"overaccent\" data-semantic-type=\"operator\">ˆ</mo></mover></mrow>$widehat{P}$</annotation></semantics></math></mjx-assistive-mml></mjx-container>) in the Loess Plateau from 1999 to 2018. We examine the spatiotemporal patterns of RWUP by employing a coupled land-atmosphere model with scenario experiments, and reveal the mechanisms through which revegetation influences water resources balance. We quantify the maximum vegetation carrying capacity (MVC) under optimal RWUP regulation. Our findings show that revegetation could increase both <i>P</i> and <span data-altimg=\"/cms/asset/43a43458-c21d-49ed-9ec6-77a9e","PeriodicalId":23799,"journal":{"name":"Water Resources Research","volume":"54 1","pages":""},"PeriodicalIF":5.4,"publicationDate":"2026-01-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146034168","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}
Process-based land surface models (LSMs) are widely used for global water cycle and runoff assessments, but when integrated with hydrodynamic models, the streamflow simulations exhibit significant uncertainties in uncalibrated mode, limiting their effectiveness in local hydrology applications. The calibration of LSMs against observed streamflow across large basins and regions is computationally prohibitive and sometimes degrades performance of other variables. In contrast, deep learning models, particularly Long-Short Term Memory (LSTM) networks, have shown promising results in streamflow simulations, but they are often limited by poor reproducibility of other water cycle variables. This study presents a hybrid modeling framework that integrates process-based models with deep learning to improve daily streamflow simulations without requiring basin-specific calibration. The framework is showcased on a national scale using a multi-model hydrologic ensemble from the Indian Land Data Assimilation System (ILDAS). It is integrated with a proposed two-stage post-processor, which pairs a residual error prediction LSTM with an auto-regressive meta-learning LSTM to predict 1-day ahead streamflow. Trained on multi-decadal data from 220 catchments across India, the framework improves Kling-Gupta Efficiency in 208 catchments, raising the national median from 0.18 (uncalibrated) to 0.60. It also reduced peak flow timing error and peak mean absolute percentage error by 25% in 135 catchments. During monsoon and post-monsoon periods, residual error interquartile range (IQR) decreased by 66.3% and 81.7%, respectively. This approach has the potential to integrate LSMs with deep learning for more accurate and locally relevant streamflow predictions, while enhancing other water cycle variables through methods like data assimilation.
{"title":"Locally Relevant Streamflow by Integrating a Land Surface Model Ensemble With a Two-Stage LSTM Post-Processor","authors":"Bhanu Magotra, Manabendra Saharia","doi":"10.1029/2024wr039792","DOIUrl":"https://doi.org/10.1029/2024wr039792","url":null,"abstract":"Process-based land surface models (LSMs) are widely used for global water cycle and runoff assessments, but when integrated with hydrodynamic models, the streamflow simulations exhibit significant uncertainties in uncalibrated mode, limiting their effectiveness in local hydrology applications. The calibration of LSMs against observed streamflow across large basins and regions is computationally prohibitive and sometimes degrades performance of other variables. In contrast, deep learning models, particularly Long-Short Term Memory (LSTM) networks, have shown promising results in streamflow simulations, but they are often limited by poor reproducibility of other water cycle variables. This study presents a hybrid modeling framework that integrates process-based models with deep learning to improve daily streamflow simulations without requiring basin-specific calibration. The framework is showcased on a national scale using a multi-model hydrologic ensemble from the Indian Land Data Assimilation System (ILDAS). It is integrated with a proposed two-stage post-processor, which pairs a residual error prediction LSTM with an auto-regressive meta-learning LSTM to predict 1-day ahead streamflow. Trained on multi-decadal data from 220 catchments across India, the framework improves Kling-Gupta Efficiency in 208 catchments, raising the national median from 0.18 (uncalibrated) to 0.60. It also reduced peak flow timing error and peak mean absolute percentage error by 25% in 135 catchments. During monsoon and post-monsoon periods, residual error interquartile range (IQR) decreased by 66.3% and 81.7%, respectively. This approach has the potential to integrate LSMs with deep learning for more accurate and locally relevant streamflow predictions, while enhancing other water cycle variables through methods like data assimilation.","PeriodicalId":23799,"journal":{"name":"Water Resources Research","volume":"179 1","pages":""},"PeriodicalIF":5.4,"publicationDate":"2026-01-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146021781","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}
