Advances in Water Resources最新文献

{"title":"STAMNet—A spatiotemporal attention module and network for upscaling reactive transport simulations of the hyporheic zone","authors":"Marc Berghouse ,&nbsp;Rishi Parashar","doi":"10.1016/j.advwatres.2025.104951","DOIUrl":"10.1016/j.advwatres.2025.104951","url":null,"abstract":"<div><div>Reactive transport (RT) simulations are important tools for understanding and predicting phenomena in the subsurface. However, RT is computationally intensive and complex simulations can be numerically unstable. Here, we present STAMNet, a low-parameter attention-based suite of neural nets that can upscale and upsample reactive transport simulations, applied to example problem of bioremediation in the hyporheic zone. We show that a simple MLP offers 30x speedup over standard multiphysics RT simulations and can accurately (<span><math><mrow><mo>≈</mo><mn>90</mn><mtext>%</mtext></mrow></math></span> <span><math><msup><mrow><mi>R</mi></mrow><mrow><mn>2</mn></mrow></msup></math></span>) predict the output of multiple variables of a 1x20 meter RT simulation by using the output from a 1 × 2 m simulation as input. We add efficient channel attention to our optimized MLP which significantly improves the mean average error but does not affect the <span><math><msup><mrow><mi>R</mi></mrow><mrow><mn>2</mn></mrow></msup></math></span>. We further develop a novel spatiotemporal attention module (STAM), which results in improvements both in mean square error and <span><math><msup><mrow><mi>R</mi></mrow><mrow><mn>2</mn></mrow></msup></math></span> (92.5%). Finally, we present a network architecture that utilizes STAM to accurately (99.9% <span><math><msup><mrow><mi>R</mi></mrow><mrow><mn>2</mn></mrow></msup></math></span>) upsample simulations in two dimensions. Specifically, our model allows for the 2x upsampling of simulations in the <span><math><mi>x</mi></math></span> and <span><math><mi>y</mi></math></span> dimensions to convert a coarse-grained input into a fine-grained output. These models have potential use for Monte-Carlo-style investigations of bioremediation and the work presented serves as a proof-of-concept for accurate prediction of large sets of spatiotemporal outputs.</div></div>","PeriodicalId":7614,"journal":{"name":"Advances in Water Resources","volume":"200 ","pages":"Article 104951"},"PeriodicalIF":4.0,"publicationDate":"2025-03-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143687827","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":"Uncertainty assessment of solute concentration in natural aquifers sampled from observation wells","authors":"Felipe P.J. de Barros ,&nbsp;Jinwoo Im","doi":"10.1016/j.advwatres.2025.104949","DOIUrl":"10.1016/j.advwatres.2025.104949","url":null,"abstract":"<div><div>The subsurface environment’s complex heterogeneous structure poses challenges for accurately modeling transport phenomena due to limited data and measurement errors, leading to uncertainties in solute transport predictions. This study proposes a computational framework to semi-analytically compute the cumulative distribution function (CDF) of solute concentration in heterogeneous aquifers. We investigate how hydrogeological heterogeneity and sampling volume affect concentration uncertainty, as measurements are typically taken at observation wells defined by their sampling volume. Our framework estimates the CDF for a conservative solute, considering hydraulic conductivity heterogeneity, local-scale dispersion, and sampling dimensions. The CDF solution is applicable to point-source injections and scenarios with low to moderate heterogeneity. The CDF solution is verified against high-resolution numerical simulations in a 3D heterogeneous aquifer setting. Results show that sampling volume significantly impacts the concentration CDF tails, crucial for evaluating extreme event probabilities, with sensitivity to sampling volume decreasing as travel distance increases.