Advances in Water Resources最新文献

{"title":"Integrating Hidden Markov and Multinomial models for hydrological drought prediction under nonstationarity","authors":"Marcus Suassuna Santos ,&nbsp;Louise J. Slater","doi":"10.1016/j.advwatres.2025.104974","DOIUrl":"10.1016/j.advwatres.2025.104974","url":null,"abstract":"<div><div>Understanding the drivers of drought variability is crucial for developing effective adaptation and management strategies. This study develops a two-step modelling approach to characterize and predict hydrological droughts in a nonstationary context. First, a multivariate Hidden Markov Model (HMM) is used to classify low-water level time series into Dry, Normal, and Wet years, identifying Dry years as hydrological droughts. Second, a Multinomial Logistic Regression model (MLR) is proposed to predict low-water level class transitions, incorporating external variables into the transition matrix estimates. Precipitation thresholds for annual minima are also derived, with uncertainties and sensitivities assessed via bootstrap resampling. Our framework was successfully applied to the Paraguay River basin (PRB), where long-term changes in hydrological variables are frequent. The HMM transition matrix reveals a long persistence of years in each water level class and an inhomogeneity between two periods (1901–1960 and 1961–2024). The second period exhibits more extended runs of wet, dry, and non-dry years, suggesting a change in the driving dynamics. A multi-annual hydrological drought lasting for 13 years (1961–1973) was identified, followed by a stretch of 46 years (1974–2019) with no droughts in the study area. Our simulations indicate that the 46-year period with no drought had only a 4 % probability of occurrence. Precipitation is the primary predictor of regime shifts, but the class transition probabilities and precipitation thresholds are non-homogeneous and conditional on the current low-water level regime. We identify precipitation thresholds for initiating transitions between Dry, Wet and Normal years, conditioned on the current water levels: in a normal year, precipitation below 1040 mm triggers a hydrological drought, while in a drought year, precipitation above 1180 mm triggers a return to normal conditions. The research advances nonstationary extreme event analysis by proposing an efficient new approach to estimate inhomogeneity in hydrological drought occurrence; identify long persistence of hydrological drought episodes and their associated probabilities; define precipitation thresholds; and reveal the importance of coupled drivers of low water level shifts.</div></div>","PeriodicalId":7614,"journal":{"name":"Advances in Water Resources","volume":"200 ","pages":"Article 104974"},"PeriodicalIF":4.0,"publicationDate":"2025-04-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143791914","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}

{"title":"Evaporation characteristics and salt deposition dynamics from a homogeneous porous medium consisting of mono-disperse glass beads under controlled IR heating from above","authors":"Shivani Chauhan,&nbsp;Navneet Kumar","doi":"10.1016/j.advwatres.2025.104964","DOIUrl":"10.1016/j.advwatres.2025.104964","url":null,"abstract":"<div><div>The present study experimentally investigated the evaporation-precipitation dynamics at sub-millimetric to millimetric scales in 1 M NaCl salt-DI water solution-based homogeneous porous media consisting of nearly mono-disperse glass beads (ranging from 0.10 to 2.50 mm) under controlled infrared heating from above, mimicking realistic field scenarios. Three diagnostic tools were employed simultaneously: mass loss measurements using a precision weighing balance, surface temperature measurements using an IR camera, and optical imaging (at times with fluorescein dye) for visualization purposes. In all the cases, salt precipitates around <span><math><mrow><mo>∼</mo><mn>0.97</mn></mrow></math></span> of saturation. Experimental results show a strong interdependence between evaporation and salt deposition, significantly influenced by the particle sizes. For 0.70–0.85 mm case, evaporation characteristics with and without salt were found to be similar. For 0.40–0.60 mm and 0.70–0.85 mm cases, a constant evaporation rate persisted for <span><math><mo>∼</mo></math></span>4–5 h at <span><math><mo>∼</mo></math></span>1,000W/m² of incident heat flux, deviating from the regular ‘No Salt’ nature under external heating. In finer particles (0.10–0.30 mm), rapid salt deposition led to a steep fall in evaporation rate, while in medium and coarser sizes, it initially increased linearly, then exponentially during Stage 1 due to emerging precipitation sites within interconnected voids forming distinct wet patch, visible in IR imaging. Despite variations, <span><math><mo>∼</mo></math></span>80 % of the exposed surface was covered during Stage 1 (except the 2.00–2.50 mm case due to weak capillary forces) in all the cases. Unlike pure water cases, the evaporative capillary length shows a non-linear trend with particle size with maximum value appearing for 0.70–0.85 mm case, interestingly similar to the value for the ‘No Salt’ case.</div></div>","PeriodicalId":7614,"journal":{"name":"Advances in Water Resources","volume":"201 ","pages":"Article 104964"},"PeriodicalIF":4.0,"publicationDate":"2025-04-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143851982","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}

