ABSTRACT. Natural runoff is a key element for water cycle in river basin and hydrological calculation. Due to complex terrain, terrible weather, economic constraint or political effect, construction and management of hydrometric stations and discharge measurement is a rather tough thing in regions with none or less data. To solve this problem, river cross-section is initially classified as triangular-type and power-law-type, and then, according to hydraulic analysis, a regular formula associating water surface width and discharge has been deduced for both types. Combining remote sensing technique of real-time, efficient, large-scale and mass-data characteristics, a discharge estimating method for regions with none or less data has been presented. Further, it was tested with model experiment and field data. The results show that, for 56 estimated discharge in experiments, the averaged relative error is 18.77% and the number that less than 20% is 40, whose ratio is 71.43%. For the estimated discharge according to field data, the averaged relative error is 20.71%. The ratio of relative error less than 20% is 64.66%, while that less than 30% is 86.03%. It indicates that the proposed method is an effective and accurate way to estimate natural runoff, providing a solution for discharge estimation in regions with none or less data.

ABSTRACT. Auxiliary spillways are critical hydraulic components engineered to manage extreme flood discharges, mitigating dam overtopping risks and maintaining structural integrity. Given the rising frequency of high-magnitude hydrologic events attributed to climate change, it is imperative to achieve a high-fidelity understanding of the complex turbulent flows within these conduits to inform robust design and operational protocols. When operating under open-channel conditions, spillway conduits exhibit highly non-linear hydraulic phenomena, including free-surface fluctuations, three-dimensional turbulent shear, and non-uniform energy dissipation. This study employs a dual-methodology approach, integrating physical scale modeling with numerical simulation. A 3D Computational Fluid Dynamics (CFD) model was developed to replicate the boundary conditions of an experimental dam auxiliary spillway. The investigation quantitatively evaluates key hydraulic parameters, specifically: free-surface profiles and water-level variations; three-dimensional velocity vectors across the flow field; and pressure distributions along the conduit invert. The study aims to validate the CFD model’s predictive accuracy against laboratory measurements while leveraging numerical insights to characterize internal flow structures and turbulence mechanisms that remain elusive in physical modeling alone. Results indicate that the CFD simulations successfully captured the dominant hydraulic features observed in the physical model. The high degree of correlation between numerical and experimental datasets confirms that CFD is a reliable, high-resolution engineering tool for the hydraulic analysis and optimization of auxiliary spillways.

ABSTRACT. Flood risk is increasing under changing climatic conditions, requiring accurate prediction of river water levels during extreme events. In distributed rainfall–runoff and inundation models, river roughness critically controls simulated flood water levels. However, roughness coefficients are commonly treated as spatially uniform and temporally static, despite strong variability induced by in-channel vegetation and floodplain dynamics. Although satellite-derived vegetation indices provide spatially continuous information, their hydraulic linkage to river roughness at the basin scale remains insufficiently validated. This study develops a vegetation-informed dynamic roughness formulation for improved flood water level prediction. Satellite-derived datasets were integrated to estimate basin-scale dynamic river roughness. Sentinel-2 imagery was used to derive vegetation indices, MODIS LAI products were employed to train a machine-learning-based downscaling model for high-resolution LAI estimation, and ICESat-2 ATL08 data were used to extrapolate canopy height along the river corridor. The resulting LAI and canopy height maps were linked to river roughness through a vegetation-dependent resistance formulation. The dynamic roughness scheme was implemented in a distributed Rainfall–Runoff–Inundation (RRI) model. Simulations using static roughness were compared with vegetation-driven dynamic roughness simulations for three major flood events in the Yamato River basin (area: 1070 km²; main channel length: 68 km). Water level time series were evaluated against observations. Moderate relationships were identified between vegetation indices and river roughness. LAI showed a positive correlation with roughness within a limited range (LAI = 0.5–1.5; r = 0.45), while canopy height exhibited a negative correlation (r = −0.56). Hydraulic simulations indicated that dynamic roughness reduced RMSE by approximately 0.32 m relative to static simulations. Peak water levels decreased by an average of 0.73 m due to submergence-induced resistance changes. However, peak level reproducibility relative to observations improved only marginally, highlighting the complexity of vegetation–flow interactions and the need for further refinement of dynamic roughness formulations.

ABSTRACT. In the Takatsu River, located in Masuda City, Shimane Prefecture, Japan, changes in sediment dynamics have been observed as a result of postwar afforestation. Specifically, a reduction in sediment supply from mountainous areas and an increase in short-duration, high-intensity rainfall events have acted as major factors contributing to hydrological and geomorphological polarization. These changes have led to alterations in the riffle–pool structure, resulting in a noticeable decrease in the number of riffles. In the lower reaches of the Abashiri River system lies Lake Abashiri, a brackish lake characterized by a strong and stable halocline. Such brackish lakes (coastal lagoons) are typically surrounded by human living areas, and their watersheds are strongly influenced by agricultural, forestry, livestock, and other industrial activities, as well as by daily human life. Similarly, a decline in the population of sweetfish, once abundant in the Takatsu River, has been observed, indicating that the riverine environment is undergoing significant change. Consequently, river improvement projects must be implemented in a manner that minimizes further environmental impacts. The objective of this study is to predict riverbed variation using the two-dimensional numerical analysis model Nays2DH in order to support environmentally conscious river planning. As a first step, the characteristics of riverbed variation in the Takatsu River are investigated. The Takatsu River is a first-class river without dams and is characterized by pronounced meandering channel morphology. A sequence of riffle–pool structures extends from approximately 3.4 km upstream from the river mouth. The computational domain was defined as a 3.2 km reach from 3.0 km to 6.2 km upstream of the river mouth, encompassing notable features such as the Enkō riffle. The simulation results indicate a maximum deposition depth of 102 cm and a maximum erosion depth of 86 cm. The locations where significant riverbed variation occurred were already characterized by elevated bed levels prior to the simulation, suggesting that initial topographic conditions played a major role. In other sections, changes in channel width upstream and downstream caused increased flow velocity, which was considered to be the primary factor driving riverbed change. To improve the reproducibility of actual riverbed variation, future work will focus on modifying locations where elevated riverbed conditions existed prior to the simulation and examining their influence on sediment dynamics.

ABSTRACT. Flow development in rough-bed open channels strongly influences turbulence production, momentum transfer, and energy dissipation, yet integrated numerical experimental validation of such flows remains limited. This study presents a coupled numerical and experimental investigation of subcritical open-channel flow over rough boundaries. Three-dimensional numerical simulations were performed using ANSYS FLUENT, employing the Renormalization Group (RNG) k-ε turbulence model to resolve turbulence characteristics and the Volume of Fluid (VOF) method to capture free-surface behavior. Realistic bed conditions were represented using a sand-grain roughness formulation. Complementary laboratory experiments were conducted in a controlled rectangular flume, where an Acoustic Doppler Velocimeter (ADV) was used to measure streamwise velocity profiles and turbulence statistics, including lateral and vertical Reynolds stresses, at multiple downstream locations. Direct quantitative comparisons reveal good agreement between numerical predictions and experimental measurements, demonstrating the robustness of the adopted CFD framework in reproducing rough-bed flow development. Results show that both bed roughness and discharge significantly affect velocity distribution and Reynolds stress evolution. The validated numerical-experimental approach provides a reliable framework for improved prediction and design of rough-bed open-channel flows in natural and engineered systems.

ABSTRACT. During extreme flood events, complex flow structures develop around hydraulic structures, highlighting the need for detailed analysis of flow behavior and hydraulic characteristics around bridges. In this study, a three-dimensional numerical model was applied to simulate overtopping and pressurized flow around a bridge under extreme hydrologic conditions, and the resulting flow characteristics were analyzed. An unsteady Reynolds-averaged Navier-Stokes (URANS) type a non-linear k-ε turbulence model was employed, and the free surface was captured using the density function method. The computational domain had a length of 1.6 m, a width of 0.3 m, and a height of 0.2 m, with a bed slope of 1/2000. A bridge abutment with a length of 0.1 m, width of 0.1 m, and height of 0.05 m, together with a bridge deck having a length of 0.1 m, width of 0.3 m, and thickness of 0.025 m, was installed at a location 0.5 m downstream from the inlet boundary. The flow rate Q=0.0085 m³/s was imposed at inlet boundary, and no-slip boundary conditions were applied at the abutments, bridge deck, and walls. The computational mesh consisted of 2.4 million cells with grid spacing ranging from 0.0025 to 0.004 m. The numerical model was validated against free surface data from previous hydraulic experiments. The root mean square error (RMSE) and mean absolute percentage error (MAPE) were 0.0064 m and 2.13%, respectively, indicating good agreement between numerical predictions and experimental data. As flow rate increased, the water depth difference between upstream and downstream of the bridge became larger, and strong shear layers and recirculation zones developed due to the plunging overtopping flow downstream of the bridge. Hydraulic jump-like phenomena were observed around the bridge structure. During overtopping conditions, the maximum velocity ratio reached 2.0, while pressurized flow conditions produced maximum velocity ratios in the range of 1.8-2.7. In particular, a submerged orifice-type pressurized flow resulted in velocity ratio 2.7. For water depth ratios ranging from 0.15 to 0.48, and a linear relationship between these parameters was identified. The results of this study are expected to provide fundamental data for hydraulic stability assessment in bridge design and maintenance under extreme flood conditions.

Acknowledgements This work was supported by Korea Environment Industry & Technology Institute (KEITI) through Research and Development on the Technology for Securing the Water Resources Stability in Response to Future Change, funded by Korea Ministry of Climate, Energy, Environment(MCEE)(RS-2024-00397820).

ABSTRACT. Hydraulic modeling of river channels with vegetation patches requires high geometric fidelity to capture local flow heterogeneity, because these features generate complex turbulence and wake patterns. In particular, adequate resolution for numerical modeling must be refined to resolve the interfaces between vegetated and non-vegetated zones, where flow separation and localized blockage occur. However, coarse Digital Elevation Models (DEMs) often oversimplify these distinct topographic features through spatial averaging, thereby distorting simulated drag forces and flow diversion. To identify an appropriate terrain resolution for capturing these localized effects, this study evaluates how DEM resolution affects two-dimensional flow simulation results. The HEC-RAS 2D simulations were based on a large-scale outdoor flume experiment featuring eight evenly spaced willow patches along the channel centerline. The terrain was reconstructed from high-density 3D point cloud data and resampled into DEMs at resolutions of 0.1 m, 0.01 m, and 0.001 m. Simulations were conducted under two flow conditions (high and low discharge) and four vegetation scenarios; group-vegetated, single-dense vegetated, single-sparse vegetated, and non-vegetated. Spatially distributed Manning’s n values were applied using the momentum-based model to represent vegetation-induced flow resistance. In the non-vegetated case, model outputs were insensitive to DEM resolution primarily. In contrast, vegetated scenarios exhibited clear resolution-dependent behavior. Water surface level differences reached up to 0.03 El.m, while local velocity deviations near the vegetation patches varied by up to ±0.5 m/s between the coarsest and finest DEMs, with the high-resolution models demonstrating superior agreement with the experimental observations. Sensitivity tended to stabilize beyond the 0.01 m resolution, indicating a practical threshold for representing vegetation–flow interactions. These preliminary findings highlight that capturing the precise topography of vegetation patches is critical for reliable hydraulic modeling. Although terrain is not the sole factor, neglecting resolution distorts the simulation of local flow dynamics. Coarse terrain data may be sufficient for bare or uniform channels. Still, higher-resolution DEMs are essential to minimize prediction errors and consistently reproduce the complex hydrodynamics induced by vegetation patches.

ABSTRACT. Gated-circular structures are widely used in open channels, conduits, drainage control systems, and particularly in water storage tunneling systems for extreme rainfall events. The hydraulic characteristics at their inlets play a critical role in determining the discharge under various flow types. Variations in headwater, tailwater levels, and gate openings can induce different flow types which have distinct head-discharge relationships; thus, adaptive discharge computing is required to predict operational storage performance during extreme rainfall. However, the head-discharge relationship and the mechanisms governing flow type transitions at gated circular structures vary considerably depending on specific geometric boundary conditions and gate control, particularly when tailwater effects are present. Consequently, a systematic framework is required to reliably calculate discharge for hydraulic design and operational methods in water storage tunneling systems that utilize gated circular structures. The objective of this study is to investigate discharge coefficient characteristics and transition thresholds associated with four representative flow types: Uncontrolled Free Flow, Uncontrolled Submerged Flow, Controlled Free Flow, and Controlled Submerged Flow. To achieve these objectives, a laboratory-scale hydraulic model consisting of an upstream reservoir, a circular conduit, and a downstream outlet structure was constructed. Hydraulic experiments were conducted by systematically varying the upstream head, gate opening, and downstream tailwater depth. Experimental observations were used to analyze discharge behavior, and transition thresholds between flow types were identified, enabling a systematic classification. This study presents a quantitative hydraulic framework for classifying flow types at gated circular structures based on experimentally identified transition thresholds and for analyzing flow-dependent discharge coefficient characteristics. The proposed framework distinguishes submergence ratio characteristics and head-dependent variations in discharge coefficients, thereby improving the reliability of discharge estimation and supporting efficient hydraulic performance. These findings contribute to enhancing the operation of both general and gated circular hydraulic structures and provide an important foundation for improving mitigation against flooding caused by extreme climate change.

ABSTRACT. Great-depth stormwater detention and drainage facilities are designed to temporarily store and discharge stormwater to mitigate urban flooding, particularly given the increasing frequency and intensity of extreme rainfall events. During operation, sediments transported by stormwater accumulate within the system, progressively reducing storage capacity and impairing hydraulic efficiency. These sediments often contain contaminants such as heavy metals and organic matter, which can be released into nearby rivers and groundwater during drainage, potentially exerting adverse effects on receiving water area. Consequently, effective management strategies for accumulated sediments are essential to ensure stable and sustainable water system operations. While periodic dredging is required to maintain functional performance, it generates large volumes of sediment waste, highlighting the need for sustainable resource management. Rather than relying on simple disposal, it is critical to assess the physicochemical characteristics of dredged sediments to determine their suitability for recycling. Specifically, evaluating both total and labile contaminant concentrations is necessary, as labile fractions are directly related to contaminant mobility, bioavailability, and potential environmental risk. This study aims to establish a sustainable management plan for dredged sediments from great-depth facilities by assessing their contaminant characteristics and reuse feasibility. Mixed sediments collected from the maintenance to the ventilation shafts were analyzed for total and labile concentrations of heavy metals and organic contaminants. Results were compared with relevant regulatory standards to evaluate environmental safety. Furthermore, recycling potential was assessed by examining the sediments' applicability as construction materials and soil amendments. This study provides a scientific basis for efficient sediment management and supports the development of sustainable recycling strategies within an integrated stormwater management framework.

ABSTRACT. The Loess Plateau in China suffers from severe soil erosion, with steep and unstable slope. The unique topographical features in this region leads to significant heterogeneity in spatial erosion-deposition pattern, which bring challenges to the simulation of flood and sediment processes under heavy rainfall. In this study, a distributed flood and sediment process model for the Loess Plateau was proposed based on infiltration excess runoff and soil erosion module. The erosion process of three geomorphic units (i.e., slope, gully and groove) is physically generalized and simulated by energy balance in soil erosion. The proposed model is applied to three nested catchments in Wuding River Basin, an important sediment-laden tributary of the Yellow River in the Loess Plateau. The Wuding River Basin covers 3,0261km2 with nested catchments of the Xiaoli River, Dali River, and its own. The area of Xiaoli River and Dali River is 867 km2 and 3893 km2, respectively. The nested catchments facilitate attempts to expand the spatial scale of model applications. The Nash-Sutcliffe efficiency coefficient of runoff simulation in the Xiaoli River, Dali River, and Wuding River was higher than 0.62, 0.56, and 0.52. The simulated flood peak value, and peak time of the simulated flood were in good agreement with measured flood events in nested catchments. However, the accuracy of sediment concentration was relatively low, especially in the Wuding River. This is due to the simplification of sediment transport process and the insufficient consideration of gravity erosion and soil and water conservation measures. Therefore, the model will add gravity erosion module and soil and water conservation measures to improve simulation accuracy and efficiency in the future, providing an effective tool for simulating flood and sediment dynamics in the Loess Plateau.

ABSTRACT. In the Kizu River basin within the Yodo River basin, significant riverbed lowering has occurred as a result of reduced sediment supply caused by past gravel extraction and dam construction. This riverbed lowering has led to fragmentation between the “riverside pools” (still water areas) in the floodplain and the main channel. Riverside pools are essential for fish spawning and juvenile growth sites. However, reduced water exchange between these pools and the main channel results in deteriorated water quality, accumulation of aquatic organic matter, and increasing threats to the aquatic ecosystem. This study investigates whether the “Seigyu”, a traditional Japanese crib spur dyke, can enhance inflow into riverside pools and thereby improve their environmental conditions. Using iRIC (ver. 4.0) software, two-dimensional numerical simulations flow and riverbed changes were conducted for the reach downstream of the Tamamizu Bridge on the Kizu River. Inflow rates into the riverside pools were compared under model peak discharge conditions of 500 m³/s and 1,000 m³/s, both with and without the installation of the “Seigyu” structure. Installation of “Seigyu” increased the threshold discharge required for inflow into the riverside pools. Without "Seigyu", inflow began at a discharge of 481 m³/s, whereas a higher discharge of 535 m³/s was required when “Seigyu” was installed. This increase is attributed to “constriction resistance (an orifice-like effect)” generated by “Seigyu” at the entrance of the riverside pools, which narrows the effective flow passage. Furthermore, because “Seigyu” is a permeable structure, it dissipated flow energy, thereby reducing the pressure gradient necessary to induce flow into the riverside pools. Regarding riverbed changes, the installation of "Seigyu" restricted the effective channel width, resulting increased flow velocity in the main channel. This led to enhanced scour and channel deepening in the main channel, suggesting a potential acceleration of channel “polarization.” In contrast, stable sediment deposition was observed in the area behind "Seigyu", indicating beneficial effects such as bank protection and the provision of a refugia for aquatic organisms during floods events. “Seigyu”, a traditional Japanese crib spur dyke, was found to be highly effective for “defensive” functions, such as bank protection and the creation of habitats. However, the results indicate that it is insufficient for achieving the “offensive” objective of actively inducing inflow into riverside pools. Future research should focus on identifying optimal placement and configurations of “Seigyu” structures, including arrangements such as staggered layouts in order to enhance their effectiveness for environmental restoration.

ABSTRACT. The intricate morphodynamic behavior of meandering rivers is controlled by the interplay of sediment transport, flow hydraulics, and bank erosion processes. The primary cause of meander bend formation and evolution is curvature-induced secondary circulation, which promotes inner-bank deposition and increases outer-bank erosion, resulting in progressive channel migration and alternating bar-pool sequences. Many current models are based on simplified curvature-based formulations that do not adequately represent turbulence structure, sediment redistribution, and dynamic width adjustment, despite the fact that considerable progress has been made through theoretical analyses, laboratory experiments, and numerical simulations. Additionally, vertical flow structures are suppressed by the majority of two-dimensional depth-averaged methods, while fully three-dimensional models are computationally demanding and infrequently integrated with full morphodynamic coupling. In order to increase the physical realism of meander evolution simulations, this work creates an improved two-dimensional morphodynamic model. The framework combines the depth-averaged shallow-water equations with a physics-based bank migration algorithm that permits variable channel width, a hybrid eddy-viscosity turbulence formulation, a curvature-driven secondary flow representation, and flexible sediment transport models for both bedload and suspended load. The Exner sediment continuity equation is used to calculate bed evolution, and an adaptive curvilinear grid is used to preserve geometric consistency and numerical stability during channel deformation. A structured curvilinear mesh with explicit time-stepping and finite-difference discretization is used to solve the model while adhering to morphodynamic and hydrodynamic stability constraints. Realistic development of alternating bars and pools, curvature-controlled redistribution of shear stress, outer-bank intensification of erosion, and progressive morphodynamic adjustment toward quasi-equilibrium conditions are all demonstrated by simulations performed on a laboratory-scale sine-generated channel. Strong agreement between the results and well-known benchmark studies attests to the validity of the suggested framework. In addition to offering practical relevance for river engineering, flood hazard mitigation, and sustainable channel restoration, the developed modeling approach offers a strong basis for future extension to fully three-dimensional simulations.

ABSTRACT. Transient mixed flows pose significant safety risks to long-distance water conveyance systems, yet their prediction faces a critical bottleneck, namely the efficiency–accuracy dilemma. Conventional Preissmann Slot Methods (PSM) often suffer from prohibitive computational costs due to stability constraints, or they compromise physical accuracy, which limits their applicability in large-scale engineering practice.

To address this issue, this study introduces a Weighted-Average Preissmann (WAP) slot method. Unlike static approaches, the WAP model dynamically adjusts the slot geometry to achieve a target wave speed. This mechanism enables a smooth, exponential transition of wave speeds based on the local pressure head, thereby relaxing the stiff Courant–Friedrichs–Lewy (CFL) constraint. This physics-based flexibility further facilitates coupling with Adaptive Mesh Refinement (AMR), allowing the use of coarser meshes in transition zones.

The model was validated against analytical solutions and laboratory experiments of rapid filling pipes reported in the literature. The results indicate that the WAP method effectively mitigates numerical oscillations while yielding pressure predictions comparable to those of conventional PSM approaches. Furthermore, in the examined rapid-filling scenarios, the coupled WAP–AMR framework demonstrated substantial gains in computational efficiency. These findings suggest that the proposed method offers a promising and robust approach for simulating hydraulic transients.

ABSTRACT. Experimental Evaluation of Velocity Distribution and Bed Shear Stress in a Meandering Compound Channel Understanding flow structure and bed shear stress in meandering compound channels is essential for interpreting hydraulic behavior in complex channel geometries. In this study, laboratory-scale hydraulic experiments were conducted in a meandering compound open channel to experimentally evaluate the spatial characteristics of three-dimensional velocity distributions and bed shear stress. The experiments were performed in a rigid-bed meandering compound channel consisting of a main channel and adjacent floodplains. Bed shear stress was directly measured using a load-cell-based shear force sensor. Velocity measurements were obtained using a micro Acoustic Doppler Velocimeter (ADV) at multiple vertical and transverse locations across both the main channel and the floodplain. Based on time-resolved velocity measurements acquired at high sampling frequency, depth-wise mean velocities were calculated, and turbulence-related quantities such as Reynolds stresses and turbulent kinetic energy were derived. Bed shear stress was primarily estimated by applying the logarithmic velocity profile (log-law) to depth-wise mean velocity distributions and was used as the main shear stress estimation method in this study. The log-law-based shear stress was then compared with directly measured bed shear stress obtained from the load-cell sensor to examine their differences and characteristics. In addition, Reynolds stress-based and turbulent kinetic energy-based shear stresses derived from ADV measurements were evaluated and compared to assess the applicability and limitations of different shear stress estimation methods. The results show pronounced lateral and vertical non-uniformity of velocity distributions near the interface between the main channel and the floodplain. Localized increases in bed shear stress were observed due to secondary flow effects induced by channel curvature. Differences in estimated shear stress among the applied methods were identified, highlighting the importance of combining direct shear force measurements with velocity-based estimation approaches under complex flow conditions in meandering compound channels. The experimental results provide a comprehensive dataset of velocity structure and bed shear stress behavior in meandering compound channels and can serve as fundamental experimental reference data for shear stress evaluation in complex open-channel flows.

This work was supported by the Korea Environmental Industry & Technology Institute (KEITI) through the “Research and Development on the Technology for Securing Water Resources Stability in Response to Future Change” Program, funded by the Korea Ministry of Climate, Energy and Environment (MCEE) (Grant number RS-2024-00397820).

ABSTRACT. Accurately predicting flow resistance in vegetated compound channels requires upscaling (i.e., averaging and parameterization) from plant/patch-scale drag to reach-scale hydraulic friction. Momentum-based vegetation-resistance models embody different upscaling assumptions, spanning reach-averaged momentum balance with canopy-averaged drag, patch/stand-scale momentum closure yielding reach-scale friction, and reach-scale lumped formulations that account for foliage–stem effects. Despite their analytical advances, their relative performance under full-scale floodplain-vegetation conditions—where flow partitioning and main-channel–floodplain interactions are pronounced—remains insufficiently validated against direct reach-scale measurements. Here, we use full-scale compound-channel experiments with floodplain vegetation patches, in which the water-surface slope was directly measured to derive a reference reach-scale friction factor. We then apply representative momentum-based models to identical configurations and benchmark predicted friction factors against the measured reference to quantify model bias and identify key controlling parameters. This comparison provides practical guidance for selecting and calibrating vegetation-resistance parameterizations (closures) in compound-channel modeling and design. Full-scale experiments were conducted in an asymmetric concrete compound channel (S₀ = 0.0015). The 180 m-long test reach included a 60 m-long vegetated floodplain section (main channel width = 4.04 m; floodplain width = 4.14 m), where 0.77 m-high foliated, branched model vegetation was installed in a staggered array at a density of 5.56 stems m⁻². Two flow conditions (Q = 1.67 and 2.29 m³s⁻¹) were tested while maintaining a constant floodplain flow depth of 0.38 m by adjusting the downstream gate. Reach-scale bulk flow resistance was computed from the measured water-surface slope (pressure sensors), discharge (ADCP), vegetated-floodplain approach velocity (ADV), and channel geometry (total station). Vegetation-induced resistance was obtained by separating boundary resistance from the bulk resistance and was benchmarked against estimates from representative momentum-based models. The bulk friction factors were 0.098 and 0.113 at Q = 1.67 and 2.29 m³s⁻¹, respectively, with vegetation-induced friction accounting for ~63% of the bulk values (0.062 and 0.070). When applied to the same configurations, momentum-based models—both those assuming rigid vegetation and those incorporating flexible reconfiguration—produced vegetative friction factors that were approximately 4–6 times larger than the experimental estimates. Overall, the results demonstrate a substantial difference between reach-scale friction factors derived from full-scale measurements and those inferred from momentum-based vegetation-resistance parameterizations under vegetated floodplain compound-channel conditions.

Acknowledgement: This research was funded by the Korea Environment Industry & Technology Institute (KEITI) through the Climate Resilient R&D Project for Water-Related Disaster Management, funded by the Korea Ministry of Climate, Energy, Environment (MCEE)(RS-2022-KE002091).

ABSTRACT. The long-wave equations are widely used to simulate flows in rivers and shallow lakes. However, challenges remain in the treatment of boundary conditions and, particularly, in accurately reproducing short-wavelength waves and complex hydraulic phenomena such as flood inundation. The purpose of this study is to compare the reproduction accuracy of various numerical schemes and to establish a more accurate and stable numerical solution method for the long-wave equations by combining these schemes. In this study, one-dimensional long-wave equations were employed for flow simulations. First, the accuracy and stability of the Constrained Interpolation Profile (CIP) scheme and the Ultimate Quickest (UQ) scheme for the advection term were compared using a simulation of the advection of a rectangular water mass. Next, the characteristics of the explicit (EP) scheme and the novel implicit (NiP) scheme, which utilize the upwind difference method for the advection term, were investigated through simulations of hydraulic jump and dam-break problems. Furthermore, dam-break phenomena were simulated using composite schemes that combined the advection schemes (CIP and UQ) with the time integration schemes (EP and NiP). Error evaluations were conducted by comparing the results of these composite schemes with theoretical solutions. The comparison of the CIP and UQ schemes showed that the CIP scheme achieved higher accuracy in regions with steep gradients, while the UQ scheme performed better in relatively flat regions. Overall, the UQ scheme was found to provide higher accuracy due to the occurrence of aliasing in some parts of the CIP scheme. In the comparison between the EP and NiP schemes, the NiP scheme showed closer agreement with theoretical solutions and exhibited higher stability for both hydraulic jump and dam-break problems. Finally, the error evaluation of the composite schemes revealed that introducing either the CIP or UQ scheme for the advection term significantly improved the reproduction accuracy of both the EP and NiP schemes in dam-break simulations. Among these, the UQ-based composite schemes produced the most accurate results. These findings demonstrate that combining fundamental time integration schemes (EP and NiP) with advanced advection schemes (CIP and UQ) enables higher accuracy and stability in numerical solutions of the long-wave equations.

ABSTRACT. In response to the increasing occurrence of extreme rainfall and flooding associated with climate change, the demand for underground drainage tunnels has continued to grow. In underground drainage tunnels, the flow initially behaves as open-channel flow with a free surface at the inlet, and gradually transitions to pressurized pipe flow as time progresses. To establish a fundamental basis for predicting unsteady two-phase flow behavior in such systems and for reflecting these characteristics in engineering design, a reliable numerical framework is required. In this study, three-dimensional two-phase flow simulations were conducted using OpenFOAM. The Reynolds-Averaged Navier–Stokes (RANS) equations were employed as the governing momentum equations, and turbulence was modeled using the RNG k-ε model. The computational domain consisted of a circular pipe with a diameter of 0.025 m and a length of 8 m. Numerical simulations were performed by applying various combinations of superficial liquid and gas Reynolds numbers at the inlet boundary. The characteristics and frequency of slug flow were analyzed. The numerical results demonstrated that the translation and development of slugs could be qualitatively reproduced through comparison with hydraulic experimental observations. The liquid slug length showed a decreasing trend with increasing superficial liquid Reynolds number, and flow features such as air pockets and liquid droplet entrainment were clearly observed. The slug frequency exhibited a relatively consistent trend at a superficial gas Reynolds number of 1,377, whereas no clear trend was identified at 4,320. The results indicate that the proposed three-dimensional numerical model provides a useful basis for predicting complex unsteady two-phase flow behavior in underground drainage systems.

Acknowledgements This work is financially supported by Korea Ministry of Climate, Energy, Environment(MCEE) as 「Climate Resilient R&D Project for Water-Related Disaster Management)(RS-2024-00397821)」

ABSTRACT. Artificial reefs are widely recognized as multifunctional marine structures that provide habitats for marine organisms, deter illegal and unlicensed fishing activities, conserve coastal fishery resources, and enhance the income of local fishing communities. However, when artificial reefs are deployed on natural seabed terrains, local scour around the structures can critically influence their structural stability and long-term performance. Therefore, integrated hydraulic model experiments and numerical modeling are required prior to installation to comprehensively analyze the hydrodynamic characteristics and potential local scour behavior around artificial reefs. This study investigates the local scour characteristics of three types of artificial reefs planned for installation in the coastal waters near Yeonpyeongdo and Daecheongdo Islands in the northwestern West Sea of Korea through a combined approach of hydraulic model experiments and numerical modeling. The hydraulic model experiments were conducted in a wave flume using a three-dimensional physical model at a scale of 1:50 under various irregular wave conditions corresponding to the 50-year and 100-year return-period design waves for the coastal waters of Yeonpyeongdo and Daecheongdo Islands. In the numerical modeling, a new OpenFOAM-based approach incorporating the Immersed Boundary Method (IBM) was developed to accurately represent submerged structures with complex geometries. The model was implemented in the foam-extend (version 4.0) environment and coupled with olaFlow, which is widely used for free-surface wave simulations, to simulate wave generation, propagation, and interactions with the structures. The developed numerical model successfully represented the complex geometries of the artificial reefs within the computational domain and showed good qualitative agreement with the hydraulic model experiments in reproducing the flow behavior around the structures. This study presents an integrated experimental–numerical framework for evaluating local scour characteristics around artificial reefs and is expected to provide fundamental reference data for the design and layout optimization of artificial reefs in future coastal applications.

ABSTRACT. Estimating turbulent flow and bed shear stress on riverbeds is crucial for understanding sediment transport in rivers. This understanding significantly impacts flood control and environmental management, as water flow and bed shear stress shape river morphology. However, directly measuring these parameters in flooding rivers is challenging due to the large amounts of sediment and debris being transported and changing bed morphology during floods. As a result, researchers have proposed indirect evaluation methods that utilize pressure measurements on the riverbed. This study explores the relationship between pressure fluctuations on riverbed roughness and the surrounding turbulent flow structures through laboratory experiments. A channel was designed to replicate the river bed of gravel-bed rivers. A laboratory flume experiment using Particle Image Velocimetry (PIV) was conducted to analyze the flow field around the bed roughness. The roughness unit for pressure measurement, 0.2 m in diameter, equipped with three pressure sensors—positioned at the front (P1), top (P2), and back (P3) in relation to the flow direction, was used for the flume experiment. The study provided several significant findings. First, a strong spatial correlation was observed between the pressure measured by the sensors and the velocity field, specifically at a scale of 1/10 of the roughness unit diameter. Second, the front and back pressures were influenced by the circulatory flows at the upstream and downstream of the roughness measuring unit. As a result, these pressures were sensitive to the arrangement of bed roughness. In contrast, the top pressure was relatively unaffected by the bed roughness configuration, and the variation in pressure fluctuation was linked to bed shear stress. The ratio of pressure fluctuation variation to bed shear stress was found to be consistent with previous studies. Field measurements of bed shear stress allow for a comprehensive analysis of sediment transport. This includes determining whether the supply of sediment is adequate or if sediment is limited by comparing the estimated bed load with empirical bed-load formulas.

ABSTRACT. In recent decades, flooding increased profoundly mostly due to hydrological characteristics changes in catchments, population growth and urbanization. On July 4, 2022, a devastating flood ravaged 40 villages and 3 fatalities occurred at Kampung Iboi, Baling Kedah causing 3546 people evacuated with RM25.91 million of total losses. Rainfall station at Kampung Iboi recorded rainfall around 36 mm in 3 hours, which was considered low and does not seems leading to serious debris flow, while Malaysian Meteorological Department’s radar observations demonstrations significant rainfall received at the peak of Mount Inas. In view of the uncertainties, this study aimed to assess the Kupang River channel capacity, flood scenario and possible magnitude of floods during the flood event on July 4, 2022 as well as to propose strategic measures in minimizing the effects of future flooding for Kupang River. The methodology involves data collection and analysis, hydrodynamic model using InfoWorks Integrated Catchment Modelling (ICM), model verification and simulation for the flood event. River channel capacity and the estimation of flood volume were also calculated. The simulation results highlight Kupang River experienced flood seriously from CH 500 to CH1050 and higher flood depth was observed near the right bank from CH 500 to CH 1000 due to lower ground level. Hence, an action plan in guiding the policy makers and stakeholders especially on river improvement, monitoring and enforcement of Erosion and Sedimentation Control Plan (ESCP) as one of the Best Management Practices (BMPs) for Kupang River Catchment is imperative and urgent.