</div></div>","PeriodicalId":7614,"journal":{"name":"Advances in Water Resources","volume":"199 ","pages":"Article 104949"},"PeriodicalIF":4.0,"publicationDate":"2025-03-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143643460","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":"Pore-scale study on solute dispersion in the aqueous phase within unsaturated porous media","authors":"Ruichang Guo ,&nbsp;Lingzao Zeng ,&nbsp;Qingqi Zhao ,&nbsp;Cheng Chen","doi":"10.1016/j.advwatres.2025.104957","DOIUrl":"10.1016/j.advwatres.2025.104957","url":null,"abstract":"<div><div>Solute dispersion in the aqueous phase within unsaturated porous media has critical implications to various natural and engineered systems, such as nutrient and contaminant transport in the vadose zone. This work developed pore-scale lattice Boltzmann (LB) modeling to simulate water-air multiphase flow and solute transport in the aqueous phase, which unraveled the role of the water saturation and Peclet (Pe) number on solute dispersivity in unsaturated three-dimensional porous media. The Pe number and water saturation influenced the solute concentration profile along the longitudinal direction in a coupled manner. Under a low Pe number, the solute concentration profile approximately followed the classic Gaussian distribution regardless of the water saturation. Under a high Pe number, the solute concentration profile became skewed and the role of the water saturation was noticeable. Under a moderate water saturation, a significant portion of the solute particles were trapped at the trailing edge, leading to the non-Fickian transport behavior. In contrast, with a high water saturation, the enhanced water connectivity in the pore space mitigated the trapping of solute particles at the trailing edge, leading to the classic Fickian dispersion. This study also showed the non-monotonic dependence of longitudinal dispersivity on the water saturation, which was caused by a competing process between pore water connectivity and flow field heterogeneity. The pore-scale simulation confirmed that the dispersivity is not a constant but depends on the Pe number nonlinearly. Therefore, there is a complicated coupling between the dispersivity, water saturation, and Pe number for multiphase flows in porous media. This study developed a six-parameter empirical model to account for the role of the water saturation and Pe number on the longitudinal dispersivity in unsaturated flows. This empirical model can be used to predict the solute dispersion coefficient in unsaturated porous media, which is critical to large-scale contaminant transport modeling.</div></div>","PeriodicalId":7614,"journal":{"name":"Advances in Water Resources","volume":"199 ","pages":"Article 104957"},"PeriodicalIF":4.0,"publicationDate":"2025-03-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143705386","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":"Effects of wettability heterogeneity on multiphase flow: From pore-scale mechanisms to cross-scale insights","authors":"Jingrui Liu ,&nbsp;Kang Duan ,&nbsp;Rihua Jiang ,&nbsp;Qiangyong Zhang","doi":"10.1016/j.advwatres.2025.104956","DOIUrl":"10.1016/j.advwatres.2025.104956","url":null,"abstract":"<div><div>The macroscopic behavior of multiphase flow systems is governed by interfacial dynamics, which are a strong function of the synergy between viscous forces and wettability. However, the effects of wettability heterogeneity, an inherent feature of natural porous media, remain poorly understood. In this study, we incorporated spatial-related wettability into phase field model and systematically examined the synergy between capillary number (<em>Ca</em>) and wettability heterogeneity on CO₂-water multiphase flow. We developed three phase diagrams to illustrate the evolution of pore-scale displacement mechanisms driven by <em>Ca</em>-wettability-heterogeneity synergy, identifying a novel mechanism