{"title":"Péclet number and transport length dependences of dispersion and dispersivity coefficients during the transition to Fickian transport in homogeneous sands","authors":"Kuldeep Singh ,&nbsp;Victor Obi","doi":"10.1016/j.advwatres.2025.104975","DOIUrl":"10.1016/j.advwatres.2025.104975","url":null,"abstract":"<div><div>This experimental study systematically investigates the influence of the Peclet number (<em>Pe</em>) and transport length on the transition to Fickian transport in homogeneous sand packs. Through Darcy column experiments with varying lengths and two distinct sediment sizes (<em>d</em>₅₀), we analyzed breakthrough curves (BTCs) to quantify non-Fickian characteristics and transport parameters. The dispersion coefficient exhibited an asymptotic transition to steady-state values between transport lengths of 0.91 m and 1.83 m, coinciding with a shift from heavy-tailed residence time distributions (RTDs) towards inverse Gaussian (Fickian) behavior. Non-Fickian attributes (quantified by skewness) scale with <em>Pe</em> via a power-law relationship, with exponents decreasing as transport length increases. During non-Fickian transport, the dispersion coefficient exhibited nonlinear power-law relationships with <em>Pe</em>, with the power-law exponent converging to ∼1 as transport length increased, consistent with hydrodynamic dispersion theory in the mechanical transport regime. The dispersivity coefficient (α) showed weak <em>Pe</em> dependence <em>only</em> in the non-Fickian regime and became <em>Pe</em>-independent under Fickian conditions. No significant length scale dependence of α was observed between 0.18 m and 1.83 m. This study demonstrates that extrapolating dispersivity from shorter length scales can be unreliable, as convergence to Fickian behavior requires transport lengths or solute transport representative elementary volume (REV) of at least 1 m These findings emphasize the need for longer experimental setups to determine asymptotic transport coefficients consistent with Fickian solute transport theory in porous media.</div></div>","PeriodicalId":7614,"journal":{"name":"Advances in Water Resources","volume":"200 ","pages":"Article 104975"},"PeriodicalIF":4.0,"publicationDate":"2025-04-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143823812","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}

{"title":"Flow in a falaj (qanat) in an unconfined aquifer system considering the head loss inside the falaj, arbitrary trajectory of falaj, and areal recharge","authors":"Mohammad M Sedghi ,&nbsp;Azizallah Izady ,&nbsp;Ali Al-Maktoumi ,&nbsp;Mingjie Chen ,&nbsp;Hongbin Zhan","doi":"10.1016/j.advwatres.2025.104973","DOIUrl":"10.1016/j.advwatres.2025.104973","url":null,"abstract":"<div><div>A falaj (plural <em>aflaj</em>) (also known as qanat) is a type of horizontal or nearly horizontal well that extracts water from an aquifer by gravity. Despite their importance, aflaj with arbitrary trajectories has not yet been mathematically modeled. Moreover, the available analytical models do not include the effects of several head losses, including head loss inside the falaj due to friction with the inner falaj wall, change of flow direction from the aquifer to the falaj, acceleration due to a change in velocity and minor loss caused by change of the falaj direction. Driven by these knowledge gaps, this study aims to develop a semi-analytical solution for the discharge variations of a falaj with an arbitrary trajectory subjected to arbitrary areal recharge. To obtain the solution, the point sink/source solution of groundwater drawdown in an unconfined aquifer is modified to simulate the drawdown due to the specified flux line sink/source of arbitrary trajectory that is known as a snake well. Then, the specified-flux line sink/source solution is converted into an equivalent specified-head line sink/source via well-screen segmentation. Finally, the Darcy–Weisbach equation, along with equations associated with the minor head loss due to the change of flow direction, is employed to simulate the head variation inside the falaj. This study finds that the hydraulic diffusivity of the aquifer and the length of the falaj significantly affect discharge; the trajectory of the falaj affects its late-time discharge, and head loss in falaj tends to mask the influences of its geometric parameters. Furthermore, the presented model can be employed to determine if it is possible to reduce the discharge of a falaj during the wet season when the groundwater is not needed without affecting the dry season discharge.</div></div>","PeriodicalId":7614,"journal":{"name":"Advances in Water Resources","volume":"201 ","pages":"Article 104973"},"PeriodicalIF":4.0,"publicationDate":"2025-04-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143842869","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}