ABSTRACT. Accurate prediction of local scour development is a critical factor in ensuring the stability of hydraulic structures and preventing engineering disasters. While conventional numerical models often simplify riverbeds by assuming a single representative grain size, natural riverbeds consist of various sizes of sediments where erosion and deposition behaviors vary significantly depending on the particle size. Neglecting this non-uniformity can lead to inaccurate estimations of local scour depth and morphological changes. This study presents a two-dimensional coupled hydro-morphological model developed to evaluate the impact of grain-size distribution on riverbed dynamics around hydraulic structures. The numerical framework employs a finite volume method based on a triangulated irregular network to handle complex geometries and utilizes a Riemann solver. The model couples the shallow water equations with an advection-diffusion equation and a bottom evolution equation, explicitly accounting for grain-size-dependent settling velocities and critical shear stresses for multiple sediment fractions. This approach allows for the analysis of not only bed elevation changes but also the spatial variation of bed material composition. To validate the model’s performance, simulations of dam-break flows and local scour around bridge piers are conducted. The study benchmarks the proposed multi-fraction model against both experimental data and results from single-grain simulations. Preliminary assessments indicate that single-grain models tend to overestimate erosion depth over narrow areas. In contrast, the proposed model is expected to reproduce more realistic, gentler erosion patterns over broader regions by capturing the differential transport of varying particle sizes. The final presentation will demonstrate the comparative accuracy of the model, highlighting that incorporating grain-size distribution is essential for reliable stability assessments in hydraulic engineering.

ABSTRACT. Recent climate change has intensified variability in the global ocean environment, resulting in gradual increases in both the frequency and severity of extreme waves. The seas around South Korea exhibit various types of extreme waves generated not only by typhoons but also by strong winter monsoons over the Yellow Sea and the East Sea. In particular, non-typhoon extreme waves generated by winter monsoon winds are characterized by sudden occurrence and have a short propagation time to the coast, making accurate prediction and timely response challenging. Such extreme waves were found to cause approximately four incidents of human casualties and property damage per year on average, highlighting the importance of developing extreme wave forecasting system. Existing operational wave forecast models in South Korea rely on default wave parameter configurations, thereby limiting their ability to accurately predict non-typhoon extreme wave generation and propagation. Such extreme waves are driven by diverse meteorological conditions and exhibit a range of spectral shapes and wave climates, indicating the necessity for wave forecast model calibrations optimized for different types of extreme waves. In the present study, a new definition of non-typhoon extreme waves is proposed based on extreme value analysis and characteristics of historical non-typhoon extreme waves are assessed using buoy observations. Wave climates and associated meteorological conditions during the non-typhoon extreme wave events are analyzed and the historical events are classified into regional wind-sea-dominant and swell-dominant extreme waves. Finally, pre-optimized ADCIRC+SWAN model configuration sets resulting in the best performance for simulating different types of non-typhoon extreme waves are developed. These pre-optimized models will be incorporated into the operational wave forecast system to enhance the prediction accuracy of non-typhoon extreme waves. In addition, a framework of wave model calibration developed herein can be applied to other wave climates worldwide.

ABSTRACT. During dam discharge, supersaturated total dissolved gas (TDG) is generated in the plunge pool and transported downstream for a long distance. Fish living in supersaturated TDG water may suffer from gas bubble disease and even death. Material transport in large deep reservoirs is influenced by the operation of upstream and downstream hydropower stations, resulting in highly complex transport characteristics. It is necessary to study the transport process of supersaturated TDG in large deep reservoirs to determine the transport time and improve the effectiveness of ecological management measures. This study utilized a laterally averaged numerical model to investigate the factors influencing the supersaturated TDG transport time in a large deep reservoir. The Baihetan (BHT)–Xiluodu (XLD) hydropower stations were selected as the research object. Simulations under various BHT–XLD joint operation strategies were conducted to analyze the supersaturated TDG transport time in the XLD Reservoir. It is revealed that the transport time of supersaturated TDG is associated with the discharge flow and reservoir characteristics. A power function relationship was identified between the transport time and the transport distance as well as the discharge flow. The relationship among the transport time, water depth and volume follow a linear function. Furthermore, a quantitative relationship between the transport time and these four influencing factors was established. The results provide a scientific basis for the accurate prediction of supersaturated TDG transport process in large deep reservoirs and the formulating effective regulation and early warning schemes.

ABSTRACT. Vegetation (riparian trees such as willow) colonizing bars and islands is a key factor controlling channel pattern and braided river morphodynamics. In steep gravel-bed rivers in cold regions of Japan, willows often establish rapidly after floods on emergent bar surfaces and gradually expand into low-flow channels. This process is known to gradually narrow channel width, and channel planform shifts from multi-thread to single-thread in several rivers. However, dynamic vegetation effects such as vegetation invasion, growth, and mortality due to aging or flush away by floods can influence channel morphodynamics, particularly during large floods. This study investigates how flood-induced vegetation loss influences channel planform dynamics in braided rivers through a combined approach of flume experiments and numerical analysis. 　The experimental flume was 11.7 m long and 3.0 m wide with a bed slope of 1/100, and the bed material was uniform sediment with a mean grain size of 0.77 mm. A sinuous channel with two wavelengths and a meander angle of 28.7° was prepared as the initial condition. A constant discharge of 0.00276 m³/s was supplied for 5 hours with sediment feed from upstream. Two cases were examined: Case 1 without vegetation and Case 2 with bentgrass (outside the low-flow channel before the run). Results showed that a braided channel was formed in the no-vegetation case, whereas a multi-thread channel with relatively high sinuosity was formed in the bentgrass case. Although channel deformation due to bar migration appeared in both cases, the bentgrass case showed a larger lateral channel migration across the flume width. 　Next, numerical analyses were conducted using a two-dimensional morphodynamic model (iRIC Nays2DH) with an added vegetation flush away model. The results, comparing a no-vegetation case with densely vegetated cases, matched well with the experiment. When vegetation was flushed away during the flood at the cut bank, the location became fixed and channel sinuosity gradually increased. In contrast, when vegetation was strong (not flushed away) during the flood, lateral channel migration was suppressed. When the vegetated area and vegetation density were small, the vegetation effect was minor and the channel tended to braid. These results show that flood-induced vegetation loss plays an important role in controlling channel mobility and channel planform dynamics in steep gravel-bed braided rivers.

ABSTRACT. Cadmium (Cd) enters sediments through both natural and anthropogenic sources and can be taken up and accumulated by benthic organisms, potentially posing adverse effects on biodiversity and ecosystem health. Assessing Cd bioaccumulation in benthic organisms is therefore essential for the protection of aquatic ecosystems and for evaluating potential benthic ecological risks. In this study, a biotic ligand–toxicokinetics (BL–TK) model was developed to predict Cd bioaccumulation in the benthic midge Glyptotendipes tokunagai, accounting for both dissolved Cd uptake and sediment-bound Cd ingestion pathways. Model parameters for each exposure pathway were estimated using an interspecies extrapolation approach. This approach enables the estimation of model parameters for the target species by using data from previously studied reference species based on phylogenetic and physiological similarities. For the uptake pathway, Chironomus riparius was selected as the reference species, whereas Hydropsyche californica was used for the ingestion pathway. For the uptake pathway, the maximum Cd uptake rate (Jmax) and the conditional binding constants for Cd and Ca (KCd and KCa) were estimated as 2.10 µmol·g-1·h-1, 1,763 L·mol-1, and 3,460 L·mol-1, respectively. For the ingestion pathway, the particle ingestion rate constant (kig) and assimilation efficiency (AE) were estimated as 0.00604 g·g-1·h-1 and 96%, respectively. In addition, the elimination rate constant (ke) and growth rate constant (g) were estimated as 0.018 h-1 and 0.00024 h-1, respectively. Model validation results showed that the predicted Cd bioaccumulation agreed well with experimentally measured values, with deviations within one order of magnitude. The developed predictive model can be used to estimate the biota–sediment accumulation factor (BSAF), a quantitative indicator of contaminant accumulation between sediments and benthic organisms, and to support site-specific ecological risk assessments.

ABSTRACT. During floods, floating debris often accumulates at hydraulic structures and alter the flow in ways that are not well captured by simplified hydraulic indicators. In practice, similar water levels or discharges can lead to very different accumulation outcomes. From our perspective, this inconsistency points to missing three-dimensional controls that govern accumulation stability and failure. The objective of this study is therefore not to document debris effects in general, but to identify which hydraulic processes actually control whether an accumulation remains stable or breaks down under increasing discharge. We employ a physics-based, open-source three-dimensional numerical model developed in OpenFOAM to simulate floating debris transport and accumulation at a hydraulic structure. The model setup follows a laboratory flume configuration used in controlled debris-feeding experiments and reflects observed accumulation geometries rather than idealized debris representations. Debris motion and trapping are resolved dynamically over a range of discharges. Model behavior is evaluated through direct comparison with measured water surface elevations, three-dimensional velocity fields, and indicators of flow regime, using the experimental data as a reference rather than a calibration target. The simulations reproduce the main hydraulic responses observed in the experiments, including debris-induced backwater, local acceleration near flow constrictions, and discharge-dependent changes in flow regime. More importantly, they reveal that accumulation stability is closely linked to localized flow contraction beneath the debris carpet. In these regions, near-bed shear stress and turbulence increase sharply and unevenly. We interpret the spatial pattern of this shear stress, rather than bulk flow depth or mean velocity, as the primary factor controlling the transition from stable accumulation to squeezing and eventual failure. By resolving these three-dimensional processes, the proposed framework provides a practical basis for assessing debris-related hydraulic impacts and supports more reliable flood-risk evaluation at debris-prone structures.

”This work is financially supported by Korea Ministry of Climate, Energy, Environment (MCEE) as ｢Research and Development on the Technology for Securing the Water Resources Stability in Response to Future Change (RS-2024-00332494)｣.”

ABSTRACT. Abstract Sand dams are receiving renewed attention as a sustainable water resource infrastructure solution for water-scarce and underserved regions, including developing countries with limited access to centralized water supply systems. This study provides a comprehensive review of the evolution of sand dam technologies by comparatively analyzing the traditional Kenyan-type sand dam, the Korean-type bypass-type multistage sand-filled dam (BMSD), and a hybrid system integrated with an underground dam. The analysis highlights a significant structural transition from simple floodwater retention facilities toward hybrid water management platforms that enhance storage capacity by collecting subsurface flow through the riverbed and raising groundwater levels via underground dam integration. From the perspectives of structural design, hydrological performance, and operational strategy, the three sand dam types were systematically compared and evaluated. Based on this comparative assessment, the study proposes policy-oriented strategies for deploying sand dams in water-supply underserved areas, emphasizing their role as decentralized water supply infrastructure that can complement conventional centralized water systems. Owing to their technical simplicity, adaptability to local conditions, and low operational requirements, sand dams are particularly suitable for application in developing countries facing financial, technical, and institutional constraints. In conclusion, sand dams represent an efficient and resilient water management strategy that can improve basic water accessibility, strengthen local water security, and support climate change adaptation in water-vulnerable regions.

Acknowledgements Research for this paper was carried out under the KICT Research Program (project no. 20250223, Research on Establishing a Foundation for Responding to Current Issues and Challenges in Water Management) funded by the Ministry of Science and ICT.

ABSTRACT. In recent decades, global warming caused by the increase of carbon dioxide (CO2) in the atmosphere has become a serious environmental issue. Coastal ecosystems, especially seagrass meadows, are known as important "Blue Carbon" sinks. However, their ability to store carbon changes greatly over time and space because of the complex interactions between water movement and biological activities.

The key issue addressed in this study is how physical forces from the ocean and land compete with the biological activities of seagrass to change the local CO2 levels. To explore this, field measurements were conducted during the summer seasons of 2021 to 2025 in Ashikita Port, Japan, to investigate the factors controlling the partial pressure of CO2 (pCO2) in an eelgrass (Zostera marina) habitat. Our methodology involved collecting water samples at different tidal stages—high, slack, and low tides. We measured salinity, water temperature, and carbon parameters, including Total Alkalinity (TA) and Dissolved Inorganic Carbon (DIC), which were used to calculate the pCO2 in the water.

The results show that pCO2 dynamics in Ashikita Port are driven by two main processes working together. First, freshwater from nearby rivers plays a major role. When river flow increases, it dilutes the seawater and lowers the concentrations of TA and DIC. This physical process directly reduces pCO2 levels, regardless of any biological activity. Second, we found clear evidence of biological influence during low-tide periods when the water is relatively still. During these times, eelgrass absorbs CO2 through photosynthesis, leading to a further drop in pCO2. In conclusion, both river discharge and seagrass photosynthesis are essential for controlling CO2 levels in this area. These findings emphasize that we must consider both water movement and ecological functions to accurately evaluate the carbon budget of coastal seagrass meadows.

ABSTRACT. With the recent progression of climate change, concerns are mounting regarding rising river water temperatures alongside increases in air and sea temperatures. In addition to global warming, previous studies have also suggested that agricultural water use and extensive paddy fields may contribute to such rising temperatures during periods of low flow. However, while existing studies in Japan have formulated the thermal balance of river water, insights into the variability characteristics of water temperature remain limited. Therefore, this study aimed to examine the patterns of water temperature variation in the middle reaches of the Nagara River. We analyzed how air temperature, precipitation, and discharge affect water temperatures, based on data observed in the main channel, tributaries and paddy-related waterways. The results showed that while water temperature generally increased downstream, local deviations from this trend were observed at tributary confluences. The Itadori and Mugi Rivers had a cooling effect in spring and summer but a warming effect in autumn. These patterns are presumed to occur because the Itadori and Mugi Rivers have abundant spring water and exhibit smaller temperature variability compared to the main river. In contrast, the Tsubo River and drainage channels receive a significant amount of return flow from paddy fields, resulting in markedly higher temperatures; this explains why the warming effect disappears after autumn. Indeed, comparisons between irrigation canals and drainage channels revealed that drainage water temperatures frequently exceeded even air temperature before entering the river. Site-specific analysis identified three primary variability patterns: diurnal variations (daily cycle), seasonal variations (annual cycle), and flow-dependent variations caused by changes in discharge. Diurnal variation is driven by solar radiation. Seasonal variation is influenced by the watershed land use. Regarding flow-dependent variations, water temperature was found to approach air temperature during periods of low flow regardless of the season. This is attributed to the reduced heat capacity of the river water, which increases the influence of sensible heat transfer. Furthermore, agricultural water use causes a dual warming effect by reducing discharge in the main channel and increasing the inflow of warm paddy return flow. As for precipitation, the cooling effect appeared with a time lag, coinciding with the subsequent discharge recession; this delay is attributed to advection from upstream reaches. Future studies should aim to increase the number of monitoring points, particularly in tributaries and upstream reaches, incorporate parameters required for heat balance analysis, and conduct quantitative evaluations.

ABSTRACT. Upstream-downstream cascade effects in river networks create spatial dependencies in aquatic ecosystem health that have direct implications for watershed management, yet the relative contributions of land use versus hydrological connectivity to these patterns remain poorly quantified. This study investigated scale-dependent spatial autocorrelation and upstream-downstream cascade effects in three biological indicators—Trophic Diatom Index (TDI), Benthic Macroinvertebrate Index (BMI), and Fish Assessment Index (FAI)—across 757 sub-basins in five major river systems of South Korea. Using network-based spatial weights that explicitly account for river topology and flow direction, we quantified Moran's I at national and basin scales, upstream-downstream correlations, and the proportion of spatial autocorrelation attributable to land use versus pure hydrological connectivity.

Results revealed strong indicator-specific patterns. TDI exhibited the highest spatial autocorrelation (Moran's I = 0.559, p < 0.001) and upstream-downstream correlation (r = 0.740), followed by BMI (I = 0.368; r = 0.541) and FAI (I = 0.051; r = 0.134). Basin-level analysis revealed substantial heterogeneity, with the Han River basin showing the strongest cascade effects (TDI I = 0.690) and smaller basins showing weaker patterns. Land use regression explained 63.5% of TDI variance but only 4.1% of FAI variance. After controlling for land use, 43–74% of spatial autocorrelation persisted, representing pure cascade effects attributable to hydrological connectivity independent of land cover. Sensitivity analysis across different network topological distances (1–5 orders) confirmed the robustness of these patterns, though coefficient of variation increased for FAI due to its inherently weak spatial structure.

These findings demonstrate that upstream-downstream cascade effects vary substantially among biological indicators, reflecting differences in dispersal ability, generation time, and habitat specificity. The persistence of significant spatial autocorrelation after land use adjustment confirms that hydrological connectivity creates ecological linkages beyond those mediated by spatially structured land use. For watershed management, our results suggest that upstream restoration efforts would most effectively improve downstream conditions for diatom and macroinvertebrate communities, while fish assemblage recovery requires local-scale habitat interventions.

ABSTRACT. Restoring longitudinal river connectivity is a primary goal in global aquatic ecosystem management. However, conventional structural indices like the Dendritic Connectivity Index(DCI) often misrepresent reality by assuming complete disconnection. Conversely, detailed assessments are prohibitively expensive for basin-wide application. This study introduces a practical, two-tiered assessment framework utilizing a rapid physical screening tool(Crapid) to support decision-making in data-scarce regions. We applied this framework to 89 structures in three Korean small and medium-sized streams—Gokseongcheon, Samcheok-osipcheon, and Hancheon—to derive practical restoration strategies.

The assessment quantified the disparity between structural potential(DCI) and functional connectivity(DCIm). While structural connectivity ranged from 3.53% to 9.80%, functional connectivity dropped significantly to 0.53%–4.43%. This gap highlights "functional limitations" driven by ineffective fishways and management deficits. Crucially, scenario analysis revealed that removing barriers based solely on physical attributes (e.g., low-head dams) does not guarantee recovery if downstream functional bottlenecks persist.

By identifying priority "hotspots," the framework empowers managers to optimize limited budgets by distinguishing necessary structural removal from effective functional retrofitting. This study suggests a scientific, cost-effective strategy for establishing rational restoration priorities, particularly for Korean streams where high barrier density poses significant challenges.

Acknowledgement: This work was supported by Korea Environmental Industry&Technology Institute(KEITI) through Research and Development on the Technology for Securing the Water Resources Stability in Response to Future Change, funded by Korea Ministry of Climate, Energy, Environment(MCEE)(2480000388).

ABSTRACT. This study investigates the coupled effects of channel sinuosity and particle density on microplastic transport in meandering open-channel flows using a fully three-dimensional (3D) numerical framework. A 3D RANS model with the SST k–ω turbulence closure is integrated with a Lagrangian particle-tracking model to resolve flow structures and particle dynamics under varying sinuosity conditions. Results show that increasing sinuosity intensifies secondary flows and promotes the transverse velocity gradients, enhancing velocity heterogeneity. The interplay between curvature-induced flow structures and density-dependent settling governs particle retention and breakthrough behavior, leading to pronounced non-Fickian transport. Dense particles exhibit strong near-bed trapping in high-sinuosity channels, significantly flattening retention-time tails. Overall, the study demonstrates that sinuosity-driven flows and particle density jointly control anomalous microplastic transport in meandering channels.

ABSTRACT. The increasing frequency of droughts due to climate change has made securing stable water resources critical in the Asia-Pacific region. While sand dam-Underground dam integrated systems show promise as adaptation strategies, research on their groundwater quality impacts remains limited. This study analyzed groundwater quality variations in an integrated system in Chuncheon, Korea, to evaluate its drought response effectiveness.

Field monitoring was conducted from August to November 2025 at seven locations: stream water, sand dam inlet and interior, Underground dam, and water tanks. Parameters measured included temperature, dissolved oxygen (DO), pH, electrical conductivity (SPC), total dissolved solids (TDS), turbidity, major ions, trace metals, and bacterial counts. All monitoring points consistently exhibited a Ca-HCO₃ water type, unchanged from pre-construction 2020 data, indicating minimal external contamination. Temperature ranged 7.9-19.1°C seasonally, while DO maintained high levels (7.87-12.23 mg/L). pH ranged 6.78-8.12, and SPC values (27.9-118.8 µS/cm) reflected pristine water quality. Natural sand filtration significantly reduced bacterial counts. Major ions met drinking water standards: chloride (1.4-2.5 mg/L), sulfate (3-6 mg/L), and nitrate-nitrogen (1.7-2.6 mg/L) showed improvement from 2020. Trace metals remained below regulatory limits, confirming minimal non-point pollution.

This study demonstrates that the integrated system effectively maintains groundwater quality stability while securing water resources for drought response. This represents a practical climate change adaptation strategy for drought-vulnerable regions. Future work should include dry season monitoring to characterize seasonal groundwater variations and develop optimized operational strategies.

Funding: The Research for this paper was carried out under the "2025 Groundwater Basic Survey Project (Project No. 20250137-001)", funded by the Ministry of Climate, Energy and Environment.

ABSTRACT. To investigate the impact of water and sediment regulation on carbon, nitrogen, and phosphorus distribution in the water-sediment system of Xiaolangdi Reservoir, stratified water and sediment samplings were carried out at 10 typical cross-sections before (May) and after (November) the regulation in 2025. The spatial–temporal distribution characteristics of carbon, nitrogen, and phosphorus in water and sediments were analyzed. The results showed that the predominant forms of nitrogen and phosphorus in both water and sediments remained unchanged before and after the regulation. In the water column, dissolved total nitrogen was predominantly composed of nitrate nitrogen, while total phosphorus was mainly in particulate form. In sediments, total phosphorus was primarily inorganic, while total nitrogen was predominantly organic. After the water and sediment regulation, the concentrations of carbon, nitrogen, and phosphorus in both water and sediments generally increased, with the exception of a slight decrease in sediment organic matter content. Correlation analysis indicated that ,there was no significant correlation between the concentrations of carbon, nitrogen, and phosphorus in bottom water and those in sediments before the regulation. However, after the regulation, concentrations of calciumbound phosphorus and inorganic phosphorus in sediments showed significant negative correlations with dissolved total phosphorus in bottom water. It suggested that the water and sediment regulation might enhance the adsorption and precipitation of phosphorus from the water column.

ABSTRACT. The incorporation of carbon-based materials into additive manufacturing has enabled the development of functional composite filaments for fused deposition modelling (FDM), with growing potential in water treatment and environmental remediation applications. This review systematically examines the transformation of carbonaceous fillers, particularly graphene-enhanced biochar, into printable composite filaments and evaluates their suitability for adsorption-based pollutant removal in aqueous systems. A structured literature screening approach was adopted to analyse the relationships between material formulation, filament fabrication, and printing parameters, alongside their influence on functional performance. Critical factors including filler–polymer ratio, dispersion quality, and melt-extrusion behaviour were assessed in relation to both printability and adsorption efficiency. The findings indicate that optimal filler loading is required to balance processability and mechanical integrity, as excessive filler content leads to agglomeration, increased melt viscosity, and structural defects that may limit effective surface exposure. Processing parameters such as nozzle temperature, print speed, and layer height significantly influence polymer chain mobility and interlayer bonding, thereby affecting structural stability and durability in aqueous environments. The results further demonstrate that filler characteristics govern rheological behaviour, crystallisation, and surface functionality, which are essential for enhancing adsorption mechanisms. Overall, this review establishes a processing structure–property framework that links material design with functional performance, providing guidance for the development and optimisation of graphene-enhanced biochar composites as sustainable and scalable solutions for water treatment.

ABSTRACT. Seasonal hypoxia is a critical environmental issue in semi-enclosed coastal seas, where bottom dissolved oxygen (DO) dynamics are governed by the combined effects of stratification, biogeochemical processes, and benthic–pelagic interactions. Although three-dimensional hydro-ecological models are widely applied to assess hypoxia under present and future climate conditions, bottom DO is often insufficiently reproduced, partly due to simplified ecosystem representations. This study aims to improve the simulation of bottom DO in the Ariake Sea, Japan, by refining ecosystem-related processes within a coupled hydro-ecological modeling framework and by quantitatively evaluating model performance using observational data. A Delft3D-FLOW and Water Quality (WAQ) coupled model was applied to the Ariake Sea, and the ecosystem module was enhanced to better represent biological and sediment-related processes influencing near-bottom oxygen dynamics. Model outputs were compared with in situ observations of bottom DO, temperature, and salinity at multiple monitoring stations. Model performance was evaluated using standard statistical metrics, including bias, root mean square error (RMSE), correlation coefficient, Nash–Sutcliffe efficiency (NSE), and Kling–Gupta efficiency (KGE). In addition, event-based metrics focusing on low-oxygen conditions were employed to assess the model’s ability to reproduce the occurrence, duration, and intensity of hypoxia. The results show that temperature and salinity are well reproduced, indicating that the physical structure controlling stratification and mixing is realistically simulated. For bottom DO, the model captures the observed temporal variability with a correlation coefficient exceeding 0.8 and a small mean bias, while overall performance indices indicate moderate skill. Event-based analysis further demonstrates improved representation of hypoxic conditions during stratified summer periods, particularly in terms of hypoxia duration and minimum DO levels. These improvements suggest that refined ecosystem processes play an important role in reducing discrepancies between observed and simulated bottom DO. This study highlights the importance of ecosystem representation in coastal oxygen modeling and provides a flexible framework for future model developments targeting additional biological and biogeochemical processes. The results contribute to more reliable assessments of hypoxia dynamics and coastal ecosystem responses under ongoing climate change.

ABSTRACT. As ecological stream restoration projects have expanded in urban areas, accessibility between residential areas and streams has emerged as a key factor shaping urban environmental quality. However, even for the same stream, residents’ perceived convenience of use may vary depending on residential proximity, yet previous studies in Korea have not sufficiently analyzed such differences. This study quantitatively examines the economic benefits of stream accessibility capitalized into housing asset values by applying a hedonic price model to the Seoho Stream ecological restoration project in Suwon, which was implemented between 2006 and 2015. The model incorporates housing and locational–environmental characteristics, including the number of bathrooms, distance between apartments and the Seoho Stream, and the green space ratio. Using apartment transaction data, the marginal willingness to pay (MWTP) is estimated across distance bands from the stream. MWTP is defined as the additional amount individuals are willing to pay as the distance between apartments and the Seoho Stream decreases. The results based on a semi-log specification indicate that economic benefits reflected in housing asset values increase significantly with closer proximity to the stream. The estimated MWTP amounts to approximately KRW 2.47 million for apartments located within 750 m of the stream, KRW 3.94 million within 500 m, and KRW 4.54 million within 250 m. Notably, apartments located within 250 m exhibit a sustained increase in MWTP over a long period compared to other distance bands. This finding suggests that residences in close proximity to the stream more strongly experience the positive effects of ecological stream restoration, such as enhanced waterside accessibility, improved views, and increased environmental amenity.

ABSTRACT. Most current research on constructed wetlands concentrates on nitrogen removal under static conditions, focusing on carbon-nitrogen (C/N) ratios and vegetation species. However, in practical engineering, findings from static experiments often differ considerably from actual requirements under dynamic water quality fluctuations. Without precise control strategies adapted to dynamic C/N conditions, constructed wetlands may suffer from unstable denitrification efficiency and poor shock-load resistance, which have become a major bottleneck restricting large-scale applications. Carbon availability and vegetation are key factors controlling nitrogen removal in horizontal subsurface flow constructed wetlands, but their relative contributions to different nitrogen transformation pathways remain poorly quantified. We set up a series of wetlands operated at various influent C/N ratios. Results show that nitrogen removal responds strongly to the combined regulation of C/N and vegetation, with optimal performance at high C/N ratios. C/N ratio and vegetation significantly shape microbial community composition, while vegetation stabilizes microbial networks and improves interspecific cooperation, thus promoting nitrogen transformation. Enhanced network robustness supports more efficient nitrogen cycling under carbon-rich conditions. A higher C/N ratio generally boosts nitrogen removal via microbial community restructuring and functional gene regulation, favoring dominant denitrifying bacteria and upregulating denitrification genes including nirS, nrfA, and nxrAB. This study reveals a multi-level regulatory mechanism: C/N ratio interacts with vegetation to restructure microbial communities and functions, enhancing nitrogen removal and favoring the denitrification pathway.

ABSTRACT. Urban road infrastructures represent a functionally distinct form of impervious surface, characterized by elongated spatial structures, intensive traffic activity, and pronounced anthropogenic pollutant inputs. Compared with conventional urban catchments, road environments are often associated with pollutant profiles dominated by traffic-related contaminants (e.g., metals and microplastics), rather than traditional organic indicators. However, existing non-point source pollution management frameworks, exemplified by regulatory and planning systems in the Republic of Korea, have not been explicitly tailored to the physical and pollutant characteristics of road infrastructures. Current practices largely rely on generalized low impact development approaches and bulk-based indicators, while process-oriented and road-specific management strategies remain underdeveloped. Road-deposited sediments serve as a key interface linking traffic activity, surface accumulation processes, and runoff-driven pollutant transport. Despite their critical role, sediment dynamics have rarely been systematically incorporated into policy and engineering frameworks for road-related non-point source pollution control. This study integrates policy analysis, management practice review, and experimental evidence to elucidate the role of road-deposited sediments in pollutant build-up and wash-off processes on impervious road surfaces. The results indicate that overlooking sediment-related heterogeneity may constrain the effectiveness of existing management approaches. Furthermore, this study highlights implications for road-oriented stormwater management and sediment-informed design and assessment frameworks.

Acknowledgement: This research was supported by the Major Project of the Korea Institute of Civil Engineering and Building Technology [grant numbers: 2025-0238 and 2025-0284].

ABSTRACT. In the Kushiro Wetland, a large-scale restoration project has been underway since 2005. In 1960s, the area surrounding the wetland was developed, and it caused significant environmental changes in the wetland. The nature restoration project was launched to preserve the wetland, and several restoration works has been carried out. The old channel restoration work is one of the works. The abandoned channel remaining near the current one is revived and that lead the nature environment. The first site of the old channel restoration was implemented at Kayanuma in major river of Kushiro River in 2010. Fish habitat was improved and increase number of fish was seen there. The second site of this restoration work was implemented in March 2025 at Numaoro, a tributary of the Kushiro River. Before restoration, the straightened river channel at Numaoro was wide and shallow, with uniform flow conditions, providing poor habitat for fish. By redirecting the flow into the former meandering channel, more diverse flow conditions were created, which were expected to improve habitat quality for aquatic organisms. In addition to improving hydraulic conditions, suppression of water temperature increases was anticipated. As the previous channel was wide and shallow and warmed up quickly due to solar radiation and heat exchange with the air, the restored channel was expected to help prevent excessive temperature increases by providing meandering, narrower, and deeper conditions. This will improve habitat for fish that prefer cool water, leading to an increase in the number of salmon species. This study evaluates the effect of old channel restoration on suppressing water temperature increases based on field investigations conducted before and after implementation, focusing on the first year after conversion.

ABSTRACT. The construction of multi-functional weirs under the Four Major Rivers Project has fundamentally modified the hydraulic environment of the Nakdong River. Increased water depth and reduced flow velocity have altered riverine conditions, creating favorable environments for cyanobacterial blooms. These blooms not only disrupt aquatic ecosystems but also pose potential risks to public health through the production of cyanotoxins. However, the mechanisms governing cyanobacterial occurrence remain difficult to quantify due to the complex interactions among hydraulic, water quality, and watershed factors. In recent years, machine learning techniques have been introduced as effective tools for analyzing nonlinear and multivariate environmental systems. The predictive performance of such models is highly dependent on the selection of input variables, making feature composition a critical issue in cyanobacterial bloom analysis. The Nakdong River is regulated by eight multi-functional weirs arranged longitudinally along the main channel, resulting in distinct hydraulic regimes and basin characteristics within each weir section. In this study, cyanobacterial occurrence characteristics were analyzed separately for each weir section, and key influencing factors were identified through a data-driven approach. Machine learning models were applied to evaluate bloom occurrence conditions under different combinations of water quality and hydraulic variables. Particular attention was given to hydraulic factors associated with altered flow regimes, including differences in hydraulic residence time among weir sections. The influence of input variable selection on prediction accuracy was also examined to assess the robustness and applicability of the models. The results provide insight into section-specific cyanobacterial dynamics and demonstrate the potential of machine learning-based approaches for improving predictive capability. Ultimately, this study aims to support adaptive and anticipatory management strategies for cyanobacterial blooms in the Nakdong River.

ABSTRACT. As the non-stationarity of hydrological patterns intensifies due to climate change, significant shifts are occurring in the discharge and water quality patterns of river basins. Consequently, predicting the relationship between these variables has become increasingly complex. Traditional static regression models face limitations in capturing the characteristics of streamflow and water quality parameters that evolve over long periods. To address these limitations, this study introduces and evaluates the applicability of Dynamic Linear Models (DLMs). The DLM framework, designed with a state-space architecture where parameters vary over time, offers the advantage of adapting flexibly to changing hydrological regimes. This study utilized long-term monitoring data in the Nakdong River basin from the WEIS (Water Environment Information System) to construct a DLM and evaluate its suitability by simulating the time-varying relationship between discharge and water quality. Particular emphasis was placed on analyzing the structural changes in pollutants (BOD, TOC, T-P, etc.) in response to shifting flow characteristics over the past decades. Prior to model construction, annual variations in discharge and water quality at key monitoring stations were analyzed, confirming significant temporal changes. Based on the annual DLM, optimal model coefficients were determined, and the temporal fluctuations of model coefficients were examined. The results demonstrated that DLMs outperform traditional models in tracking the sensitivity of water quality parameters to discharge, particularly during extreme weather events and seasonal transitions. By decomposing time-series data into trend, seasonal, and regression components, the study successfully identified how intensifying climate variability has modified flow-water quality relationships. Furthermore, it was confirmed that the baseline of pollutant concentration responses to discharge fluctuations is shifting, suggesting that watershed management must account for these dynamic sensitivities. As the study area is subject to both natural and anthropogenic hydraulic changes, further research is required to determine whether DLMs can effectively detect changes in water quality and mass transport characteristics resulting from variations in hydraulic residence time.

ABSTRACT. Accurately determining aquatic vegetation resistance remains challenging due to vegetation bending and resistance reconfiguration under water flow. This study conducted laboratory experiments to explore the bending and resistance reconfiguration for single submerged flexible vegetation under water flow, and corresponding factors influencing resistance have been analyzed and quantified. Four diameters and five heights of vegetation were tested to observe the changes in bending posture of vegetation under a wide velocity range (0.05–1.85 m/s), with the resistance including drag force and skin friction measured. By integrating the relationship between flow velocity and vegetation resistance, the results indicate that resistance changes from a quadratic law to a linear relationship with velocity as both the slenderness Cauchy number CYS and the vegetation bending angle θ increase, and the deviation was caused by the streamlined structure and resistance reconfiguration. The skin friction acted on flexible vegetation became significant under both low and high flow velocities, and thus should be considered in related analysis. Furthermore, the calculated drag coefficient Cd, which accounts for micro-element bending angle dθ, presents a trend of first decreasing and then increasing as the Reynolds number Re and CYS increase, whereas no statistically significant variation trend is observed with the Froude number Fr. The drag coefficient factor αb and the effective length le of the vegetation combined with CYS were used to compare drag force between flexible and rigid vegetation of the same scale, and to explain the resistance reconfiguration effect. Finally, a new empirical formula was established by integrating Re, CYS and the bending angle of flexible vegetation. Compared with traditional empirical formulas based solely on Re, the new formula presents superior predictive performance in calculating Cd of flexible vegetation for both high and low Cd ranges. This study provides valuable references for understanding the interaction mechanism between flexible vegetation and water flow, and for improving resistance calculations in open channels featured with vegetation.