termed hybrid filling. Furthermore, we distinguished two types of wetting pinning: p-pinning, resulting from adhesion on patches, and b-pinning, caused by wettability contrasts between patches. These pinnings can enhance the effects of capillary force and contribute to energy dissipation, playing a key role in <em>Ca</em>-wettability-heterogeneity synergy. By clarifying how this synergy affects macroscopic flow characteristics, we established multidimensional diagrams to offer cross-scale insights for Geological Carbon Sequestration (GCS). This work links pore-scale dynamics with large-scale flow behavior, facilitating the design of more effective GCS strategies.</div></div>","PeriodicalId":7614,"journal":{"name":"Advances in Water Resources","volume":"199 ","pages":"Article 104956"},"PeriodicalIF":4.0,"publicationDate":"2025-03-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143641681","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":"Spatial variations of velocity and pressure fields induced by large-scale (single stalk) and small-scale (sediment) roughness elements","authors":"Angel Monsalve ,&nbsp;William Jeff Reeder ,&nbsp;Katherine Adler ,&nbsp;Jose Roberto Moreto ,&nbsp;Xiaofeng Liu ,&nbsp;Daniele Tonina","doi":"10.1016/j.advwatres.2025.104954","DOIUrl":"10.1016/j.advwatres.2025.104954","url":null,"abstract":"<div><div>Characterizing velocity and pressure fields in aquatic systems is crucial for understanding fundamental processes such as sediment transport, hyporheic flow, air-water exchange, and habitat quality. While large obstacles like vegetation stalks are known to create significant localized pressure gradients, the role of small-scale bed roughness in generating local pressure gradients remains poorly understood. Here, we explore flow dynamics around a vertical cylinder (simulating vegetation) over a coarse granular bed using stereo particle image velocimetry (SPIV) with refractive index-matched (RIM) fluid, integrated with Large Eddy Simulations (LES). Our combined approach reveals an important phenomenon: while large obstacles like vegetation stalks create localized pressure gradients, bed roughness elements generate frequent pressure fluctuations across the entire streambed. Although previous studies have focused primarily on large-obstacle effects, our findings show that grain-scale pressure variations generate stronger local gradients (up to ±250 mmH₂O/m) than those from large obstacles (±50 mmH₂O/m), and their widespread occurrence throughout the bed surface may collectively have substantial effects on hyporheic exchange. By quantifying pressure fields at both large and small scales, we demonstrate that bed roughness elements create persistent pressure gradients that, due to their widespread occurrence, may significantly influence surface-subsurface water interactions. Our results highlight the importance of considering grain-scale roughness effects when studying hyporheic processes in natural streams.</div></div>","PeriodicalId":7614,"journal":{"name":"Advances in Water Resources","volume":"199 ","pages":"Article 104954"},"PeriodicalIF":4.0,"publicationDate":"2025-03-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143679613","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":"Lattice Boltzmann modelling of capillarity, adsorption and fluid retention in simple geometries: Do capillary and film water have equal matric suction or not?","authors":"Zi Li ,&nbsp;Zhenlei Yang ,&nbsp;Sergio Galindo-Torres ,&nbsp;Ling Li","doi":"10.1016/j.advwatres.2025.104950","DOIUrl":"10.1016/j.advwatres.2025.104950","url":null,"abstract":"<div><div>The pore water retained in unsaturated soil includes film water attached on the solid surface and capillary water in corners or pores, which are mainly controlled by adsorptive force from the solid surface and capillary force from the water-gas interface, respectively. The soil water retention (SWR) curve represents