{"title":"The rigorous upscaling of advection-dominated transport in heterogeneous porous media via the Method of Finite Averages","authors":"Kyle Pietrzyk","doi":"10.1016/j.advwatres.2025.104958","DOIUrl":"10.1016/j.advwatres.2025.104958","url":null,"abstract":"<div><div>Systems involving advection-dominated transport through heterogeneous porous and fractured media are ubiquitous in subsurface engineering applications. However, upscaling such systems continues to challenge rigorous modeling efforts, particularly when advection is stronger than diffusion at fine spatial scales (i.e., when the Péclet number is greater than one at length scales that characterize a system’s unit-cells, representative elementary volumes, or averaging regions). In this work, we propose and validate a strategy for extending the Method of Finite Averages (MoFA), a rigorous upscaling methodology for heterogeneous porous media, to upscale transport systems experiencing stronger advection than diffusion at fine scales (i.e., fine-scale Péclet numbers greater than one). We detail the strategy, the physical conditions under which it can be applied while retaining <em>a priori</em> modeling error guarantees, and implement the strategy to obtain a MoFA model for advective-diffusive transport that accommodates advective physics at fine spatial scales. We then perform two numerical experiments considering systems with system-scale Péclet numbers of 300 and 1000 — which correspond to fine-scale Péclet numbers of 30 and 100, respectively — to verify that the error guarantees are met under the strategy. After, we conduct a numerical study to demonstrate the strategy’s advantages over the original MoFA methodology. The results suggest that rigorously-upscaled transport models for heterogeneous porous media experiencing advective physics at finer spatial scales can be derived through MoFA and resolved orders of magnitude faster than their pore-scale counterparts. The results also suggest that the presented strategy is limited to modeling shallow concentration gradients when there are large differences between the time scales related to advection and a system’s temporally-varying boundary conditions. This limitation hinders the strategy’s practicality in modeling more advective systems, and as such, opportunity exists for developing additional strategies that accommodate rapidly-varying boundary conditions — and consequentially, steeper concentration gradients — while modeling advective systems with MoFA.</div></div>","PeriodicalId":7614,"journal":{"name":"Advances in Water Resources","volume":"200 ","pages":"Article 104958"},"PeriodicalIF":4.0,"publicationDate":"2025-04-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143816118","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}