ABSTRACT. Abstract: This study aims to improve the accuracy of automated juvenile ayu sweetfish counting using the YOLOv8-l object detection model. Previous research has demonstrated that image preprocessing using Contrast Limited Adaptive Histogram Equalization (CLAHE) can mitigate the degradation of detection accuracy caused by poor underwater image quality. However, evidence suggests that the effectiveness of this method is highly dependent on the clipLimit parameter. In this paper, we propose a method to dynamically determine the optimal parameters for each target image and evaluate the resulting improvements in confidence scores. Furthermore, we conducted additional evaluations on nighttime urban scenes to verify the general versatility of this approach beyond aquatic environments.

Optimization Method for clipLimit: The proposed system consists of three distinct phases:

Training Phase: For the original training images, five variations were generated with different clipLimit values (1.0, 1.5, 2.0, 4.0, and 8.0). Five separate YOLOv8-l models were then trained to generate teacher data, where the average confidence score across all classes was recorded for each image. Finally, a ResNet18 model was trained as a regression problem to learn the relationship between images and their predicted confidence scores for each clipLimit.

Inference Phase: For each test image, five variations were prepared using the same clipLimit settings. The ResNet18 regression model predicted the expected confidence for each case, and the clipLimit yielding the highest predicted score was dynamically selected for the final YOLOv8-l inference.

Evaluation Phase: Object detection was performed on all five fixed clipLimit variations and the proposed dynamic method to compare the overall mean confidence scores.

Experimental Datasets:

Underwater Dataset: 669 images (Training: 569, Testing: 100) of sweetfish captured via underwater cameras in a fishway.

Nighttime Urban Dataset: 4,626 images (Training: 4,490, Testing: 136) from Roboflow, featuring persons, bicycles, and cars in low-visibility conditions.

Results and Discussion: Evaluations were conducted at confidence thresholds of 0.2, 0.3, and 0.4.

Underwater Results: The proposed method achieved scores of 0.744, 0.768, and 0.740, outperforming the best fixed-parameter results (0.700, 0.711, and 0.712, respectively).

Nighttime Results: The proposed method reached 0.522, 0.549, and 0.547, exceeding the maximum fixed-parameter scores (0.500, 0.508, and 0.513). Across both datasets, dynamically selecting the clipLimit on a per-image basis consistently improved detection reliability.

Conclusion: This study confirms that deep learning-based clipLimit optimization effectively enhances object detection confidence in challenging visual environments, including underwater and nighttime conditions.

ABSTRACT. Cascade small hydropower development disrupts river hydrological and hydrodynamic processes, leading to alterations in benthic macroinvertebrate communities and affecting ecosystem structure and function. To elucidate the impact mechanisms on headwater streams, this study compared two adjacent, physically similar streams: Tongluowan Stream (impacted by cascade hydropower) and Yashankeng Stream (natural reference). Benthic macroinvertebrate communities and environmental parameters were monitored during the wet (June) and dry (December) seasons of 2022. The results demonstrated that hydropower development not only reduced macroinvertebrate density and diversity but also shifted their seasonal dynamics. In the regulated river, benthic density decreased by approximately 50% in the dry season compared to the wet season, whereas no significant seasonal difference was observed in the natural stream, diversity was lower in the dry season in the regulated river but higher in the natural river. Furthermore, hydropower operations shifted the dominant taxa from rheophilic, clean-water indicators to pollution-tolerant species preferring stagnant habitats. Flow velocity, discharge, and substrate type were identified as the primary environmental drivers structuring these communities. This study enhances the understanding of the ecological impacts of small hydropower cascades on headwater ecosystems.

ABSTRACT. In the Yodo River, located in the center of the Kinki District, Japan, past river improvement projects primarily focused on flood control and water utilization, resulting in enhanced flood safety. However, these interventions have also led to the deterioration of the riverine environment, including a reduction in lentic water areas known as Wandos. Wandos are backwater habitats connected to the main channel and are recognized for their environmental diversity and high biodiversity within the Yodo River system. To address this issue, the Yodo River Office launched a plan to double the number of Wandos. In addition, the Shimeno area was selected as a model site for the redevelopment of Yodogawa River Park, owing to long-term, community-based activities carried out by river rangers and local citizen groups around the Wandos. The Yodo River Office subsequently initiated the Shimeno Waterfront Development Project, which aims to create high-quality fish habitats while improving public accessibility from the floodplain through gently sloped banks. As part of this project, two new Wandos with open mouths and a range of water depths were constructed upstream of an existing Wando. This study aimed to investigate the flow conditions under which water enters newly constructed Wandos from the main channel. Two-dimensional numerical simulations of flow and riverbed changes were conducted using iRIC Nays2DH (ver. 4.0) under three discharge scenarios—small- (maximum flow rate: 521 m3/s), medium- (maximum flow rate: 1,375 m3/s), and large-scale floods (maximum flow rate: 6,343 m3/s)—to determine the discharge thresholds for water inflow into the Wandos and to evaluate their suitability as habitats for aquatic organisms. The results showed that water inflowed into the open mouth between the newly constructed No.1 Wando and the existing No.2 Wando when discharge exceeded 1,000 m³/s. In contrast, no inflow was observed at the open mouth between the newly constructed No.0 Wando and No.1 Wando. During small- and medium-scale floods, backflow from No.1 Wando toward No.0 Wando was observed. This backflow is considered to promote sediment deposition within the Wandos by transporting sediment in an upstream direction. Progressive sediment deposition may reduce water depth within the Wandos, potentially decreasing fish habitat availability and disrupting ecological conditions. These findings highlight the importance of ensuring appropriate water inflow through Wando open mouths. Currently, additional two-dimensional flow analyses are being conducted to evaluate scenarios involving flow control structures at the open mouths, and detailed results will be reported in future studies.

ABSTRACT. The Seto Inland Sea (SIS), the largest enclosed sea area in Japan, has undergone oligotrophication. This has resulted in poor growth of nori (Porphyra spp.), a seaweed traditionally consumed in Japan, and has become serious concerns. To address this problem, some coastal managers have introduced nutrient supply measures. Under these measures, wastewater with higher nutrient concentrations than usual is discharged from industrial and sewage facilities. In contrast to oligotrophication, low achievement rates of environmental standards for chemical oxygen demand (COD), an indicator of organic pollution, have been reported. This study assessed both beneficial and adverse impacts of nutrient supply measures on the water quality of the SIS using three-dimensional hydrodynamic and water quality simulations. Numerical simulations were conducted for the Pacific Ocean–SIS region using the three-dimensional hydrodynamic model SCHISM and the water quality model CE-QUAL-ICM. We conducted simulations under two scenarios to evaluate the impact of nutrient supply measures; one scenario involved implementing measures, while the other without implementing any measures. Dissolved inorganic nitrogen (DIN), the primary limiting nutrient for production in the SIS, and COD were selected as evaluation indicators. To estimate biological responses, a nori growth model was developed and integrated into the water quality model. Using this model, we calculated the nitrogen-to-carbon ratio, closely related to the color and quality of nori. This allowed us to evaluate the impact of nutrient supply measures. Nutrient supply measures increased DIN and COD concentrations in Harima Nada, located in the eastern SIS, and in the adjacent Osaka Bay, while their impact on other sea areas was negligible. This result reflected the relatively large discharge in DIN discharge associated with the measures in Harima Nada. In Harima Nada and Osaka Bay, COD concentrations continued to rise even after measures were completed. Since organic pollution control is still needed in the inner part of Osaka Bay, nutrient supply measures should be implemented and considered in coordination with adjacent sea areas. In the northeastern part of the Harima Nada, the increase in DIN concentration was mainly confined to an area within 2 km of the coastline. Changes in other sea areas were minor, indicating that the impact was limited in both spatial extent and magnitude of concentration increases. However, during the period when the nitrogen-to-carbon ratio of nori decreases, its ratio and color were successfully restored through the measures. This result demonstrated the effectiveness of nutrient supply measures.

ABSTRACT. Vegetation within river channels strongly influences flow structures and sediment transport, with significant implications for channel stability and flood risk. While existing studies often focus on the immediate vicinity of vegetation, the downstream impacts of patches under mobile-bed conditions remain poorly understood. This study quantifies how a finite emergent vegetation patch alters three-dimensional flow and bed morphology both locally and over an extended downstream distance. In this study, flume experiments were conducted in a straight channel with a mobile sediment bed and a square patch (length L=0.34 m) of emergent rigid cylinders. Using acoustic Doppler velocimetry (ADV) and ultrasonic distance sensors, we characterized mean flow, turbulent structures, and bed elevation changes across a domain extending from 3L upstream to 10L downstream. Results reveal distinct morphodynamic signatures: localized scour at the upstream edge and lateral boundaries, and a prominent depositional ridge in the wake, likely driven by secondary flows. Notably, bed disturbances persisted throughout the 10L downstream domain, indicating that patch-induced effects are far-reaching. These findings emphasize that river restoration and flood-mitigation strategies need to account for these far-field impacts to ensure long-term geomorphic stability.

ABSTRACT. This study focuses on bed disturbances induced by flow and sediment transport during flood events and aims to quantify the flood disturbance intensity on the river bed at a watershed scale by evaluating its spatiotemporal distribution. Flood disturbances during flood events can enhance the physical heterogeneity of river habitats, such as riffles and pools, when moderate in magnitude. However, extreme floods substantially alter channel morphology and strongly affect aquatic communities inhabiting river channels. Even within a single watershed, spatial variations in precipitation distribution cause differences in flood conditions, and consequently, flood disturbances vary among river segments. Therefore, understanding the responses of river environments to flood disturbances requires an integrated evaluation at the watershed scale that combines upstream, downstream, and tributary reaches, rather than focusing solely on a single river section. First, a simplified disturbance intensity index acting on each river segment within the watershed was developed. In selected reaches with detailed topographic data, two-dimensional unsteady flow simulations (iRIC Nays2DH) were conducted to analyze the spatial distribution of bed shear stress under a wide range of discharges from low to high flows. The discharge–shear stress relationships, aggregated and averaged for smaller reach segments, were fitted with a power-law function (τ = αQ^β). The parameters α and β were then estimated based on river geometry and scale, enabling the simplified disturbance intensity index to be applied to ungauged river segments across the watershed. Flood events from 2006 to 2023 (April–November) were simulated using the Rainfall–Runoff–Inundation (RRI) model for the Nagara River basin. By combining the RRI-based discharge simulations with the simplified index (τ = αQ^β), the spatiotemporal distribution of flood disturbance intensity at watershed scale was quantified. The results demonstrate that the discharge–shear stress relationships derived from the Nays2DH simulations were well represented by the power-law expression (τ = αQ^β), enabling robust representation of the disturbance intensity acting on individual channel segments. The RRI simulations further indicate pronounced spatiotemporal variability in flood magnitudes acting on different channel segments across the watershed. These findings highlight the feasibility of assessing the spatiotemporal distribution of flood disturbance intensity at a watershed scale by integrating hydraulic and hydrological modeling. At the conference, detailed analyses of the spatiotemporal distribution of flood disturbance intensity and its relationship with river environment dynamics in the Nagara River system will be presented.

ABSTRACT. Seasonal lakes experience periodic water-level fluctuations, driving complex hydrodynamic interactions at the sediment-water interface (SWI). Such dynamic environments pose significant challenges for water quality management, particularly regarding the long-term fate of adsorptive pollutants. Using a generalized section of the Poyang Lake floodplain as a prototype, this study presents a combined experimental and numerical investigation into the migration of adsorptive pollutants across the SWI under these periodic fluctuations. Experimental results indicate that rapid adsorption accelerates the initial solute exchange, increasing the SWI flux by 56.1% during the first fluctuation cycle compared to non-adsorbing solutes. Subsequently, the interplay between groundwater flow and sediment adsorption significantly suppresses deeper solute transport in the pore water. This interaction retains the pollutants in a droplet-shaped plume, effectively reducing the peak pore-water concentration by 86.8%. Following validation against experimental data, a solute transport model was utilized to simulate field-scale migration. The simulations demonstrate that higher adsorption capacities lead to a highly concentrated accumulation of pollutants within the shallow lakebed layer (0-1m). Specifically, as the adsorption coefficient (KD) increases from 0.0021 to 0.0169 m3 kg-1, the peak solute mass in this shallow layer surges by 578.72%, and the proportion of retained shallow solute mass at time t = 8T increases by 78.49%. These findings provide a mechanistic foundation for understanding and modeling contaminant retention, and its potential secondary release—in seasonal lake systems driven by periodic hydrological forcing. Ultimately, this work highlights the critical need to incorporate transient hydrodynamic boundaries and sediment adsorption kinetics into ecological risk assessments and targeted remediation strategies.

ABSTRACT. Successful implementation of In-situ Chemical Oxidation (ISCO) relies heavily on the efficient delivery of oxidants and the management of subsurface hydraulic changes. This study presents a field-scale application of electrical resistivity tomography (ERT) to monitor the injection of persulfate for remediating a site contaminated with Total Petroleum Hydrocarbons (TPH). The site exhibited contamination levels, with maximum TPH concentrations of 12,079 mg/kg in soil and ranging from 0.5 to 1.1 mg/L in groundwater. Unlike traditional point-sampling methods, which often fail to capture subsurface heterogeneity, this study utilized time-lapse 2D borehole ERT. The acquired data were visualized using fence diagrams to effectively map the spatiotemporal distribution of the injected oxidant between boreholes. The high conductivity of the persulfate solution created a sharp contrast against the background formation, enabling the tracking of fluid transport pathways within the cross-hole sections. Initial monitoring phases successfully delineated the radius of influence and identified preferential flow paths governed by the site's hydraulic conductivity. Significantly, the continued time-lapse monitoring revealed an unexpected evolution in subsurface properties. As the oxidation process progressed, a distinct zone of high resistivity emerged, contradicting the initial trend of conductivity increase due to ionic injection. Integrated geochemical analyses, including X-ray diffraction (XRD) and X-ray fluorescence (XRF), confirmed that this anomaly was caused by the precipitation of secondary minerals, specifically sodium sulfate and calcium oxalate. These precipitates likely induced pore clogging, thereby altering the local hydraulic permeability. Despite these complex hydro-geochemical interactions, the pilot test achieved a TPH reduction of approximately 60% in the target zone. This study highlights the critical role of geophysical monitoring in environmental hydraulics. By capturing both the fluid transport and the subsequent geochemical clogging effects through time-lapse fence diagrams, ERT provides essential data for adaptive management of ISCO projects, ensuring that hydraulic blockages do not hinder remediation efficiency.

Funding: The research for this paper was carried out under the KICT Research Program (Project no. 20250284–001, Development of Digital Urban Flood Control Technology for the Realization of Flood Safety City) funded by the Ministry of Science and ICT.

ABSTRACT. Estuarine Reservoirs are vital resources serving multiple functions, including domestic and agricultural water supply and ecosystem conservation. However, interest and investment in their water quality management remains insufficient. This is primarily due to the long hydraulic retention time of pollutants, management responsibilities among local governments in catchment area, budgetary and Operational personnel, and a lack of public awareness.

This study explores the feasibility of improving water quality in large-scale agricultural Estuarine Reservoirs using a coagulation-filtration system. A pilot with a capacity of 300 m3/day was installed at the Sagi Pumping Station in Ganwol Esturine Reservoir (Seosan City, Chungcheongnam-do). The system's performance, target water quality attainment, and maintenance feasibility were evaluated based on operational data collected from January 2024 to the present. Influent and effluent samples were analyzed for eight parameters, including COD, TOC, SS, T-N, T-P, DTN, DTP, and DTOC.

The filtration system features radially arranged filter media with varying packing densities. To enhance the removal efficiency of particulate matter, Poly Aluminum Chloride (PAC) was utilized as a coagulant at a dosage of 10 mg/L. Following coagulant injection, removal efficiencies improved: T-P increased from 15.9% to 60.4%, SS from 54.8% to 92.3%, and Chl-a from 6.3% to 82.2%. However, the treatment effect on dissolved substances was negligible. In addition, a plan has been prepared to be used in Estuarine Reservoir with a large amount of water.

Phosphorus is a key driver of algal blooms and eutrophication in public water bodies and acts as a limiting nutrient for phytoplankton. Therefore, phosphorus control is critical. Future research should focus on optimizing facility scale, securing operational funding, and exploring integrated systems, such as constructed wetlands, to simultaneously manage nitrogen and phosphorus through ecological mechanisms.

ABSTRACT. The Jakasetia area, with a long history of recurrent flooding, is managed through a polder system incorporating drainage networks, retention ponds, and flood pumps. Operating this infrastructure, especially pumping and monitoring facilities, generates heavy dependence on conventional electricity. Such reliance imposes financial burdens on government and community budgets and exposes the system to risks from supply disruptions. Addressing these issues requires renewable energy solutions, with solar panels emerging as a practical alternative to enhance energy efficiency, sustainability, and long-term resilience. The central research problem examined in this study is the technical and economic feasibility of applying solar energy for polder operations.

This research adopts a quantitative approach beginning with calculations of daily power demand based on the installed electrical load. The analysis estimates the solar panel capacity required and evaluates financial feasibility using Break Even Point (BEP), Net Present Value (NPV), and Benefit Cost Ratio (BCR) indicators. Results indicate that an initial investment of IDR 222.55 million is necessary for solar panels and supporting components. Financial analysis shows the BEP is reached in the twelfth year. The NPV of benefits, amounting to IDR 1.17 billion, surpasses the NPV of costs at IDR 901.09 million. Consequently, a BCR of 1.30 confirms the investment’s economic feasibility in solar panels’ 25 years life time.

Overall, the study demonstrates that solar panel adoption is technically and economically viable for flood control infrastructure. Beyond cost savings, solar integration offers a strategic solution to reduce reliance on conventional electricity and accelerate the transition toward clean energy. This research contributes to scientific advancement by presenting an integrated analytical model linking energy demand with economic feasibility. The findings provide valuable insights for policymakers, strengthen evidence for renewable energy investment, and serve as a reference for future research and sustainable energy policy in Indonesia.

ABSTRACT. Since the 2000s, the expansion of riparian vegetation cover along Korean rivers has become increasingly evident, in association with changes in hydrological and climatic conditions. Although this transition has often been attributed to anthropogenic river regulation, climatic factors particularly variations in rainfall and temperature also play a critical role in governing riparian vegetation dynamics. However, these climate-driven vegetation processes are not yet fully incorporated in most eco-morphodynamic models, highlighting the need for improved representation in future fluvial simulations. This study enhances an eco-morphodynamic model by incorporating precipitation- and temperature-based controls on riparian vegetation germination, growth, and mortality. The revised model includes rainfall-triggered germination, temperature-driven early germination, and temperature-dependent growth adjustments, with temperature prioritized when both climatic conditions are satisfied. Comparative simulations show that the enhanced model more realistically captures vegetation dynamics and their feedbacks on fluvial morphology. These results emphasize the importance of explicitly accounting for climatic influences in eco-morphodynamic models and provide insight into potential interation between fluvial processes and vegetation under climate change.

ABSTRACT. Microplastics (MPs) have emerged as critical environmental pollutants threatening aquatic ecosystems, with their frequent detection in aquatic organisms and drinking water raising concerns about potential risks to human health. MPs are not naturally generated but are primarily introduced into river systems through rainfall–runoff processes and effluents from wastewater treatment plants. Consequently, rivers act as major transport pathways conveying MPs from terrestrial environments to marine ecosystems. Effective management of MP pollution therefore requires a quantitative understanding of MP inflow, transport, and spatial distribution under riverine flow conditions. However, studies that continuously quantify MP loads entering rivers through rainfall–runoff processes at the catchment scale remain limited. This study aims to estimate MP inflow loads from a catchment to a river and to analyze their relationship with land-use characteristics. The Water Quality module of the Storm Water Management Model (SWMM) was applied to the Gyeongan Stream catchment in South Korea to simulate rainfall-driven MP inflows. MP inflows were simulated using a build-up/washoff approach, and model parameters were calibrated using observed MP data measured in the Gyeongan Stream. Sensitivity analysis identified the maximum buildup during antecedent dry periods as the most influential parameter governing MP wash-off loads. Accordingly, its calibration range was determined based on previously reported MP measurements in urban areas characterized by mixed residential, transportation, and industrial land uses. Other parameters associated with rainfall-driven wash-off processes were calibrated within ranges commonly adopted for suspended solids. Model performance was evaluated by comparing simulated and observed MP loads at monitoring locations in 2021, yielding coefficients of determination (R²) greater than 0.8 for polypropylene (PP), polyethylene (PE), and polyester. Using the calibrated model, MP inflows from the catchment to the Gyeongan Stream were estimated for the four-month flood season. The results showed a pronounced first-flush effect, with MP inflows increasing rapidly at the onset of rainfall events and remaining at relatively low levels during subsequent dry periods. When normalized by rainfall days, the average daily MP inflow was estimated to be 2.0–10.2 particles/m²/day. Spatial analysis further indicated that downstream urban areas contributed the highest MP loads, corresponding to regions with dense residential development and transportation infrastructure, whereas less-developed areas with comparable catchment sizes exhibited substantially lower loads. These findings suggest that highly urbanized areas with intensive transportation activities represent major contributing zones for MP runoff within the Gyeongan Stream catchment.

ABSTRACT. Microorganisms such as bacteria, algae, etc. are ubiquitous in natural aquatic systems, playing an important role in geophysical processes. When these organisms adhere to the sediment bed, they produce Extracelluar Polymerix Substances (EPS), which eventually lead to the formation of biofilm within the sediment bed. The EPS bind sediment grains together and reduce the bed porosity, which in turn transform non-cohesive sediment into biologically cohesive sediment. This biological transition of sediment bed affects the deposition and resuspension of the deposited microplastics. Therefore, this study performed laboratory experiments to investigate the effect of EPS within the sediment bed on microplastic deposition. Considering that vegetated channels exhibit higher EPS concentration and are considered as hotspots of microplastic deposition compared to bare beds, both bare and vegetated channels were exmained. Fluorescent particles were used to represent microplastics and their deposition rate was compared under a wide range of conditions including microplastic size, EPS concentration, and flow rate. Our results demonstrated that as turbulence intensity increased, particles were more easily resuspended, decreasing deposition rate. Under the same turbulence level, smaller particles exhibited higher deposition rates compared to larger ones. This was because, once the smaller particles settled on the channel bed, they experienced the stronger hiding effect between sediment grains, providing larger resistance to resuspension compared to larger particles. With increasing EPS concentration, the deposition rate decreased for all particle types. This trend was more pronounced for smaller particle because the increase in particle exposure due to EPS is more significant for the smaller particles compared to larger ones. Using EPS concentration, flow characteristics, and particle properties, a physical-based model for particle deposition probability was developed, which is expected to provide insights for developing effective strategies to manage and protect aquatic vegetated systems from the microplastic pollution in natural streams.

ABSTRACT. The 216 MW Upper Trishuli–1 Hydroelectric Project (UT-1), situated in Nepal’s dynamic Himalayan river system, represents a climate-resilient and environmentally integrated approach to large hydropower development under complex hydro-environmental conditions. Himalayan rivers are increasingly affected by climate-driven hydrological variability, extreme flow events, high sediment loads, and seismic sensitivity, posing significant challenges for dam safety, river hydraulics, and long-term sustainability.

UT-1’s dam, intake, and underground structures were engineered to accommodate steep hydraulic gradients, sediment-rich flows, and geological uncertainty. Extensive geotechnical investigations, adaptive design strategies, and rigorous construction quality control enabled the safe completion of major underground works, including a 9.5 km headrace tunnel. These measures demonstrate robust engineering responses to climate-sensitive and geologically complex environments.

Environmental hydraulics is central to UT-1’s design and operation. A scientifically defined environmental flow regime maintains downstream flow continuity, ecological processes, and river health under variable hydrological conditions. Integrated sediment management and river-training measures address sediment transport dynamics, flood resilience, and long-term operational safety. Ecohydraulic interventions, including fish passage facilities, habitat enhancement, and watershed conservation, illustrate how hydropower infrastructure can be harmonized with riverine ecosystems.

The project adopts international best practices in environmental and social governance, incorporating comprehensive Environmental and Social Impact Assessments and a structured Free, Prior, and Informed Consent (FPIC) process with Indigenous Peoples. Benefit-sharing mechanisms and cultural heritage protection under the Indigenous Peoples Plan ensure inclusive and sustainable development outcomes.

UT-1 also demonstrates a milestone financing and compliance framework involving multiple international financial institutions, strengthening adherence to global dam safety, environmental, and climate-risk standards. Overall, the project offers valuable lessons for IAHR practitioners by showcasing how high-head Himalayan hydropower can integrate engineering reliability, environmental hydraulics, sediment management, and community resilience in the era of climate change.

ABSTRACT. Large transportation hub stations in recent years have been exposed to a range of disaster risks, in which flood related incidents can directly lead to loss of life and severe property damage. In stations that include deep underground tunnels such as those on the GTX network, long escalator sections can act as pathways for external water inflow, potentially having a critical impact on passenger evacuation and rescue operations. Also, in these intense urban flood events, floodwater entering underground spaces can carry suspended solids and fine debris that increase risk to pedestrians and hinder movement. Therefore, the investigation of inundation in subway stations is necessary to ensure the safety of citizens against flooding. Considering these factors, this study analyzes flooding around stairs using a three-dimensional OpenFOAM numerical model, focusing inflow characteristics such as the depth and velocity of floodwater. A three-dimensional computational mesh was developed for a long escalator connected to a deep underground tunnel, reflecting the geometry and slope observed at subway stations and a GTX facility. Flooding scenarios were established by assuming initial inflow depth and boundary conditions at the upstream escalator. To incorporate sediment transport, a concentration field was introduced and transported within the water phase by using the simulated velocity field. Simulation results showed gravitational acceleration and a rapid velocity increase in the initial section, after which the rate of increase decreased and velocity gradually converged toward a maximum value. Also, the drag force applied to human models was calculated using the simulations, which can provide inundation risk values to people in subway stations. In the escalator connected to the platform, the high inflow velocity may raise concerns for evacuation safety. These findings indicate that study of flow in stair structures in underground subway stations are a key factor for both flood risk and impacts on evacuation routes.

ABSTRACT. Recent studies have highlighted the role of freshwater lakes and reservoirs as "Freshwater Carbon" ecosystems that absorb carbon dioxide (CO2) from the atmosphere. The potential for carbon capture and storage in freshwater carbon is enormous compared to that in the blue carbon ecosystem. However, research worldwide is scarce regarding the factors affecting CO2 absorption and release, and a standardized method for quantitatively estimating carbon dioxide absorption remains undecided. Lake metabolism, which refers to the activity of phytoplankton and aquatic plants, is typically evaluated using hourly changes in dissolved oxygen (DO) to determine Gross Primary Production (GPP), Ecosystem Respiration (ER), and Net Ecosystem Production (NEP). While previous studies have shown that GPP, ER, and NEP tend to increase with the size of the lake, no studies have been conducted using pH as an indicator for metabolism assessment. The purpose of this study is to clarify the relationship among pH, DO, aquatic metabolism, and the partial pressure of carbon dioxide (pCO2) in freshwater, and specifically to determine the effect of pH on water metabolism. Observations were conducted in two Japanese reservoirs for several months. Lake metabolism was evaluated using ecosystem respiration and ecosystem photosynthesis. The relationships among pH, DO, metabolism, and CO2 in the water were then analyzed based on the observation results and correlation analysis. The results demonstrated a strong relationship between DO and pH, which could be classified into two distinct regions. Furthermore, the findings indicated that photosynthetic activity may increase with an increase in pH value. This outcome holds significant implications for our understanding of carbon dioxide absorption in freshwater ecosystems.

ABSTRACT. Benthic bioturbation, characterized primarily by burrowing activities, modifies the sediment-water interface structure, and the resulting hydrodynamic changes expose the hyporheic zone to complex solute transport dynamics. This biological disturbance amplifies sediment permeability, causing significant fluctuations in pore water exchange, profoundly affecting the crucial material cycling and self-purification within aquatic ecosystems. The regulatory mechanisms governing these processes remain largely unclear, leading to uncertainties in understanding aquatic biogeochemical cycles. This study combines indoor circulating flume experiments with a coupled hydrodynamic-multiphysics model to investigate how benthic bioturbation characteristics, specifically burrow depth, dune height, and bioirrigation, affect hyporheic exchange and solute transport in the sediment. The results indicate that the intensity and efficiency of hyporheic exchange are significantly influenced by both burrow structures and overlying water velocity. A higher overlying water velocity accelerates solute transport by elevating the interfacial pressure gradient. Specifically, the downstream maximum solute flux reaches approximately 1.7 times the inflow flux, while the windward side's flux is 1.6 times that of the leeward side. Solute transport in the hyporheic zone exhibits a positive correlation with the product of the dune Reynolds number and relative burrow depth, while burrow shape itself exerts a negligible effect compared to biological roughness. Furthermore, the study demonstrates that bioirrigation induces complex local circulation patterns, and a 2D simplified model effectively simulates these dynamics when the length-to-depth ratio exceeds 2.5. These findings provide new insights into the complex internal coupling mechanisms of benthic bioturbation within aquatic environments and offer a scientific basis for enhancing ecological restoration and aquatic environment management under biotic disturbance.

ABSTRACT. Large watersheds are often subjected to extensive research to elucidate the impact of surface water quality on human and ecosystem health, arising from various land use practices and diverse sources of contamination. Several fingerprinting techniques have been used to analyze the contaminants in water bodies. Dissolved ion concentrations can be employed as a fingerprint technique to identify the sources of water contamination within a catchment basin. Therefore, the objective of our study was to characterize the dissolved ion distribution in Kushiro River catchment basin, which is characterized by exceptionally high dissolved ion concentrations. Inductively coupled plasma mass spectrometry (ICP-MS) was employed to determine the concentrations of major ions (F-, Cl-, NO2-, Br-, NO3-, PO4-, SO4-, Li+, Na+, NH4+, K+, Mg2+, and Ca2+) in river water samples from 11 sampling stations across Kushiro River catchment basin. The measurements of both cations and anions were repeated three times and the average value was used for the analysis of dissolved ions in the watershed. The 11 sampling locations were subsequently classified into five distinct groups based on land use, surface soil type, and vegetation cover to facilitate precise analysis of the ion distribution using neural networks. The dissolved ions from each group were fed into a neural network layer as inputs, followed by the application of a series of equations to estimate the dissolved ion transportation rates from each group to the downstream end of Kushiro River catchment basin. Our neural network model identified Group 1, situated within a caldera lake region, as having the highest transport rate of dissolved ions within the catchment basin. This may be attributed to the geological characteristics of the underlying rock formation. To validate our neural network model, two principal components were employed to visualize and interpret our dataset. Compositional similarities and seasonal variations in ion distribution were identified, as well as the key variability patterns, thereby revealing underlying correlations among the dissolved ions. This comprehensive analytical framework provides a robust and insightful tool for determining ion distribution within catchment basins with significant implications for environmental monitoring and sustainable resource management.

ABSTRACT. River network connectivity is a fundamental attribute of healthy freshwater ecosystems, yet its large-scale spatial characteristics remain insufficiently quantified at the national level. This study presents a peninsula-scale assessment of river connectivity across the Korean Peninsula using index-based connectivity metrics, with the aim of identifying spatial patterns of fragmentation and regions with high restoration priority.

River network and barrier datasets were compiled to represent longitudinal connectivity conditions across major basins. Connectivity was evaluated using a suite of graph- and network-based indices, including the Dendritic Connectivity Index (DCI), Population Connectivity Index (PCI), Catchment Area-based Fragmentation Index (CAFI), and the Integral Index of Connectivity (IIC). These indices collectively capture structural and functional aspects of connectivity under current barrier configurations.

The results reveal pronounced spatial heterogeneity in river connectivity across the peninsula. Highly fragmented conditions were observed in lowland and urbanized basins with dense transverse structures, whereas relatively intact connectivity persisted in mountainous headwater regions. Differences among indices highlight the importance of multi-metric evaluation, as each index emphasizes distinct network properties. The proposed index-based framework enables rapid, scalable diagnosis of river fragmentation and provides a quantitative basis for setting national-scale river restoration and management priorities.

ABSTRACT. Downstream environmental degradation caused by sediment trapping by dams has become an increasingly critical issue in river management, leading to growing demands for sediment replenishment from reservoirs and for corresponding sediment management strategies in downstream reaches. Contemporary river management often relies on localized and forceful geomorphic interventions, which can provide short-term stabilization but frequently result in repeated maintenance and limited long-term effectiveness. In contrast, historical river management in Japan was founded on a dynamic perspective of channel morphology. Rather than fixing channel forms, traditional practices allowed river channels to shift and evolve, while guiding these changes away from areas of human settlement and concentrated assets. In the Edo period, flood control was referred to as kawa-yoke (“river exclusion”), a term that reflects a management philosophy emphasizing channel guidance through selective interventions rather than rigid control. Under this framework, river managers actively utilized natural physical and biological processes to influence sediment transport and channel morphology. This study investigates the river management concepts underlying traditional river works through a systematic analysis of pre-modern technical documents. The analysis focuses on descriptions of spur dikes and related structures, such as Seigyu, which were widely used in pre-modern Japanese rivers. Historical records indicate that these structures were designed to promote sediment deposition both directly, by acting as hydraulic roughness elements during floods, and indirectly, by facilitating the establishment of in-channel vegetation. Moreover, planning and evaluation were conducted at relatively large temporal and spatial scales, accounting for channel changes over multiple flood events and over river reaches extending several kilometers． These traditional structures functioned as devices that intervened in the interactions among flow, sediment, and vegetation, enabling gradual geomorphic change through limited structural input. Such an approach contrasts with many contemporary practices that prioritize immediate, localized control. Re-evaluating traditional Japanese river engineering from a process-based perspective provides valuable insights for modern sediment management, particularly under increasing demands for sediment restoration downstream of dams and the development of adaptive, long-term river management strategies.

