Water Resources Research最新文献

{"title":"Competing Effects of Vegetation Greening-Induced Changes in Summer Evapotranspiration and Precipitation on Water Yield in the Yangtze River Basin Based on WRF Simulations","authors":"Guoshuai Liu, Weiguang Wang","doi":"10.1029/2024wr038663","DOIUrl":"https://doi.org/10.1029/2024wr038663","url":null,"abstract":"Remarkable vegetation greening has been observed in the Yangtze River Basin (YRB) during the past two decades, triggering noteworthy hydrological consequences. Previous studies have assessed the hydrological effect of vegetation greening but ignored the vegetation-precipitation feedbacks from land-atmosphere interactions. To address this knowledge gap, here we conduct coupled land-atmosphere model simulations prescribed with satellite vegetation observations to investigate how vegetation greening in the YRB affects regional hydrological cycles through vegetation physiological processes and biophysical feedbacks, with potentially competing effects on water yield (WY) by altering evapotranspiration (ET) and precipitation. Over the 2001–2020 period, the leaf area index in summer shows a significant increasing trend at a rate of 0.34 m² m⁻² decade⁻¹ (<i>P</i> &lt; 0.01). This vegetation greening causes a substantial rise in ET, primarily due to increased plant transpiration and canopy evaporation, along with reduced soil evaporation attributed to enhanced root water uptake and shading of the soil surface. Moreover, the modeled results indicate that vegetation greening is the key driver for the observed ET enhancement. In addition, vegetation greening induces increases in precipitation by modulating moisture flux convergence, which although statistically insignificant, provides considerable water to compensate for the enhanced ET. For the cumulative effects of vegetation greening from 2001 to 2020 at the basin scale, the increased precipitation (approximately, 101 mm) outpaces the increased water consumption (approximately, 93 mm), resulting in an insignificant effect on WY. Our findings underscore the importance of considering vegetation-precipitation feedbacks in evaluations of the hydrological response to natural or deliberate vegetation changes.","PeriodicalId":23799,"journal":{"name":"Water Resources Research","volume":"34 1","pages":""},"PeriodicalIF":5.4,"publicationDate":"2025-03-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143653953","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}

{"title":"Inertial Flow-Driven Enhancement of Solute Mixing and Partitioning at Rough-Walled Fracture Intersections: Experimental and Numerical Investigations","authors":"Dahye Kim, In Wook Yeo","doi":"10.1029/2023wr035609","DOIUrl":"https://doi.org/10.1029/2023wr035609","url":null,"abstract":"This study investigates the impact of the transition from viscous linear to inertial nonlinear flows on solute mixing and partitioning at rough-walled fracture intersections, using direct observations of flow dynamics and solute partitioning processes through microscopic particle image velocimetry. It is generally known that mixing at fracture intersections decreases when transport shifts from diffusion-dominated to advection-dominated processes, but this trend holds only in viscous linear flows. The experimental results conducted in this study reveal that in inertial flows, significant changes in flow structures occur at rough-walled fracture intersections, including the straightening and stretching of main streamlines and the formation of fully developed eddies. Fluid stretching and the formation of eddies contribute to advection-driven diffusive mixing. The straightened streamlines deliver solutes to the outflow leg along a direct path. More importantly, fully developed eddies generate spiral advective paths that reconnect to the main flow channels, enhancing solute redistribution at the intersection. Microscopic measurements and quantitative analyses show that flow nonlinearity—including the formation of eddies, along with enhanced flow straightening and stretching—contributes to increased flow heterogeneity and solute redistribution at fracture intersections. This phenomenon appears as an increase in “apparent” mixing at rough-walled fracture intersections.","PeriodicalId":23799,"journal":{"name":"Water Resources Research","volume":"43 1","pages":""},"PeriodicalIF":5.4,"publicationDate":"2025-03-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143640788","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}

{"title":"Improving Water Table Kinematic Conditions With Unsaturated Flow Insights","authors":"Jun-Hong Lin, Ying-Fan Lin","doi":"10.1029/2024wr038724","DOIUrl":"https://doi.org/10.1029/2024wr038724","url":null,"abstract":"Analytical models interpreting aquifer pumping test data often rely on water table kinematic conditions that assume instantaneous gravity drainage, leading to underestimation of specific yield during the drainage process. This study derives a new water table condition based on a coupled saturated-unsaturated flow model that fully accounts for both unsaturated and saturated flow dynamics. The new condition incorporates the hydraulic properties of the unsaturated zone, providing a more accurate representation of physical processes while maintaining mathematical tractability. Applied to a groundwater flow model for a pumping problem, the drawdown solution is derived using integral transformations. The proposed model is validated using field data from a series of pumping tests at the Boise Hydrogeophysical Research Site in Idaho. The results demonstrate that the new water table condition provides more reliable estimates of specific yield, effectively addressing the underestimation issue associated with existing models. Moreover, the model requires no additional empirical parameters, making it a practical tool for characterizing unconfined aquifer properties.","PeriodicalId":23799,"journal":{"name":"Water Resources Research","volume":"55 1","pages":""},"PeriodicalIF":5.4,"publicationDate":"2025-03-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143640785","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}