{"title":"Interaction of λ-cyhalothrin with chitosan and quartz sand: Attachment, transport, and cotransport in porous media","authors":"Evangelia A. Xenou ,&nbsp;Anastasios A. Malandrakis ,&nbsp;Vasileios E. Katzourakis ,&nbsp;Constantinos V. Chrysikopoulos","doi":"10.1016/j.advwatres.2025.104965","DOIUrl":"10.1016/j.advwatres.2025.104965","url":null,"abstract":"<div><div>The insecticide λ-cyhalothrin a type II synthetic parathyroid. It is a hydrophobic and highly effective broad-spectrum insecticide commonly used in pest management. However, its presence in subsurface formations poses environmental toxicity risks and potentially may have adverse effects on humans. Chitosan, a polymer with unique physicochemical and absorption properties, is utilized in numerous various industrial applications. In this study, the interactions of λ-cyhalothrin with chitosan in the presence and absence of quartz sand were investigated under static, dynamic (batch), and column transport conditions at 25 °C. The experimental adsorption data were effectively modeled using a pseudo-second-order adsorption kinetic model, while the transport experiments were simulated with an advection dispersion model incorporating two-site linear or nonlinear particle attachment. The results indicate that the observed λ-cyhalothrin adsorption onto either Chitosan or quartz sand is governed by chemisorption. Cotransport experiments revealed bidirectional interactions, where chitosan forms aggregates with λ-cyhalothrin, enhancing chitosan attachment and significantly reducing λ-cyhalothrin retention. This, highlights chitosan's potential as a remediation agent for mitigating pesticide contamination in porous media.</div></div>","PeriodicalId":7614,"journal":{"name":"Advances in Water Resources","volume":"200 ","pages":"Article 104965"},"PeriodicalIF":4.0,"publicationDate":"2025-03-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143759851","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Investigation of particle-bed pore effect on turbulent structure in oscillatory boundary layer by RIM-PIV measurement","authors":"Eiji Harada,&nbsp;Takumi Tazaki,&nbsp;Hitoshi Gotoh","doi":"10.1016/j.advwatres.2025.104963","DOIUrl":"10.1016/j.advwatres.2025.104963","url":null,"abstract":"<div><div>In surf/swash sediment transport in coastal processes, the influence of pore flow near the surface of the sediment bed cannot be ignored. Understanding how the pore flow affects the turbulent structure of the oscillating boundary layer is crucial for comprehending the mechanisms of bottom sediment transport. However, measurement is difficult and the turbulent structure in the porous sublayer is not well understood. As water is usually used as the fluid in hydraulic experiments, turbulent structures, including pore flow, can be measured by RIM-PIV (refractive index matching - particle image velocimetry) if the sediment bed is formed using a material with the same refractive index as water. Hydrogel particles have the same refractive index as water, but are soft and lightweight, making them unsuitable for forming stable sediment beds. Hence, in this study, sediment beds were formed using MEXFLON Ⓡ (NOK Corporation) particles and the turbulent structure of the oscillatory boundary layer, including the porous sublayer, was measured by RIM-PIV. Measurements of the turbulent structure, including the effect of infiltration and exfiltration caused by pore flow, indicated the presence of anisotropic fluctuating velocities and intermittent turbulent structure. The magnitude of the stress fraction, which indicates momentum exchange due to fluctuating velocities in the porous sublayer, was slight compared to the roughness and interface layers, but contributed strongly to the development of turbulent structures in the oscillatory boundary layer, especially in the acceleration phase of the forward flow. It was suggested that the presence of the porous sublayer causes strong nonlinearities in the turbulent structure of the boundary layer.</div></div>","PeriodicalId":7614,"journal":{"name":"Advances in Water Resources","volume":"200 ","pages":"Article 104963"},"PeriodicalIF":4.0,"publicationDate":"2025-03-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143769342","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}

{"title":"Emergence of efficient channel networks in fluvial landscapes","authors":"Dnyanesh Borse ,&nbsp;Basudev Biswal","doi":"10.1016/j.advwatres.2025.104962","DOIUrl":"10.1016/j.advwatres.2025.104962","url":null,"abstract":"<div><div>It is believed that channel networks across fluvial landscapes are self-organized into fractal patterns in order to minimize energy expenditure, as evidenced by the similarities between computer-generated optimal channel networks (OCNs) and real networks. However, the specific mechanisms driving such energy minimization remain largely elusive. Here, we propose that the energy minimization tendency of channel networks results from the ‘hierarchical’ role of randomness in their evolution – flow direction is more likely to change at lower flow-accumulation areas, and vice versa. The proposed probabilistic growth model then employs a power function to simulate channel-network evolution, where exponent (η) quantifies the role of randomness such that greater η would ensure less randomness and more network stability. Any positive η leads to an asymptotic decrease in energy expenditure implying that energy minimization is merely a consequence of adaptive evolutionary processes rather than an intended goal. Interestingly, the most efficient networks are consistently observed to emerge at η = 0.5, implying a delicate balance between randomness and the forces resisting change as a necessary condition for the emergence of the most efficient networks. Thus, the proposed framework holds promise for explaining the evolution of other tree-like networks in nature and for developing more efficient optimization methods for practical applications.</div></div>","PeriodicalId":7614,"journal":{"name":"Advances in Water Resources","volume":"200 ","pages":"Article 104962"},"PeriodicalIF":4.0,"publicationDate":"2025-03-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143816017","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}