the fundamental characteristic of unsaturated soil, in which the connected capillary and adsorptive water are conventionally presumed to have equivalent suction values as matric suction. Here, a long-range adsorptive fluid-solid interaction force is developed in the mesoscopic multiphase lattice Boltzmann model (LBM) framework to model the macroscopic processes of capillarity and adsorption. The pressure tensors of capillary and film water derived based on mechanical equilibrium and the results of numerical simulations combine to show that the adsorptive suction is much higher than the capillary suction, not following the classical relationship. We attribute this inequality to the different adsorptive interaction potentials incorporated in the capillary and film water pressures, due to the fluid density profiles varying differently with the separation distance from solid surface, and, from the perspective of thermodynamic equilibrium, the deviation of film water and capillary water densities from free water density. The film thickness almost does not change for the given radii of meniscus curvature in simple geometries (i.e., slits and corners). The adsorption effects on the matric suction upscaled from the intrinsic phase average method and on the equivalent pore size distribution are investigated for both single-sized slits and complex pore networks. The findings reveal the influences of capillarity and adsorption on the shape of SWR curve, and help establish the SWR function with accurate physical meanings in the field of soil physics and hydrology and measure the disjoining pressure isotherm properly in colloid and interface chemistry.</div></div>","PeriodicalId":7614,"journal":{"name":"Advances in Water Resources","volume":"199 ","pages":"Article 104950"},"PeriodicalIF":4.0,"publicationDate":"2025-03-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143620567","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":"Impact of saturation on continuum-scale conductivity and tracer dispersion in heterogeneous porous media","authors":"Doron Kalisman,&nbsp;Brian Berkowitz","doi":"10.1016/j.advwatres.2025.104953","DOIUrl":"10.1016/j.advwatres.2025.104953","url":null,"abstract":"<div><div>This study investigates the interplay among water saturation, hydraulic conductivity, and mechanical dispersion in heterogeneous porous media at the continuum scale. Mechanical dispersion of dissolved chemical tracers is influenced directly by water velocity variability, which is governed by the porous structure, the distribution of the water phase within it, and its corresponding conductivity field. Previous studies have either examined the relationships between these factors in fully saturated conditions, or in partial saturation but without considering continuum scale heterogeneity of the media. Through numerical simulations, the analysis here demonstrates how variations in saturation affect the hydraulic conductivity field and, consequently, mechanical dispersion. The study reveals that longitudinal spreading of the tracer plume, when scaled for varying transport times and velocities, shows a non-monotonic relationship with saturation, being least pronounced at an intermediate degree of saturation. These insights contribute to a more nuanced understanding of tracer transport in partially saturated, heterogeneous media, with implications for environmental and engineering applications.