Matthew Tiller, Lucy Reading, Marc Miska, Prasanna Egodawatta
Rainfall infiltration is a key hydrological process influencing agriculture, pollutant transport, and flood modeling. Accurate prediction of rainfall losses, defined as rainfall that does not contribute to surface runoff is critical in rainfall-runoff models. Rainfall–runoff models are typically calibrated using historical data to estimate loss parameters, which often deviate from physically realistic infiltration behavior as they compensate for other sources of error and uncertainty in the model. This study addresses this gap by investigating infiltration losses based on physical attributes under controlled rainfall conditions. Seventy-five sites in southeastern Queensland, Australia, were subjected to rainfall of approximately 60 mm/hr for 1-hr, allowing detailed analysis of infiltration responses. Key predictors of infiltration included grass cover, leaf litter, soil organic carbon, and bulk density, while slope had minimal predictive power. Findings indicate that, during short, high-intensity rainfall events, initial losses were relatively low, with runoff beginning within 10–30 min, while continuing loss rates exceeded expectations within the first hour. Multiple Linear Regression (MLR) techniques were used to develop prediction equations for several loss models, including lumped loss, initial loss–continuing loss, and Horton infiltration. These equations explained approximately 60% of the variance between observed and predicted losses. The equations provide a practical tool for estimating infiltration losses in ungauged catchments. The prediction equations are suitable for 1-hr, 60 mm/hr intensity rainfall events, with limited applicability to longer, low-intensity rainfall. The results offer insights for improving flash flood predictions, particularly in ungauged catchments experiencing intense, short-duration storms.
{"title":"Predicting Rainfall Infiltration Losses: A Rainfall Simulation Study of Land Cover, Slope and Soil Type","authors":"Matthew Tiller, Lucy Reading, Marc Miska, Prasanna Egodawatta","doi":"10.1029/2025wr040920","DOIUrl":"https://doi.org/10.1029/2025wr040920","url":null,"abstract":"Rainfall infiltration is a key hydrological process influencing agriculture, pollutant transport, and flood modeling. Accurate prediction of rainfall losses, defined as rainfall that does not contribute to surface runoff is critical in rainfall-runoff models. Rainfall–runoff models are typically calibrated using historical data to estimate loss parameters, which often deviate from physically realistic infiltration behavior as they compensate for other sources of error and uncertainty in the model. This study addresses this gap by investigating infiltration losses based on physical attributes under controlled rainfall conditions. Seventy-five sites in southeastern Queensland, Australia, were subjected to rainfall of approximately 60 mm/hr for 1-hr, allowing detailed analysis of infiltration responses. Key predictors of infiltration included grass cover, leaf litter, soil organic carbon, and bulk density, while slope had minimal predictive power. Findings indicate that, during short, high-intensity rainfall events, initial losses were relatively low, with runoff beginning within 10–30 min, while continuing loss rates exceeded expectations within the first hour. Multiple Linear Regression (MLR) techniques were used to develop prediction equations for several loss models, including lumped loss, initial loss–continuing loss, and Horton infiltration. These equations explained approximately 60% of the variance between observed and predicted losses. The equations provide a practical tool for estimating infiltration losses in ungauged catchments. The prediction equations are suitable for 1-hr, 60 mm/hr intensity rainfall events, with limited applicability to longer, low-intensity rainfall. The results offer insights for improving flash flood predictions, particularly in ungauged catchments experiencing intense, short-duration storms.","PeriodicalId":23799,"journal":{"name":"Water Resources Research","volume":"284 1","pages":""},"PeriodicalIF":5.4,"publicationDate":"2026-01-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146042631","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}
Zi Wu, Jie Zhan, Zhenduo Zhu, Man Zhang, Lu Chang, Xudong Fu