ABSTRACT. Accelerated urbanisation in tropical cities like Kuala Lumpur increases polluted stormwater runoff, degrading urban rivers. Conventional drainage offers limited treatment. Hybrid systems integrating engineered devices with Nature-based Solutions (NbS) are advocated for both flood and water quality goals, but their performance is rarely assessed against regulatory standards. This study uses the Malaysian Department of Environment’s (DOE) Water Quality Index (WQI), the national water classification standard, as a comprehensive, regulation-aligned metric to evaluate a pilot-scale hybrid stormwater treatment system. The system was installed in a highly impervious commercial catchment in greater Kuala Lumpur. The treatment train comprised four consecutive units: (1) a Continuous Deflective Separation (CDS) unit for coarse pollutants; (2) a vetiver-based floating treatment wetland for organics and nutrients; (3) a palm-kernel-shell biochar filter for dissolved organics; and (4) a zeolite bed for ammoniacal nitrogen removal. Grab sampling and IoT sensors provide data for stage-wise WQI calculation. The influent stormwater WQI was 68.4 (Class III, lightly polluted). The final effluent of WQI was 82.7, a 14.3-point (20.9%) improvement that elevated the water by one WQI class, meeting national discharge requirements. Stage-wise analysis showed contributions of 4.2 (CDS), 3.8 (wetland), 4.1 (biochar filter), and 2.2 (zeolite bed) WQI points. The system remained resilient despite occasional ammonia increases during wet weather, likely from upstream sewage overflows. This study demonstrates that CDS-NbS systems can achieve regulatory water-quality standards under tropical monsoon conditions. Applying to the official DOE WQI provides a critical link between engineering performance and regulatory decision-making, supporting the wider adoption of hybrid grey–green infrastructure across rapidly urbanising regions.

ABSTRACT. Nature-based Solutions (NbS) are increasingly recognized as sustainable approaches to address the growing challenges of urban water pollution, particularly in relation to sewage-contaminated runoff under changing climatic and urbanization pressures. This study investigates the application of biochar, specifically in the form of biochar socks, as an innovative NbS for the treatment of sewage-impacted water in urban and peri-urban environments. Biochar socks, permeable fabric tubes filled with engineered biochar offer a flexible, low-cost, and easily deployable solution for improving water quality in drainage systems where sewage intrusion from leakages or overflows is common. The research evaluates the effectiveness of biochar socks in removing key sewage-related contaminants. Experimental investigations were conducted using simulated sewage-contaminated water to assess treatment performance under controlled flow conditions. The pollutant removal efficiency was examined through column and flow-through studies, targeting organic matter (COD and BOD), nutrients (nitrogen and phosphorus), suspended solids, and selected microbial indicators. Results demonstrate that biochar socks function as efficient passive filtration and adsorption systems. The porous structure and high surface area of biochar enable strong adsorption of organic pollutants and nutrients, while also facilitating the physical trapping of suspended solids. Furthermore, the biochar matrix provides a favorable environment for microbial interactions, contributing to the reduction of sewage-associated contaminants. The findings highlight the potential of biochar socks as a practical and scalable NbS for decentralized sewage water treatment, particularly in regions with aging or inadequate sewer infrastructure. Their modular design allows for easy integration into existing drainage networks, enhancing water quality without the need for complex treatment systems. Overall, this study demonstrates that biochar socks offer a sustainable, cost-effective, and environmentally friendly solution for mitigating sewage pollution, supporting circular economy principles through the valorisation of biomass into high-value environmental materials.

ABSTRACT. Nature-based solutions (NbS) are increasingly recognized as essential for sustainable river basin management by providing ecological, hydrological, and societal benefits. Selecting the suitable location for any NbS can maximize their effectiveness as well as can balance the trade-off between economic and environmental benefits. Therefore, the site selection needs distinct framework that integrates the properties of study area, ecological, economic and social factors, and stakeholder input. In this context, Geographic Information Systems (GIS) combined with Multi-Criteria Decision Analysis (MCDA) methods have emerged as powerful tools for mapping and evaluating NBS in many river basins.

Among various NBS strategies, wetland restoration plays a crucial role in flood regulation, water purification, biodiversity conservation, and climate resilience. Effective wetland management supports ecological health while promoting social and economic benefits, thereby contributing to multiple Sustainable Development Goals (SDG 1, 2, 5, 6, 8, 9, 11, 13, 14, 15) simultaneously. However, wetlands are increasingly threatened by land-use change, urban expansion, pollution, and climate change, leading to significant degradation of their ecological functions.

This study focuses on identifying suitable sites for NbS, particularly for wetland restoration in the agriculture-dominated Thamirabarani River Basin in southern India using an integrated GIS–MCDA framework. The analysis incorporates spatial datasets along with stakeholder-informed weighting of criteria and compares different MCDM approaches, including Analytic Hierarchy Process (AHP), VIKOR, and Technique for Order Preference by Similarity to Ideal Solution (TOPSIS). The parameters considered in this study includes topography, Land Use and Land Cover (LULC), soil characteristics, proximity to drainage networks, and hydrological conditions. These factors were standardized and integrated using weighted overlay analysis to generate a composite suitability index that classifies areas into different suitability categories for wetland restoration.

The results shows that the proposed framework effectively identifies spatially distributed zones suitable for wetland restoration. Areas characterised by low slopes, poorly drained soil, lower agricultural intensity, and proximity to drainage networks show higher suitability for wetland restoration and management. The generated suitability maps provide a spatial decision-support tool for identifying priority locations where wetland-based NBS can enhance hydrological regulation and ecosystem services. This study highlights the importance of integrating advanced GIS and MCDA methodologies with stakeholder engagement to support decision-making for NBS planning. This approach can be adapted to other river basins and extended to evaluate additional nature-based interventions for sustainable watershed management.

ABSTRACT. Large-scale human activities (e.g., dredging, reclamation) and natural disasters (e.g., floods) often elevate suspended sediment concentrations (SSC) in riverine, estuarine, and coast, threatening aquatic ecosystems and complicating effective management. Rapid and spatially detailed monitoring of sediment dispersion is therefore essential for proactive disaster response and resilience building. However, traditional ship-based sampling and fixed-point sensors lack spatial coverage, while satellite remote sensing suffers from low temporal resolution and weather sensitivity. Consequently, real-time detection of localized and sudden sediment events remains a critical gap in current monitoring frameworks. This study introduces an integrated surface SSC monitoring approach combining RGB imagery from Unmanned Aerial Vehicles (UAVs) with real-time data collected by moored sensor buoys. Multiple buoys were deployed in a tributary of the Mangyeong River to continuously record turbidity, water temperature, and position. Concurrently, UAVs captured orthorectified images that were spatially aligned with the buoy observation domain to ensure geometric accuracy. Analysis focused on the red-band Pixel Intensity Ratio (PIR), which exhibited strong sensitivity to sediment optical properties. A robust linear relationship (R² = 0.84) was identified between PIR and in situ SSC, demonstrating that quantitative SSC estimation is achievable using standard RGB imagery. The resulting two-dimensional SSC maps effectively visualize sediment generation, transport, and dilution processes. The proposed framework overcomes spatial limitations of point-based monitoring, providing rapid identification of sediment diffusion and pollution hotspots during disaster-related events. By facilitating timely situational awareness and data-driven decision-making, this method enhances socio-environmental resilience against water-related hazards and supports adaptive river and coastal management. Future work will integrate acoustic and optical profiling techniques, along with multispectral imaging, to extend sediment monitoring into three-dimensional domains.

ABSTRACT. The swash zone is a highly energetic region characterized by repeated uprush and backwash, generating complex turbulent flows. These hydrodynamic processes significantly influence coastal morphological evolution and stability. This study investigates the hydrodynamic characteristics of swash-swash interactions and bed instability by reproducing swash flows using a double dam-break mechanism. The double dam-break setup reproduces swash-swash interactions on a sloping bed by controlling the release timing of two installed gates. The numerical model is based on OpenFOAM and solves the Reynolds-Averaged Navier-Stokes (RANS) equations with two immiscible fluids using a k-ω Shear Stress Transport (SST) turbulence closure. The gate movements are effectively implemented using the overset mesh method within the overInterDyMFoam solver. The numerical results were validated by comparing them with laboratory data for swash flows on a rough immovable planar beach. Good agreements were obtained for free surface elevation, velocity, and horizontal pressure gradient. Based on the numerical results, swash-swash interactions were classified into Wave Merging (WM), Wave Capturing (WC), and Backwash-Uprush Interaction (BUI), and their flow characteristics were investigated. During the backwash stage of WM type, the potential for bed instability is primarily driven by shear stress. During BUI type shows significant potential for bed instability induced by pressure gradients, accompanied with the highest turbulent kinetic energy and intense vorticity. These findings provide a qualitative classification of swash-swash interactions under idealized conditions, offering a foundation for extending research to permeable beds and sediment transport processes.

ABSTRACT. In this study, future wave characteristics along the Korean Peninsula were evaluated using wind field data derived from climate change scenarios and the SWAN model. Several researchers (e.g., Li et al., 2022 and Shimura et al., 2020) investigated future changes of regional wave climate in the East Asian seas based on the RCP scenarios presented in the IPCC fifth assessment report. These studies suggested an overall decreasing trend in the annual mean significant wave height in the sea waters around China and Japan. However, research on future wave climate based on regional wave modeling remains relatively limited around the Korean peninsula. In this respect, this study analyzed future projections of wave climate in the coastal regions surrounding the Korean Peninsula by carrying out high-resolution wave simulation with the future wind fields based on the IPCC sixth assessment report. Future wave predictions were conducted using two regional climate models, HadGEM3-RA and CCLM, based on SSP scenarios provided by the National Institute of Meteorological Sciences of Korea. Both models feature a spatial resolution of 25 km, and simulations were performed for the period from 2020 to 2100. Two contrasting scenarios were considered: SSP1-2.6, a low-emission scenario assuming carbon neutrality and sustainable growth, and SSP5-8.5, a high-emission scenario assuming continued fossil fuel reliance. Wind field data extracted from the HadGEM3-RA and CCLM models were used as inputs for the SWAN model to derive long-term changes in wave characteristics, including significant wave height and direction. Annual mean significant wave heights were calculated to analyze long-term trends, and seasonal variations were examined at three-month intervals. Additionally, extreme significant wave heights, defined as the 99.9th percentile, were estimated to assess potential maximum wave conditions. These results are expected to provide essential baseline data for understanding climate-induced variations in the future wave environment and for developing strategies to mitigate coastal disaster risks.

ABSTRACT. Wave reflection and transmission characteristics of permeable low-crested structures are commonly evaluated under conditions in which water levels on the seaward and landward sides are identical. In tidal environments, however, water level differences frequently develop across such structures. This study investigates how water level differences on either side of a permeable low-crested structure influence wave reflection and transmission under irregular wave conditions, in comparison with cases where water levels on both sides are the same. Hydraulic model experiments were conducted in a two-dimensional wave flume using a rubble-mound permeable low-crested structure. Irregular waves were generated under controlled conditions, and three water level configurations were considered: identical water levels on both sides of the structure, a higher water level on the seaward side, and a higher water level on the landward side. Wave reflection and transmission coefficients were evaluated from wave measurements obtained in front of and behind the structure, and their relative tendencies were examined with respect to the identical water level condition. The results show that the presence of a water level difference leads to systematic changes in wave reflection and transmission behavior under irregular waves. When the seaward water level is higher, wave transmission becomes more pronounced compared to the identical water level condition, indicating enhanced wave energy passage through the permeable core. In contrast, when the landward water level is higher, wave reflection is relatively emphasized, while wave transmission remains restrained. These contrasting responses indicate that the direction of the water level difference governs internal flow patterns and wave energy dissipation processes within permeable low-crested structures. The findings demonstrate that wave reflection and transmission characteristics under irregular wave conditions cannot be adequately described by assuming identical water levels alone. Considering water level differences across permeable low-crested structures is therefore essential for reliable hydraulic performance assessment and for the design of coastal structures exposed to tidal asymmetry and water level variability.

ABSTRACT. The potential risk of coastal hazards has been escalating due to global warming-induced sea-level rise and increasing sea surface temperatures. This vulnerability is further exacerbated by rapid industrialization and population concentration in coastal urban areas, highlighting the urgent need for robust structural and institutional disaster mitigation measures. Traditional coastal defense strategies primarily employ grey infrastructure such as concrete breakwaters and revetments. However, these rigid structures often struggle to adapt to long-term environmental shifts and entail significant costs for maintenance and post-disaster recovery. Building upon this rationale, Nature-Based Solutions (NbS) have emerged as sustainable alternatives or complements to conventional methods. NbS leverage natural elements, such as coastal vegetation, to enhance ecosystem resilience while providing socio-economic benefits, including wave energy dissipation and carbon sequestration. This study quantitatively evaluates the applicability of nature-based coastal protection by conducting hydraulic experiments on the wave attenuation performance of rigid artificial vegetation. The experiments were performed in a wave flume (50 m x 1.2 m x 1.6 m) under both regular and irregular wave conditions, comparing a non-vegetated scenario with three distinct vegetation density configurations. A spatial array of five wave gauges was utilized to capture precise wave transformations across the vegetation zone. The results demonstrate that increasing vegetation density leads to a substantial reduction in wave height, with the wave height decay coefficient reaching up to 0.16 under specific irregular wave conditions. These findings provide a quantitative benchmark for the performance of NbS and underscore the efficacy of nature-based coastal defense systems. By integrating marine ecosystem conservation with disaster risk reduction, this research contributes to the development of economically feasible and sustainable coastal management strategies in an era of climate uncertainty.

ABSTRACT. Baisha Bay sandy beach is continuously affected by several typhoons in a short period of time. Based on the measured beach profile data and field investigation before and after typhoons, the response of the beach to continuous typhoons was studied in the paper. The results showed that beach profiles eroded in general, and the basic response characteristics of beach profiles to continuous typhoons were erosional downcutting of the beach profiles and steepened beach slope. The response characteristics of beaches at different locations to typhoons were different. A typical storm profile was formed after the post-storm action in the eastern beach area, with severe local erosion and obvious distribution of underwater sand bars, while such characteristics were not evident in other sections. The typhoons’ response intensity of the southern beach profile was the least. Surficial sediments at each beach were slightly coarsened and the sorting was weakened. The dynamic response mechanism of the beach under continuous typhoons is complex. The nearshore geology and geomorphology, the typhoon dynamic process and coastal engineering all have certain influence on the beach storm effect.

ABSTRACT. The interface between terrestrial aquifers and coastal water bodies—the subterranean estuary—is increasingly recognized as a critical flux for nutrient and solute transport. While the biogeochemical magnitude of Submarine Groundwater Discharge (SGD) has been well-constrained, often exceeding riverine inputs in specific hydrogeologic settings, its role as a physical driver of estuarine hydrodynamics remains systematically under-represented in standard oceanographic models. This study presents a comprehensive, coupled surface water-groundwater modeling analysis quantifying the impact of SGD on intertidal mixing and estuarine circulation across a range of estuary types and different tidal regimes (micro-, meso-, and macrotidal). Equivalent non-SGD models are conducted by setting impermeable boundary at the benthic domain. Results indicate that SGD exerts a nonlinear control on estuary salinity, functioning primarily as a salinity buffer with different time lags. Our results suggested that omitting SGD from hydrodynamic models leads to a systematic overestimation of average salinity and an underestimation of circulation paths, with profound implications for nutrient budgeting and eutrophication management.

ABSTRACT. In hydrodynamic and water-quality simulations targeting coastal areas, hydrostatic approximation is often used. While the hydrostatic approximation has low computational cost, non-hydrostatic models have been reported to be useful for analyzing vertical phenomena, such as the upwelling of bottom nutrients due to estuarine circulation, the upwelling of bottom hypoxia, and the reproduction of density structures. Its use is limited temporally and spatially due to high computational costs, and few studies have discussed their effects at the bay scale and on an annual basis when using non-hydrostatic models. This study conducted annual hydrostatic and non-hydrostatic current and water-quality simulations in Osaka Bay to evaluate the effects of the no use of the hydrostatic condition on currents and material cycles. This study performed simulations using a non-hydrostatic flow model kinaco (Matsumura and Hasumi, 2008) and RCA an water quality model (HydroQual, Inc., 2004), comparing hydrostatic (HM) and non-hydrostatic models (NHM). The kinaco model enables high-speed non-hydrostatic simulations and allows selection between hydrostatic and non-hydrostatic conditions via options. RCA represents carbon, nitrogen, phosphorus, and oxygen cycles, enabling reproduction of complex coastal material cycles. The calculation period was one year of 2012, with 2011 as a pre-conditioning calculation. Parameters were kept unchanged between HM and NHM. Considerable differences in both water temperature and DO were observed during the density-stratified summer period. The NHM model shows lower water temperature and DO near the eastern shoreline during the northeast wind, indicating it captures the observed values well. While the HM model showed vertical mixing, resulting in uniform water temperature and DO, the NHM model showed no vertical mixing, revealing upwelling of bottom layer water near the coastline. The NHM is considered capable of representing coastal upwelling that occurs along the eastern shore. We also analyzed the phosphorus cycle in the eastern bay during early September. Inorganic phosphorus flux (I-P) is greater in NHM than in HM, with transport particularly strengthened from the lower central bay to the lower eastern bay and from the lower eastern bay to the upper eastern bay. The increase in I-P transport from the lower eastern bay to the upper layer in NHM, compared to the transport in HM, was estimated to equal half of the supply from rivers. Therefore, careful consideration is required when deciding whether to employ HM or NHM.

ABSTRACT. The Cilacap estuary and its link to the Segara Anakan Lagoon constitute a dynamic coastal system wherein tidal forces and river discharge interact to influence circulation, sediment transport, and long-term morphological development. Despite comprehensive research on the lagoon, the tidal characteristics and asymmetry at the Cilacap river mouth are still inadequately comprehended, especially concerning their impact on the eastern lagoon. This study seeks to examine tidal asymmetry in the Cilacap estuary and evaluate its hydrodynamic consequences for the Segara Anakan system. The primary concerns examined encompass (1) the identification of predominant tidal constituents and nonlinear tidal distortion, (2) the characterization of flood-ebb asymmetry in current magnitude and duration, and (3) the assessment of tidal processes' impact on residual circulation and net transport. Field observations were performed at four stations ranging from 2 km offshore to 5 km upriver, utilizing ADCP and AWAC instruments, along with a coastal tide gauge. Harmonic and statistical analyses, encompassing skewness, probability density functions, and cross-correlation, were employed to quantify tidal characteristics and asymmetry. A two-dimensional hydrodynamic model was created utilizing the DHI MIKE 21 Flexible Mesh system and validated against empirical water levels and currents (correlation coefficients > 0.8; normalized RMSE < 20%). The findings reveal a mixed semi-diurnal tidal regime primarily influenced by M2 and K1 constituents. Overtides (M4, MS4) intensify landward, indicating nonlinear tidal distortion. Velocity skewness values (0.25–0.99) affirm a flood-dominant asymmetry regarding peak magnitude, despite the increased frequency of ebb events near the estuarine mouth. Cross-correlation analysis indicates a lag of 1.8 to 2.3 hours between tidal elevation and current response. and continuous inward residual flows, signifying a predominance of net flooding and landward transport. The findings indicate that tidal asymmetry in the Cilacap estuary significantly influences circulation patterns and sediment dynamics in the eastern Segara Anakan Lagoon, offering essential information for estuarine management and morphological evaluation.

ABSTRACT. Rip currents pose persistent hazards at energetic beaches, motivating the development of prediction systems that can operate across diverse wave and water level conditions. In South Korea, operational interest accelerated after several high impact “mega rip” events at Haeundae Beach, which prompted the development of a pre simulation (PS) approach for rip current prediction. The PS framework performs ensembles of phase resolving FUNWAVE simulations over a wide range of incident wave conditions and tidal elevations, and extracts scenario dependent indicators of rip current likelihood to build a structured database. This database has supported look up style guidance and also provides a strong foundation for data driven pattern discovery. This paper introduces the operational PS based prediction concept, which is currently deployed across 10 beaches with reported accuracy exceeding 84 percent, and presents an ML PS framework that learns statistical relationships between environmental forcing (waves and tides) and PS derived rip current metrics for a target beach. The proposed approach enables rapid prediction without running a numerical model in real time, while retaining the physics embedded in the pre simulated library. A forward looking pathway for broader applicability is also discussed. By incorporating bathymetric information as model inputs during training, the ML PS framework can be extended toward transfer learning, allowing a model trained on representative morphologic settings to be adapted to new locations. Practical considerations for deployment are summarized, highlighting ML PS as a scalable bridge between physics based simulation and operational rip current forecasting.

ABSTRACT. Tsunami-induced coastal inundation poses a critical threat to freshwater resources in coastal regions by triggering rapid salinization of coastal aquifers. Unlike gradual seawater intrusion driven by long-term sea-level rise, tsunami inundation introduces seawater abruptly onto the land surface, promoting vertical penetration into the aquifer and altering groundwater flow regimes. This study experimentally investigates the mechanisms of saltwater intrusion and the subsequent recovery processes in a coastal aquifer subjected to tsunami-like inundation conditions. A series of controlled sand-tank experiments were conducted to simulate coastal aquifers bounded by an impermeable vertical structure. Steady-state seawater–freshwater equilibrium conditions were first established, after which episodic inundation events with varying inundation heights and distances were imposed to represent tsunami flooding scenarios. High-resolution image-based analysis was employed to track the spatiotemporal evolution of the seawater body within the aquifer, enabling quantitative assessment of salinization intensity and recovery dynamics. The results reveal that tsunami inundation primarily induces vertical seawater intrusion from the ground surface, leading to rapid expansion of the saline zone beyond the initial equilibrium state. The extent of salinization increases with inundation distance, while inundation height governs the degree of interaction between infiltrated seawater and the pre-existing saline wedge. Recovery processes are governed by groundwater hydraulic gradients and density-driven flow, with restoration times strongly dependent on inundation distance rather than inundation height. Longer inundation distances significantly prolong recovery, indicating persistent salinity impacts even after surface flooding has receded. These findings demonstrate that tsunami-induced salinization follows distinct mechanisms compared to conventional coastal seawater intrusion and that recovery timescales can be substantially extended depending on inundation characteristics. The experimental insights provide a physical basis for assessing groundwater vulnerability to extreme coastal hazards and highlight the necessity of incorporating episodic inundation effects into coastal aquifer management and resilience planning under future tsunami and compound disaster scenarios.

ABSTRACT. With the rapid growth of the offshore aquaculture industry and its expansion from coastal to offshore farming, there is a growing need for a reliable engineering approach to evaluate the structural design stability of large-scale offshore aquaculture systems against various oceanic external forces under extreme marine environments. Evaluation of dynamic response, performance of mooring system, and marine environmental loads is necessary for the safe design and long-term operation of offshore aquaculture structures. This study aims to analyze the behavior of a 35-meter-class offshore aquaculture system in preliminary, before evaluating the offshore environment (200-meter-class) through hydraulic experiments, in order to understand the physical behavior characteristics under extreme wave conditions, load variations due to seaweed growth, and the interaction between the structure and the mooring system. Furthermore, based on various previous studies, this research will provide critical foundational data for establishing design guidelines for offshore aquaculture and management facilities in Korea. The hydraulic experiment was performed to evaluate the dynamic behavior of an offshore aquaculture structure under various wave conditions. A simplified model of the aquaculture structure was fabricated using seaweed samples collected from the Tongyeong Marine Living Resources Station of the Korea Institute of Ocean Science and Technology. The physical experiment was carried out in a wave flume(50m(L)x1.2m(W)x1.6m(H)). Four buoys were connected in the direction of the flume, and samples were attached to them. The tension in each mooring line was individually measured using load cells, and the entire process was recorded using a video camera. The tension measured at each buoy was analyzed by location, and spectral analysis was applied to the recorded video to quantitatively evaluate the nonlinear behavior characteristics resulting from surge and heave motions of the buoys. It is expected that this will contribute to establishing design improvements and reinforcements to ensure the structural stability of offshore aquaculture systems, as well as to verifying the performance of the mooring system.

ABSTRACT. Local pier scour is a major contributor to bridge instability, motivating prediction tools that quantify both expected scour depth and its uncertainty. This paper proposes an interpretable, data-driven workflow that integrates CatBoost for deterministic estimation and NGBoost for probabilistic modeling. The models are trained using 552 laboratory observations of pier scour, represented by four dimensionless predictors (y/b, V/Vc, b/d50, and Fr). Predictive skill is assessed with common regression metrics, while feature contributions are examined via SHAP-based interpretation. To connect the predicted scour depths to risk-based design, a Monte Carlo reliability procedure is employed to compute the reliability index β and the probability of exceedance, using a simple multiplicative adjustment. Results on a held-out test set indicate that CatBoost achieves marginally better point-estimation accuracy, whereas NGBoost provides calibrated uncertainty bounds, with empirical coverages close to the nominal 68% and 95% levels. Overall, the proposed framework supports accurate, explainable, and uncertainty-aware scour assessment for target-reliability and risk-informed bridge design.

ABSTRACT. Coastal risk assessment and coastal infrastructure design in low-lying delta cities are often based solely on projected sea level rise, assuming that land surfaces remain stable. However, in many coastal cities, such as Jakarta, this assumption does not hold true due to significant land subsidence. As a result, sea level rise relative to land surfaces also occurs. Relative Sea Level Rise is controlled by the combined effects of sea level rise and land subsidence factors. These two processes occur simultaneously but are still often analyzed separately, which can lead to an underestimation of future coastal flood risks and inappropriate coastal protection planning. Extreme hydrometeorological events on the coast that often occur due to global warming can be evaluated using Extreme Value Analysis (EVA). This study evaluates coastal risk areas through an integrated assessment of Relative Sea Level Rise (RSLR) by combining satellite observations of sea level rise and land subsidence. Land subsidence is estimated using Sentinel-1 Interferometric Synthetic Aperture Radar (InSAR) data to obtain the rate of vertical land deformation along coastal areas, while sea level variations are obtained from satellite altimetry data representing long-term sea level changes in nearby coastal waters. Both datasets were temporally aligned and referenced to the same vertical reference system to form a consistent RSLR time series. In addition to long-term trend analysis, this study applied Extreme Value Analysis (EVA) to characterize extreme RSLR events by extracting maximum values and modeling the probability of extreme events and their recurrence periods. These recurrence interval estimates are then used as a quantitative basis for coastal hazard assessment and determining the design elevation of coastal infrastructure, making the proposed approach more representative of the extreme conditions relevant to future coastal adaptation planning and evaluation of the reliability of coastal protection structures.

ABSTRACT. Climate change is intensifying water-related natural hazards such as extreme precipitation, storm surges, coastal flooding, and sea-level rise, posing increasing risks to nuclear power plant (NPP) safety. Flooding at NPP sites can lead to loss of off-site power, damage to safety-related structures, systems, and components (SSCs), and reduced cooling capacity, potentially resulting in plant shutdowns or severe accidents. According to NOAA and NRC assessments, a significant proportion of nuclear power plants are located in regions vulnerable to hurricane-induced storm surges, high flood risks, or sea-level rise, with climate change expected to exacerbate these hazards. Therefore, comprehensive evaluation of compound flood hazards at NPP sites is a critical safety concern. This study presents a simulation-based framework for assessing inundation risks at coastal nuclear power plant sites under extreme water-related hazards. In particular, the research focuses on inundation caused by extreme-frequency wave overtopping associated with storm surges. A three-dimensional Lagrangian-based numerical model, DualSPHysics, was employed to simulate external flooding processes and inundation behavior within a nuclear power plant site. Extreme-frequency wave conditions corresponding to a 1-in-1,000,000-year return period were derived using national deep-water design wave specifications provided by the Ministry of Oceans and Fisheries of the Republic of Korea. These conditions were applied to wave estimation models, and overtopping discharges were calculated using the EurOtop overtopping formulation. The resulting overtopping rates were subsequently used as input for three-dimensional inundation simulations. High-resolution terrain and structural data were constructed to accurately represent the nuclear power plant site, including detailed digital elevation models and three-dimensional building geometries. The simulation domain incorporated realistic topography and structural configurations to capture the complex interactions between floodwaters and plant facilities. Numerical simulations were performed with fine particle resolution to evaluate inundation depth, flow behavior, and potential impacts on safety-related facilities. The results demonstrate the applicability of Lagrangian-based three-dimensional flood modeling for evaluating compound flood hazards at nuclear power plant sites under extreme climate scenarios. This approach provides valuable insights for flood risk assessment, safety evaluation, and the development of enhanced flood protection strategies for nuclear facilities facing increasing climate-driven hazards.

ABSTRACT. Tsunami-induced inundation can generate large drifting debris, such as containers, which may collide with coastal structures and cause severe secondary damage. This study investigates the effectiveness of a submerged horizontal plate (SHP) in reducing debris impact forces under tsunami-like wave-induced inundation, based on hydraulic experiments conducted in a wave flume. The experiments were performed in a two-dimensional wave flume, where solitary waves were generated to reproduce tsunami-like inundation. Drifting containers were allowed to collide with a fixed column located 100 cm landward of the shoreline, representing debris impact within the inundated region. The experimental setup considered structures with either a vertical revetment (VR) or a wave-absorbing revetment (WAR), and comparative tests were conducted with and without the installation of the SHP. Containers representing partially loaded and fully loaded conditions were tested, and different initial debris positions relative to the shoreline were examined. The analysis focused on the maximum horizontal impact force measured during debris–structure collision. The experimental results indicate that the installation of the SHP leads to a systematic reduction in debris impact forces for both VR and WAR. Compared to cases without the SHP, the maximum horizontal impact force is reduced by approximately 25–35% for VR and by about 15–25% for WAR. Although fully loaded containers generate larger absolute impact forces than partially loaded containers, the relative reduction associated with the SHP remains comparable across container loading conditions. Variations in the initial debris position influence the magnitude of the impact loads but do not alter the overall reduction trend induced by the SHP. Overall, the results demonstrate that the SHP consistently reduces debris impact forces measured in the inundated region under tsunami-like wave conditions, regardless of revetment type or container loading condition.

ABSTRACT. River-derived sediment plumes are an essential component of the coastal sediment budget. Coastal areas in Suruga Bay, Japan, have experienced substantial coastal erosion over the past decades. In this study, we analyzed the spatial patterns of total suspended matter (TSM) in Suruga Bay to evaluate the influence of river plumes on sediment transport within the bay. TSM data were obtained from Sentinel-3 OLCI images spanning eight years (2017–2024). After cropping each image to the Suruga Bay region, we compiled only those scenes containing more than 70% valid (cloud-free) pixels. A self-organizing map (SOM) was applied to classify dominant plume patterns. The reconstructed spatial distributions revealed a consistently skewed plume transport originating from the Oi and Abe Rivers. The dominant patterns indicated southward-directed transport driven by the bay’s counter-clockwise circulation, with plumes converging toward the sediment trail emanating from the western bay mouth near Cape Omaezaki. These results suggest that the plume pathways identified from satellite observations can serve as an important reference for understanding and assessing ongoing coastal erosion in Suruga Bay.

Acknowledgements: This research was supported by the National Research Foundation of Korea (NRF) grant funded by the Korean government (MSIT) (No. RS-2024-00336456).

ABSTRACT. Wave overtopping at coastal structures is a major concern for infrastructure safety and disaster risk management. Despite its importance, continuous field-scale measurements of individual wave overtopping height and volume remain scarce, limiting the validation of overtopping prediction methods and the development of operational monitoring systems. This study presents a novel, fully automated field observation framework for real-time quantification of wave overtopping at a prototype breakwater located at Yeondaepu breakwater, Jeju Island, South Korea. The monitoring system consisting of multiple LiDARs and CCTVs, operate continuously to capture overtopping processes at the scale of individual waves. Three LiDAR sensors are mounted on a rigid frame that is installed along the breakwater crest to measure overtopping depth on the crest as well as record its temporal evolution. Beneath the LiDAR frame, an overtopping tank is installed to directly collect and measure the volume of each overtopping wave splash, enabling synchronous geometric and volumetric measurements. In parallel, a high-resolution CCTV camera is installed behind the measurement system to record overtopping events. An object detection model (YOLO) is applied to the CCTV footage to automatically detect overtopping events in real time, which are represented by bounding boxes in image space. The CCTV detected overtopping events are subsequently calibrated using concurrent LiDAR-derived overtopping heights and measured overtopping volumes from the tank. Through this multi-sensor calibration approach, video-based detections are converted into physically meaningful estimates of individual wave overtopping height and volume without manual intervention. The object of the proposed system was to demonstrate the operational viability of a continuous, field-scale, event-based overtopping monitoring by combining direct measurements and artificial intelligence. The resulting dataset provides a valuable basis for improving overtopping prediction models, supporting real-time hazard assessment, and advancing resilient coastal infrastructure design.

ABSTRACT. This study presents a three-dimensional hydrodynamic modeling framework for assessing the hydrodynamic impacts of developing offshore wind farms on the inter-Korean border waters. A high-resolution numerical model based on the finite-difference estuary model has been configured for the Gyeonggi Bay, including the border water, incorporating bathymetry, tidal/wave/wind forcing, and detailed boundary conditions. Wind turbine structures are parameterized as enhanced drag. This technique treats the turbine foundation structure as porous and applies a corrected drag coefficient and an increased roughness coefficient. The model domain extends from the open ocean boundary to nearshore areas with grid resolution ranging from 250 m to 25 m. Model validation will be conducted using available tide gauge data as well as moored ADCP and CTD measurements in the bay. Scenarios involving different turbine arrangements and densities will be simulated to study the impact on changes in tidal current patterns, stratification structures, and estuary circulations. This methodology provides a systematic approach to assessing the environmental impact assessment of offshore renewable energy infrastructure in estuarine zones.