{"title":"On Robustness of the Explanatory Power of Machine Learning Models: Insights From a New Explainable AI Approach Using Sensitivity Analysis","authors":"Banamali Panigrahi, Saman Razavi, Lorne E. Doig, Blanchard Cordell, Hoshin V. Gupta, Karsten Liber","doi":"10.1029/2024wr037398","DOIUrl":"https://doi.org/10.1029/2024wr037398","url":null,"abstract":"Machine learning (ML) is increasingly considered the solution to environmental problems where limited or no physico-chemical process understanding exists. But in supporting high-stakes decisions, where the ability to <i>explain</i> possible solutions is key to their acceptability and legitimacy, ML can fall short. Here, we develop a method, rooted in formal <i>sensitivity analysis</i>, to uncover the primary drivers behind ML predictions. Unlike many methods for <i>explainable artificial intelligence</i> (XAI), this method (a) accounts for complex multi-variate distributional properties of data, common in environmental systems, (b) offers a global assessment of the input-output response surface formed by ML, rather than focusing solely on local regions around existing data points, and (c) is scalable and data-size independent, ensuring computational efficiency with large data sets. We apply this method to a suite of ML models predicting various water quality variables in a pilot-scale experimental pit lake. A critical finding is that subtle alterations in the design of some ML models (such as variations in random seed, functional class, hyperparameters, or data splitting) can lead to different interpretations of how outputs depend on inputs. Further, models from different ML families (decision trees, connectionists, or kernels) may focus on different aspects of the information provided by data, despite displaying similar predictive power. Overall, our results underscore the need to assess the explanatory robustness of ML models and advocate for using model ensembles to gain deeper insights into system drivers and improve prediction reliability.","PeriodicalId":23799,"journal":{"name":"Water Resources Research","volume":"214 1","pages":""},"PeriodicalIF":5.4,"publicationDate":"2025-03-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143640786","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}

{"title":"Linking Stochastic Resonance With Long Short-Term Memory Neural Network for Streamflow Simulation Enhancement","authors":"Xungui Li, Jian Sun, Qiyong Yang, Yi Tian, Xiaoli Yang","doi":"10.1029/2024wr039659","DOIUrl":"https://doi.org/10.1029/2024wr039659","url":null,"abstract":"The accuracy of peak streamflow simulation is often lower than that of normal streamflow simulation, posing a significant challenge. This study introduces stochastic resonance (SR) to enhance simulation accuracy, utilizing its ability to leverage noise energy to improve correlations between streamflow and meteorological factors. The proposed SR-LSTM model, validated across major Chinese basins, demonstrates that SR effectively enhances the accuracy of streamflow simulations. By using SR, the Nash-Sutcliffe efficiency increased from 0.70 to 0.79, and the kling-gupta efficiency improved from 0.69 to 0.82. Furthermore, this study utilizes the global Caravan streamflow data set (including CAMELES, CAMELESBR, CAMELESAUS, and LamaH) comprising 1,244 station data points to validate the applicability of SR-LSTM. Results indicate that SR improves accuracy at approximately 70% of 1,244 stations, particularly in regions with high-quality data. Comparative analysis shows that incorporating SR enhances the performance of deep learning models, highlighting its potential for improving both global and peak streamflow simulation accuracy. These findings underscore the effectiveness of SR in enhancing streamflow simulation accuracy.","PeriodicalId":23799,"journal":{"name":"Water Resources Research","volume":"124 1","pages":""},"PeriodicalIF":5.4,"publicationDate":"2025-03-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143635172","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}