Drifting fish eggs are a type of fish egg with a slightly higher density than water, requiring floating for successful hatching. While it is acknowledged that interaction with the riverbed surface can cause mortality of the eggs, the impact of this process on their downstream transport remains unclear. In this paper, we theoretically explore the transport of drifting fish eggs in turbulent open channel flows, taking into account both gravitational settling and riverbed mortality effects. This is done by incorporating a vertical drift term in the governing advection-diffusion equation and an absorbing boundary condition for the riverbed surface, respectively. For the first time, we derive an analytical solution by the method of separation of variables for the vertical distribution of eggs during transport. Our analysis shows that in principle, settling can lead to egg accumulation near the riverbed, reducing the population's mean velocity, while conversely, riverbed mortality can decrease near-bed accumulation and accelerate drifting to some extent. However, by estimating values of the mortality rate parameter in the real rivers, we conclude that while it can significantly affect the population size, it has a negligible effect on the vertical concentration distribution in practice, allowing for a considerable simplification of the analytical solution. Furthermore, we deduce an analytical solution for the mean velocity of the egg population, indicating variations of the deceleration rate compared to mean flow velocity, which is capable of assisting in the identification of spawning grounds. The obtained analytical solutions are validated by various numerical and experimental results.
{"title":"Effects of Gravitational Settling and Riverbed-Induced Mortality on the Transport of Drifting Fish Eggs in Rivers","authors":"Zi Wu, Jie Zhan, Zhenduo Zhu, Man Zhang, Lu Chang, Xudong Fu","doi":"10.1029/2025wr041343","DOIUrl":"https://doi.org/10.1029/2025wr041343","url":null,"abstract":"Drifting fish eggs are a type of fish egg with a slightly higher density than water, requiring floating for successful hatching. While it is acknowledged that interaction with the riverbed surface can cause mortality of the eggs, the impact of this process on their downstream transport remains unclear. In this paper, we theoretically explore the transport of drifting fish eggs in turbulent open channel flows, taking into account both gravitational settling and riverbed mortality effects. This is done by incorporating a vertical drift term in the governing advection-diffusion equation and an absorbing boundary condition for the riverbed surface, respectively. For the first time, we derive an analytical solution by the method of separation of variables for the vertical distribution of eggs during transport. Our analysis shows that in principle, settling can lead to egg accumulation near the riverbed, reducing the population's mean velocity, while conversely, riverbed mortality can decrease near-bed accumulation and accelerate drifting to some extent. However, by estimating values of the mortality rate parameter in the real rivers, we conclude that while it can significantly affect the population size, it has a negligible effect on the vertical concentration distribution in practice, allowing for a considerable simplification of the analytical solution. Furthermore, we deduce an analytical solution for the mean velocity of the egg population, indicating variations of the deceleration rate compared to mean flow velocity, which is capable of assisting in the identification of spawning grounds. The obtained analytical solutions are validated by various numerical and experimental results.","PeriodicalId":23799,"journal":{"name":"Water Resources Research","volume":"4 1","pages":""},"PeriodicalIF":5.4,"publicationDate":"2026-01-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146034139","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}
Marie-Madeleine Stettler, Marco Dentz, Olaf A. Cirpka
Spatial Markov models (SMM) are an efficient approach to simulate transport in heterogeneous media across scales. They represent particle transport by equidistant spatial transitions with correlated random velocities, which renders the associated transition times correlated random variables. While SMM perform excellently for modeling purely advective transport, incorporating diffusion remains challenging. So far, applying SMM to advective-diffusive transport in porous media has been mostly restricted to using empirical transition matrices based on numerical simulations. Using a transition matrix for advective-diffusive transport obliterates the fundamental differences between the two transport processes and is conflicting with the goal of replacing explicit transport simulations by a SMM. Here, we present an advective-diffusive SMM that conceptualizes diffusion as jumps between advective trajectories, that is, diffusion competes with advection for changing the particle velocity. At each particle transition, a random diffusion time is compared to the current advection time. If the advection time is shorter than the diffusion time, the particle remains on its current SMM trajectory and the longitudinal velocity correlation is kept. If the diffusion time is shorter, the particle velocity is reset. Breakthrough curves and their first and second moments calculated with the advective-diffusive SMM are in agreement with three-dimensional numerical simulations in heterogeneous log-conductivity fields with isotropic, exponential covariance function with variances up to five.