{"title":"Microfluidic study of CO2 diffusive leakage through microfractures in saline aquifers for CO2 sequestration","authors":"Wei Yu ,&nbsp;Jack H.Y. Lo ,&nbsp;Abdulrauf R. Adebayo ,&nbsp;Mohamed Gamal Rezk ,&nbsp;Ahmed Al-Yaseri ,&nbsp;Zuhair AlYousef","doi":"10.1016/j.advwatres.2025.104960","DOIUrl":"10.1016/j.advwatres.2025.104960","url":null,"abstract":"<div><div>CO₂ diffusive leakage, or diffusive transport, through intrinsic or induced caprock fractures poses a significant concern for the security of CO₂ sequestration in saline aquifers. Although this issue has garnered considerable interest and has been the subject of many numerical analyses, experimental studies remain limited. We present the first experimental investigation of CO₂ diffusive leakage through microfractures in a generalized microfluidic system that represents the key features of the system under realistic CO₂ sequestration conditions. Our findings reveal two-stage depletion kinetics of trapped CO₂ in porous media, driven by dissolution and diffusion through fractures. The first stage is characterized by the rapid dissolution of CO₂ into nearby brine, while the second stage exhibits a steady leakage rate as CO₂ diffuses through the fractures into a water sink, driven by the solubility limit, assuming stable microfracture structures and negligible advection. Between these two stages, there is a transition period during which CO₂ saturation remains stable. Two key parameters are proposed to quantify the diffusive leakage process: the transition time and the steady-state leakage rate. The transition time <span><math><mrow><mn>0.1</mn><mfrac><msup><mrow><mi>l</mi></mrow><mn>2</mn></msup><mi>D</mi></mfrac><mspace></mspace></mrow></math></span> defines the timescale for the onset of a diffusive leakage event, where <em>l</em> represents the fracture length and <em>D</em> the gas diffusivity. The steady-state leakage rate is primarily governed by aquifer conditions and fracture properties, which scales as <span><math><mfrac><mrow><mi>D</mi><msub><mi>C</mi><mn>1</mn></msub></mrow><mi>l</mi></mfrac></math></span>​​, where <em>C</em>₁​ is the solubility limit. Our theoretical predictions align well with the experimental results. Additionally, the effects of temperature, pressure, salinity, and storage depth on CO₂ diffusivity and solubility are explored through sensitivity analysis. Despite the simplifications in our experimental design and modeling, our study lays the foundation for future research by progressively incorporating additional complexities. These findings provide broader implications for assessing leakage risks in subsurface geological gas storage, such as H₂ and CH₄.</div></div>","PeriodicalId":7614,"journal":{"name":"Advances in Water Resources","volume":"200 ","pages":"Article 104960"},"PeriodicalIF":4.0,"publicationDate":"2025-03-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143777468","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}

{"title":"FluidNet-Lite: Lightweight convolutional neural network for pore-scale modeling of multiphase flow in heterogeneous porous media","authors":"Mohammed Yaqoob ,&nbsp;Mohammed Yusuf Ansari ,&nbsp;Mohammed Ishaq ,&nbsp;Unais Ashraf ,&nbsp;Saideep Pavuluri ,&nbsp;Arash Rabbani ,&nbsp;Harris Sajjad Rabbani ,&nbsp;Thomas D. Seers","doi":"10.1016/j.advwatres.2025.104952","DOIUrl":"10.1016/j.advwatres.2025.104952","url":null,"abstract":"<div><div>Modeling breakthrough patterns in heterogeneous porous media during two-phase fluid flow presents unique challenges due to computational complexity and data scarcity. Current deep learning approaches, primarily generative adversarial network (GAN) based, focus on homogeneous media, limiting their practical application in real-world heterogeneous pore systems. In this work, we introduce <em>FluidNet-Lite</em>, a lightweight Convolutional Neural Network for pore-scale modeling in heterogeneous porous media. Departing from generative task frameworks, we reformulate breakthrough pattern prediction as an innovative pixel-wise classification task, significantly reducing model complexity. By integrating two essential physical parameters—viscosity ratio (<span><math><mi>M</mi></math></span>) and contact angle (<span><math><mi>θ</mi></math></span>), our approach improves predictive accuracy and embeds critical physics-based dependencies directly into the learning process. A Grain-Weighted Adaptive Loss (GWAL) function further enforces fluid flow principles, enhancing model consistency with physical laws. <em>FluidNet-Lite</em> achieves state-of-the-art performance with an Intersection over Union (IoU) of 0.92 and a Structural Similarity Index Measure (SSIM) of 0.89. It is 94% lighter and 48% more computationally efficient than GAN-based alternatives, reducing VRAM usage by 40% and inference time by 30%. Demonstrating robust generalization across interpolation, extrapolation, and unseen test samples, <em>FluidNet-Lite</em> sets a new benchmark for lightweight, physics-informed modeling in heterogeneous porous media fluid dynamics, as evidenced by its superior performance and efficiency improvements over conventional approaches. We also publish a comprehensive dataset and codebase to support future research in lightweight architectures for deep learning-based surrogate modeling of pore-scale immiscible displacement patterns.</div></div>","PeriodicalId":7614,"journal":{"name":"Advances in Water Resources","volume":"200 ","pages":"Article 104952"},"PeriodicalIF":4.0,"publicationDate":"2025-03-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143724971","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}