دانشگاه شهرکرد پژوهش آب ایران 2008-1235 20 3 2026 09 23 Modeling Streamflow During the Snowmelt-Dominated Period Using LSTM and ERA5-LAND Reanalysis Data (A Case Study: The Snow-Covered Bazoft Basin) مدل‌سازی جریان دوره غالب ذوب برف با استفاده از LSTM و داده‌های بازتحلیل ERA5-LAND (مطالعه موردی: حوضه برفگیر بازفت) 1 14 116623 10.22034/iwrj.2026.14932.2634 FA علی فتاحی نافچی دانشجوی دکتری دانشکده مهندسی آب و محیط زیست، دانشگاه شهید چمران اهواز، اهواز، ایران محمدرضا شریفی دانشیار، دانشکده مهندسی آب و محیط زیست، دانشگاه شهید چمران اهواز، اهواز، ایران 0000-0001-6870-2023 امین خرمیان استادیار، دانشکده مهندسی آب و محیط زیست، دانشگاه شهید چمران اهواز، اهواز، ایران 0009-0008-4055-7117 Journal Article 2026 02 16 Extended Abstract Introduction Mountain snow reserves are a crucial component of the hydrological cycle in high-altitude basins, supplying a substantial portion of river flow during the warm season. Forecasting discharge during the snowmelt-dominated period is therefore essential for water resource management and flood mitigation, a challenge intensified by climate change impacts on melt timing. While physically-based and conceptual temperature-index models are common, they face limitations due to data scarcity or an inability to capture complex, non-linear melt-runoff processes. Consequently, this study employed a data-driven approach using a Long Short-Term Memory (LSTM) deep learning network to model the non-linear temporal dynamics of snowmelt. Specifically, an LSTM model was developed and evaluated to forecast daily discharge during the snowmelt period in the data-scarce, snow-fed Bazoft Basin, Iran. To address the lack of in-situ data, meteorological time series from the ERA5 reanalysis product were utilized as model inputs. The results demonstrated the LSTM model's effectiveness in learning the long-term dependencies between snowmelt processes and streamflow, offering a robust framework for seasonal runoff prediction in snow-dominated regions to support sustainable water management. Materials and Methods This study was conducted in the snow-covered Bazoft Basin, located in the northern Karun River catchment, Iran (31.6°–32.65° N, 49.56°–50.46° E). The basin experiences snow accumulation from December to March, with the snowmelt-dominated period occurring from mid-February to April, during which snowmelt is the primary contributor to streamflow. Daily streamflow data from the Landi hydrometric station were used as the target variable. Due to the scarcity of ground-based meteorological stations in this mountainous region, this study utilized the ERA5-LAND reanalysis dataset from the European Centre for Medium-Range Weather Forecasts (ECMWF). ERA5-LAND provides improved spatial resolution (9 km) compared to ERA5 (31 km) and has been extensively validated in previous studies over Iran, demonstrating high correlation with ground observations, particularly for temperature (R² > 0.95). The following daily variables were extracted: air temperature (T), precipitation (P), and snow water equivalent (SWE). The Gamma Test (GT), a non-parametric method for identifying the optimal input combination, was employed to select the most relevant input variables. Four different input combinations were evaluated: M1 (Q and SWE), M2 (Q, T, and SWE), M3 (Q, P, and SWE), and M4 (Q, P, T, and SWE). The combination with the minimum Gamma value and V-ratio was selected as the optimal input set. A Long Short-Term Memory (LSTM) neural network was developed to model the non-linear, long-term temporal dependencies inherent in the snowmelt-runoff process. The LSTM architecture includes forget, input, and output gates that regulate information flow, enabling the network to retain relevant information over extended time steps. The model was implemented using the Keras framework. The dataset was randomly divided into training (80%) and testing (20%) subsets. Various hyperparameters were optimized, including the number of hidden layers (1–3), the number of hidden units (10–40), and seven optimization algorithms (Adam, Adamax, SGD, RMSprop, Nadam, Adagrad, and Adadelta). Model performance was evaluated using the Nash-Sutcliffe Efficiency (NSE), Root Mean Square Error (RMSE), and Coefficient of Determination (R²). Results and Discussion The Gamma Test results identified the optimal input combination for LSTM modeling. Among the four evaluated combinations, M2 (discharge, temperature, and snow water equivalent) exhibited the minimum Gamma value (0.00151) and V-ratio (0.00412), indicating the smoothest input-output relationship with the lowest irreducible noise. Consequently, M2 was selected as the most appropriate input set for subsequent LSTM modeling. Evaluation of LSTM