{"title":"Spatial Markov Model of Advective-Diffusive Transport in Heterogeneous Domains","authors":"Marie-Madeleine Stettler, Marco Dentz, Olaf A. Cirpka","doi":"10.1029/2025wr041175","DOIUrl":"https://doi.org/10.1029/2025wr041175","url":null,"abstract":"Spatial Markov models (SMM) are an efficient approach to simulate transport in heterogeneous media across scales. They represent particle transport by equidistant spatial transitions with correlated random velocities, which renders the associated transition times correlated random variables. While SMM perform excellently for modeling purely advective transport, incorporating diffusion remains challenging. So far, applying SMM to advective-diffusive transport in porous media has been mostly restricted to using empirical transition matrices based on numerical simulations. Using a transition matrix for advective-diffusive transport obliterates the fundamental differences between the two transport processes and is conflicting with the goal of replacing explicit transport simulations by a SMM. Here, we present an advective-diffusive SMM that conceptualizes diffusion as jumps between advective trajectories, that is, diffusion competes with advection for changing the particle velocity. At each particle transition, a random diffusion time is compared to the current advection time. If the advection time is shorter than the diffusion time, the particle remains on its current SMM trajectory and the longitudinal velocity correlation is kept. If the diffusion time is shorter, the particle velocity is reset. Breakthrough curves and their first and second moments calculated with the advective-diffusive SMM are in agreement with three-dimensional numerical simulations in heterogeneous log-conductivity fields with isotropic, exponential covariance function with variances up to five.","PeriodicalId":23799,"journal":{"name":"Water Resources Research","volume":"31 1","pages":""},"PeriodicalIF":5.4,"publicationDate":"2026-01-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146034150","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}
Small (<10 ha) and medium (10–100 ha) inland waterbodies are crucial for water storage and flood regulation, necessitating an improved understanding of their water volume. Traditional water volume measurement methods, modeling techniques, and advanced altimetry missions struggle to capture the non-linear volume changes in these waterbodies, leading to inaccurate and temporally sparse volume estimates. Using 86 in situ bathymetry and spatio-temporal water spread area data, the study addresses this gap by developing a machine learning model that estimates monthly volume changes (1988–2023) in 914 waterbodies of the Adyar-Chennai basin, India. The machine learning model demonstrated superior performance (R2 = 0.94), outperforming global (R2 = 0.57) and regional models (R2 = 0.24). The water volume of small and medium waterbodies in the Adyar-Chennai basin nearly halved from ∼102.28 (95% CI: 93.38–114.28) million cubic meters (MCM) in January 1988 to ∼40.13 (32.25–61.42) MCM in December 2023, which is possibly driven by urbanization, vegetation loss, and increasing vapor pressure. The first-ever future volumes were projected for these waterbodies (R2 = 0.62). While analyzing small and medium waterbodies' flood mitigation potential, the peak flood rate in the basin increased by 50% in their absence, highlighting their crucial role in flood control. To completely mitigate floods in the basin, we propose (a) creating 90 new waterbodies and (b) deepening existing waterbodies by 1 m. Suitable sites for creating new waterbodies were identified using hydrology and a land-use-based novel tankshed overflow index. This research advances water volume estimation and flood mitigation strategies of small and medium inland waterbodies.