ABSTRACT. The coastal areas are also becoming susceptible to erosion and flooding caused by waves as the impacts of climate change and rising sea levels become more significant. The breakwaters are common forms of protective coastal constructions, though the traditional designs usually offer minimal energy dissipation during the accelerated waves. Combining natural features like vegetation and artificial buildings in the coastal area has become a promising strategy towards enhancing coastal resilience and sustainability. This study presents a numerical investigation of wave interaction with a vegetation-integrated curved breakwater system. The objective is to evaluate the influence of vegetation density on wave attenuation and hydrodynamic flow characteristics around the structure. A two-dimensional numerical wave tank is developed using computational fluid dynamics simulations in ANSYS Fluent. Regular waves are generated at the inlet boundary using a multiphase Volume of Fluid (VOF) model to capture the interaction between water and air phases. Vegetation has been modeled as a set of cylindrical elements before the curved breakwater to explore the impact of vegetation densities on energy dissipation of waves. The system performance is compared based on the reduction of the wave height, transmission coefficient, and dissipation of the energy. The results show that the vegetation has a significant impact on the breakwater performance. For the configuration without vegetation, the transmission coefficient was observed to be approximately 0.62, whereas the introduction of vegetation reduced the transmission coefficient to 0.41–0.46, depending on vegetation density. In addition, the vegetation-integrated configuration resulted in an increase in wave energy dissipation of about 30–35% compared with the conventional breakwater case. These findings demonstrate that the use of vegetation as an extension of the coastal protection structures is a practical approach to increase the wave attenuation and the general effectiveness of the breakwaters. The suggested solution identifies the possibilities of integrating environmental factors along with artificial coastal infrastructure to come up with sustainable and resilient coastal defense systems.

ABSTRACT. This study examines the geometric characteristics of tidal turbine arrays optimized for power extraction in an idealized shallow channel. Numerical experiments are performed using OpenTidalFarm, a PDE-constrained, gradient-based optimization framework. To alleviate the computational cost of identifying a global optimum, a quasi-global optimum is introduced as a representative local solution. The resulting optimal layouts exhibit systematic geometric transitions governed by a nondimensional parameter E, defined as the ratio of the minimum hypothetical linear fence length to the lateral extent of the farm domain. For E≤1, the optimal configuration forms a linear fence, whereas larger values of E yield parabolic and V-shaped arrangements. As the number of turbines increases, the optimal layout departs from a single continuous fence. Quantitative results demonstrate that appropriate tuning of optimization constraints can enhance power production by up to 50%.

ABSTRACT. The integral time scale (ITS) characterizes the dominant energy-containing temporal scale in turbulent flows and is widely used in turbulence analysis and modeling. In experimental measurements, however, ITS estimation is often affected by spectral distortions originating from measurement noise and finite resolution. In particular, high-frequency noise floors in power spectral density (PSD) can significantly bias ITS estimates, even when their energy level is several orders of magnitude smaller than that of the turbulent signal. In this study, we systematically investigate the influence of spectral noise on ITS estimation using a frequency-domain framework. A parametric PSD model is constructed to represent an ideal turbulent spectrum combined with a high-frequency noise floor. By varying the noise level over a wide range, its effect on the autocorrelation function and the resulting ITS is quantitatively assessed. The results demonstrate that even a weak noise floor can substantially reduce the estimated ITS, while the zero-crossing time of the autocorrelation function remains largely unchanged, indicating that conventional time-domain indicators may fail to detect noise-induced bias. Based on these findings, a practical noise elimination methodology is proposed. The approach identifies the transition between the inertial subrange and the noise-dominated region using slope-based regression in logarithmic spectral space. The noise floor is then replaced by an extrapolation of the inertial-range scaling, yielding a refined PSD suitable for robust ITS evaluation. To improve stability, ensemble-averaged spectra are employed to mitigate logarithmic fluctuations. The proposed method is validated using velocity data derived from direct numerical simulation (DNS), in which controlled white noise is artificially injected. While the original distorted spectra cannot be fully recovered, the refined spectra yield substantially improved agreement with the underlying turbulent time scales compared to conventional estimation approaches. The results highlight the critical role of spectral noise treatment in integral-scale estimation and provide a practical framework for improving ITS accuracy in experimental turbulence measurements.

ABSTRACT. Urban flooding represents a complex disaster in which extreme rainfall, urban topography, drainage capacity, land use, and climate change interact to generate cascading impacts on critical infrastructure. During flood events, power outages and communication failures frequently compromise sensing, monitoring, and coordination capabilities, thereby limiting the effectiveness of emergency response. Although flood forecasting and hydraulic modeling technologies have advanced significantly, their operational value is constrained when supporting infrastructure cannot be sustained under inundated conditions. This study proposes a resilience-oriented framework for enhancing urban flood response through the strategic deployment of distributed power systems as disaster response infrastructure. Rather than prioritizing large-scale centralized generation, the framework emphasizes small-scale, site-adaptive power sources capable of maintaining electricity supply to essential field assets, including sensors, CCTV systems, communication relays, and early warning devices during flood events. In particular, the study considers flow-induced energy resources available in rivers, drainage channels, and flood conveyance facilities as viable contributors to distributed emergency power supply. The primary objective of the research is to identify optimal deployment locations for distributed energy systems from a disaster response perspective. The proposed approach integrates flood hazard characteristics (e.g., inundation depth and flow velocity), infrastructure vulnerability, functional importance of response facilities, accessibility, and system survivability under flood conditions. Emphasis is placed not on maximizing energy production efficiency, but on enhancing operational continuity, rapid recoverability, and direct linkage with on-site response activities. By leveraging localized hydraulic conditions, such as sustained flow during flood events, flow-driven energy systems can complement conventional backup power sources and reduce reliance on fuel-based generators. This integration enables continuous operation of monitoring and communication infrastructure, thereby strengthening the connection between real-time flood forecasting outputs and field-level decision-making. The results suggest that distributed energy systems—particularly those utilizing in-situ hydrodynamic conditions—can be repositioned as functional components of urban flood response infrastructure rather than conventional energy assets. The proposed framework provides practical insights for disaster resilience planning, urban infrastructure design, and policy development, offering a scalable strategy for incorporating flow-based distributed energy into flood response systems. This research is anticipated to offer implications for the resilient urban systems literature by reframing energy infrastructure as an enabler of adaptive, sustained, and decentralized disaster response.

Acknowledgments This research was supported by the Korea Institute of Science and Technology Information (KISTI) under grant K25L1M4C4.

ABSTRACT. Submerged aquatic vegetation (SAV) is a major source of hydraulic resistance in rivers, wetlands, and shallow aquatic systems. Accurate prediction of vegetation-induced drag is therefore essential for modeling flow resistance and water-level variations in vegetated channels. However, vegetation is often idealized as rigid or quasi-static in hydraulic models, despite the fact that natural vegetation exhibits substantial flexibility and flow-induced motion. This study examines how the flexibility of submerged aquatic vegetation modifies drag force using three-dimensional fluid–structure interaction (FSI) simulations. The simulations resolve the fully coupled dynamics between the flow field and a flexible vegetation element subjected to steady open-channel flow. Under hydrodynamic loading, the vegetation undergoes pronounced streamwise bending driven by mean drag forces, while simultaneously exhibiting lateral oscillations associated with vortex shedding. In contrast to rigid vegetation, these oscillations actively interact with the surrounding flow, altering vortex formation and shedding characteristics. This interaction strengthens wake unsteadiness and promotes the development of a more coherent and intensified wake downstream of the vegetation. The intensified wake produced by flexible vegetation leads to a larger pressure difference between the upstream and downstream sides of the element. In aquatic environments, where fluid viscosity is relatively low, drag on submerged bodies is dominated by pressure (form) drag rather than viscous shear stress. The enhancement of vortex shedding and wake strength associated with vegetation flexibility therefore significantly increases pressure-based drag. As a result, the total drag acting on flexible vegetation is greater than that acting on rigid vegetation under identical flow conditions. These results demonstrate that vegetation flexibility can amplify, rather than reduce, hydraulic drag through unsteady flow–structure interactions. Neglecting deformation and oscillatory motion may lead to systematic underestimation of drag in vegetated flow models. The findings underscore the need to incorporate flexibility-induced effects into drag formulations for submerged aquatic vegetation to improve hydraulic predictions in natural and engineered aquatic systems.

ABSTRACT. Climate-driven increases in localized intense rainfall are raising the likelihood of incidents at aging reservoirs, where overtopping, piping, internal erosion, and slope instability remain key failure modes. In Korea, many small and mid-sized reservoirs have been in service for decades and are managed by local governments under limited staffing and budgets. Safety management therefore still relies largely on visual checks and periodic inspections, which can miss early abnormal behavior and delay decisions during fast-developing events. This demand-driven regional cooperation project addresses this gap by developing and field-demonstrating a digital system that links continuous monitoring with early warning and emergency response support for aging reservoirs. System design was informed by a review of national inspection rules and instrumentation guidelines, including safety grading procedures. Conventional inspection items—such as settlement, displacement, cracking, seepage, and surface erosion—were mapped to measurable indicators suitable for sensor monitoring, including vibration, tilt, soil moisture, and visible surface changes. Based on this mapping and commonly reported weak points around spillways and conduits, a multi-sensor AIoT configuration was deployed at an operational reservoir testbed. The field system integrates accelerometers, inclinometers, soil-moisture sensors, surveillance cameras, and device temperature/humidity sensors, with data transmitted to a server for near real-time monitoring. For practical operation, incoming measurements are summarized in 10-minute windows using minimum, maximum, mean, standard deviation, and RMS statistics and displayed through a monitoring interface, while raw data are archived for detailed diagnostics and model development. To improve reliability during storms, the platform includes solar-assisted backup power, an internal battery designed for at least 24 hours of continued operation during primary power loss, and local storage to buffer data during communication failures. Sensors were installed in September 2025, and baseline data are being accumulated across seasons to build a reference dataset for anomaly screening. Current work focuses on data quality control, preprocessing, and pattern characterization; subsequent work will evaluate unsupervised anomaly detection models (K-means, Isolation Forest, LSTM autoencoder) and compare performance to identify a practical approach for real-time early warning. This study was funded by the major project of the Korea Institute of Civil Engineering and Building Technology (KICT) (grant number 20250369-001).

ABSTRACT. This study investigates long‑term soil moisture variability on the semi‑arid Mongolian Plateau and evaluates the accuracy of satellite‑derived estimates. Because coarse global products perform poorly in arid regions, independent validation is required. Ground measurements at 3 cm were compared with AMSR2 data from 2015–2025, and MODIS NDVI and snow cover were examined. AMSR2 captured seasonal variations but overestimated moisture in dry periods, underestimated values above 15%, and showed no clear spring snow‑cover influence.

ABSTRACT. The observation of three-dimensional wave fields is fundamental to marine studies, including coastal processes monitoring. Traditional point-based sensors, such as buoys, lack the spatial continuity required to capture localized wave events, while space-based systems utilizing radar or LiDAR often involve high costs and complex calibration techniques. This study addresses these limitations by demonstrating an easy-to-use and calibration-free methodology for 3D wave field reconstruction. Also presented is the application of the technique to wave breaking analysis in real-world coastal environments.

Field experiments were conducted at Manlipo Beach, South Korea. Two unsynchronized drone cameras and two mobile phone cameras were used to capture stereo imagery of the surf zone while an in-situ Aquadopp acoustic Doppler current profiler recorded validation data. The data was processed using the SWASSZ (Stereo Wave Analysis in Surf-Swash Zone), a novel calibration-free framework that utilizes a structure-from-motion and multi-view stereo (SfM-MVS) approach.

The method performed temporal synchronization by maximizing feature matches between frames and utilized ground control points surveyed via RTK GPS for georeferencing. This process transformed 3D point clouds from arbitrary camera coordinates into real-world coordinates.

The reconstruction achieved a georeferencing accuracy of order of O(10 cm). Validation against the Aquadopp measurements confirmed the reliability of the stereo-derived surface elevation time series. The resulting point clouds allowed for the extraction of spatial profiles to analyze wave breaking characteristics. We computed the surf similarity parameter at identified crests using the extracted breaking wave height.

The resulting values were in the range of spilling and plunging breakers. These quantitative findings were consistent with visual observations of actual breaking waves. This study shows that calibration-free stereo photogrammetry provides a practical, high-resolution tool for coastal engineering and sciences.

ABSTRACT. Soil moisture plays a critical role in regulating the exchange of water and energy between the land surface and the atmosphere, influencing both runoff and infiltration processes. Accurate soil moisture information is essential for monitoring droughts and early detection of levee failures caused by rapid moisture fluctuations. However, existing microwave-based satellite soil moisture products often lack sufficient accuracy and spatial detail over the Korean Peninsula due to complex land-cover characteristics and dense vegetation. To address these limitations, this study proposes a high-resolution soil moisture estimation framework optimized for Korean terrain. A deep learning-based speckle suppression technique was applied to Sentinel-1 SAR backscatter data to enhance signal quality while preserving fine-scale surface features. Using data from 2020 to 2025, we developed an empirical Multiple Linear Regression (MLR) model that integrates VV-polarized backscatter coefficients, short-term cumulative rainfall (1–3 days), and the number of consecutive dry days. This approach effectively captures rapid wetting and drying dynamics following precipitation events. Model performance was validated using in-situ measurements from the Rural Development Administration (RDA) in the Chungcheong region. The results demonstrated strong temporal consistency in reproducing soil moisture variations across various land-cover types, including croplands and forests. The proposed framework shows significant potential to support drought monitoring, levee stability assessment, and hydro-meteorological hazard prediction in Korea.

ABSTRACT. Climate change is intensifying hydrological extremes, increasing the frequency and severity of floods, droughts, and compound hydroclimatic hazards worldwide. Traditional hydrological prediction systems often rely on single-sensor observations, region-specific calibration, or event-specific modeling frameworks, which limit their scalability and transferability under rapidly changing environmental conditions. Recent advances in Earth observation (EO) satellite systems and artificial intelligence (AI) foundation models provide a transformative opportunity to address these challenges by enabling unified, data-driven, and physically consistent hydrological prediction frameworks.

This study introduces an integrated framework that combines multi-satellite EO data with foundation AI models to support next-generation hydrological prediction. Diverse satellite observations, including optical, synthetic aperture radar (SAR), passive microwave, and GNSS reflectometry data, are jointly used to represent key hydrological states such as surface water dynamics, soil moisture variability, vegetation stress, and precipitation-related signals. By learning generalized spatiotemporal representations of hydro-environmental processes, the developed model provides a common predictive backbone that can be adapted to different hydrological hazards.

We demonstrate how the developed framework can be applied to both drought and flood prediction. For drought conditions, the model captures soil moisture depletion, vegetation response, and land–atmosphere feedback signals that precede rapid drought intensification, supporting early warning of emerging dry conditions. For flood events, the framework leverages surface water signals, antecedent wetness conditions, and precipitation-related information to improve detection of flood-prone states and spatial flood extent evolution. This unified approach allows knowledge learned from one hydrological regime or hazard type to inform prediction in others, enhancing robustness under data-sparse or rapidly changing environments.

Beyond predictive capability, the framework enhances understanding of hydrological processes by highlighting dominant drivers across land and atmosphere domains. The integration of EO satellites and foundation AI models supports the transition from isolated hazard modeling toward unified hydro-environmental intelligence systems. This paradigm contributes to climate adaptation, disaster risk reduction, and water resources management by providing scalable, transferable, and data-rich decision-support tools. The study highlights how next-generation EO and AI technologies can redefine hydrological forecasting in the era of climate change.

ABSTRACT. X-band marine radar provides continuous, wide-area observations and is widely used for spatial wave monitoring in coastal and offshore environments. However, radar-derived wave spectra are frequently contaminated by non-wave components, which degrade the reliability of wave parameter estimation and motivate the need for improved processing for coastal monitoring and hazard-related applications.

In this study, two deep learning–based denoising approaches for improving wave parameter estimation from X-band marine radar data are examined. Two alternative approaches are systematically examined. In the first approach, denoising is applied directly in the frequency–wavenumber (spectral) domain using two-dimensional spectral slices extracted from 3D Fast Fourier Transform (3D FFT) representations of radar time-series data. In the second approach, denoising is performed in the time domain prior to spectral analysis, after which the denoised signals are transformed into the spectral domain for wave parameter estimation. Both approaches employ denoising autoencoder models and are designed to suppress non-wave energy while preserving physically meaningful wave structures. These approaches are evaluated using synthetic radar datasets generated from physically consistent sea surface simulations with known reference wave fields.

The results show that spectral-domain denoising more effectively preserves coherent wave energy structures and produces wave spectra that are closer to the reference wave field. As a result, dominant wave parameter estimates obtained from the spectral-domain approach are more stable, with errors generally below 10%, while time-domain denoising exhibits larger and more variable errors, particularly under higher noise conditions. These results demonstrate the advantage of applying deep learning techniques directly in the spectral domain and highlight the potential of the proposed framework for improving radar-based wave monitoring in coastal observation and disaster mitigation systems.

Acknowledgements: This research was supported by KIOST (PEA0331) and the Korea Institute of Marine Science & Technology Promotion (KIMST) funded by the Ministry of Oceans and Fisheries, Korea (RS-2021-KS211502, RS-2022-KS221620).

ABSTRACT. Most agricultural reservoirs in Korea were constructed before the 1970s and have since undergone progressive aging. Many small-scale reservoirs, in particular, are ungauged and lack monitoring facilities for fundamental specifications and water level measurements. These reservoirs serve dual purposes of water supply and flood control, playing a critical role in mitigating natural disasters such as droughts and floods. However, increasing climate variability and fluctuating water demands have heightened the need for quantitative assessment of available storage and improved decision-making for reservoir operations. This study proposes a methodology to estimate the stage–storage relationships of agricultural reservoirs by extracting time-series water surface areas from Sentinel-2 imagery and integrating them with topographic data, thereby evaluating the applicability of satellite-based approaches for storage estimation and operational threshold determination. Sentinel-2 Level-2A surface reflectance data were used to compute the Normalized Difference Water Index (NDWI) to extract water bodies over multiple time periods. The resulting temporal water surface areas were combined with a Digital Elevation Model (DEM) to estimate embankment elevations and shoreline boundaries and to establish cumulative area–elevation relationships. In-situ observations were further used for elevation correction and bias adjustment, enabling derivation of the storage–elevation relationship. The proposed framework demonstrates the feasibility of estimating hydrological characteristics of ungauged reservoirs using satellite imagery and spatial information and is expected to contribute to improved diagnosis of water supply capacity and enhanced reservoir operation decision-making.

ABSTRACT. This study utilized dual-polarization radar data and applied hydrometeorological classification technology to identify grid-based dry and wet snow estimation areas. Then, a snow depth and snow load were estimated using a snow type-specific snow depth algorithm. This resulted in the creation of a nationwide snow load map. Through this process, the potential of radar-based snow load maps for disaster management was confirmed. The results were formatted for storage and management, enabling future analysis and validation. The ability to monitor the spatial distribution and temporal changes of radar data remains a powerful tool. The provision of snow depth and snow load maps will enable visual identification of regional risks and facilitate management.

Acknowledgements Research for this paper was carried out under the KICT Research Program (project no. 20260156-001, Development of IWRM-Korea Technical Convergence Platform Based on Digital New Deal) funded by the Ministry of Science and ICT.

ABSTRACT. Accurate estimation of reservoir inflow is essential for effective water resources management; however, continuous inflow measurements are often unavailable for many small and medium-sized reservoirs. At KRC standard test-bed(Baekma Reservoir), inflow discharge has been monitored at a 10-minute interval using water level and velocity gauges, providing high-resolution reference data. In contrast, rainfall information commonly available for surrounding reservoirs is limited to coarser temporal resolutions, such as daily or hourly observations from meteorological stations.

This study aims to improve inflow estimation accuracy by integrating high-resolution inflow measurements with lower-resolution rainfall data and by establishing a methodology applicable to nearby ungauged reservoirs. Using daily rainfall data as the primary forcing variable, temporal alignment and scaling techniques are explored to capture the rainfall–inflow response characteristics observed in the 10-minute inflow records. Event-based analyses are conducted to identify representative lag times and response patterns between rainfall and reservoir inflow.

The proposed approach is evaluated by comparing estimated inflows against observed 10-minute inflow data at Baekma Reservoir, with a focus on accuracy improvement relative to conventional aggregation methods. The results are expected to demonstrate that reliable inflow estimates can be achieved even with coarse temporal rainfall data, provided that appropriate multi-temporal integration strategies are applied. The findings offer a practical framework for inflow estimation at reservoirs where direct flow measurements are unavailable, contributing to improved reservoir operation and regional water management.

Acknowledgments: This work was supported by Korea Environment Industry & Technology Institute(KEITI) through Climate Resilient R&D Project for Water-Related Disaster Management, funded by Korea Ministry of Climate, Energy and Environment(MCEE).(2022003610002)

ABSTRACT. In developing countries, the expansion of automatic water level gauging stations is often constrained by high installation costs and maintenance challenges. Conventional staff gauges rely on manual visual readings, which limits efficient data collection and prevents real-time monitoring. To address these limitations, this study developed a cost-effective smart staff gauge capable of real-time data acquisition and validated its performance through field implementation in Lao PDR. The proposed smart staff gauge adopts a wireless data transmission system powered by primary batteries, enabling easy installation and low-maintenance operation. Two measurement configurations were developed: a pulse-based resistive method that measures water resistance and a laser-based method utilizing a compact float. To evaluate field applicability, a total of four smart staff gauges were installed along the main stem of the Mekong River in Lao PDR. Considering both structural durability and production cost, the device housing was fabricated using acetal resin. Wireless communication networks were selected based on local connectivity conditions, and measurement accuracy and operational practicality were systematically assessed. Field validation results demonstrated that time-series water level data from both smart staff gauge types exhibited consistent and reliable hydrological patterns. The average measurement discrepancy between the two devices was within ±2 mm. Although the resistive-type gauge showed slightly higher variability compared to the laser-type gauge, comparisons with manual observations recorded by the Department of Meteorology and Hydrology (DMH), Lao PDR confirmed measurement accuracy within ±2 mm, indicating a reliable performance level. User evaluations further revealed high satisfaction in terms of ease of use and reduced maintenance costs, with overall operational efficiency surpassing that of conventional staff gauges. This study presents a practical solution for enhancing hydrological data availability and flood response capacity in developing countries. The developed smart staff gauge shows strong potential for further integration with IoT technologies and real-time monitoring systems, paving the way toward more precise and automated water level observation frameworks.

ABSTRACT. Effective climate adaptation and coastal management require accurate, high-frequency, and long-term shoreline monitoring to assess sea-level rise impacts and support decision-making on mitigation. While traditional field surveys using Real-Time Kinetics Global Positional System (RTK-GPS) and Unmanned Aerial Vehicle (UAV) photogrammetry have been effective, they remains labour-intensive and cost-prohibitive for continuous monitoring. Satellite-Derived Shoreline mapping with publicly-accessible satellite such as Sentinel-2 and Landsat provides a low-cost and scalable alternative. However, their applications to complex and urban shorelines, such as those in Singapore, are constrained by the relatively low spatial resolution of their sensors so far.

This study addresses this gap through the application of a Super-Resolution (SR) framework to Sentinel-2 multispectral imagery to enhance shoreline mapping of urban coastlines with high spatial complexity, reconstructing the red (R), green (G), blue (B) and near-infrared (NIR) bands from 10 m to 3 m resolution. To train the SR model, a paired dataset was developed using commercial satellite, PlanetScope. These 3m-resolution PlanetScope images served as the High-Resolution (HR) ground truth and were spatially degraded using bicubic interpolation to create synthetic 10 m Low-Resolution (LR) counterparts. By training on these degraded PlanetScope pairs, the model learns to recover fine-grained coastal features, which is then applied to real-world Sentinel-2 imagery to achieve sub-pixel shoreline precision. Shorelines were subsequently derived using the Normalized Difference Water Index (NDWI) combined with three automated thresholding methods-Otsu, Fixed Zero Value, and Weighted Peaks.

The evaluation follows a two-tiered approach. First, the SR model performance was quantified by comparing outputs from synthetic datasets with 3m-resolution PlanetScope imagery using metrics such as Peak-Signal-to-Noise-Ratio, Mean Squared Error, and Error Relative Global Dimensionless Synthesis (ERGAS). Subsequently, the positional accuracy of shorelines extracted from native 10m-resolution Sentinel-2 and 3m-SR-enhanced imagery are benchmarked against two sets of ground truth shorelines, 3m-resolution PlanetScope imagery and UAV-derived orthomosaics (<0.1m Ground Sample Distance) using Root Mean Squared Error and mean offset metrics.

The results obtained so far indicate that the SR-enhanced imagery yielded systemic improvement in positional accuracy, overcoming the "staircase" effect that is present in standard Sentinel-2 imagery due to pixelation along curved or diagonal coastal segments. At the same time, SR framework exhibited performance degradation at the sharp, high-contrast interfaces of engineered structures, such as piers and narrow jetties. These findings suggest that while SR techniques effectively mitigate the "mixed pixel" problem in natural and semi-engineered small-scale zones, further refinement is needed for complex shoreline infrastructure.

ABSTRACT. The objective of this study is to ensure the reliability of the "Small River Smart Measurement Management System Construction Project" initiated by the Ministry of the Interior and Safety by objectively verifying the performance of non-contact measurement devices. While automated water level and discharge monitoring systems are becoming increasingly vital due to rising flood risks in small rivers, a comprehensive performance verification framework for field-deployed devices remains insufficient.

Currently, although calibration systems exist for individual non-contact flowmeters (Image and electromagnetic), there is a lack of unified standards or type-approval systems for evaluating the performance of the entire integrated system. This absence of verification raises concerns that installing uncertified equipment could undermine the reliability of the smart management infrastructure in the future.

Performance verification was conducted using a full-scale channel at the Andong River Experiment Center of the Korea Institute of Civil Engineering and Building Technology (KICT). A total of 15 companies participated (8 image velocimetries, 6 electromagnetic flowmeters, and 1 combined system), completing 19 experimental sessions across three discharge scenarios. Reference velocity was measured using pre-calibrated ultrasonic flowmeters, while reference discharge was determined by calculating the mean of 16 repeated measurements using an Acoustic Doppler Current Profiler (ADCP) via the moving-boat method. Additionally, a real-time computational system was implemented to allow immediate re-testing if anomalies or missing data were detected in the reference or test values.

Analysis revealed a Mean Absolute Relative Error (MARE) of 20.9% for surface velocity and an average Absolute Relative Error (ARE) of 22.7% for discharge across all devices. This study is significant for implementing a type-approval performance evaluation that assesses the system's capacity to reflect actual flow variations. These findings provide critical foundational data for advancing smart measurement systems and establishing standardized performance evaluation criteria for small rivers.

ABSTRACT. This paper reports on the observed water temperature data measured at the mouth of Tanabe Bay in Wakayama Prefecture, Japan over the past decade.

Tanabe Bay is located in the southwest area of the Kii Peninsula, Japan, bordering the sea area that connects Seto Inland Sea to the Pacific Ocean.

The sea area including Tanabe bay is affected by Kuroshio (Japan Current), rapid changes in water temperature have been observed frequently.

The water temperature data presented here was measured between 2011 and 2024.

Water temperature measurements are performed at two depths (5 m and 10 m deep) at the offshore observation tower located at the mouth of Tanabe Bay.

The water temperature measurement sampling frequency is 1 Hz and continuous observation has been conducted.

Occasional data gaps occur due to equipment malfunction or other problems.

In the following data analysis, missing values are not imputed or otherwise processed and are excluded from calculations.

The followings are the main results obtained by data analysis of water temperature data.

10-minute averaging data at 1 hour intervals is used for data analysis. The mean acquisition rate is 97–98%.

Annual mean water temperature changes show a weak decline (5m deep) and nearly flat (10m deep).

Monthly average temperature variations showed essentially no change except winter season in 2018.

Seasonal variations in water temperature revealed a development of temperature gradients during summer months and small temperature differences in other seasons.

Maximum water temperatures in summer exhibit year-to-year variation. However, relatively high water temperature conditions have continued since 2022 with small fluctuations.

Sudden temperature shifts, both upward and downward, occur more frequently in summer season.

ABSTRACT. River levees serve as primary disaster prevention structures that protect human lives and property from floods and inundation, playing a crucial role in river disaster mitigation. Although the magnitude and frequency of river disasters have been increasing due to climate change, a systematic monitoring system for levees has yet to be fully established. To address these limitations, this study aims to analyze the surface cover conditions of river levees using SAR (Synthetic Aperture Radar) imagery. To classify the surface cover conditions of the levee, a section of the levees located along the Miryang River in Miryang City, Gyeongsangnam-do, South Korea, was selected as the study area. Sentinel-1 SAR imagery with VV polarization was used, and the analysis period was from March 2021 to December 2022. VV polarization, characterized by vertical transmission and reception, is sensitive to scattering responses from rough surfaces and thus was used for backscattering coefficient analysis. Analysis results showed that the backscattering coefficients observed at the levee exhibited two distinct characteristics, divided into sections with roads on the levee crown and sections with trees. In road sections, the average backscattering coefficient ranged from approximately –15 dB to –10 dB, with some locations showing values as low as –26 dB. In contrast, tree-covered sections exhibited relatively high backscattering coefficients –6 dB or more, Sections where roads and trees coexisted showed a tendency for the scattering characteristics of both surface types to be mixed. This indicates that backscattering decreases in road sections due to their relatively smooth surface properties, whereas different scattering characteristics were observed in tree sections due to branches and trunks. Therefore, systematic classification and analysis of levee surface cover conditions are required for the quantitative evaluation of SAR-based levee monitoring systems, and the results of this study are expected to provide fundamental data for the development of SAR-based levee monitoring systems.

Acknowledgements Research for this paper was carried out under the KICT Research Program (project no. 20250285-001, Development of infrastructure disaster prevention technology based on satellites SAR) funded by the Ministry of Science and ICT.

ABSTRACT. The global progression of climate change has intensified the focus on freshwater carbon cycles as a critical component of greenhouse gas dynamics. Accurate monitoring of lake water quality is essential for evaluating these cycles; however, traditional field surveys are often resource-intensive. This study aims to develop a high-accuracy prediction model for lake water quality, specifically dissolved inorganic carbon (DIC), pH, and dissolved oxygen (DO), by integrating multispectral satellite data with deep learning. Focusing on Lake Suwa, Japan, multispectral data were retrieved from Sentinel-2 via Google Earth Engine (GEE) for the period from January 2016 to March 2024. Input variables included various satellite bands and spectral indices, while ground-truth data (water temperature, pH, and DO) were provided by the Lake Suwa Environmental Research Center. Total alkalinity (TA) was determined based on field observations to calculate DIC. To account for seasonal variations, the dataset was bifurcated into summer and winter periods for separate deep learning training. The results demonstrate that the proposed deep learning model can predict DIC, pH, and DO with high precision. Sensitivity analysis revealed that Band 5 (B5) was a primary predictor for summer DIC and pH, likely due to its sensitivity to active aquatic vegetation. Conversely, the Normalized Difference Chlorophyll Index (NDCI) played a more significant role in winter and for DO predictions, reflecting the seasonal shift toward phytoplankton dominance. These findings suggest that seasonal segmentation significantly enhances the model's ability to capture complex limnological dynamics, providing a robust tool for remote lake management without the necessity of exhaustive field surveys.

ABSTRACT. This study proposes an image-based 3D technique for non-contact grain size distribution analysis of riverbed gravel using multi-view images captured in the field. Conventional sieve analysis and 2D image analysis face limitations, such as intensive labor requirements and the inability to account for particle height information. To address these issues, 3D reconstruction techniques based on COLMAP and Instant-NGP were applied to multi-view images taken with a smartphone to restore the spatial geometry of riverbed gravel. The reconstructed high-density point clouds and mesh models were converted into analytical data including depth information through scale calibration, enabling the automated detection of individual particle outlines. Based on the detected data, the projected area and weight of each particle were calculated to derive the final grain size distribution curves. By automating grain size estimation without physical sampling, this method reduces labor and costs while significantly enhancing field applicability. These findings are expected to serve as an efficient non-contact alternative for river surveys and sediment analysis, with potential for expansion to various riverbed conditions in the future.

ABSTRACT. Hyperspectral imaging has emerged as a promising non-contact technique for assessing the surface condition of hydraulic structures by providing detailed spectral information over large areas. This study investigates the applicability of hyperspectral imaging for condition assessment of dam structures with a focus on safety-relevant surface deterioration. Hyperspectral images were acquired using a ground-based hyperspectral camera to capture surface features associated with cracking, efflorescence, and intact concrete. The acquired data were analyzed to identify characteristic spectral responses of anomalous surface conditions by comparing reflectance patterns across multiple wavelength bands. Based on these spectral differences, areas exhibiting potential deterioration were detected and spatially mapped to identify zones requiring further inspection. The identified anomalous regions were examined in terms of their relevance to dam safety by considering possible deterioration mechanisms and exposure conditions. The results demonstrate that hyperspectral imaging enables effective detection and localization of surface anomalies that may be difficult to identify through conventional visual inspection, particularly over large or inaccessible dam surfaces. By providing objective and spatially continuous information, the proposed approach supports safety-oriented condition assessment and contributes to informed inspection and maintenance planning for dam safety management.

ABSTRACT. Overtopping-induced levee failure represents a major hazard in reservoir and flood management, as internal failure mechanisms often progress without visible surface deformation. The inability to detect internal instability at an early stage significantly limits the effectiveness of conventional inspection-based warning systems. This study aims to develop and validate a sensor-based monitoring system for early warning of levee failure by analyzing internal physical responses observed in hydraulic model experiments. The experimental channel was 30 m long and 2 m wide, within which a 1 m high levee model was constructed on a 0.3 m thick sandy foundation layer. To monitor internal behavior in real time, six pore water pressure sensors and six earth pressure sensors were embedded at the bottom of the levee and connected to a multi-channel data logger system. The experimental procedure consisted of a steady seepage phase followed by overtopping induced through increased inflow discharge. During the steady seepage stage, pore water pressure gradually shifted toward more negative values due to progressive infiltration, while earth pressure exhibited a slow increasing trend associated with effective stress development. Upon initiation of overtopping, pore water pressure showed rapid responses, whereas earth pressure changes remained relatively small, indicating a temporary hydraulic loading stage. Notably, prior to any visible slope failure, distinct precursor signals were identified. Pore water pressure reached extreme negative values and subsequently began to recover, while earth pressure experienced abrupt decreases followed by rapid reversals to positive values. These responses reflect internal soil erosion, loss of load-bearing capacity, and stress redistribution within the levee body. Although no external collapse was observed, the sensor data revealed significant internal degradation, confirming that internal erosion is a critical precursor to levee failure. The results demonstrate that combined monitoring of pore water pressure and earth pressure provides reliable early warning indicators prior to visible failure. This study supports the development of real-time sensor-based monitoring systems for proactive disaster mitigation and levee safety management.

Acknowledgement

This work was supported by Korea Environment Industry & Technology Institute (KEITI) through Research and Development on the Technology for Securing the Water Resources Stability in Response to Future Change Program, funded by Korea Ministry of Climate, Energy and Environment (MCEE) (RS-2024-00332877)

ABSTRACT. River channel behavior and morphological evolution are governed by the spatial configuration of water bodies, exposed sediments, and riparian vegetation. Reliable land-cover information is therefore a critical prerequisite for interpreting channel dynamics and vegetation-induced hydraulic responses. Nevertheless, land-cover datasets generated from medium-resolution satellite imagery frequently exhibit limitations in delineating river corridors, especially within transitional environments such as sandbars and riparian margins, where mixed-pixel artifacts and stair-stepped boundaries are prevalent.