hyperparameters revealed that a single hidden layer architecture outperformed deeper configurations (2-3 layers), achieving the lowest RMSE (0.220) and MAE (0.125). Regarding hidden units, 20 neurons yielded optimal performance. Among seven optimization algorithms compared, Adamax demonstrated superior results with RMSE and MAE values of 0.37 and 0.18, respectively, and was therefore selected as the optimizer for the final model. The LSTM model with M2 inputs (Q, T, SWE) demonstrated exceptional performance in simulating daily streamflow during the snowmelt-dominated period. During the training phase, the model achieved NSE=0.994, R²=0.991, and RMSE=0.08. In the testing phase, performance remained excellent with NSE=0.994, R²=0.991, and RMSE=0.174. Visual comparison of observed and simulated hydrographs for the 2021-2022 and 2022-2023 water years showed that the LSTM accurately captured both low-flow conditions and peak discharges during snowmelt events. Scatter plots further confirmed this accuracy, with points closely clustered around the 1:1 line and R² values of 0.997, demonstrating that the model effectively reproduced the variance in observed streamflow across the entire flow regime. The superior performance of the LSTM model can be attributed to its inherent capability to capture long-term temporal dependencies through memory cells and gating mechanisms, which is particularly advantageous for snowmelt-runoff processes characterized by lagged responses and memory effects. The inclusion of SWE as an input variable proved crucial, as it directly represents the accumulated snowpack available for melt, while temperature controls the melt energy. Notably, the M2 combination (Q, T, SWE) outperformed combinations including precipitation (M3 and M4), suggesting that during the snowmelt-dominated period, the hydrological signal is primarily governed by snowmelt dynamics rather than concurrent rainfall. Comparison with previous studies demonstrated the competitiveness of the proposed approach. The obtained NSE (0.994) substantially exceeded values reported for traditional temperature-index models (NSE~0.80) and was higher than ANN-based snowmelt runoff predictions (NSE~0.93). The results were comparable with other LSTM applications in hydrological modeling (NSE~0.99). However, this study specifically contributed to the limited literature on LSTM application for snowmelt-dominated period streamflow modeling in data-scarce mountainous regions. The successful integration of ERA5-LAND reanalysis data addresses the critical challenge of data scarcity in mountainous basins. The high accuracy achieved demonstrated that reanalysis products can effectively substitute for ground-based meteorological observations when the latter are unavailable, provided they are appropriately validated for the study region. Conclusion This study demonstrated that the LSTM model, forced with ERA5-LAND reanalysis data (temperature and snow water equivalent) together with antecedent discharge, accurately simulated daily streamflow during the snowmelt-dominated period in the data-scarce Bazoft Basin (NSE=0.994, RMSE=0.174). The optimal performance was achieved with a single hidden layer, 20 hidden units, and the Adamax optimizer. The superior performance of the temperature-SWE-discharge combination over precipitation-inclusive inputs confirmed the dominance of snowmelt processes during the study period. This framework offers an effective solution for seasonal runoff prediction in snow-dominated mountainous regions lacking dense ground-based observations, supporting informed water resource management and planning. پیش‌بینی جریان رودخانه در حوضه‌های برفگیر طی دوره غالب ذوب برف، برای مدیریت منابع آب، دارای اهمیت است. مدل های یادگیری ماشین بعنوان مدل های داده محور، بواسطه آموزش روابط بین ورودی‌ها و خروجی‌ها، قابلیت زیادی برای شبیه سازی روابط پیچیده و غیر خطی دارند. شبکه‌های عصبی مصنوعی معمولی، بدلیل عدم ذخیره اطلاعات زمانی عوامل ورودی، قابلیت چندانی در مدل سازی سری زمانی فرایند پیچیده متاثر از ذوب برف را ندارند. رویکرد یادگیری عمیق LSTM، تحت عنوان حافظه بلند ترم های کوچک، برخی محدودیت های مدل های یادگیری ماشین را بواسطه خصوصیت ذخیره سازی اطلاعات زمانی، برطرف نموده است. در تحقیق حاضر، مدل LSTM برای مدل‌سازی جریان روزانه در دوره غالب ذوب برف و با استفاده از داده های بازتحلیل ERA5-LAND در حوضه آبریز برفی بازفت بکارگرفته شد. بر اساس آزمون گاما، بهترین ترکیب متغیرهای ورودی به مدل، شامل دما، آب معادل برف بازتحلیل‌ شده از پایگاه ECMWF و دبی پیشین تعیین شدند. نتایج مدل LSTM با یک لایه پنهان و بهینه ساز Adamax، مقدار NSE وRMSE، بترتیب با ۹۹۴/۰ و 174/0را در مرحله آزمون نشان داد. از اینرو مدل LSTM، بهمراه داده های باز تحلیل، بخوبی قادر به شبیه سازی جریان روزانه در دوره غالب ذوب برف، می باشد.