{"title":"Evaluating the Potential to Quantify Salmon Habitat via UAS-Based Particle Image Velocimetry","authors":"Lee R. Harrison, Carl J. Legleiter, Brandon T. Overstreet, James S. White","doi":"10.1029/2024wr038045","DOIUrl":"https://doi.org/10.1029/2024wr038045","url":null,"abstract":"Continuous, high-resolution data for characterizing freshwater habitat conditions can support successful management of endangered salmonids. Uncrewed aircraft systems (UAS) make acquiring such fine-scale data along river channels more feasible, but workflows for quantifying reach-scale salmon habitats are lacking. We evaluated the potential for UAS-based mapping of hydraulic habitats using spectrally based depth retrieval and particle image velocimetry (PIV) by comparing these methods to a more well-established flow modeling approach. Our results indicated that estimates of water depth, depth-averaged velocity, and flow direction derived via remote sensing and modeling techniques were comparable and in good agreement with field measurements. Predictions of spring-run Chinook salmon (<i>Oncorhynchus tshawytscha</i>) juvenile rearing habitat produced from PIV and model output were similar, with small errors relative to direct field observations. Estimates of hydraulic heterogeneity based on kinetic energy gradients in the flow field were generally consistent between PIV and flow modeling, but errors relative to field measurements were larger. PIV results were sensitive to the velocity index <span data-altimg=\"/cms/asset/c96feada-812b-4550-a25f-c8092e43d6b6/wrcr70081-math-0001.png\"></span><math altimg=\"urn:x-wiley:00431397:media:wrcr70081:wrcr70081-math-0001\" display=\"inline\" location=\"graphic/wrcr70081-math-0001.png\">\u0000<semantics>\u0000<mrow>\u0000<mo stretchy=\"false\">(</mo>\u0000<mi>α</mi>\u0000<mo stretchy=\"false\">)</mo>\u0000</mrow>\u0000$(alpha )$</annotation>\u0000</semantics></math> used to convert surface velocities to depth-averaged velocities. Sun glint precluded PIV analysis along the margins of some images and a large degree of overlap between frames was thus required to obtain continuous coverage of the reach. Similarly, shadows cast by riparian vegetation caused gaps in spectrally based bathymetric maps. Despite these limitations, our results suggest that for sites with sufficient water surface texture, UAS-based PIV can provide detailed hydraulic habitat information at the reach scale, with accuracies comparable to traditional field methods and multidimensional flow modeling.","PeriodicalId":23799,"journal":{"name":"Water Resources Research","volume":"24 1","pages":""},"PeriodicalIF":5.4,"publicationDate":"2025-03-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143635130","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}

{"title":"The Swimming Performance and Transcriptomic Insights Into Diverse Gene Regulation in Grass Carp Brain Under Water Velocity Stress","authors":"Mian Adnan Kakakhel, Nishita Narwal, Alam Khan, Majid Rasta, Liming Liu, Ihsan Ali, Yujiao Wu, Shi Xiaotao","doi":"10.1029/2024wr037990","DOIUrl":"https://doi.org/10.1029/2024wr037990","url":null,"abstract":"By linking gene regulation to swimming performance under different water flow conditions, the study could reveal how the fish adapt to their environments, providing insights into evolutionary biology and ecology. The current study observed significant variations in swimming performance under various water flow velocities and examined the associated gene regulation. Grass carp were subjected to controlled water velocities to measure the critical swimming speed (<i>U</i>_<i>crit</i>), which showed that the swimming performance was increased based on body length; however, a reduction in swimming performance was observed as the water flow increased (<i>p</i> &lt; 0.05). Additionally, brain samples were collected for transcriptomic analysis, which revealed that differentially expressed genes (DEGs) were functionally annotated revealing key pathways associated with changed behavior patterns. The Enrichment analysis showed significant variation in all groups including behavior (<i>p</i> &lt; 0.05***), skeletal system development (<i>p</i> &lt; 0.05***), hormone activity (<i>p</i> &lt; 0.05***), muscle contraction (<i>p</i> &lt; 0.05**), locomotion (<i>p</i> &lt; 0.05*), and swim bladder development (<i>p</i> &lt; 0.05*) were found the major regulators of behavior in grass carp under water velocities. Moreover, some genes were identified and found significantly different for enzymes and hormones, which could play a potential role during swimming performance such as gene-ca7 (<i>p</i> &lt; 0.005***). The current study provides evidence of the neurogenetic mechanism underlying the changed swimming activity of grass carp under water velocity, which could have important implications for understanding the impact of hydrodynamics and the fish.","PeriodicalId":23799,"journal":{"name":"Water Resources Research","volume":"7 1","pages":""},"PeriodicalIF":5.4,"publicationDate":"2025-03-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143618920","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}