</div></div>","PeriodicalId":7614,"journal":{"name":"Advances in Water Resources","volume":"199 ","pages":"Article 104953"},"PeriodicalIF":4.0,"publicationDate":"2025-03-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143679608","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":"Derivation of weakly hydrodynamic models in the Dupuit–Forchheimer regime","authors":"Martin Parisot","doi":"10.1016/j.advwatres.2025.104933","DOIUrl":"10.1016/j.advwatres.2025.104933","url":null,"abstract":"<div><div>The current study is dedicated to the formal derivation of a hierarchic of asymptotic models that approximate the groundwater waves problem within the Dupuit–Forchheimer regime, over a regular, non-planar substratum. The derivation methodology employed bears resemblance to the techniques utilized in hierarchic of asymptotic models for approximating the water waves problem in the shallow water regime. Mathematically speaking, the asymptotic models manifest as nonlinear, non-local diffusion equations. We identify an energy dissipation law inherent to these models, thereby bolstering the physical validity and confidence in the proposed framework. A numerical strategy is proposed that preserved at the discrete level the energy dissipation.</div></div>","PeriodicalId":7614,"journal":{"name":"Advances in Water Resources","volume":"199 ","pages":"Article 104933"},"PeriodicalIF":4.0,"publicationDate":"2025-03-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143611035","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":"Assessment of hybrid RANS/LES models for the prediction of the flow and scour around a wall-mounted cylinder","authors":"Alban Gilletta ,&nbsp;Cyrille Bonamy ,&nbsp;Marie Robert ,&nbsp;Julien Chauchat","doi":"10.1016/j.advwatres.2025.104934","DOIUrl":"10.1016/j.advwatres.2025.104934","url":null,"abstract":"&lt;div&gt;&lt;div&gt;In many environmental flow situations, a solid body emerges from a sediment bed. This may occur in natural systems, for example when a tree-root emerges from the bed, or around a man-made structure such as a bridge pier or a wind turbine foundation. When this situation occurs, various flow-structure interactions such as the horseshoe vortex or vortex shedding lead to the scour process, namely the erosion pattern around the solid obstacle. In this context, accurate numerical prediction of the flow-structure interactions is of primary importance. In this contribution, three-dimensional hydrodynamic simulations are performed using the single-phase incompressible flow solver pimpleFOAM (OpenFOAMv2206plus). Two hybrid RANS-LES turbulence models available in OpenFOAM are tested: the non-zonal &lt;span&gt;&lt;math&gt;&lt;mrow&gt;&lt;mi&gt;k&lt;/mi&gt;&lt;mo&gt;−&lt;/mo&gt;&lt;mi&gt;ω&lt;/mi&gt;&lt;/mrow&gt;&lt;/math&gt;&lt;/span&gt;SST-SAS model and the zonal &lt;span&gt;&lt;math&gt;&lt;mrow&gt;&lt;mi&gt;k&lt;/mi&gt;&lt;mo&gt;−&lt;/mo&gt;&lt;mi&gt;ω&lt;/mi&gt;&lt;/mrow&gt;&lt;/math&gt;&lt;/span&gt;SST-IDDES model. As a first benchmark, a plane channel flow configuration at large bulk Reynolds number (&lt;span&gt;&lt;math&gt;&lt;mrow&gt;&lt;mi&gt;R&lt;/mi&gt;&lt;msub&gt;&lt;mrow&gt;&lt;mi&gt;e&lt;/mi&gt;&lt;/mrow&gt;&lt;mrow&gt;&lt;mi&gt;b&lt;/mi&gt;&lt;/mrow&gt;&lt;/msub&gt;&lt;mo&gt;=&lt;/mo&gt;&lt;/mrow&gt;&lt;/math&gt;&lt;/span&gt;1.9 &lt;span&gt;&lt;math&gt;&lt;mo&gt;×&lt;/mo&gt;&lt;/math&gt;&lt;/span&gt; 10&lt;sup&gt;4&lt;/sup&gt;) is investigated to evaluate the mesh requirements and to determine a set of acceptable numerical parameters. These simulations show that while the &lt;span&gt;&lt;math&gt;&lt;mrow&gt;&lt;mi&gt;k&lt;/mi&gt;&lt;mo&gt;−&lt;/mo&gt;&lt;mi&gt;ω&lt;/mi&gt;&lt;/mrow&gt;&lt;/math&gt;&lt;/span&gt;SST-IDDES model allows to simulate some features of the fluctuating motions with fewer grid requirements than LES, the &lt;span&gt;&lt;math&gt;&lt;mrow&gt;&lt;mi&gt;k&lt;/mi&gt;&lt;mo&gt;−&lt;/mo&gt;&lt;mi&gt;ω&lt;/mi&gt;&lt;/mrow&gt;&lt;/math&gt;&lt;/span&gt;SST-SAS model behaves as a RANS model. Concerning the wall-mounted cylinder flow at high Reynolds number (&lt;span&gt;&lt;math&gt;&lt;mrow&gt;&lt;mi&gt;R&lt;/mi&gt;&lt;msub&gt;&lt;mrow&gt;&lt;mi&gt;e&lt;/mi&gt;&lt;/mrow&gt;&lt;mrow&gt;&lt;mi&gt;b&lt;/mi&gt;&lt;/mrow&gt;&lt;/msub&gt;&lt;mo&gt;=&lt;/mo&gt;&lt;/mrow&gt;&lt;/math&gt;&lt;/span&gt;1.8 &lt;span&gt;&lt;math&gt;&lt;mo&gt;×&lt;/mo&gt;&lt;/math&gt;&lt;/span&gt; 10&lt;sup&gt;5&lt;/sup&gt;), the conclusions are quite different, the &lt;span&gt;&lt;math&gt;&lt;mrow&gt;&lt;mi&gt;k&lt;/mi&gt;&lt;mo&gt;−&lt;/mo&gt;&lt;mi&gt;ω&lt;/mi&gt;&lt;/mrow&gt;&lt;/math&gt;&lt;/span&gt;SST-SAS model switches to LES in the region of interest namely the horseshoe-vortex region and the wake thanks to its non-zonal property. The &lt;span&gt;&lt;math&gt;&lt;mrow&gt;&lt;mi&gt;k&lt;/mi&gt;&lt;mo&gt;−&lt;/mo&gt;&lt;mi&gt;ω&lt;/mi&gt;&lt;/mrow&gt;&lt;/math&gt;&lt;/span&gt;SST-SAS model provides better predictions than the &lt;span&gt;&lt;math&gt;&lt;mrow&gt;&lt;mi&gt;k&lt;/mi&gt;&lt;mo&gt;−&lt;/mo&gt;&lt;mi&gt;ω&lt;/mi&gt;&lt;/mrow&gt;&lt;/math&gt;&lt;/span&gt;SST-IDDES model in terms of amplitude of the bimodal oscillation and vortex merging processes in the horseshoe vortex at lower CPU cost than LES. After this benchmarking and assessment of the hybrid RANS-LES model, a new &lt;span&gt;&lt;math&gt;&lt;mrow&gt;&lt;mi&gt;k&lt;/mi&gt;&lt;mo&gt;−&lt;/mo&gt;&lt;mi&gt;ω&lt;/mi&gt;&lt;/mrow&gt;&lt;/math&gt;&lt;/span&gt;2006-SAS turbulence model has been developed for the two-phase flow solver &lt;em&gt;sedFOAM&lt;/em&gt;. The model is then applied to a scour configuration around a wall-mounted cylinder and successfully represents the upstream scour depth. The simulations improve the resolution of the complex horseshoe-vortex dynamics which is directly in","PeriodicalId":7614,"journal":{"name":"Advances in Water Resources","volume":"199 ","pages":"Article 104934"},"PeriodicalIF":4.0,"publicationDate":"2025-03-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143643459","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":"Transfer learning for geological carbon storage forecasting using neural operator","authors":"Andres Nunez ,&nbsp;Siddharth Misra ,&nbsp;Yusuf Falola","doi":"10.1016/j.advwatres.2025.104948","DOIUrl":"10.1016/j.advwatres.2025.104948","url":null,"abstract":"<div><div>Geological carbon storage (GCS) is critical for sequestering CO₂ deep underground. GCS projects may face environmental challenges, such as leakage risks, adverse pressure buildup, and groundwater contamination. Numerical simulators play a vital role in accurate forecasting but can be computationally expensive. In this work, we leveraged an updated Fourier Neural Operator (FNO) which includes data sparsity management, to learn to rapidly forecast pressure and CO₂ phase saturation distributions in a geological carbon storage (GCS) reservoir. Compared to commercial reservoir simulators, FNO-based forecasting offers accurate prediction while reducing the computational time by a factor of 40, enabling high volume of forecasting in less time. Additionally, we applied transfer learning (TL) to further reduce the data and computational requirements of the FNO-based forecasting across a wide array of scenarios. Specifically, we demonstrated the usefulness of TL in accurately predicting the pressure and CO₂ saturation distributions for uncertain and variable geological and operational conditions. The results of this study indicate that the improved FNO workflow reduces the computational time by approximately 97 %, and the relative mean error for predicting both CO₂ saturation and pressure distributions is &lt;1 %. Generally, the use of TL effectively transfers knowledge from a pre-existing model to other related tasks. TL significantly reduces the required training data by 78 % while maintaining a relative mean error below 5 %. Although, the results in this work can be further improved, this study demonstrates the potential of integrating FNO and TL to reduce computational time and data requirements for CO₂ forecasts during GCS projects, providing a more efficient and faster approach.</div></div>","PeriodicalId":7614,"journal":{"name":"Advances in Water Resources","volume":"199 ","pages":"Article 104948"},"PeriodicalIF":4.0,"publicationDate":"2025-03-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143620566","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}