This study applies a deep learning–based framework for high-resolution riverine land-cover mapping to enhance the representation of fluvial surface characteristics for hydraulic applications. PlanetScope imagery with a 3 m spatial resolution was employed to capture fine-scale channel features and adjacent floodplain areas, while Dynamic World land-cover products derived from Sentinel-2 imagery were utilized as initial reference annotations. To compensate for the inherent inconsistencies of medium-resolution labels in spatially heterogeneous river settings, a label-refinement strategy based on weak supervision was adopted to suppress noise within mixed and boundary pixels.

A U-Net–based semantic segmentation architecture was implemented to discriminate primary riverine surface classes, including water, bare substrates, and vegetation. The classification results demonstrate enhanced spatial coherence and more precise boundary delineation relative to the original reference data, particularly along channel edges, sandbars, and narrow flow paths, thereby alleviating boundary distortions commonly observed in coarse-resolution land-cover products.

The resulting high-resolution land-cover dataset offers spatially detailed information that can serve as essential input for both hydraulic and morphological analyses, contributing to improved quantitative assessment of riverine environments.

Funding: This work is financially supported by Korea Ministry of Environment(MOE) as 「Research and Development on the Technology for Securing the Water Resources Stability in Response to Future Change(RS-2024-00332494)」

ABSTRACT. Coastal erosion along the east coast of the Korean Peninsula has intensified under changing wave climate and coastal development. Shoreline evolution is governed by sediment transport and coastal morphodynamics. Effective coastal management therefore requires analysis at the scale of littoral cells rather than administrative units. Direct observation of sediment transport remains limited due to the lack of long-term in situ measurements, particularly along the inter-Korean coast. Existing littoral cell classifications along the East Sea are largely constrained by political boundaries. These boundaries fail to represent continuous sediment transport systems. This limitation is more pronounced in regions where field observations are unavailable. A data-driven approach based on remote sensing is therefore required. This study presents a methodological framework for littoral cell classification using satellite-derived shoreline dynamics and unsupervised learning. The study area covers the entire east coast of the Korean Peninsula including North Korea. Multi-temporal Sentinel-2 imagery is used as the primary data source. Coastlines are extracted from time-series satellite imagery using an automated shoreline detection approach. Shoreline variability is analyzed to derive dimensionless indices representing relative sediment transport intensity and morphodynamic response. These indices capture spatial differences in shoreline mobility associated with alongshore sediment transport processes. The indices are standardized and projected into a reduced feature space using Uniform Manifold Approximation and Projection. This method is selected to preserve both local and regional structure in spatially complex coastal data. Unsupervised clustering is then applied to group coastal segments with similar sediment transport characteristics. Cluster structures are examined using cluster validity indices including v-measure. The clustering output provides an objective basis for delineating candidate littoral cell boundaries. The extracted boundaries are interpreted with reference to coastal geomorphology such as headlands embayments and major coastal structures to assess physical consistency. The proposed framework supports data-driven identification of sediment transport systems without reliance on administrative boundaries or in situ measurements. The approach is suitable for data-limited coastlines and transboundary regions. It provides a methodological foundation for large-scale morphodynamic assessment and may inform future coastal management and climate adaptation strategies along the East Sea.

ABSTRACT. The Korean Peninsula contains shared rivers, such as the Imjin River and the Bukhan River, which are used jointly by both South and North Korea. The Imjin River basin, with a total area of ​​approximately 8,117.5 km2, is occupied by South Korea (37.1%) and North Korea (62.9%). The construction of the Hwanggang Dam and the diversion of the basin to the Yesong River have reportedly reduced the Imjin River's flow by approximately 940 million m³ annually. However, access to gauge data for the Hwanggang Dam limits quantitative analysis of its reservoir characteristics. The purpose of this study is to estimate a relationship between the reservoir water level and water storage capacity of the Hwanggang Dam, located in an ungauged basin, using global topographic data. The Hwanggang Dam is a multipurpose dam located approximately 27 km north of the Imjin River Demilitarized Zone. It is known to be built 34 m high on a terrain approximately 80 m above sea level, with an estimated water storage capacity of approximately 350 million m³. For topographic analysis, the National Geographic Information Institute of Korea's DEM (2013) and the Japan Aerospace Exploration and Land Survey (JAXA)'s AW3D30 DSM (2024) were utilized. Contours extracted at 1 m intervals based on the AW3D30 data revealed the lowest water level to be 81 m above sea level, and the highest elevation at the dam crest to be 115 m. A new DEM was constructed by combining the contour lines from the 80.5–102 m section extracted from the NGII DEM with the contour lines from the AW3D30 DSM (103–115 m). The relationship between the reservoir area and reservoir volume was derived by calculating the area for each contour line and the volume between contour lines. As a result, unlike the reservoir capacity presented in the existing literature, the effective reservoir capacity from the estimated dead water area (less than 85 m above sea level) to the highest water level (112 m above sea level) was analyzed to be approximately 268 million m³. The water level-reservoir storage relationship developed in this study is expected to be utilized as basic data for estimating the ungauged inflow of the Hwanggang Dam and analyzing changes in the flow rate of shared rivers in the future.

Acknowledgements This research was supported by a grant (Research on Establishing a Foundation for Responding to Current Issues and Challenges in Water Management) of KICT.

ABSTRACT. Ensuring drinking water quality is essential for public health and requires strict compliance with regulatory standards. In Indonesia, the permissible limit for total mercury (THg) in drinking water is 1 µg/L, as stipulated in the Regulation of the Minister of Health of the Republic of Indonesia No. 2 of 2023, consistent with international guidelines. This study was initiated after preliminary results from Laboratory X indicated potential mercury exceedances in the UGM Drinking Water System (SPAM TOYAGAMA). The objectives were to evaluate mercury testing procedures, validate analytical results, and assess water quality relative to regulatory standards. Water samples were analyzed using cold vapor atomic absorption spectrometry (CV-AAS) across three accredited laboratories employing different instruments: Laboratories Y and Z used the AULA GOLD 254 mercury analyzer; Laboratory Z additionally used the Agilent GTA 120; and Laboratory A employed the Agilent VGA 77. Analytical results were validated following SNI 6992:2-2004 and SNI 6989:78-2019. All samples contained mercury below the 1 µg/L threshold, indicating no contamination in the drinking water. Initial discrepancies were attributed to non-standard testing methods. These findings highlight the critical importance of adhering to standardized protocols at every stage of analysis to ensure reliable data and support effective drinking water quality assurance.

ABSTRACT. Remote ocean observation and tracking face challenges from rapidly changing, doubly selective signal transmission influenced by platform movement (buoys, autonomous underwater vehicles, ships), sea conditions, and extended multipath delays. In such circumstances, typical multicarrier signaling methods, like orthogonal frequency-division multiplexing (OFDM), are very susceptible to Doppler-caused inter-carrier interference and coherence loss. This reduces the quality of extracted observation data (channel behavior, Doppler signatures, reflectivity changes) and creates a difficult compromise between overhead and accuracy. Therefore, a crucial need is for a waveform and receiver able to maintain observation precision despite significant Doppler and delay spread while still providing a dependable communication link for data transmission and command. We suggest an affine frequency division multiplexing (AFDM)-centered ocean observation and communication system that specifically aims for Doppler-resilient observation by utilizing AFDM's affine time-frequency design. In essence, the transmitter organizes pilot signals and data onto an AFDM frame using precisely chosen affine parameters to manage time-frequency interaction and maintain recoverability in quickly fluctuating channels. The receiver conducts AFDM-domain processing to simultaneously assess the ocean channel/reflectivity behavior and equalize the data stream. This allows observation outputs and communications to be restored within a streamlined process, even with real-world discrepancies and limited pilot signal overhead. In typical open-sea situations with dual selectivity, the suggested AFDM method gives a smaller sensing/estimation error. This is demonstrated by a decreased NMSE in the retrieved channel/reflectivity profile and better stability in tracking Doppler effects, compared to standard multicarrier signaling using equal bandwidth and power, with the expected precision, recall, and F1 score above 95%, in each. Regarding communication, these same conditions lead to enhanced demodulation, shown by a lower EVM of nearly 1% and a BER dropping to 10-3 to 10-4 across the evaluated SNR range, proving that the sensing strength doesn't sacrifice link dependability. This approach suggests AFDM is a promising waveform choice for future ocean surveillance systems. It allows for more reliable sensing in high-mobility settings while also ensuring consistent data transmission in tough marine conditions.

ABSTRACT. Effectively integrating satellite-based soil moisture data into high-resolution, physically based hydrological models, such as the Weather Research and Forecasting Hydrological model (WRF-Hydro), is a significant challenge due to persistent spatiotemporal scale mismatches. While data assimilation (DA) can improve streamflow forecasts, balancing predictive skills and computational cost becomes critical when using large satellite datasets. Prior work typically examined either the impacts of spatial or temporal resolution on DA performance, yet the joint effects and the associated accuracy-efficiency trade-off remain unexplored. To address this gap, here we systematically evaluate how spatiotemporal resolution shapes streamflow prediction skill and computational demand by assimilating Soil Moisture Active Passive (SMAP) data using an Ensemble Adjustment Kalman Filter (EAKF) within WRF-Hydro. A comprehensive DA experiments were designed using multiple spatial resolutions (400 m, 1 km, 9 km, and 36 km) and various temporal configurations, including assimilation frequencies (three-day, weekly, biweekly, monthly) and seasonal windows (winter-, spring-, summer-, fall-only). Spatial resolution had no consistent effect, whereas temporal configuration showed a clear pattern. Winter-only assimilation substantially improved overall and low-flow accuracy, whereas summer-only assimilation degraded performance. The results show that the most efficient DA strategy is not one of high frequency, but one tailored to specific hydroclimatic conditions. The findings offer valuable guidance for advancing resource-efficient DA strategies for hydrological forecasting.

ABSTRACT. In this study, the time series of soil moisture at 10 cm and 30 cm depths were analyzed at the multiple agricultural monitoring sites to examine the vertical coupling between the surface and subsurface soil layers. The study sites represent a range of soil textures and profile characteristics, and soil moisture variations were evaluated together with the site-specific physical soil properties. The time-series comparisons during the growing season indicate that soil moisture at the 10 cm and 30 cm depths generally exhibits consistent temporal behavior, although the strength of coupling and overall performance vary among the sites. At some locations, larger uncertainties were observed in subsurface soil moisture variations, suggesting that vertical water transfer can be influenced by heterogeneity in soil profile characteristics. Nevertheless, across most sites, near-surface soil moisture was found to explain subsurface soil moisture dynamics at an average level, indicating that surface-based information can provide a reasonable representation of root-zone moisture conditions. This study provides an observation-based assessment of the applicability of surface soil moisture data for interpreting subsurface soil moisture dynamics and highlights its potential usefulness for root-zone soil moisture analysis and related hydrological and agricultural applications.

ABSTRACT. In recent years, the need for precise detection and analysis of suspended-sediment concentration layers in underwater environments has increased due to climate change and elevated sediment inflows. Conventional RGB imagery has limitations for quantitative analysis because light scattering and absorption vary with water depth, while acoustic (ultrasound-based) sensors often struggle to detect fine particles. In contrast, underwater hyperspectral imaging provides continuous spectral information across hundreds of wavelengths, enabling the capture of subtle spectral signatures associated with concentration gradients. This capability offers strong potential for quantitatively analyzing turbid layers whose boundaries are often indistinct.

This study proposes an underwater hyperspectral–based methodology for quantitative detection and classification of suspended-sediment concentration layers through an indoor water-tank experimental design. The experiment incorporates radiometric calibration using reference panels, consideration of underwater illumination conditions, and regression-based modeling. The proposed framework can be further extended to analyses of the spatial distribution of concentration layers, classification of bed material, and condition assessment of underwater structures.

This work was supported by the Climate Change Adaptation Water Disaster Management Technology Development Program (R&D), funded by the Ministry of Climate, Energy and Environment and administered by the Korea Environmental Industry & Technology Institute (KEITI) (RS-2023-00218973).

ABSTRACT. In artificial reservoirs, meander bends and tributary junctions often generate complex hydraulic structures. In particular, interactions between flow conditions and water-intake facilities can strongly influence the stability and safety of water supply systems. This study investigates the three-dimensional flow field of Lake Paldang, a critical drinking-water source serving approximately 25 million people in the Seoul metropolitan area of Korea. The objective is to clarify how seasonal variations in inflow structures during flood and normal-flow periods modify bend hydraulics, tributary mixing processes, and their implications for intake management and hydraulic resilience. Detailed field surveys were conducted along major bend cross-sections using an Acoustic Doppler Current Profiler (ADCP) and multi-parameter water-quality sondes (YSI-EXO2). Spatially distributed velocity and water-quality profiles were synthesized from cross-sectional velocity distributions, electrical conductivity fields, and the positions of maximum-velocity lines (MVL). These datasets were used to analyze secondary-flow structures, mixing behavior, and transport pathways of water masses. Results show that during flood conditions, increased discharge from the main tributaries strengthens curvature-induced momentum, expands lateral separation of the MVL, and produces a dominant single secondary-circulation cell, resembling meandering river behavior. These conditions imply enhanced potential for outer-bank erosion and inner-bank deposition. Conversely, under normal-flow conditions, overall velocities weaken and multiple small-scale secondary cells develop, highlighting lacustrine (lake-like) hydraulic characteristics. Moreover, under specific regimes, pollutant-rich tributary inflows tend to migrate along the surface toward intake facilities, suggesting potential risks for water-quality management. Overall, the findings demonstrate that artificially regulated reservoirs can alternate between riverine and lacustrine states depending on flow regimes, directly affecting sedimentation, mixing efficiency, and intake stability. Continuous, high-resolution monitoring and adaptive operational strategies are therefore essential to enhance the resilience of water-supply systems to hydraulic disturbances and water-quality incidents.

This research was funded by the Korea Environment Industry & Technology Institute (KEITI) through the Smart Water-supply Service Research Program, funded by the Korea Ministry of climate, Energy, Environment (MCEE) (RS-2024-00397970)

ABSTRACT. Nearshore bathymetry is fundamental for interpreting a wide range of coastal processes, yet repeated surveys at the temporal resolution required by morphodynamically active coasts remain expensive. Video-based kinematic depth inversion offers an efficient alternative by estimating wave frequency and wavenumber from planform imagery and inverting the dispersion relation for depth. Its linear formulation, however, becomes biased in shallow water because finite-amplitude waves propagate faster than linear theory predicts, leading to systematic depth overestimation. To mitigate this bias, we introduce a bidirectionally coupled framework that integrates nonlinear kinematic depth inversion with the phase-averaged spectral wave model SWAN. The depth inversion supplies updated bathymetry and boundary conditions for SWAN, and SWAN returns a spatially varying significant wave height field used to quantify amplitude dispersion and correct the inversion. The corrected depth is obtained through a shallow water amplitude correction that is smoothly activated toward the nondispersive limit, and the coupling is iterated until the bathymetric updates converge. The framework is evaluated at Mallipo Beach, South Korea, using ten 10 minute UAV video cases that span calm to energetic wave conditions. Neglecting nonlinearity produces depth overestimation that systematically scales with the inferred degree of nonlinearity, whereas incorporating nonlinearity reduces the mean bias by as much as 90 percent in shallow water. Relative to an independent ground truth bathymetry, the nonlinear coupled inversion achieves an average error of 4.5%. The proposed method improves the accuracy of depth inversion in the shoaling and surf zones, where the seabed evolves rapidly and accurate bathymetric estimation is required.

ABSTRACT. Numerous experimental studies have been conducted over the past decades to understand the behavior of bed-load transport in rivers. In recent years, visualization-based experiments using PTV (Particle Tracking Velocimetry) have enabled detailed tracking of the motion of individual bed-load sediment particles. However, conventional PTV methods have difficulty detecting non-spherical natural sediment particles and thus fail to achieve sufficiently high accuracy in tracking bed-load sediment particle motion. Therefore, in this study, YOLO (You Only Look Once), a deep learning–based object detection technique, was applied to PTV to improve the tracking accuracy of bed-load sediment particles. PTV analysis was performed using high-speed camera images of a bed-load transporting flow in open-channel experiments. The PTV algorithm can be broadly divided into three stages: image preprocessing, detection, and tracking. In this study, the detection process was replaced with YOLO. By training YOLO on images containing low-intensity or overlapping particles that are difficult to detect with conventional PTV, we sought to improve detection accuracy. A total of 150 images extracted from the full set of experimental images were used for training, and 30 images were used for validation. The frequency distributions of particle velocity and travel distance of bed-load sediment obtained using the conventional and proposed PTV methods were compared. Compared with the conventional method, the proposed YOLO-based method demonstrated a substantial increase in the number of detected particles and higher discrimination accuracy for closely spaced or overlapping particles that are difficult to detect using the conventional method. Furthermore, the proposed method enabled continuous detection of the same particles over time, demonstrating its ability to accurately capture time-series data on particle behavior. We plan to apply the proposed method to flows involving mixed-size bed-load transport in future work.

ABSTRACT. With the rapid advancement of marine exploration technology, the volume of underwater topographic data has experienced exponential growth, making the effective transformation of massive datasets into intuitive visualization models a key factor in improving observation efficiency. However, previous visualization methods struggle to maintain high-resolution details while providing smooth interactive experiences. Particularly in Virtual Reality (VR), the extreme geometric complexity of raw meshes often leads to excessive GPU load, resulting in latency and motion sickness. To address this performance conflict, this study proposes to design an optimized rendering pipeline that utilizes mesh simplification algorithms to pre-process high-precision raw data. Furthermore, a view-dependent Level of Detail (LOD) management system and an out-of-core visualization framework are planned, aiming to resolve memory bottlenecks through dynamic data loading techniques. The expected contribution of this research is to mitigate the latency issues associated with large-scale data, serving as a reference for future development of immersive virtual observation platforms to enhance the analysis efficiency of complex underwater environments and the accuracy of engineering decision-making in hydro-environment projects.

ABSTRACT. This study analyzed the influence of wind on water surface current velocity in a bridge section of a river. To conduct an experiment on the vertical distribution of wind along a bridge section, wind sensors were installed at heights of 1.48 m, 6.70 m, and 11.30 m above the water surface upstream of the Hongcheon Bridge on the Hongcheon River, and observations were conducted for 24 hours. As a result, the vertical distribution of wind blowing in the same direction as the flow from the upstream of the river (tailwind) was similar to the logarithmic distribution, but the vertical distribution of wind blowing from the downstream of the river (headwind) showed a parabolic shape due to the influence of the bridge deck, piers, and the water surface. The effect of wind on surface current velocity was analyzed by installing a wind sensor 0.10 m above the water surface and simultaneously observing wind speed and surface current velocity using a radar anemometer. As a result, the effect of wind on surface current velocity was analyzed to be 5-8% of wind speed for headwinds in the opposite direction to the current, 5-8% for tailwinds in the same direction as the current at speeds above 3 m/s, and 1-2% for speeds below 3 m/s. The results of the wind impact experiment were applied to the Hongcheon-gun (Hongcheon Bridge) flow observation station in the Hongcheon River and the flow velocity before and after wind impact correction was compared (2025.09.13.~09.16., 09.25.~09.26.). As a result, it was analyzed that the RMSE decreased from 0.0801 and 0.0143 before correction to 0.0587 and 0.0139 after correction.

Acknowledgements This work was supported by Korea Environment Industry & Technology Institute (KEITI) through Research and development on the technology for securing the water resources stability in response to future change Program, funded by Korea Ministry of Climate, Energy and Enviroment (MCEE)(RS-2024-00336020)

ABSTRACT. This study, linked to the development of a technology for detecting embankment leaks based on satellite SAR data, aims to secure baseline data and ensure stable operation of soil moisture measurement equipment by selecting a ground Test-bed. The Test-bed(KICT River Experiment Center) was selected through a literature review and operational case study, and the stability of the measurement data was confirmed through the installation of TDR measurement equipment. Four types of soil moisture measurement devices were installed at two locations, buried at depths of 10 cm and 30 cm, with 10 minute observation intervals. Over a five-month operation period, field soil samples were collected four times, and soil moisture content was compared. Results showed satisfactory measurements, ranging from ±1.72 to ±4.35%(average ±2.87%). Therefore, ground observation technology for satellite observation calibration has been stably secured, and future operations are planned after selecting an actual measurement site(embankment).

ABSTRACT. Key issues: Healthy freshwater is the most important resources for living things. Freshwater bodies are facing decreasing water quality in fact of eutrophication and phytoplankton due to the anthropogenic factors rather than natural environmental factors. Purpose: This study was carried out in two dams, Daecheong and Yeongju Dam, in Republic of Korea to investigate the seasonal pattern of concentration of chlorophyll-a and to compare the performance of machine learning algorithms and contribution of band features to predict the concentration of chlorophyll-a in the water bodies. Methodology: Regarding the in-situ data, comprehensive hydrometeorological in-situ dataset developed by Water Environment Information System under Ministry of Environment was applied for Daecheong dam and concentration of chlorophyll-a were measured by spectral fluorometry (FluoroProbe by bbe–biological.biophysical.engineering) which includes global position system (GPS) and auto signal transmitting system driving the boat across the Yeongju Dam. Concentration of chlorophyll-a was analyzed by means of remote detection approach using in-situ data, Sentinel-2 Level 2A and PlanetScope satellite images comparing performance of five common AI machine learning algorithms (random forest, gradient boosting, extreme gradient boosting, supported vector machine and k-nearest neighbors) by cross-validation and influence of different bands by feature engineering approach. Results: In this study, summer and fall seasons are the favourable seasons increasing the concentration of chlorophyll-a. According to both cross-validation and feature engineering approach. A result showed that random forest (RF) model is the best fitted model for using Planet scope satellite images with large enough dataset but best model for Sentinel-2A with small or large dataset. Red band (650-680 nm wavelength) and Red Edge band (697-713 nm of wavelength) are the most effective common band contributing to the best spectral model to evaluate chlorophyll-a using Sentinel-2A and PlanetScope images respectively. Conclusion: This study highlighted to consider the better analysis approach depending on using available facilities rather than zero information and no effort to prevent before being worst. The information of this study may be implicated to monitor the phenomenon of eutrophication and to evaluate the water quality of inland freshwater bodies for long-term.

ABSTRACT. With the increasing frequency of extreme rainfall and flooding events, water infrastructure has been increasingly exposed to stronger and more frequent environments stresses. Such repetitive external loads accelerate the formation and propagation of cracks in structures, leading to reduced durability and potential safety issues. In this study, a multi-stage artificial intelligence vision-based virtual sensor approach is proposed to automatically identify cracks in water infrastructure using building crack image data. First, the region of interest (ROI) is defined through wall segmentation based on YOLO11, followed by crack candidate detection using the same YOLO11 framework. Next, a ResNet18-based binary classification model determines the actual presence of cracks, and finally, a YOLOv11-based segmentation model is applied to precisely identify the crack geometry. The Proposed multi-stage AI analysis framework enables efficient crack identify in water infrastructure and provides a scientific foundation for the early identification of potential structural damage.

ABSTRACT. Accurate and continuous discharge measurement is a cornerstone of rational water resources management. Recently, image-based discharge measurement technology using CCTV has been introduced as an innovative hydrological observation tool. However, conventional analysis methods, such as cross-correlation method, brightness gradient tensor method, and Fast Fourier Transform (FFT) method etc., face limitations when image quality degrades due to adverse weather conditions or maintenance issues. To overcome these challenges, this study proposes a deep learning-based algorithm that utilizes Generative Adversarial Networks (GANs) to automatically predict the gradients of spatio-temporal images. Through validation using long-term video data from actual storm events, it was demonstrated that the proposed technique significantly enhances the robustness and accuracy of spatio-temporal image gradient estimation for discharge measurement, even under unfavorable conditions.

ABSTRACT. Climate change has significantly intensified the frequency and severity of droughts in South Korea, leading to severe regional disparities. Notable examples include the historic 7.5-month drought in the Honam region (2022–2023) and the recent Gangneung drought in 2025. While the Integrated Multi-satellitE Retrievals for GPM (GPM IMERG) serves as a valuable alternative for drought monitoring in areas lacking ground observation networks, its native spatial resolution of 0.1° limits its utility for precise detection at the administrative level. This study aims to quantitatively evaluate whether the spatial downscaling of GPM IMERG improves the performance of drought detection in South Korea. For this analysis, we utilized GPM IMERG V07 Final Run monthly precipitation data alongside ground observations from Automatic Weather Stations (AWS) and Automated Synoptic Observing Systems (ASOS) operated by the Korea Meteorological Administration (KMA) for the period 2017–2024. Spatial downscaling was performed 1using an Attention U-Net model, which incorporates a Digital Elevation Model (DEM) and 1km Moderate Resolution Imaging Spectroradiometer Normalized Difference Vegetation Index (MODIS NDVI) as auxiliary variables. The Standardized Precipitation Index (SPI) was calculated for 1-, 3-, 6-, and 12-month durations to compare the original IMERG, downscaled IMERG, and ground observation data. Performance was assessed using various statistical metrics, including the Correlation Coefficient (CC), Root Mean Square Error (RMSE), Probability of Detection (POD), False Alarm Ratio (FAR), and Critical Success Index (CSI). The study specifically analyzed drought characteristics—such as onset, termination, duration, and severity—across five major drought events occurring between 2021 and 2025. The results demonstrate that downscaled satellite data possesses significant potential to enhance the accuracy of local-scale drought monitoring.

ABSTRACT. As the MODerate resolution Imaging Spectroradiometer (MODIS) reaches its end of life (expected data cessation by 2027), ensuring the continuity of long-term environmental records via its successor, the Suomi National Polar orbiting Partnership (SNPP) Visible Infrared Imaging Radiometer Suite (VIIRS), is critical. This study verifies the sensor compatibility between MODIS and VIIRS Normalized Difference Vegetation Index (NDVI) and evaluates the capability of the Dry Condition Index (DCI) for monitoring agricultural drought in South Korean paddy areas. Using Google Earth Engine, MODIS (MOD13A1) and VIIRS (VNP13A1) datasets were compiled for the period 2015–2024. Comparative analysis revealed high consistency between the two sensors, with correlation coefficients (R2) ranging from 0.965 to 0.990 across various provinces. During the critical agricultural drought season (February to May), the mean deviation between sensors was notably low at 0.008. To further enhance continuity, a pixel-based linear correction was applied, reducing the Root Mean Square Error (RMSE) by 7.8% to 15.2% during the spring months. Time-series analysis of the DCI successfully captured the progression of historical drought events, showing high agreement with the Standardized Precipitation Index (SPI). In known drought years such as 2015, 2017, 2022, and 2023, the DCI effectively identified regional dry conditions (DCI < 0). While recent rising NDVI trends due to climate-driven phenological shifts present a challenge for traditional baselines, these results provide an objective basis for adopting VIIRS as a reliable replacement for MODIS. This research supports the development of continuous, long term drought monitoring systems to inform effective water management policies.

Acknowledgement: This work was supported by Korea Environment Industry & Technology Institute(KEITI) through Aquatic Ecosystem Conservation Research Program, funded by Korea Ministry of Climate, Energy and Environment(MCEE) (RS-2025-02304832).

ABSTRACT. It is important to produce and utilize highly reproducible predicted rainfall and rainfall scenarios to predict flooding in rivers and urban areas. Analysis of the spatio-temporal characteristics of rainfall is especially necessary to improve the reproducibility of predicted rainfall and rainfall scenarios. The observation data to analyse rainfall characteristics have been usually point observation data of rain gauges, so it was possible to analyse the temporal characteristics of rainfall, but there were limitations in analysing spatial characteristics. The observation instruments that provides spatially dense observation data is required for analysing the spatial characteristics of rainfall, and the rain-radar is a representative instrument that provides rainfall data with high spatial resolution. The rain-radar produces rainfall data up to a radius of 150 km with a spatial resolution of 125 m in Korea. This study analysed the spatial characteristics of rainfall using rain-radar data that provides spatially dense rainfall data.

Acknowledgements : The research for this paper was carried out under the KICT Research Program (Development of Elemental Technologies for River Management based on The New Normal in Response to Water Issues) funded by the Ministry of Science and ICT.

ABSTRACT. Accurate quantification of soil moisture (SM) is fundamental to hydrological modeling, flood forecasting, and understanding the terrestrial water cycle. Although satellite missions provide global observations, their spatial resolution and subgrid representation remain key limitations, especially over heterogeneous landscapes. To overcome this gap, Core Validation Sites (CVS) are designed to capture fine-scale land surface variability and provide benchmark data for validating satellite-derived SM products, including those from Soil Moisture Active Passive (SMAP).

This study presents the development of the first NASA Korea CVS in Hampyeong-gun, Jeollanam-do (35.01 N, 126.55 E), South Korea. The site fills a major validation infrastructure gap in East Asia and plays a critical role in supporting the calibration and validation of both SMAP and the upcoming NASA-ISRO Synthetic Aperture Radar (NISAR) mission. Vegetation water content measurements collected at the site further enable improved interpretation of land surface microwave signals and provide deeper insight into regional SM dynamics.

The 1 km^2 site is characterized by a highly heterogeneous landscape composed of agricultural fields, forests, and grasslands. A particularly important hydrological feature of this CVS is the rice paddies, which account for approximately 70 percent of the agricultural area. These paddies undergo artificial inundation during summer, pronounced wetting-drying cycles that differ significantly from other land cover types. To capture these contrasts, sensors are installed in both inundated and non-saturating soils to directly observe differences during irrigation periods. In addition to in situ SM sensors, L-band radiometers are deployed to monitor area-scale SM dynamics, while a drone-based L-band radiometer provides full-site coverage and enables the creation of a super high-resolution SM dataset.

The site is structured into 25 grid cells at 200-m resolution to systematically monitor spatial variability. As of early 2026, the network operates 27 stations with 49 sensors, with expansion to 47 stations expected by spring 2026. The long-term plan targets 90 stations. This exceptionally high station density is designed to ensure statistical representativeness, reduce sampling uncertainty, and provide one of the most detailed SM ground networks in Asia.

The resulting dataset will serve as a critical reference for the calibration and validation of SM products from SMAP and NISAR. Also, the CVS establish a long-term hydrological observatory in a climatically and agriculturally important region, advancing understanding of land-atmosphere interactions, irrigation-driven hydrology, and microwave remote sensing performance over complex terrain. Ultimately, this effort strengthens international collaboration and directly contributes to improving the reliability of global satellite-based hydrological observations.

ABSTRACT. Meandering rivers are a common pattern in natural river systems, characterized by complex flow structures, intense mixing, and active bed deformation. Within meander bends, mass and momentum are continuously redistributed laterally, and the core of the maximum velocity shifts toward the deepest part of the cross-section. Flow redistribution in meander bends is primarily governed by four factors: (1) transverse bed slope, (2) changes in channel curvature, (3) the central secondary flow cell and, (4) streamwise variation of transverse bed topography. The relative contributions of these factors vary depending on the geometric characteristics of the meander bends. Although numerous previous studies have focused on small rivers with width-to-curvature radius ratio (B/R) less than 50, the flow redistribution in shallow rivers with B/R exceeding 50 is not yet fully understood. Thus, this study investigates the characteristics of flow redistribution in shallow rivers with sharp bends. Field measurements were conducted at six meander bends across three major rivers in South Korea—the Geum River, Nakdong River, and Yeongsan River—to analyze flow redistribution and quantify the dominant governing mechanisms. An integrated measurement system combining ADCP (acoustic Doppler current profiler and LISST (laser in-situ scattering and transmissometry) was employed to simultaneously obtain velocity, bathymetry, and suspended sediment concentration data. The investigated meander bends exhibit an average B/R of 0.64 and an average width-to-depth ratio (B/H) of 63, indicating sharply curved bends in shallow rivers. Analysis of the field data reveals systematic variations in flow velocity and suspended sediment concentration associated with the geometric characteristics of the meander bends. The relative contributions of the governing factors vary primarily with transverse bed slope and B/R. Using comprehensive multi-sensor field measurements, this study characterizes velocity fields and suspended sediment dynamics in large river meander bends. By linking flow and sediment characteristics to meander-bend morphology, the results enhance understanding of hydraulically and environmentally dynamic reaches and provide insights to support effective river management strategies.

ABSTRACT. Return flow from agricultural water use represents the portion of supplied water that is discharged back into rivers and can potentially be reused, making it a key component of water balance analysis and national water resources planning. In Korea, however, return flow ratios derived from basin surveys conducted in the 1970s have been routinely applied, despite significant changes in climate conditions, agricultural practices, and water use systems. The objective of this study is to re-estimate agricultural return flow based on field measurements and hydrologic–hydraulic analysis, and to propose a more reliable framework for estimating return flow ratios. Six agricultural reservoirs with storage capacities greater than 5 million m³ were selected as representative sample sites across the four major river basins of Korea (Han, Geum, Nakdong, and Yeongsan–Seomjin). For each site, meteorological conditions, land cover, soil characteristics, irrigated areas, and irrigation–drainage networks were investigated. Water level and velocity sensors were installed at key points along main and terminal canals to collect field measurements during 2024. Agricultural return flow was estimated using the EPA-SWMM model, incorporating rainfall, evapotranspiration, infiltration, supply, and drainage components within a water balance framework. Multiple scenarios reflecting different agricultural water use conditions were developed to assess variability in return flow and return flow ratios. The results indicate that return flow quantities and ratios vary considerably among reservoirs, depending on watershed characteristics, cropping patterns, and water supply systems. Comparisons with conventional return flow ratios reveal substantial discrepancies, highlighting limitations of fixed empirical values. This study demonstrates the importance of measurement-based analysis for reducing uncertainty in agricultural return flow estimation and provides essential baseline information to support future national water management planning and agricultural water policy development.