{"title":"Small and Medium-Sized Inland Waterbodies: Water Volume Predictions and Flood Implications","authors":"Ankit Sharma, Idhayachandhiran Ilampooranan","doi":"10.1029/2024wr038283","DOIUrl":"https://doi.org/10.1029/2024wr038283","url":null,"abstract":"Small (<10 ha) and medium (10–100 ha) inland waterbodies are crucial for water storage and flood regulation, necessitating an improved understanding of their water volume. Traditional water volume measurement methods, modeling techniques, and advanced altimetry missions struggle to capture the non-linear volume changes in these waterbodies, leading to inaccurate and temporally sparse volume estimates. Using 86 in situ bathymetry and spatio-temporal water spread area data, the study addresses this gap by developing a machine learning model that estimates monthly volume changes (1988–2023) in 914 waterbodies of the Adyar-Chennai basin, India. The machine learning model demonstrated superior performance (<i>R</i><sup>2</sup> = 0.94), outperforming global (<i>R</i><sup>2</sup> = 0.57) and regional models (<i>R</i><sup>2</sup> = 0.24). The water volume of small and medium waterbodies in the Adyar-Chennai basin nearly halved from ∼102.28 (95% CI: 93.38–114.28) million cubic meters (MCM) in January 1988 to ∼40.13 (32.25–61.42) MCM in December 2023, which is possibly driven by urbanization, vegetation loss, and increasing vapor pressure. The first-ever future volumes were projected for these waterbodies (<i>R</i><sup>2</sup> = 0.62). While analyzing small and medium waterbodies' flood mitigation potential, the peak flood rate in the basin increased by 50% in their absence, highlighting their crucial role in flood control. To completely mitigate floods in the basin, we propose (a) creating 90 new waterbodies and (b) deepening existing waterbodies by 1 m. Suitable sites for creating new waterbodies were identified using hydrology and a land-use-based novel tankshed overflow index. This research advances water volume estimation and flood mitigation strategies of small and medium inland waterbodies.","PeriodicalId":23799,"journal":{"name":"Water Resources Research","volume":"32 1","pages":""},"PeriodicalIF":5.4,"publicationDate":"2026-01-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146034152","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}
Zhongzheng Wang, Yixiang Gan, Jean-Michel Pereira, Scott McCue, Anna Herring, Emilie Sauret
Cyclic injection of immiscible fluids in porous media is a key process in applications such as carbon geosequestration and underground hydrogen storage, where understanding and predicting the residual trapping efficiency are critical. This study develops a modified invasion percolation algorithm incorporating a pore coupling coefficient to simulate quasi-static, cyclic fluid displacement in porous media. The coefficient captures the effects of pore-scale cooperative pore-filling mechanisms by modifying the capillary pore entry pressures based on the filling status of neighboring pores. Systematic simulations reveal that the displacement morphology and saturation hysteresis are strongly influenced by the pore coupling strength. Phase diagrams highlight regimes for which cyclic injections significantly enhance residual gas trapping. Results also establish connections between the pore coupling coefficient and physical parameters such as wettability and porosity. This work provides new insights into the pore-scale origins of saturation hysteresis and its implications for optimizing fluid injection strategies in subsurface applications.
{"title":"Saturation Hysteresis During Cyclic Injections of Immiscible Fluids in Porous Media: An Invasion Percolation Study","authors":"Zhongzheng Wang, Yixiang Gan, Jean-Michel Pereira, Scott McCue, Anna Herring, Emilie Sauret","doi":"10.1029/2025wr041271","DOIUrl":"https://doi.org/10.1029/2025wr041271","url":null,"abstract":"Cyclic injection of immiscible fluids in porous media is a key process in applications such as carbon geosequestration and underground hydrogen storage, where understanding and predicting the residual trapping efficiency are critical. This study develops a modified invasion percolation algorithm incorporating a pore coupling coefficient to simulate quasi-static, cyclic fluid displacement in porous media. The coefficient captures the effects of pore-scale cooperative pore-filling mechanisms by modifying the capillary pore entry pressures based on the filling status of neighboring pores. Systematic simulations reveal that the displacement morphology and saturation hysteresis are strongly influenced by the pore coupling strength. Phase diagrams highlight regimes for which cyclic injections significantly enhance residual gas trapping. Results also establish connections between the pore coupling coefficient and physical parameters such as wettability and porosity. This work provides new insights into the pore-scale origins of saturation hysteresis and its implications for optimizing fluid injection strategies in subsurface applications.","PeriodicalId":23799,"journal":{"name":"Water Resources Research","volume":"53 1","pages":""},"PeriodicalIF":5.4,"publicationDate":"2026-01-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146022043","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}