https://iwrj.sku.ac.ir/article_116623_c43f0c8ec2ae76b1703805a5554bcdd8.pdf

دانشگاه شهرکرد پژوهش آب ایران 2008-1235 20 3 2026 09 23 A comprehensive review of bridge pier scouring: Analysis of mechanisms and flow patterns, key factors, challenges, and protective measures مروری جامع بر آبشستگی پایه پل: تحلیل مکانیسم‌ها و الگوی جریان، عوامل کلیدی، چالش‌ها و راهکارهای حفاظتی 15 54 116690 10.22034/iwrj.2026.14891.2628 FA الهام قنبری عدیوی گروه مهندسی آب دانشگاه شهرکرد 0000-0002-7781-0138 محمد خسروی گروه مهندسی آب دانشگاه شهرکرد Journal Article 2026 01 26 Introduction Bridge pier scour is recognized as one of the most significant hydraulic challenges in the design, op-eration, and maintenance of bridge structures. It arises from the complex interplay between flow dy-namics, alluvial bed characteristics, and pier geometry. Scour occurs when alterations in flow patterns and increased bed shear stress lead to sediment erosion, ultimately compromising structural stability. In recent decades, the growing body of research on localized pier scour underscores its critical im-portance in mitigating both structural and environmental risks. Despite considerable advancements in numerical, experimental, and analytical modeling, significant uncertainties persist in predicting scour depth and extent. This study aims to provide a comprehensive review of the physical mechanisms governing scour, identify key influencing factors, and evaluate both structural and non-structural miti-gation strategies. The current research establishes a scientific framework for scour prediction, man-agement, and mitigation. By examining flow behavior around piers, sediment characteristics, and the application of numerical and artificial intelligence models, this framework enhances the safety and longevity of bridge infrastructure under variable hydrological conditions. Materials and Methods This article employs a systematic review and analytical approach to examine bridge pier scour from the perspectives of physical mechanisms, influencing parameters, and mitigation strategies. Initially, the theoretical foundations and classifications of scour types (including general, local, and combined scour) were synthesized from peer-reviewed literature. Subsequently, flow patterns around piers were analyzed using experimental data and numerical simulations (CFD and two-phase models) to elucidate the formation of vortices, wake currents, and flow separation zones. The influence of hydraulic pa-rameters (e.g., flow velocity, water depth, Froude number) and sediment properties (e.g., density, grain size, cohesion) was evaluated through a comprehensive review and categorization of empirical data from prior studies. In the second phase, scour mitigation techniques were compared across two prima-ry categories: “bed armoring” (e.g., riprap, gabions, and articulated concrete block mats) and “flow alteration measures” (e.g., collars, sacrificial piles, and submerged vanes). These approaches were as-sessed based on hydraulic efficiency, implementation cost, and long-term durability. Scour prediction methodologies—including empirical equations, single-phase and two-phase CFD models, Smoothed Particle Hydrodynamics (SPH), and machine learning algorithms—were comparatively evaluated to determine their accuracy, advantages, and limitations. Additionally, global case studies were reviewed to assess climatic and geomorphological impacts on scour behavior. The review follows a structured methodological framework comprising the following stages: (1) Literature collection and screening: Relevant studies were retrieved from reputable scientific databases using targeted keywords and fil-tered according to predefined inclusion criteria. (2) Thematic classification: Extracted literature was categorized into core themes (flow dynamics, influencing factors, mitigation strategies, predictive modeling, and future research directions). (3) Synthesis and critical analysis: Findings from diverse studies were analyzed to identify consensus, discrepancies, and overarching trends, which were then integrated into the article’s conceptual framework. (4) Comparative evaluation: Mitigation and model-ing techniques were benchmarked against key parameters, including efficiency, cost-effectiveness, structural stability, and environmental impact. (5) Identification of research gaps: The final section outlines critical knowledge gaps and future research needs, providing a forward-looking perspective on the field. Results and Discussions Comprehensive analysis of bridge pier scour mitigation techniques revealed that no single measure can entirely eliminate scour. However, a strategically