ABSTRACT. River water infrastructure—including levee junctions, drainage pump stations, sluice gates, and culvert gates—faces increasing stress under localized extreme rainfall, rapid water-level fluctuations, and the aging of mechanical and electrical components. In practice, safety management still depends heavily on periodic visual inspections. Under statutory frameworks, routine inspections are typically conducted one to three times per year depending on the safety grade, while detailed diagnoses are performed only once every four to six years. Such discrete inspection cycles are inherently limited as a continuous monitoring method and may not provide timely evidence for decision-making during fast-developing events. This study develops an operational monitoring system that complements statutory inspections by adopting a physical–virtual sensing approach to collect, manage, and analyze diversified data streams for early warning and timely operational decision-making. System development and pilot operation are being conducted at the Korea Institute of Civil Engineering and Building Technology (KICT) testbed in the Andong River Experiment Center. The physical layer includes field sensors such as tri-axial acceleration, gyroscope, magnetometer, displacement, tilt, and water-level measurements. The virtual layer provides context and broader situational awareness through video and remote observations, including CCTV/NVR streams and drone surveys. To handle heterogeneous facility types and data volumes, sensor streams are stored as raw time series and windowed summaries, while imagery is archived for review and, when available, saved with location information for mapping. All datasets are indexed with consistent metadata to enable cross-source alignment, traceable retrieval, and efficient incident review. Operational resilience is strengthened through emergency power support and local buffering to reduce data loss during outages or communication disruptions. Current work prioritizes stable deployment, data quality control, and characterization of normal patterns across seasons and operating conditions. Next, unsupervised anomaly detection models—including K-means clustering, Isolation Forest, and sequence autoencoders—will be evaluated to detect abnormal structural or operational behavior while controlling false alarms by incorporating environmental drivers such as rainfall and water level. The expected outcome is a practical physical–virtual monitoring capability that can be deployed across river water infrastructure, improving situational awareness and supporting timely, evidence-based decisions alongside existing inspection practices. This work was supported by the Korea Environment Industry & Technology Institute (KEITI) through a project funded by the Ministry of Climate, Energy, Environment (MCEE), Republic of Korea (Grant No. RS-2024-00331809).

ABSTRACT. Turbulent flow structures below the water surface, such as kolk–boil vortices and depth-scale longitudinal vortices known as cellular secondary currents, play an important role in sediment and material transport during river floods. However, direct observation of subsurface turbulent flow structures in real rivers remains extremely challenging. In contrast, recent advances in image-based measurement techniques have enabled high-accuracy and high-resolution observations of water-surface velocity fields during floods. In this study, a method for estimating turbulent flow structures below the water surface from water-surface data is proposed, based on a sequential data assimilation framework that combines a large-eddy simulation (LES) with a local ensemble transform Kalman filter (LETKF), aiming to visualize and estimate flood flows. In the proposed framework, the LES serves as the system model to simulate turbulent open-channel flow, while instantaneous water-surface velocity and pressure data are assimilated as observations. The estimation performance is evaluated using instantaneous flow fields and turbulence statistics obtained from twin experiments of a turbulent open-channel flow with a stable cellular secondary current. The results show that the proposed method can accurately reproduce instantaneous turbulent flow structures near the water surface. However, the estimation performance decreases with increasing distance from the water surface, particularly for turbulent structures smaller than the water-depth scale. On the other hand, the estimation performance does not strongly depend on the bottom condition, and the large-scale structure of the cellular secondary current can be successfully estimated from water-surface data. This study demonstrates the potential of sequential data assimilation as a physically consistent approach for visualizing and estimating subsurface turbulent flow structures from water-surface observations, providing a promising foundation for future applications to laboratory experiments and real river flood flows.

ABSTRACT. Accurate discharge estimation at automatic gauging stations remains difficult during unsteady floods because a single stage-discharge rating curve cannot represent hysteresis. This study estimated field-based Manning’s roughness coefficients and improved discharge computation using the inverse continuous slope-area (CSA) method at Naju Station on the Yeongsan River, Korea. Water-surface slope was calculated from 10-minute water-level observations at two separated locations, together with H-ADCP discharge data, for six flood events in 2020. The estimated roughness exhibited a clear stage-dependent pattern, characterized by decreasing nat low stage, convergence at intermediate stage, an inflection before rapid cross-sectional expansion, and re-increase at high stage. Representative stage-segmented nvalues were then applied to CSA-based discharge estimation. Compared with H-ADCP observations, the method yielded R^2values ranging from 0.92 to 0.99 for the six flood events. Overall, the stage-wise application of measured roughness improved CSA-based discharge estimation and provided a more reliable framework for flood discharge estimation under unsteady-flow hysteresis.

ABSTRACT. This study aims to enhance the prediction of soil water content (SWC) at field capacity (FC) and wilting point (WP) using advanced machine learning-based pedotransfer functions (ML-PTFs) tailored to the diverse soil conditions of South Korea. Traditional PTFs often fall short in heterogeneous landscapes due to their reliance on limited input features and simplified linear relationships. In this research, six ML models—linear regression (LR), K-nearest neighbors (KNN), support vector regression (SVR), multi-layer perceptron (MLP), adaptive boosting (AB), and random forest (RF)—were evaluated. Random Forest (RF) emerged as the most robust model, demonstrating superior predictive performance with the highest R² values (0.82 for FC and 0.72 for WP) and the lowest error metrics across training and testing datasets. RF's ability to model complex, nonlinear relationships, incorporating a comprehensive feature set including soil texture, chemical properties, and environmental factors, was critical to its success. The selected RF model was used to impute missing SWC data, significantly enhancing the completeness and accuracy of the RDA dataset. Subsequently, an available water capacity (AWC) map was generated for the entire country, highlighting higher AWC in the western plains and lower values in the mountainous eastern regions. The results underscore the importance of integrating extensive soil datasets with ensemble learning methods, offering valuable insights for agricultural management and water resource planning in South Korea.

Funding: This work was supported by Korea Environmental Industry&Technology Institute(KEITI) through R&D Program for Innovative Flood Protection Technologies against Climate Crisis Project, funded by Korea Ministry of Environment(MOE)(2022003460002).

ABSTRACT. Traditional numerical models often encounter limitations when capturing the nonlinear processes of aquifers. To address this, the Mean Action Time (MAT) offers a powerful, data-driven alternative to quantify system responses. This study presents MAT as a primary performance indicator that characterizes the duration of hydrological disturbances, maintaining its effectiveness even in systems with incomplete convergence or gradual transitions. By applying this metric to laboratory data simulating a coastal aquifer's freshwater-seawater interface, we demonstrate how MAT effectively quantifies response dynamics during both advancing and receding phases. The integration of MAT with the Variance of Action Time (VAT) allows for a dual-layered evaluation of both the speed and the stability of the aquifer’s reaction. Our findings suggest that this analytical framework significantly enhances predictive reliability compared to conventional approximations. This approach is particularly valuable for groundwater management, providing a scalable tool for assessing seawater intrusion risks and optimizing remediation strategies in complex hydrogeological settings.

Acknowledgements: Research for this paper was carried out under the KICT Research Program (Development of IWRM-Korea Technical Convergence Platform Based on Digital New Deal) funded by the Ministry of Science and ICT

ABSTRACT. Accurate prediction of ocean surface waves is essential for marine safety, coastal hazard mitigation, and offshore operations, particularly in marginal seas characterized by complex bathymetry and strong land–sea interactions. Conventional phase-averaged numerical wave models are physically robust but often require extensive parameter tuning and may exhibit reduced predictive skill under rapidly evolving or extreme conditions. Meanwhile, data-driven deep learning approaches have demonstrated strong short-term forecasting capabilities, yet their lack of physical consistency can limit robustness and generalization. This study proposes a physics-informed deep learning framework for spatio-temporal ocean wave prediction by integrating a Convolutional Long Short-Term Memory (ConvLSTM) network with a Physics-Informed Neural Network (PINN). The ConvLSTM component is designed to capture spatio-temporal dependencies in gridded wave and atmospheric fields, while physical constraints derived from the wave action balance equation and the linear dispersion relation are incorporated into the training process to enforce physical consistency. The framework is applied to the East Asian marginal seas surrounding the Korean Peninsula, including the Yellow Sea, East China Sea, and East/Japan Sea. Hourly ERA5 reanalysis data are used, consisting of significant wave height, mean wave period, mean wave direction, surface winds, and bathymetry. Model performance is evaluated for short-term forecast lead times of up to 12 hours, with a baseline ConvLSTM model serving as a reference. Results show that the standalone ConvLSTM model already achieves competitive prediction accuracy, consistent with previous spatio-temporal deep learning studies. The introduction of physics-based constraints further improves prediction stability and reduces spatially distributed errors, particularly in coastal and semi-enclosed seas where terrestrial influence and bathymetric variability are pronounced. The physics-informed framework also enables a more physically consistent representation of wave evolution, contributing to improved robustness across forecast lead times. These results demonstrate that physics-informed deep learning provides an effective and flexible approach for regional ocean wave prediction, combining the strengths of data-driven learning and physical modeling. The proposed framework offers a promising foundation for next-generation wave forecasting systems and near-real-time applications in operational oceanography.

ABSTRACT. Accurate estimation of coastal bathymetry is essential for understanding nearshore hydrodynamics and managing coastal erosion. Video-based depth inversion methods commonly infer water depth from surface wave motions using the kinematic dispersion relation. To improve robustness, these methods often incorporate mode decomposition techniques. Recent DMD-based frameworks have demonstrated that adaptive windowing and modal averaging can provide stable bathymetric estimates across complex wave fields. Nevertheless, a fundamental scale-dependent trade-off remains: large analysis windows reliably capture long-period wave modes propagating offshore, while smaller windows are better suited for resolving spatially localized nearshore features. Extending this DMD-based depth inversion approach, we propose a frequency-dependent wave-mode weighting framework that explicitly exploits the scale-dependent roles of individual wave modes. Unlike conventional approaches where mode weights derived from modal energy are applied uniformly across the entire wave field, the proposed method assigns weights to wave modes according to their frequency characteristics. Long-period modes are prioritized to reconstruct large-scale offshore bathymetry, ensuring stability in deeper water. In contrast, short-period modes with smaller analysis windows are weighed more heavily in shallower regions to enhance the resolution of complex nearshore features such as sandbars and trenches. The proposed framework will be tested using wave fields derived from high-resolution UAV (Unmanned Aerial Vehicle) imagery captured at Mallipo Beach, Korea. Mallipo Beach features a wide intertidal zone forming a macrotidal flat, where the seabed is exposed during low tide and submerged during high tide. This tidal setting enables direct comparison between low-tide surveys and wave-based depth inversion over the same terrain at high tide. We expect that frequency-dependent weighting will improve the spatial continuity of depth estimates across multiple spatial scales. By integrating scale-aware modal weighting into a DMD-based inversion workflow, the proposed approach aims to generate more coherent bathymetric estimates from a single video source and contribute to future coastal monitoring and numerical wave-morphology modeling applications.

ABSTRACT. Jeju Island, Republic of Korea, is a basaltic volcanic island with limited surface water resources and well-developed aquifers, resulting in high dependence on groundwater. When drought occurs concurrently with intensive pumping, seawater intrusion (SWI) vulnerability in coastal aquifers can expand spatially. This situation highlights the necessity of quantitative assessment grounded in long-term observations, interpretation of temporal change patterns, and continuous monitoring. This study applied the GALDIT to analyze long-term SWI vulnerability in Jeju. GALDIT is an overlay and index-based approach that calculates a vulnerability index using a weighted average of six hydrogeological factors: groundwater occurrence (G), aquifer hydraulic conductivity (A), groundwater level (L), distance from the shoreline (D), impact of existing SWI (I), and aquifer thickness (T). In this study, electrical conductivity (EC) was used as a proxy for the impact of existing seawater intrusion (I), representing salinity conditions in aquifer. Annual GALDIT index was calculated from groundwater level and electrical conductivity (EC) based on long-term time series data, and the temporal evolution of vulnerability distributions was tracked. Groundwater level (L) and the EC (I) factor, the two time-varying components in GALDIT, were also examined to interpret annual changes. The results showed that high-vulnerability scores repeatedly appeared in low-lying areas adjacent to the coastline, and the expansion and contraction of these zones tended to correlate with groundwater level declines during dry and drought periods. Groundwater level (L) receives the highest weighting in GALDIT, variations in groundwater level accounted for most of the changes in the GALDIT index, whereas EC (I), although limited in influence, provided useful information for refining vulnerability interpretation in highly vulnerable areas. These results provide scientific support for identifying priority areas for seawater intrusion management and designing long-term monitoring strategies in Jeju Island’s groundwater system.

Acknowledge: Research for this paper was carried out under the KICT Research Program (project no.20250258-001, Development of Elemental Technologies for River Management based on The New Normal in Response to Water Issues) funded by the Ministry of Science and ICT.

ABSTRACT. Tropical rivers are active sources of methane (CH4), and CH4 concentrations in river networks play a central role in regulating diffusive CH4 from inland waters. However, the spatial and temporal variability of CH4 concentrations in humid tropical regions remains poorly constrained. This study aims to investigate how CH4 concentrations in humid tropical rivers vary across space and time and identify the main environmental factors that govern these patterns. A comprehensive dataset of CH4 concentration measurements from tropical rivers spanning 2000 - 2025 was compiled and assigned to corresponding reaches within the Global Reach scale A-priori Discharge Estimates (GRADES). Static predictor attributes (elevation, slope, soil organic carbon, peatland, and wetland cover) and time-varying hydro-climatic variables (discharge, relative discharge, antecedent flow index, and air temperature) were extracted at the reach scale. Using a Random Forest (RF) model, monthly CH4 concentrations were predicted across all the river reaches. The model performance was evaluated using cross-validation metrics. The model exhibited strong predictive skill with (R² = 0.50 and RMSE = 0.94). Spatial analyses reveal pronounced heterogeneity in mean CH4 concentrations, with land use emerging as the dominant driver. Peatland extent explains around 20 % of the spatial variability in baseline CH4 concentrations, highlighting the importance of terrestrial-aquatic carbon linkages in tropical river systems. To assess temporal dynamics, concentration anomalies (∆CH4) were defined as deviations from reach-specific baselines and related hydrological variability. Relative discharge and antecedent wetness exert systematic but non-uniform effects on CH4 concentrations, reflecting the interplay between dilution processes and enhanced connectivity with carbon-rich riparian corridors. While hydrological modulation explains a modest fraction (~ 1%) of the temporal variability, its influence is consistent across river networks and dependent on local landscape context. Our findings suggest that the magnitude of CH4 concentration in humid tropical rivers is governed by a strong spatial baseline anchored by landscape characteristics, with hydrological variability acting as a secondary but important modulator of temporal dynamics. This concentration-focused analysis provides a robust basis for improving the assessment of riverine CH4 dynamics in data-sparse tropical regions and offers a foundation for future integration with emission estimates.

Acknowledgement: This work was supported by the Korea Environmental Industry and Technology Institute (KEITI) through the Wetland Ecosystem Conservation Technology Development Project, funded by the Ministry of Climate, Energy, and Environment (RS-2022-KE002030).

ABSTRACT. Methane (CH₄) emissions from wetland ecosystems are still a major source of uncertainty in climate change mitigation efforts. The FLUXNET-CH4 network increased the availability of methane flux observations, and it enabled the application of machine-learning models to capture complex, nonlinear relationships between methane emissions and environmental drivers. However, understanding how these models learn and represent physical processes remains challenging. Differences in model interpretation may substantially influence scientific conclusions, even when predictive performance appears similar. The purpose of this study is to develop and apply an interpretation-focused framework to examine how machine-learning models represent the main environmental drivers of methane fluxes in wetland ecosystems. Rather than identifying a single “best” model, the work addresses the key issue of how modelling approaches can be systematically compared in terms of their learned relationships and physical interpretability. Methane flux and environmental data from selected wetland sites within the FLUXNET-CH4 network are analyzed using a machine-learning modelling framework centered on Random Forest (RF), with additional algorithms considered as part of an ongoing comparative analysis. Model interpretation is performed using a combination of permutation feature importance (PFI), partial dependence plots (PDP), and SHAP values. These complementary techniques are employed to assess how models prioritize variables, represent nonlinear responses, and capture interactions among methane flux drivers. Preliminary results based on the RF model indicate that interpretation metrics provide valuable insights into model learning behavior and the relative influence of key environmental variables. Differences among interpretation techniques highlight that conclusions regarding driver importance and functional responses depend on the chosen diagnostic approach, emphasizing the need for multi-method interpretation rather than reliance on a single metric. By focusing on how models learn rather than solely on predictive performance, this work contributes to more robust and physically meaningful use of FLUXNET-CH4 data in climate- and water-related research.

ABSTRACT. The Tsuya River runs along the western edge of the Nobi Plain in central Japan. The plain drains water from canal networks that were once used to irrigate "waju" settlements. These settlements were surrounded by traditional ring levees that prevented flooding from the river. Today, there are still no continuous levees on the right side of the channel at the edge of the alluvial fan near the western mountain foothills. This study analyzed the recent floodplain environment of some semi-enclosed bodies of water in the historical Tsuya River system. Study locations were determined by referencing old maps and ortho-rectified aerial photographs. Some water bodies were connected to the right side of the main channel without levees, while others existed within waju settlements that remained within their ring levee systems on the left side. We collected bottom sediment cores to measure dry and wet densities, as well as various soil properties, in each layer at different depths. We determined the specific activities of radionuclides, such as lead (Pb) and cesium (Cs), by gamma-ray analysis. We applied the CRS model to the excess 210Pb activity derived from atmospheric fallout to estimate the depositional age and sedimentation rate of each layer of the sediment. We applied these investigation and analysis procedures to the river system to study the depositional environment under conditions of no erosion by running water or human modification. As previously reported by the authors, the sedimentation rates on the left side were slightly smaller than those on the right side. These rates demonstrate the influence of sediment inflow in the late 19th century and the 1920s. These results correspond to records of past floods and suggest the influence of Kiso River floods in May and September of 1881, as well as the Nagara and Ibi River flood in July of 1922. However, the impact of the 1959 floods was less pronounced despite heavily inundating a wide area, including the location on the right side, twice when the Makita River, a major tributary of the Ibi River, breached its levee. This explains why the location was far from the breach point and why flooding was so extensive due to backwater from the Ibi River. These findings suggest that we can study the temporal and spatial dynamics of floodplain environments by analyzing the near-past environments of semi-enclosed bodies of water using bottom sediment cores.

ABSTRACT. Constructed wetlands can function as both carbon sources and sinks, yet their net contribution to regional carbon budgets remains highly uncertain. Improving carbon budget estimates for constructed wetlands is critical for evaluating their long-term environmental impacts and informing adaptive wetland management under changing hydro-climatic conditions. Although land surface models are widely used to simulate terrestrial carbon cycling and support long-term projections, representing wetland carbon fluxes remains challenging due to simplified wetland process representations and strong site-specific parameter uncertainty. In this study, we combined multi-scale field observations with data-driven parameter optimization of a process-based land surface model to improve carbon flux simulations in a constructed riverine wetland. We first refined CLM5 by introducing a wetland-specific land unit characterized by hydric soil conditions and herbaceous vegetation. Observed soil moisture and water table depth were incorporated to better constrain hydrological controls on carbon dynamics. In addition, microbial carbon pools and associated transport pathways were implemented in the decomposition scheme based on the concept of microbial carbon use efficiency. To integrate observations quantitatively into the modeling framework, key model parameters were optimized using Bayesian Optimization for Anything (BOA), minimizing root mean square error between simulated fluxes and observations of net ecosystem exchange and methane emissions derived from monthly chamber measurements and an eddy covariance flux tower. Model performance was evaluated during a period of pronounced hydrological change following the construction of a new waterway in October 2024, which shifted wetland conditions toward a drier regime. This hydrological transition led to distinct changes in observed carbon fluxes, including reduced summer gross primary productivity, enhanced winter respiration, and elevated methane emissions. With improved structural representation and optimized parameters, the model successfully captured these shifts in flux variability under changing hydro-environmental conditions. Our results demonstrate a resilient, observation-informed modeling framework for long-term carbon budget assessment and for understanding wetland greenhouse gas responses to climate change.

This work was funded by Korea Environment Industry & Technology Institute (KEITI) (RS-2022-KE002030), and Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education (RS-2024-00413227) and the Ministry of Science, ICT and Future Planning (RS-2024-00456724).

ABSTRACT. In recent years, localized rainfall events in Japan have become increasingly frequent and severe, leading to serious urban flooding and highlighting the need for rapid and effective evacuation decisions during water-related disasters. Conventional flood management has relied heavily on structural measures and expert judgment; however, these approaches alone are insufficient under conditions of increasingly intense rainfall and anticipated workforce turnover. This study aims to develop and evaluate a meteorological information system that effectively supports earlier and more reliable issuance of evacuation information during flood events.

The key challenge addressed in this study is the difficulty of making timely crisis management decisions based on existing warning information, which is generally uniform across wide areas and does not adequately reflect local conditions. Such limitations hinder district-level prioritization of evacuation shelter operations and delay residents’ evacuation actions. To overcome this issue, the study focuses on developing a decision-support system that enables disaster management at the local municipal level by automatically collecting and analyzing more detailed, area-specific meteorological data.

As a case study for the proposed system, Iruma City in Saitama Prefecture was selected as a medium-sized city with diverse district characteristics. The system automatically collects observed and forecast rainfall data, including high-resolution short-term forecasts and lower-resolution medium-term forecasts up to 72 hours ahead. Based on analyses of past flooding events and associated rainfall conditions, thresholds for potential inundation occurrence were analyzed and established. Using these thresholds, the system generates district-specific outlooks for flood occurrence that are aligned with existing disaster warning levels, enabling intuitive and timely interpretation by decision-makers.

After its operational deployment, the proposed system was utilized during a heavy rainfall event in June 2023. As a result, Iruma City successfully provided evacuation information to citizens prior to the official meteorological warning announcement by the Japan Meteorological Agency. This experience demonstrated that the system effectively supports earlier and more localized crisis management decisions compared with conventional approaches. These findings indicate that district-level flood outlooks derived from historical event data and real-time meteorological information provide well-balanced information that is neither overly broad nor excessively localized, and are therefore critically important for effective disaster management.

In conclusion, the developed meteorological information system enhances the speed and spatial resolution of evacuation decision-making during flood events. It is expected that the integration of artificial intelligence with accumulated historical and real-time data will further improve the accuracy and efficiency of evacuation support systems.

ABSTRACT. Floods and debris flows pose significant hazards in the Hobart, Tasmania area, where steep catchments and urban development can amplify the impact of extreme rainfall events. This study uses three-dimensional computational fluid dynamics (CFD) simulations in Flow3D, with the Volume-of-Fluid (VOF) method and viscoplastic rheology, to investigate the effects of obstacle spacing on water floods and viscoplastic debris flows in channels derived from real Hobart topography. The objective is to understand how strategically placed obstacles influence flow velocity, thickness, deposition patterns, and energy dissipation, providing insight into local hazard mitigation strategies. For water-dominated floods, simulations reveal that closely spaced obstacles reduce downstream velocities and enhance energy dissipation, but excessive blockage can cause upstream water accumulation. Optimally arranged obstacles allow for effective flow control while minimizing the risk of backwater formation, reflecting realistic hydrodynamic behaviour in natural channels around Hobart. For viscoplastic debris flows, interactions with obstacles are more complex due to yield stress effects. Obstacles cause upstream accumulation, local stagnation, and increased deposition, with reductions in channel porosity intensifying these phenomena. Analysis of varying obstacle spacings allowed for the development of empirical relationships linking obstacle configuration to flow reduction and deposition patterns. These results highlight the nonlinear response of debris flows compared to water floods and underscore the importance of considering material rheology in hazard mitigation design. By combining CFD simulations with Hobart-specific topographic data, this study provides a detailed understanding of flow–obstacle interactions in both flood and debris flow scenarios. The findings support the design of site-specific mitigation measures, such as debris barriers and protective structures, tailored to the local environment. Ultimately, this research enhances predictive capabilities for managing flood and debris flow hazards in Hobart, offering guidance for resilient infrastructure planning and community protection.

ABSTRACT. Coastal erosion and flooding are intensifying under climate variability, increasing exposure of coastal communities and infrastructure. Nature-based approaches are increasingly incorporated into coastal protection, yet their application is constrained by functional latency, during which natural features require time to develop sufficient structural and hydraulic capacity. During this period, coastline may remain vulnerable to ongoing erosion and extreme events, particularly in high-energy environments where tolerance for early-stage failure is low. This paper treats functional latency as a primary design constrain and examines how geosynthetics can be used as enabling solutions within hybrid Nature-Based and Nature-Based Features (NNBF) to manage early-stage exposure. Building on synthesis of NNBF, applied literature and recent modelling studies, this paper proposes application geosynthetic sand container (GSC) systems as enabling infrastructure within hybrid Nature-Based and Nature-Based Features (NNBF) in Transitionary Resilience Framework for managing early-stage exposure within hybrid nature-based coastal protection schemes. The framework distinguishes between material pathways in which geosynthetics systems are either designed to relinquish function over time or to retain performance as long-life assets through inspection and maintenance. GSC systems are treated as engineered components with defined protective responsibilities during periods when nature-based features have not yet developed sufficient capacity. The analysis draws on modelling studies, economic assessments, and applied hybrid design literature to examine how protective responsibility can be redistributed over time through the deliberate integration of GSC systems. Conventional GSC systems are examined as long-life assets capable of providing immediate, design-controlled protection and continuing to function alongside mature nature-based features through inspection and maintenance. Two implemented coastal projects employing GSC systems demonstrate how early-stage risk can be reduced without constraining ecological development. A conceptual application is also examined for environments in which biobased biodegradable GSC systems would relinquish capacity before natural features are able to assume a primary protective role. Treating early-stage performance as a design requirement rather than an implementation detail provides a practical basis for integrating GSC systems within hybrid coastal protection, aligning short-term risk management with long-term ecological objectives.

ABSTRACT. This study investigates the influence of obstacle configuration, channel orientation (slope), and spacing on water floods and viscoplastic debris flows in idealised channels using three-dimensional computational fluid dynamics (CFD) simulations in Flow3D. The Volume-of-Fluid (VOF) method is employed to capture free-surface dynamics, while viscoplastic rheology is incorporated to model yield stress effects in debris flows. The objective is to quantify how variations in obstacle spacing and channel slope affect flow velocity, thickness, deposition, and energy dissipation under controlled, laboratory-like conditions. For Newtonian water flows, a series of simulations were performed to examine hydrodynamic responses to different obstacle spacings and channel slopes. Results indicate that decreasing porosity and increasing channel slope lead to higher flow velocities and energy dissipation, but excessive reduction in porosity can cause significant upstream water depth increases. Optimally spaced obstacles strike a balance between flow retardation and controlled energy dissipation, effectively reducing downstream flow velocities without creating undesirable backwater effects. These findings provide a useful baseline for understanding obstacle–flow interactions in inertial–gravitational flow regimes. For non-Newtonian debris flows, obstacle-induced effects are amplified by the material’s yield stress, resulting in pronounced upstream accumulation, local flow stagnation, and extensive deposition zones. Simulations show that decreasing porosity and increasing channel slope significantly modify flow thickness distributions and reduce downstream velocities. Systematic variations in obstacle spacing and channel slope were used to establish empirical relationships between flow reduction, deposition, and obstacle configuration. Overall, the study demonstrates that while water floods primarily respond through hydrodynamic adjustments to obstacles and slope, viscoplastic debris flows exhibit strong nonlinear behaviour due to yield stress and deposition effects. These insights are crucial for the design of hazard mitigation measures, such as debris flow barriers and protective structures, and for planning laboratory experiments. By combining high-fidelity CFD modelling with systematic parametric analysis, the study advances predictive understanding of both water and debris flow dynamics, supporting the development of resilient and efficient mitigation strategies.

ABSTRACT. It is concerned that ongoing global warming can alter the hydrological processes into the future. These changes can include more frequent occurrence of extreme hydrological events such as severe floods and droughts, posing challenges in water resources management. Sophisticated operation of existing reservoir systems can play a significant role in coping with such challenges. Consideration of advanced operational hydrological forecasts provided by meteorological authorities is one of the ways to improve operation of reservoir systems to derive an enhanced capability of reservoirs for managing floods as well as droughts.

Considering the circumstances described above, a method to optimize operation of multi-purpose reservoirs for flood management considering operational ensemble hydrological forecasts was developed in this study. Two kinds of operational ensemble forecasts of precipitation provided by Japan Meteorological Agency (JMA) were considered: namely, 39-hour forecast of precipitation with 21 ensemble members predicted by their Meso-scale Ensemble Prediction System (MEPS) and 11-day forecast of precipitation with 27 members with their Global Ensemble Prediction System (GEPS). Ensemble inflow prediction for the coming 11 days was then estimated from the both of the ensemble precipitation predictions by use of Hydro-BEAM (Hydrological River Basin Environment Assessment Model), a distributed rainfall-runoff model. Prior release (early drawdown of the reservoir storage) and flood control operation of the multi-purpose reservoirs were then optimized by use of sampling stochastic dynamic programming (SSDP) based on the ensemble inflow prediction for the coming 11 days in order to mitigate the flood discharge while ensuring storage water recovery for secured water supply after the flood event.

The proposed method was applied to a multi-reservoir system operated for flood control and water supply in the Chikugo River basin, Japan, demonstrating the effectiveness of the proposed method in management of large-scale flood events without sacrificing its water supply function.

ABSTRACT. Arid and semi-arid regions are increasingly exposed to severe flash flooding due to short-duration extreme rainfall, rapid runoff generation, and limited infiltration capacity, with risks intensifying where flood-prone wadis intersect with culturally and environmentally sensitive landscapes. This study focuses on the Najran–Himā cultural landscape in southern Saudi Arabia, a nationally protected heritage area recognized on the UNESCO World Heritage List, with the explicit aim of supporting and proposing the Najran–Himā Geopark through an integrated assessment of flood susceptibility and resilience. The proposed geopark encompasses extensive rock art, ancient water-harvesting systems, and geologically significant wadi networks that are increasingly accessible due to tourism development and infrastructure expansion. A key challenge addressed is the lack of hazard-informed planning frameworks that integrate UNESCO conservation requirements, geological controls, and visitor accessibility in arid geopark settings. The methodology combines remote sensing and GIS-based spatial analysis with a fuzzy analytical hierarchy process to evaluate and weight critical flood conditioning factors, including elevation, slope, drainage density, distance to wadi channels, rainfall intensity, soil texture, lithology, structural geology, land-use and land-cover, and accessibility indicators such as proximity to roads, trails, and visitor zones. Geological characteristics, particularly bedrock type, fault-controlled drainage alignment, and surface permeability, play a central role in controlling runoff concentration and flow acceleration within the wadi system. Accessibility analysis reveals that several high flood-susceptibility zones spatially coincide with UNESCO-designated heritage corridors and key areas of geopark visitation, increasing potential exposure during extreme events. Results indicate that high and very high flood susceptibility zones are concentrated along the main Najran wadi and structurally controlled tributaries where unfavorable geology, limited infiltration, and high accessibility converge. The study concludes that integrating flood hazard assessment with geological and accessibility-based conditioning factors provides a robust scientific basis for proposing and managing the Najran–Himā Geopark, enhancing visitor safety, heritage protection, and climate-resilient planning in arid regions.

ABSTRACT. The Northland region of Aotearoa New Zealand experienced two extreme rainfall events within one week in February 2023: Cyclone Gabrielle and the Mangawhai flash flood. Here, we use an event-based framework to characterise the structure and atmospheric controls of extreme rainfall in Northland and to contrast how differing rainfall regimes translate into differing flow responses. Long-term gauge records are analysed using inter-event rainfall definitions and Huff quartiles. Results indicate that the largest regional events are typically mid-event loaded and commonly occur under atmospheric river (AR) conditions. Cyclone Gabrielle was a prolonged, region-wide, AR-enhanced cyclone event, with exceptional multi-hour to multi-day accumulations and sustained high flows across southern Northland. In contrast, Mangawhai was driven by a compact, instability-favoured convective rainfall burst (~6 h) with very high sub-hourly intensities over a small footprint and a rapid, flash-flood-type response. These findings support a process-based framing of Northland extremes that distinguishes accumulation-driven cyclone/AR events from intensity-driven convective events, with implications for selecting duration-sensitive versus short-duration rainfall metrics in hazard assessment.

ABSTRACT. In recent years, extreme weather events associated with climate change have become increasingly frequent and intense worldwide, leading to severe flood disasters. Japan is no exception, as the country has experienced record-breaking heavy rainfall events almost every year, primarily driven by typhoons and localized torrential rains. In response to these challenges, Japan has shifted its flood control policy from a conventional approach focused mainly on levee reinforcement and dam construction to a basin-wide flood management strategy that incorporates agricultural land and agricultural water management facilities. Within this policy framework, growing attention has been directed toward the “Tambo Dam” approach, in which runoff control devices are installed at paddy field drainage outlets to utilize paddy fields as temporary flood storage, as well as toward the pre-release operation of irrigation ponds. Although previous studies have clarified the individual flood mitigation effects of these measures to some extent, quantitative evaluations of their integrated flood performance, including potential synergistic effects arising from the simultaneous implementation of multiple measures, remain limited. This study aims to elucidate the mechanisms through which combined flood mitigation effects are generated by the concurrent implementation of Tambo Dams and the pre-release operation of irrigation ponds, and to quantitatively evaluate their integrated flood reduction effects. To this end, we developed a “runoff component visualization model” that represents runoff components generated from rainfall over different land-use categories in terms of concentration, thereby enabling visualization of the contributions of individual measures to both drainage channel flows and inundated areas. This modeling framework allows for a quantitative assessment of the respective and combined roles of Tambo Dams and irrigation ponds in basin-wide flood mitigation. Application of the proposed model to actual river basins demonstrated that the simultaneous implementation of both measures promotes rainfall retention within paddy fields and irrigation ponds, resulting in a substantial reduction in peak river discharge. As a consequence, a clear synergistic effect was confirmed, characterized by lower water levels in drainage channels during peak flow periods and an increase in the number of paddy fields effectively contributing to runoff control as Tambo Dams.

ABSTRACT. Sediment flushing can effectively address reservoir storage loss due to siltation. However, intensified sediment flushing from the reservoir may raise the downstream riverbed, reducing channel discharge capacity and increasing flood risk in downstream areas. Current research primarily focuses on the impact of sediment flushing from reservoirs with established flushing facilities on downstream river morphology and ecological environments. However, for reservoirs without flushing facilities, there is a significant research gap regarding how to scientifically formulate future sediment flushing schedules and the effects of implementing such schedules on downstream channel morphology. To address these issues, this study proposes a method for evaluating the channel bed changes in downstream rivers after increasing sediment flushing from reservoirs without flushing facilities: (1) First, by investigating the historical measured sediment flushing efficiency of typical reservoirs to determine the maximum annual flushing efficiency that limits the sediment flushing schedule; (2) Next, by simulating and analyzing the channel bed changes in downstream rivers under different discharge conditions and increased sediment flushing schedules using a water-sediment model, determining suitable flushed sediment and flushing efficiency based on the maximum sediment accumulation thickness in the river downstream of the dam; (3) Finally, based on the suitable flushed sediment and flushing efficiency under different discharge conditions, a future sediment flushing schedule for the reservoir will be constructed, and the downstream riverbed erosion and deposition processes under this schedule will be simulated using a water-sediment model. Using a reservoir in China as a case study, a one-dimensional water-sediment model was constructed for a 167 km long section downstream from the reservoir. The aforementioned method was employed to assess the 50-year trend of riverbed erosion and deposition after increasing sediment flushing. The study found that increasing sediment flushing significantly slows down reservoir siltation, with the flushing efficiency increasing from 1.5% to 55.4%; however, intensified flushing leads to increased siltation in the near-dam section. Future reservoir operations will need to balance sediment flushing with flood safety in downstream rivers.