Machine learning models are highly potential to substitute computationally intensive numerical simulation models in saltwater intrusion (SWI) remediation optimization. However, uncertainty inherent in machine learning models can propagate through predictions into optimization, resulting in inaccurate solutions. Unlike deterministic modeling that ignores uncertainty with fixed outputs, this study proposes a computationally efficient mixed integer multiobjective stochastic optimization (MIMOSO) method, which uniquely bridges the gap between Bayesian multi-model uncertainty quantification and risk-aware decision-making. The method captures stochastic uncertainty propagation from model prediction to optimization by integrating with Bayesian model averaging (BMA). In contrast to traditional single-surrogate approaches, the proposed method incorporates multiple machine learning approaches to alleviate computational burden. The framework enables to derive optimal but robust extraction-injection strategies by considering various constraint-violation levels. Two conflicting goals are addressed: minimizing total extraction-injection and maximizing SWI remediation effect. Binary variables are introduced to control discrete operation states of the well system. The developed method is demonstrated in a “1,500-foot” sand aquifer located in Baton Rouge, USA. Results exhibit that Pareto optimal remediation strategies are identified with associated SWI risk levels. MIMOSO advances the field by simultaneously resolving computational bottlenecks through machine learning surrogates and rigorously propagating multi-source uncertainties via BMA. Compared to numerical simulation based optimization (≥2,000 hr), machine learning assisted model reduces computation time to 87 hr, achieving a 23-fold efficiency improvement. Three metrics (hypervolume, spacing, and maximum spread) validate superior performance regarding both convergence and diversity. The methodology provides a promising way for risk-aware real-world aquifer remediation design.
{"title":"A Computationally Efficient Stochastic Method for Quantifying the Effects of Multi-Surrogate Model Uncertainty on Saltwater Remediation Optimization","authors":"Yulu Huang, Jina Yin, Chunhui Lu","doi":"10.1029/2025wr041251","DOIUrl":"https://doi.org/10.1029/2025wr041251","url":null,"abstract":"Machine learning models are highly potential to substitute computationally intensive numerical simulation models in saltwater intrusion (SWI) remediation optimization. However, uncertainty inherent in machine learning models can propagate through predictions into optimization, resulting in inaccurate solutions. Unlike deterministic modeling that ignores uncertainty with fixed outputs, this study proposes a computationally efficient mixed integer multiobjective stochastic optimization (MIMOSO) method, which uniquely bridges the gap between Bayesian multi-model uncertainty quantification and risk-aware decision-making. The method captures stochastic uncertainty propagation from model prediction to optimization by integrating with Bayesian model averaging (BMA). In contrast to traditional single-surrogate approaches, the proposed method incorporates multiple machine learning approaches to alleviate computational burden. The framework enables to derive optimal but robust extraction-injection strategies by considering various constraint-violation levels. Two conflicting goals are addressed: minimizing total extraction-injection and maximizing SWI remediation effect. Binary variables are introduced to control discrete operation states of the well system. The developed method is demonstrated in a “1,500-foot” sand aquifer located in Baton Rouge, USA. Results exhibit that Pareto optimal remediation strategies are identified with associated SWI risk levels. MIMOSO advances the field by simultaneously resolving computational bottlenecks through machine learning surrogates and rigorously propagating multi-source uncertainties via BMA. Compared to numerical simulation based optimization (≥2,000 hr), machine learning assisted model reduces computation time to 87 hr, achieving a 23-fold efficiency improvement. Three metrics (hypervolume, spacing, and maximum spread) validate superior performance regarding both convergence and diversity. The methodology provides a promising way for risk-aware real-world aquifer remediation design.","PeriodicalId":23799,"journal":{"name":"Water Resources Research","volume":"13 1","pages":""},"PeriodicalIF":5.4,"publicationDate":"2026-01-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146021994","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}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：