{"title":"Physics-Informed Neural Networks for Estimating a Continuous Form of the Soil Water Retention Curve From Basic Soil Properties","authors":"Sarem Norouzi, Charles Pesch, Emmanuel Arthur, Trine Norgaard, Per Moldrup, Mogens H. Greve, Amélie M. Beucher, Morteza Sadeghi, Marzieh Zaresourmanabad, Markus Tuller, Bo V. Iversen, Lis W. de Jonge","doi":"10.1029/2024wr038149","DOIUrl":"https://doi.org/10.1029/2024wr038149","url":null,"abstract":"This paper presents a novel physics-informed neural network (PINN) approach for developing pedotransfer functions (PTFs) to predict continuous soil water retention curves (SWRCs) based on soil textural fractions, organic carbon content, and bulk density. In contrast to conventional parametric PTFs developed for specific SWRC models, the PINN learns a non-specific form of the SWRC from both measurements and physical constraints imposed during the training process. This approach allows the estimated SWRC to maintain its physical integrity from saturation to oven-dry conditions, even in scenarios with sparse data. The new approach is particularly effective for tackling the challenges encountered in developing PTFs on large SWRC data sets, which often have an imbalance toward the wet-end (&lt;span data-altimg=\"/cms/asset/9d5509fb-dfe2-44d2-9b0b-fb490535c677/wrcr70064-math-0001.png\"&gt;&lt;/span&gt;&lt;mjx-container ctxtmenu_counter=\"171\" ctxtmenu_oldtabindex=\"1\" jax=\"CHTML\" role=\"application\" sre-explorer- style=\"font-size: 103%; position: relative;\" tabindex=\"0\"&gt;&lt;mjx-math aria-hidden=\"true\" location=\"graphic/wrcr70064-math-0001.png\"&gt;&lt;mjx-semantics&gt;&lt;mjx-mrow data-semantic-children=\"5,3\" data-semantic-content=\"2\" data-semantic- data-semantic-role=\"inequality\" data-semantic-speech=\"p upper F italic less than or equals 4.2\" data-semantic-type=\"relseq\"&gt;&lt;mjx-mrow data-semantic-annotation=\"clearspeak:simple;clearspeak:unit\" data-semantic-children=\"0,1\" data-semantic-content=\"4\" data-semantic- data-semantic-parent=\"6\" data-semantic-role=\"implicit\" data-semantic-type=\"infixop\"&gt;&lt;mjx-mi data-semantic-annotation=\"clearspeak:simple\" data-semantic-font=\"italic\" data-semantic- data-semantic-parent=\"5\" data-semantic-role=\"latinletter\" data-semantic-type=\"identifier\"&gt;&lt;mjx-c&gt;&lt;/mjx-c&gt;&lt;/mjx-mi&gt;&lt;mjx-mo data-semantic-added=\"true\" data-semantic- data-semantic-operator=\"infixop,⁢\" data-semantic-parent=\"5\" data-semantic-role=\"multiplication\" data-semantic-type=\"operator\" style=\"margin-left: 0.056em; margin-right: 0.056em;\"&gt;&lt;mjx-c&gt;&lt;/mjx-c&gt;&lt;/mjx-mo&gt;&lt;mjx-mi data-semantic-annotation=\"clearspeak:simple\" data-semantic-font=\"italic\" data-semantic- data-semantic-parent=\"5\" data-semantic-role=\"latinletter\" data-semantic-type=\"identifier\"&gt;&lt;mjx-c&gt;&lt;/mjx-c&gt;&lt;/mjx-mi&gt;&lt;/mjx-mrow&gt;&lt;mjx-mo data-semantic-font=\"italic\" data-semantic- data-semantic-operator=\"relseq,≤\" data-semantic-parent=\"6\" data-semantic-role=\"inequality\" data-semantic-type=\"relation\" rspace=\"5\" space=\"5\"&gt;&lt;mjx-c&gt;&lt;/mjx-c&gt;&lt;/mjx-mo&gt;&lt;mjx-mn data-semantic-annotation=\"clearspeak:simple\" data-semantic-font=\"normal\" data-semantic- data-semantic-parent=\"6\" data-semantic-role=\"float\" data-semantic-type=\"number\"&gt;&lt;mjx-c&gt;&lt;/mjx-c&gt;&lt;mjx-c&gt;&lt;/mjx-c&gt;&lt;mjx-c&gt;&lt;/mjx-c&gt;&lt;/mjx-mn&gt;&lt;/mjx-mrow&gt;&lt;/mjx-semantics&gt;&lt;/mjx-math&gt;&lt;mjx-assistive-mml display=\"inline\" unselectable=\"on\"&gt;&lt;math altimg=\"urn:x-wiley:00431397:media:wrcr70064:wrcr70064-math-0001\" display=\"inline\" location=\"graphic/wrcr70064-math-0001.png\" xmlns=\"http://www.w3.org/1998/Math/MathML\"&gt;&lt;semantics&gt;&lt;mrow data-sem","PeriodicalId":23799,"journal":{"name":"Water Resources Research","volume":"183 1","pages":""},"PeriodicalIF":5.4,"publicationDate":"2025-03-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143618921","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}