selected combination of methods, tailored to site-specific hydraulic conditions, sediment characteristics, and economic constraints, can substantially reduce maximum scour depth. Integrating flow alteration techniques with bed armoring proves to be the most effective strategy. Among bed armoring solutions, riprap (particularly when installed in sloped configurations) can achieve scour reductions exceeding 80%. Articulated concrete blocks and gabions also serve as viable alternatives, albeit with certain design and maintenance limitations. Acoording to the analysis in the category of flow alteration measures, submerged vanes demonstrated scour depth reductions approaching 88%, while slots achieved reductions of 30–40%, depending on their geometric configuration. Sacrificial piles, despite requiring periodic replacement, reduce scour depth by up to 50%. The efficacy of collars is highly geometry-dependent: conventional collars exhib-ited limited performance, whereas modified designs incorporating fins, hooks, or polygonal shapes yielded reductions of 50–73%. Advanced configurations, such as lenticular or airfoil collars, showed highly variable effectiveness contingent upon specific design parameters. Pervious collars, which guide flow through controlled permeability, achieved approximately 78% scour reduction. Further-more, bed regulation structures like sills and curved plates dissipate incoming flow energy, resulted in 30–55% reductions and enhanced cross-sectional stability. Ultimately, the synergistic integration of flow alteration and bed armoring techniques represented the optimal mitigation approach. Optimal de-sign must account for local hydraulic conditions, sediment properties, flow intensity, and maintenance requirements to ensure both scour control and long-term structural and economic viability. In terms of predictive modeling, comparative analysis indicated that two-phase CFD models, particularly those utilizing an Eulerian–Lagrangian framework, offered superior accuracy in simulating scour initiation and evolution, despite higher computational demands. The SPH method has also demonstrated en-hanced capability in capturing complex vortex dynamics compared to conventional grid-based ap-proaches. Meanwhile, machine learning models (such as artificial neural networks and random forest algorithms) trained on extensive experimental datasets, exhibit strong potential for rapid, real-time scour depth estimation in practical engineering applications. Conclusion Bridge pier scour is a complex, multi-factorial phenomenon driven by the dynamic interaction be-tween flow hydrodynamics, pier geometry, and bed sediment properties. The integration of advanced numerical modeling, data-driven predictive tools, and site-specific mitigation strategies enables effec-tive scour management in bridge engineering. Developing a systematic framework for selecting coun-termeasures—based on hydraulic performance, long-term stability, and cost-effectiveness—represents a critical step toward enhancing the safety, durability, and environmental resilience of bridge infra-structure. Future research should prioritize investigating bridge vulnerability under climate change sce-narios and extreme flood events to improve adaptive design practices. آبشستگی موضعی اطراف پایه‌ پل به‌عنوان یک چالش بزرگ مهندسی هیدرولیک در جهان مطرح است که منجر به کاهش پایداری سازه، تهدید ایمنی و یا ایجاد خسارات اقتصادی قابل توجه می‌شود. این مطالعه اصول، سازوکارها و جنبه‌های مدیریتی پژوهش‌های کاهش و مدیریت آبشستگی پایه پل را به‌طور جامع بررسی می‌کند. محور اصلی این بررسی، تبیین الگوی جریان، فرایندهای ایجاد و گسترش آبشستگی اطراف پایه‌ها و تأثیر پارامترهای کلیدی مانند سرعت و عمق جریان، اشکال هندسی پایه و نوک پایه، دانه‌بندی و خواص رسوبی است تا یک درک منسجم از فرایندهای آبشستگی ارائه شود. همچنین نقاط قوت، محدودیت‌ها و چالش‌های عملی-اقتصادی دو رویکرد اصلی در کنترل این پدیده را توضیح می‌دهد: رویکرد افزایش مقاومت بستر در برابر جریان و رویکرد اصلاح میدان جریان. رویکرد نخست با مصالح مقاوم همچون سنگ‌چین، گابیون، کیسه‌های ژئوتکستایل پر از شن، دوغاب‌ریزی، تشک‌های بتنی مفصلی و صخره‌های مصنوعی سروکار دارد. رویکرد دوم به چگونگی به‌کارگیری سازه‌ها برای تغییر یا کنترل میدان جریان مانند طوق‌ها، شکاف‌های اسکله، شمع‌های قربانی، صفحات مستغرق و آستانه‌ها می‌پردازد. همچنین در ادامه مدل‌های پیش‌بینی آبشستگی پایه پل شامل مدل‌‌های تک‌فازی و دو‌فازی مبتنی بر دینامیک سیالات محاسباتی و روش‌های SPH، معادلات آبشستگی و مدل‌های یادگیری ماشین بررسی می‌شوند. محدودیت‌های آزمایشگاهی و میدانی، عدم قطعیت‌ها و نیاز به بهبود مدل‌سازی یادآوری شده و سپس فرصت‌های آتی با توجه به پیچیدگی موضوع مطرح می‌شود.

https://iwrj.sku.ac.ir/article_116690_c4709121f7d292d4cfa757e069f73ae7.pdf