{"title":"Determining the Relative Contributions of Runoff, Coastal, and Compound Processes to Flood Exposure Across the Carolinas During Hurricane Florence","authors":"Lauren E. Grimley, Antonia Sebastian, Tim Leijnse, Dirk Eilander, John Ratcliff, Rick Luettich","doi":"10.1029/2023wr036727","DOIUrl":"https://doi.org/10.1029/2023wr036727","url":null,"abstract":"Estimates of flood inundation generated by runoff and coastal flood processes during tropical cyclones (TCs) are needed to better understand how exposure varies inland and at the coast. While reduced-complexity flood models have been previously shown to efficiently simulate TC flood processes across large regions, a lack of detailed validation studies of these models, which are being applied globally, has led to uncertainty about the quality of the predictions of inundation depth and extent and how this translates to exposure. In this study, we complete a comprehensive validation of a hydrodynamic model (SFINCS) for simulating pluvial, fluvial, and coastal flooding. We hindcast Hurricane Florence (2018) flooding in North and South Carolina, USA using high-resolution meteorologic data and coastal water level output from an ocean recirculation model (ADCIRC). Modeled water levels are compared to traditional validation datasets (e.g., water level gages, high water marks) as well as property-level records of insured flood damage to draw conclusions about the model's performance. SFINCS shows skill in simulating runoff and coastal processes of TC flooding (peak error of 0.11 m with an RMSE of 0.92 m) at large scales with minimal computational requirements and limited calibration. We use the validated model to attribute flood extent and building exposure to flood processes (e.g., runoff, coastal, compound) during Hurricane Florence. The results highlight the critical role runoff processes have in TC flood exposure and support the need for broader implementation of models capable of realistically representing the compound effects resulting from coastal and runoff processes.","PeriodicalId":23799,"journal":{"name":"Water Resources Research","volume":"23 1","pages":""},"PeriodicalIF":5.4,"publicationDate":"2025-03-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143618917","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}

{"title":"Predicting the Evolution of Extreme Water Levels With Long Short-Term Memory Station-Based Approximated Models and Transfer Learning Techniques","authors":"Samuel Daramola, David F. Muñoz, Paul Muñoz, Siddharth Saksena, Jennifer Irish","doi":"10.1029/2024wr039054","DOIUrl":"https://doi.org/10.1029/2024wr039054","url":null,"abstract":"Extreme water levels (EWLs) resulting from cyclones pose significant flood hazards and risks to coastal communities and interconnected ecosystems. To date, physically based models have enabled accurate prediction of EWLs despite their inherent high computational cost. However, the applicability of these models is limited to data-rich sites with diverse characteristics. The dependence on high quality spatiotemporal data, which is often computationally expensive, hinders the applicability of these models to regions of either limited or data-scarce conditions. To address this challenge, we present a Long Short-Term Memory (LSTM) network framework to predict the evolution of EWLs beyond site-specific training stations. The framework, named LSTM-Station Approximated Models (LSTM-SAM), consists of a collection of bidirectional LSTM models enhanced with a custom attention mechanism layer embedded in the architecture. LSTM-SAM incorporates a transfer learning approach applicable to target (tide-gage) stations along the U.S. Atlantic Coast. Importantly, LSTM-SAM helps analyze: (a) the underlying limitations associated with transfer learning, (b) evaluate EWL predictions beyond training domains, and (c) capture the evolution of EWL caused by tropical and extratropical cyclones. The framework demonstrates satisfactory performance with “transferable” models achieving Kling-Gupta Efficiency (KGE), Nash-Sutcliffe Efficiency (NSE), and Root-Mean Square Error (RMSE) ranging from 0.78 to 0.92, 0.90 to 0.97, and 0.09–0.18 m at the target stations, respectively. We show that LSTM-SAM can accurately predict not only EWLs but also their evolution over time, that is, onset, peak, and dissipation, which could assist in operational flood forecasting in regions with limited resources to set up physically based models.","PeriodicalId":23799,"journal":{"name":"Water Resources Research","volume":"18 1","pages":""},"PeriodicalIF":5.4,"publicationDate":"2025-03-14","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143618919","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}