ABSTRACT. Climate change and rapid urbanization have led to a significant increase in both the frequency and intensity of extreme rainfall events, thereby accelerating the risk of urban flooding. As a result, large-scale stormwater detention and drainage facilities and deep underground tunnels are increasingly being promoted as key urban flood defense infrastructure. Within such systems, drop shafts serve as critical connecting elements that convey surface inflow into underground structures and play a vital role in regulating hydraulic behavior during extreme rainfall events. However, existing studies on drop shafts have predominantly focused on experimental investigations or empirical formulations, which limits rigorous quantitative evaluation of their hydraulic performance. Moreover, studies integrating hydraulic experiments with numerical analyses remain limited, which constrains the development of evaluation criteria applicable to design and operation. To address these limitations, this study conducts an integrated analysis combining hydraulic experiments and numerical modeling to evaluate the hydraulic performance of urban flood defense infrastructure and to enable precise safety assessment, with particular emphasis on drop shafts. First, five physical drop shaft models with different diameters and depths were constructed, and hydraulic experiments were conducted using water level and pressure measurements along with video recordings. Based on these experimental observations, the energy dissipation characteristics, pressure distributions and fluctuations, and rotational flow behavior were comprehensively analyzed. Subsequently, a three-dimensional numerical model was developed using FLOW-3D, and its reproducibility was verified through systematic comparison with the experimental results. Based on the experimental results, various scenarios were analyzed using the numerical model, and the hydraulic performance of drop shafts was evaluated accordingly. This study aims to verify the hydraulic performance of drop shafts through physical experiments and to analyze their behavior using numerical modeling, thereby assessing the safety of flood defense infrastructure. This integrated hydraulic–numerical analysis provides a basis for precise assessment of the safety of flood defense facilities and is expected to serve as fundamental reference data for future studies on the design and operational performance evaluation of flood defense infrastructure.

<Acknowledgment> This work was supported by Korea Environment Industry & Technology Institute(KEITI) through Climate Resilient R&D Project for Water-Related Disaster Management, funded by Korea Ministry of Climate, Energy and Environment(RS-2023-00218973)

ABSTRACT. During flood stages, local scouring around bridge piers was formed, and then riverbed degradation and collapse were confirmed. In some cases, bridge was destroyed. It was caused by the formation of flood flow impingement on the pier. Yasuda and Ishitsuka previously proposed protective works installing consecutively assembled boulders as an alternative to conventional bed protection blocks for countering localized scour around piers, demonstrating their effectiveness. These experimental results pertain to prolonged ellipse bridge pier shapes and do not demonstrate the effectiveness of riverbed protection for twin-cylinder piers. This study experimentally investigated scour countermeasures downstream of the protection works for two types of long-ellipsoidal pier shapes, including the twin-cylinder type. Installing a bed restoration section with small-sized pebbles immediately downstream of the assembled boulders resulted in confirmed localized scouring due to instantaneous high flow velocity on the pebbles. This was clarified because a deflected flow formed upon impact with the pier increases the time-averaged flow velocity near the bed surface, leading to an instantaneous increase in a high flow velocity. Therefore, a transition section installing crushed stones was established downstream of the assembled boulders section around the pier, with a bed restoration section due to small pebbles installed further downstream. The average stone sizes used were set at 0.092 m size in the assembled boulders section, 0.016 m size in the transition section, and 0.0053 m size in the bed restoration section. Results from over 30 hours of flow testing in this model confirmed that placing a transition zone with crushed stones downstream of the assembled boulders section suppressed the time-averaged flow velocity and turbulence (standard deviation) near the riverbed. This confirmed that seepage flow inside the crushed stones controlled the time-averaged flow velocity near the bed and suppressed the magnitude of instantaneously occurring high velocities. In the area where the deflected flow from floodwater impacting the bridge pier persists, undulating surfaces with irregularities form. By extending the transition zone to this undulating section, seepage flow was formed within the crushed stones, preventing scouring of the riverbed. Furthermore, the impact of instantaneous high flow velocities near the riverbed caused by the formation of deflected flow can be eliminated. Regarding pier shape differences, the flow passing over the prolonged ellipse bridge pier showed undular surface for a long region, and increased turbulence near the riverbed compared to the twin cylindrical type, indicating a greater tendency for scouring.

ABSTRACT. Cavitation erosion (CE) significantly compromises the structural integrity and operational lifespan of hydraulic machinery and discharge structures, particularly in river systems with extreme sediment loads, such as the Yellow River. While it is widely accepted that low-concentration sediment synergistically accelerates material degradation, the erosion mechanisms in hyper-concentrated flows remain under-explored. This study investigates the "particle shielding effect," a phenomenon where high concentrations of solid particles transition from being aggressive agents to acting as a protective barrier that mitigates cavitation damage. The research addresses how dense particle suspensions influence the energy transfer from collapsing cavitation bubbles to material surfaces. Experiments were conducted using an ultrasonic vibratory apparatus on two distinct materials: ductile 6061 aluminum alloy and brittle cement mortar. The solid volumetric concentration was varied across a wide range to identify the specific thresholds at which the shielding effect overrides the synergistic erosion mechanism. Post-test surface characterization was performed using Scanning Electron Microscopy (SEM) and an optical profilometer to analyze mass loss, erosion depth, and morphological evolution. The results demonstrate that as particle concentration increases, the proximity and density of solid particles near the boundary create a dynamic buffering zone. This zone effectively dissipates the impact energy of micro-jets and shock waves generated by bubble collapses. For the 6061 alloy and cement mortar, a significant transition to erosion mitigation—the shielding threshold—was observed at volumetric concentrations of 12 vol% and 18.5 vol%, respectively. The higher threshold for cement mortar is attributed to its inherent brittleness and the vulnerability of its Interfacial Transition Zones (ITZ). These findings reveal that the shielding effect is a concentration-dependent and material-specific process. This study provides a new theoretical framework for predicting material longevity in sediment-laden environments and suggests that traditional erosion models must be recalibrated for hyper-concentrated flow conditions.

ABSTRACT. Numerous weirs and drop structures have been installed throughout small rivers in South Korea primarily to secure agricultural water. Fixed weirs and drop structures, while serving their intended purpose, interrupt longitudinal flow continuity, generating extensive stagnant zones upstream. Such stagnation leads to sediment deposition, water quality degradation, and fragmentation of the river system. Field investigations conducted at the Geumseongcheon in Damyang revealed pronounced stagnant zones upstream of a fixed weir, where flow velocities ranged from 0.000 to 0.003 m/s within the same cross-section. These stagnant areas were associated with sediment accumulation, algal growth, and reduced dissolved oxygen, collectively degrading aquatic ecosystem conditions. In contrast, sections where flow was maintained through sluice gates or spillways exhibited comparatively improved hydraulic and water quality conditions. This study investigates the formation and mitigation mechanisms of upstream stagnant flow at fixed weirs using movable-bed hydraulic experiments. The experiments were designed to examine how stagnant flow responds to changes in the number of weir openings and flow redistribution. Five experimental cases were considered: a condition representing the current state of Geumseongcheon, a condition representing weir opening expansion through the addition of one opening, and three cases with varying flow discharge. Variations in discharge were used to examine whether the effects of opening expansion persist under different hydraulic regimes. Changes in depositional slope and the angle of repose of the sand layer were employed as quantitative indicators of stagnation mitigation and flow continuity recovery. Experimental results showed that under the single-opening condition, flow paths were locally concentrated, maintaining stagnation across the upstream reach. In contrast, the installation of an additional opening induced flow redistribution, causing the boundary of the stagnant zone to migrate progressively downstream. As discharge increased, the angle of repose decreased from approximately 37.6° to 15.1°, indicating gradual downstream sediment transport and corresponding stagnation mitigation. These findings demonstrate that upstream stagnation is not solely a consequence of insufficient discharge but is primarily governed by flow concentration and dispersion controlled by the number of weir openings. The movable-bed experiments suggest that expanding weir opening is a key structural measure for mitigating upstream stagnation and restoring hydraulic continuity.

ABSTRACT. With the increasing frequency and intensity of extreme flood events due to climate change, concerns regarding the hydraulic safety of river bridges have grown substantially. Under extreme hydrologic conditions, flow behavior deviates significantly from that observed during normal flow regimes, and such abnormal hydraulic characteristics can critically affect the structural stability and functional performance of bridge systems. Under extreme hydrologic conditions, the upstream water level may approach or exceed the bridge deck, resulting in pressurized or overtopping flow conditions. Such conditions significantly alter the hydraulic characteristics around the bridge, leading to increased flow velocity, flow contraction, and intensified turbulence, thereby accelerating local scour and substantially increasing the hydraulic loads acting on the bridge. Consequently, the structural integrity of the bridge may be severely compromised. In particular, pressurized and overtopping flow exhibit fundamentally different hydraulic behaviors compared to free-surface flow, highlighting the necessity for detailed and systematic investigation. In addition, extreme floods are often accompanied by the transport of various types of debris, including driftwood, vegetation, and anthropogenic materials, which tend to accumulate at bridge openings. Recent heavy rainfall events have demonstrated that large volumes of debris transported from upstream can accumulate at bridges, reducing the effective flow area and inducing backwater effects that lead to elevated water levels and structural damage. Such debris accumulation partially obstructs the flow and alters flow patterns, thereby reducing conveyance capacity and amplifying hydraulic loads acting on bridge structures. Although numerous studies have investigated debris-related hydraulic phenomena, experimental investigations focusing on debris behavior under extreme flood conditions remain limited. In particular, comprehensive studies addressing the combined effects of debris accumulation, overtopping, and pressurized flow on bridge hydraulics are still scarce. Therefore, this study aims to experimentally investigate the flow characteristics induced by debris accumulation at bridges under extreme flood conditions. Through hydraulic model experiments, this study examines changes in flow structure and hydraulic behavior associated with debris accumulation, and provides fundamental insights for bridge safety assessment, development of debris mitigation strategies, improvement of design criteria, and evaluation of scour processes under extreme flood scenarios.

ABSTRACT. When spur dikes are arranged continuously, a mixing layer is formed at the interface between the main flow region and the spur-dike field, through which mass exchange occurs. Although momentum transport at the boundary between the main flow region and the spur-dike field has been investigated in previous studies, the effects of the orientation of non-submerged spur-dike arrays on turbulence anisotropy on the upstream side of the spur-dike field have not yet been sufficiently clarified. The objective of this study is to investigate in detail the turbulence anisotropy at the boundary of the spur-dike field between the first and second spur dikes upstream. Five non-submerged spur dikes were installed along one side of the flume with a spacing of twice the spur-dike length. Five installation cases were examined: perpendicular to the main flow direction (θ = 0°), upstream-oriented (θ = −10°, −30°), and downstream-oriented (θ = 10°, 30°). For each case, a steady discharge of Q = 2 L/s was supplied, and the downstream weir was adjusted under the no-spur-dike condition to form a quasi-uniform flow with a water depth of 5cm. The cross-sectional mean velocity was10 cm/s, yielding Reynolds number 5000 based on the water depth, and Froude number of approximately 0.14. Flow measurements were conducted using stereo-PIV to obtain three velocity components in a longitudinal vertical plane. A laser light sheet was projected at approximately 10 cm from the right bank, and the longitudinal section between the first and second upstream spur dikes was captured. Tracer particles with a mean diameter of approximately 50 μm and a specific gravity of 1.01 were illuminated and recorded by two cameras. The visualization images were stored as monochrome images with a resolution of 2592 × 2048 pixels. The imaging conditions were a frame rate of 100 fps and a shutter speed of 1/1000 s. PIV analysis was performed using an interrogation window of 24 × 24 pixels with 50% overlap, and approximately 6000 sheets velocity vector fields were obtained. The results revealed that the installation of spur dikes weakens turbulence anisotropy in the flow field. Furthermore, the degree of anisotropy varies with the installation angle: the perpendicular and downstream-oriented cases exhibited similar trends, whereas cases with larger installation angles showed a more pronounced tendency toward isotropization compared with those with smaller angles.

ABSTRACT. With the intensification of climate change and the impact of human activities, extreme hydrological events frequently occur in the Poyang Lake Basin. Dynamic monitoring and prediction of soil moisture is of great significance to agricultural drought resistance and water resources management. Traditional ground observations have the problem of insufficient spatial representation, and the applicability of remote sensing soil moisture products in humid areas still needs to be improved. This study takes the Poyang Lake Basin as the research object. Based on the measured soil moisture data and SMAP satellite soil moisture data, the spatiotemporal distribution characteristics of soil moisture in the basin are analyzed, the proportional deviation of SMAP soil moisture data is corrected, and a soil moisture prediction model integrating time series decomposition and causal deep learning is constructed to improve the prediction ability of soil moisture. The research results can provide a scientific basis for agricultural drought monitoring, water resources optimization allocation, and disaster warning in the Poyang Lake Basin.

ABSTRACT. Peatlands represent critical carbon reservoirs, and their stability depends on groundwater dynamics, vegetation structure, and peat decomposition processes. Climate change alters precipitation patterns, evapotranspiration, and temperature, leading to water-table fluctuations that influence peat oxidation, plant functional responses, and carbon release. This study aims to develop an integrated modelling framework to quantify the influence of climate change on peatland water-table dynamics using a coupled hydro-mechanical approach, while incorporating plant functional types and the remaining mass of peat as key controls of carbon emission.

The study addresses the challenge of representing interactions among hydrology, peat mechanical behavior, vegetation dynamics, and decomposition within a single predictive framework. Existing models often treat peatlands as static porous systems, which overlook vegetation-mediated feedbacks and peat mass loss over time. These simplifications limit the ability to evaluate long-term peat stability, ecosystem resilience, and emission trajectories under climate change condition. The proposed methodology integrates climate forcing scenarios with a coupled hydro-mechanical model describing groundwater flow, effective stress, peat compressibility, and deformation. Water-table fluctuations are simulated under varying climatic conditions and linked to vegetation processes through plant functional types representing dominant peatland communities and their ecohydrological responses. Carbon emissions are estimated using process-based formulations that relate groundwater depth, temperature, and vegetation driven litter inputs to peat decomposition rates. The remaining peat mass is dynamically updated through time, enabling assessment of cumulative degradation and feedback to hydrological and mechanical properties.

Results show that intensified water table variability under climate change modifies vegetation composition and reduces the remaining peat mass, particularly in drying condition. Shifts in plant functional types influence evapotranspiration and organic matter inputs, further amplifying water table fluctuations and carbon emissions. The coupled framework captures these feedbacks and demonstrates improved prediction of peatland carbon dynamics compared with conventional hydrological models. In conclusion, integrating hydro-mechanical processes, plant functional types, water-table variability, and peat mass balance provides a more realistic representation of peatland response to climate change. The model offers a decision support basis for restoration planning, climate adaptation strategies, and mitigation policies aimed at maintaining peatland stability and reducing long-term carbon emissions.

ABSTRACT. Land use change and climate variability are primary drivers of hydrological processes, particularly in semi-arid regions that are highly sensitive to environmental disturbances. Kupang City, Indonesia, has experienced rapid urban expansion over recent decades, which may have altered its hydrological regime. This study examines the impacts of land use change on hydrological conditions in Kupang City by integrating land cover analysis with rainfall trend assessment. Land use changes from 1990 to 2025 were analyzed using unsupervised classification of multitemporal Landsat imagery in ArcGIS. The results show a substantial increase in built-up areas from 32.82 km² to 40.55 km², corresponding to an average growth rate of approximately 3% every five years, accompanied by a decline in vegetation cover and open land. Rainfall trends were assessed using the MAKESENS method, which combines the Mann–Kendall test and Sen’s slope estimator, based on data from two rainfall stations (Eltari and Lasiana). Analysis of 22 years of data from the Eltari station indicates increasing rainfall trends in July, September, and November, with a statistically significant trend observed in November (Z = 1.80, α = 0.10). In contrast, the Lasiana station, using 38 years of data, exhibits substantial increasing trends in May, August, September, and October, with the highest level of significance recorded in October (Z = 2.96, α = 0.01). These increasing rainfall trends are associated with long-term temperature increases, which enhance evaporation processes and atmospheric moisture availability. Overall, the findings indicate that intensive land use change, combined with evolving rainfall dynamics, has contributed to significant alterations in hydrological conditions within the semi-arid environment of Kupang City.

ABSTRACT. Soil moisture is a critical component of the land atmosphere system and plays a vital role in agricultural productivity, drought monitoring, and water resource management. In this study, an Artificial Neural Network (ANN) based modeling framework is developed to estimate soil moisture over India using causative parameters like skin temperature, surface air temperature, precipitation and soil moisture extracted from six GCM from the Coupled Model Intercomparison Project Phase 6 (CMIP6) for five climatically diverse Indian regions like Baramati, Rajkot, Vijayapura, Mawsynram, and Jaisalmer. The analysis covers the historical period 1940–2014, with ERA5-Land reanalysis soil moisture used as the reference dataset. Future soil moisture projections for 2015–2100 are analyzed under different Shared Socioeconomic Pathway (SSP) scenarios. Satellite based soil moisture products from GLEAM (2010–2014) are further employed to validate both ANN outputs. Model performance is evaluated against reanalysis data using Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and the Pearson correlation coefficient (r). Results demonstrate high statistical reliability of the CMIP6 GCM particularly in humid and semi-humid regions. Mawsynram exhibits the strongest model agreement, with correlation coefficients consistently exceeding 0.84, reflecting the ability of global models to capture intense and persistent hydrological cycles. Baramati and Vijayapura also show strong correspondence, with r values ranging from 0.70 to 0.82, indicating effective representation of soil moisture dynamics under monsoon-dominated conditions. The NORWAY model performs best for Baramati, while GFDL shows peak performance for Mawsynram. For Vijayapura and Rajkot, GFDL and MPI-ESM1-2-HR emerge as representative models, yielding r values between 0.67 and 0.78 and comparatively lower RMSE. In contrast, reduced correlations in the arid region of Jaisalmer highlight persistent limitations of global climate models in resolving localized, convective desert rainfall processes, leading to higher uncertainty in soil moisture simulations. These spatial discrepancies clearly identify geographic zones where improved model parameterization is required. Future soil moisture projections for 2015–2100 under multiple SSP scenarios indicate pronounced drying trends over interior semi-arid regions, particularly Vijayapura and Baramati, within the next 10–20 years, increasing susceptibility to agricultural drought. Conversely, high-rainfall regions such as Mawsynram show increasing soil moisture, suggesting elevated flood and runoff risk under high-emission pathways. Overall, the study confirms the strong applicability of ANN assisted CMIP6 model for regional soil moisture assessment, while systematically identifying model strengths, regional limitations, and future hydro-climatic risks, thereby providing a reliable scientific basis for climate resilience and water resource planning in India.

ABSTRACT. Based on the InVEST model and ArcGIS software, this study quantitatively analyzed the spatio-temporal evolution of land use and habitat quality in the Yellow River Basin (Inner Mongolia section) from 1990 to 2020. The geographical detector method was employed to identify the driving factors influencing the spatial distribution of habitat quality and to assess the degree of correlation between various factors and habitat quality. The results indicate that: 1) From 1990 to 2020, land use in the study area was dominated by grassland, desert, and farmland, with the most frequent conversions occurring between cultivated land and grassland. 2) The habitat quality indices for the years 1990 to 2020 were 0.5856, 0.5845, 0.5777, 0.5777, and 0.5725, respectively, indicating a slight declining trend. Spatially, the overall pattern of habitat quality remained relatively stable, with high-value areas primarily located in grassland regions and low-value areas concentrated in the Kubuqi Desert and scattered farmland zones. 3) Single-factor driving force analysis using the geographical detector revealed that the influencing factors on the spatial distribution of habitat quality ranked as follows: precipitation > elevation > temperature > slope > population density > GDP > total night-time light intensity.

ABSTRACT. Reliable precipitation estimation over catchment area of a reservoir is essential for safe dam operation, optimum utilization of available water, flood control, and long-term sustainability of water resources. Though General Circulation Models (GCMs) provide valuable large scale climate information however their coarser spatial resolution and inherent biases limit their direct application for regional precipitation analysis. This study investigates the capability of four machine learning techniques- Artificial Neural Network (ANN), Support Vector regression (SVR), Random Forest (RF), and Extreme Gradient Boosting (XGboost)- to estimate daily precipitation over catchment area of Koyna reservoir using data of 15 individual GCMs as predictors. Here, daily historical observed average precipitation values for the period of 1982-2014 were used as reference dataset, while corresponding climate variables for selected 15 GCMs-temperature, specific humidity, wind speed, sea level pressure- at 4 nearest grid points around selected study area were employed as inputs to the machine learning models. Performance of each technique was evaluated using Root Mean Squared Error (RMSE), Mean Absolute Error (MAE) and Correlation coefficient (r). Among the selected techniques, XGBoost and random Forest demonstrated better performance with RMSE values ranging between 6.0-8.0 mm/day, MAE values ranging between 4.0-5.5 mm/day and correlation coefficient ranging between 0.75-0.8. ANN and SVR yielded moderate performance with RMSE values ranging between 8.0-10.0 mm/day, MAE values ranging between 6.5-8.0 mm/day and correlation coefficient ranging between 0.65-0.75. The study indicated that machine learning based techniques for precipitation estimation can be effectively used for climate change impact assessment study at the catchment scale. The findings demonstrate the effectiveness of machine learning-based approaches for precipitation estimation at the catchment scale and highlight their application in climate change impact assessments and hydrological studies for similar catchments under the jurisdiction of the Water Resources Department.

ABSTRACT. Extreme precipitation in Japan is intensifying under climate warming, challenging the adequacy of existing design rainfall estimates that rely on deterministic, envelope-curve methods. This study develops a probabilistic framework for rare extreme rainfall estimation by integrating Stochastic Storm Transposition (SST) with the 5 km d4PDF large-ensemble climate dataset for the Arakawa watershed in Japan. The objective is to quantify present and future 48-hour basin-average rainfall across very low annual exceedance probabilities (AEPs) and to assess implications for design standards under climate change. d4PDF precipitation from 12 ensembles over 60 years (720 model-years) is used to construct storm catalogs, and SST is applied independently to each ensemble to preserve internal climate variability. Synthetic rainfall fields are generated by applying uniform storm transposition to multiple historical storms and realizations per ensemble, and extreme rainfall is estimated for AEPs down to 10⁻⁵ under historical, +2 K, and +4 K climates. Results show pronounced inter-ensemble variability in rare extremes. Under the historical climate, 48-hour depths at AEP 10⁻³ range from about 320 to 470 mm across ensembles, while at 10⁻⁴–10⁻⁵, ensemble-mean depths increase to roughly 360–610 mm, with maxima approaching 700 mm. These upper-tail estimates are comparable to or exceed Japan’s current MLIT design rainfall for the Arakawa basin (658 mm for a 1000-year, 48-hour event), but correspond to substantially rarer probabilities, indicating consistency with, rather than contradiction to, existing standards within a probabilistic SST framework. Climate warming leads to strong, nonlinear amplification of rare extremes. Under the +2 K scenario, basin-average 48-hour rainfall increases reach up to approximately 40% at the lowest exceedance probabilities. Under +4 K warming, amplification rates reach up to ~30% per Kelvin in the extreme tail, substantially exceeding Clausius–Clapeyron scaling (7% K⁻¹) as well as its doubled and tripled benchmarks. The widening inter-ensemble spread and divergence in the extreme tail under warming reflect a strongly nonlinear sensitivity of PMP-scale rainfall to climate change. Overall, this study demonstrates that large-ensemble SST provides a robust and climate-consistent framework for probabilistic estimation of rare extreme precipitation, offering critical insights for interpreting design rainfall under changing climate conditions and supporting climate adaptation strategies in Japan.

ABSTRACT. The Noelmina River Basin is one of the major river systems on Timor Island, Indonesia, and is highly vulnerable to drought due to its semi-arid climate. This study assesses drought characteristics in the Noelmina River Basin using the Standardized Precipitation Evapotranspiration Index (SPEI) at a one-month time scale (SPEI-1). The analysis is based on 21 years of rainfall data (1998–2018) obtained from 14 rain gauge stations. Spatial analysis and drought mapping were conducted using ArcGIS to identify the distribution and severity of drought events across the basin. The results show that SPEI-1 values range from −2.00, indicating extremely dry conditions, to 2.00, indicating extremely wet conditions. Severe and extreme drought events predominantly occurred in June, July, and October, with SPEI-1 values below −2, and were spatially widespread across the basin. In contrast, near-normal to extremely wet conditions were observed in January and February, with SPEI-1 values ranging from −0.99 to above 2.00. The watersheds experiencing the highest drought severity include the Biboko, Nitas, Oebobo–Liliba, Kasmuti, and Noelnunkurus watersheds. These findings provide valuable insights into the temporal and spatial variability of drought in the Noelmina River Basin and can support water resources planning and drought mitigation strategies in semi-arid regions.

ABSTRACT. Sediment transport is an important factor governing river morphodynamics, and uncontrolled or excessive sediment transport can directly affect the stability of hydraulic structures, such as bridge pier scour and reservoir capacity reduction. Accurate prediction of sediment transport is a fundamental challenge in river hydraulics due to the complex and nonlinear interactions among hydraulic and sediment-related variables.

Although continuous efforts have been made to analyze sediment transport for engineering applications, its complex physical mechanisms have led to a strong reliance on empirical formulas. In recent years, machine learning techniques have been increasingly applied to sediment transport problems; however, the suitability of different machine learning models and their advantages over conventional linear approaches remain insufficiently quantified.

This study aims to identify an appropriate machine learning model for sediment transport prediction through a systematic performance comparison among artificial neural networks (ANN), support vector machines (SVM), random forest regression (RF), and linear regression (LR). Identical input datasets and training–testing conditions were applied to all models to ensure a fair comparison. Prediction performance was evaluated using multiple statistical metrics, including root mean square error, coefficient of determination, and correlation coefficient.

The comparative results highlight the limitations of the linear regression model in representing nonlinear sediment transport processes and demonstrate the applicability of machine learning–based approaches. Although ANN and SVM achieved relatively high prediction accuracy, their performance exhibited greater sensitivity and lower stability across different input configurations. In contrast, the random forest model consistently provided robust and stable predictions with high accuracy, and was therefore selected as the optimal model for sediment transport prediction in this study.

These findings provide quantitative evidence supporting the applicability of machine learning approaches for sediment transport prediction and offer practical guidance for selecting suitable models in river engineering applications. The comparative framework presented in this study contributes to improving model reliability and promoting data-driven approaches in sediment transport analysis.

ABSTRACT. The primary objective of this research is to investigate the application of Physics-Informed Neural Networks (PINNs) as computationally efficient surrogate models for solving the one-dimensional Saint-Venant Equations (SVE). This investigation is motivated by the increasing global frequency of climate-induced hydrological disasters, such as floods and droughts, which necessitates the development of robust, rapid AI-based alternatives to conventional, time-intensive hydraulic models for predicting flood wave propagation.

The study addresses the need for a faster, mesh-free modeling framework for hydro-environmental systems. However, it also confronts a key limitation of the current PINN methodology: its intrinsic tendency to smooth out sharp spatial gradients and discontinuities (shocks) inherent in hyperbolic shallow water flows. This phenomenon compromises the model's reliability in accurately capturing the wavefront's dynamics, an issue that established numerical techniques, like the Finite Difference Method (FDM), are currently superior at resolving.

The core approach involves utilizing the Physics-Informed Neural Network (PINN), a method that represents a significant shift from purely data-driven modeling. The PINN rigorously integrates the governing physical laws—specifically the continuity and momentum equations of the SVE—directly into the neural network’s loss function. This structure ensures that the model’s predictions are inherently constrained by fundamental hydrodynamic principles, providing a mesh-free, differentiable framework for global flow analysis. This is benchmarked against traditional numerical methods, which rely on discrete grids and can suffer from numerical diffusion.

The findings conclude that while PINNs offer compelling advantages in terms of computational speed and a mesh-free, differentiable architecture, they currently do not possess the shock-capturing reliability of established numerical solvers. Therefore, the most impactful role for PINNs is as a complementary technology. They excel at effectively integrating physical rigor with the predictive power and computational efficiency of modern data-driven approaches, thus pointing toward a more robust hybrid modeling future for complex hydro-environmental systems.

ABSTRACT. Advancements in measuring technologies, including Surface Image Velocimetry, surface-velocity radar, and Acoustic Doppler Velocimetry, make real-time hydrometric monitoring increasingly feasible. AI-driven modeling further enables data-driven analysis for predicting future flood hazards. However, flood-management officials continue to face challenges in interpreting and operationalizing these data to identify vulnerabilities and to plan improvements in information-based prediction techniques. Moreover, the limited use of measured data for ungauged basins, representing the majority of small stream catchments, hinders accurate vulnerability assessment. To address these challenges, this study develops a real-time, data-driven flood early warning system and co-creates a web-based platform using a user-centered design approach. The system integrates vulnerability information to support decision-making and resilience planning. For gauged sections, measured data are used to construct rainfall–discharge nomographs and rating curves, and future discharge and water-surface elevation are predicted using an AI-driven prediction model. For ungauged sections, the system applies the Manning formula and scenario-based rating curves generated using the HEC-RAS model. To ensure accurate predictions, the study employs a Generative Pre-trained Transformer (GPT) architecture to develop rainfall–discharge nomographs and rating curves by learning from measured rainfall, discharge, and water-depth data from small streams. The system subsequently fine-tunes these nomographs and rating curves using real-time measurements prior to predicting discharge and water-surface elevations. The developed system is applied to 13 small streams to assess flood vulnerabilities and evaluate its decision-support capabilities. For predicting water-surface elevation in ungauged sections, the discharge predicted from the fine-tuned nomograph at the gauged section is used, under the assumption that this discharge remains constant throughout the channel. The prediction results closely reproduce measured discharge and water-surface elevations, achieving accuracies exceeding 90%. For ungauged sections, predicted water-surface elevations are compared with levee heights, and the results indicate no potential overflow locations, consistent with observations from actual flood events. Overall, the findings demonstrate that the developed system effectively integrates real-time monitoring data into a practical flood early-warning tool that enhances vulnerability assessment, supports decision-making, reduces potential flood damages, and strengthens resilience planning for small streams in Korea.

ABSTRACT. The accurate prediction of the water level of the locks approach channels is of great engineering significance and economic value for ensuring the safety of ship navigation, improving the efficiency of lock operation, and optimizing the dispatch management of locks. When multi-lane locks operate at high frequencies, multiple periodic "discharge waves" will be superimposed and reflected in the approach, resulting in a very complex water level fluctuation process in the channel. Currently, the prediction methods for such water level changes mainly rely on numerical simulation and physical model experiments. However, numerical simulation often has difficulty accurately capturing the nonlinear characteristics of water level changes caused by lock discharge; while physical model experiments are difficult to meet the engineering requirements for rapid and real-time prediction of the downstream water level of locks. Therefore, for this high-frequency and instantaneous water level fluctuation process directly caused by the operation of multi-lane locks, there is still a lack of sufficiently fine, fast and practical prediction modeling methods. In recent years, machine learning and artificial intelligence technologies have continuously permeated into traditional water conservancy engineering research fields. Among them, the Long Short-Term Memory neural network (LSTM) has demonstrated superior performance in processing long-term dependencies in time series in surface and groundwater level predictions. Therefore, based on the water level observation data of the lower approach of Gezhouba multi-lane locks in China, A LSTM machine learning model that integrates the multi-head self-attention mechanism to predict the water level fluctuations caused by multi-lane locks proposed in this paper. The calculation results show that the maximum absolute error (MAE) of the model is 8 cm, and at the same time, its root mean square error(RMSE) is also relatively small.The prediction results were in good agreement with the measured values. The model not only effectively improves the accuracy and stability of the downstream water level prediction of the locks but also has a fast training speed, and can provide more precise and rapid decision support for ship navigation safety management and intelligent dispatching of lock operations.

ABSTRACT. Industrialization, urbanization, and various anthropogenic activities have led to significant changes in river water quality. River water quality is highly dynamic and exhibits complex patterns due to spatial characteristics and temporal variability. Conventional river monitoring has primarily relied on periodic sampling followed by laboratory-based analyses; however, such approaches have inherent limitations in capturing the spatiotemporal variability of river systems. In recent years, the increasing use of in situ sensors has enabled real-time measurement of multiple water quality parameters, allowing the acquisition of high-resolution and high-frequency data. Advanced sensor-based water quality monitoring facilitates real-time analysis and prediction through artificial intelligence techniques and plays a critical role in effective river management. Nevertheless, sensor-based monitoring systems are inevitably affected by data quality issues, such as outliers and missing values, caused by sensor malfunctions, extreme weather conditions, and signal interference. These data quality problems not only degrade the accuracy of analysis and prediction models but also reduce the reliability of decision-making and evaluation processes. Recent studies have emphasized that data quality has a substantial impact on model performance and that data errors are a major source of uncertainty in environmental decision-making. In this study, various data preprocessing techniques, including outlier detection and missing value imputation, were applied to sensor-based river monitoring data. Comparative analyses were conducted between preprocessing approaches based on single variables and those utilizing multiple variables. Furthermore, the performance improvements of analysis and prediction models were evaluated using the preprocessed datasets. The results demonstrate that data quality plays a crucial role in enhancing the reliability of river monitoring and prediction models. As sensor-based monitoring systems continue to expand, the advancement of data preprocessing techniques is expected to contribute significantly to the development of intelligent river management systems.

This work was supported by the Ministry of Climate, Energy, Environment through the project ‘Research and Development on Technology for Securing Water Resources Stability in Response to Future Changes (grant number RS-2024-00332114)’.

ABSTRACT. In river management, flow and water quality are essential components for maintaining ecosystem health, ensuring the sustainability of water resources, and responding to climate change–related impacts such as floods and rainfall events. As urbanization and industrialization accelerate, increasing pollutant inflows within river basins have raised persistent concerns regarding water-quality degradation and alterations to aquatic ecosystems. In response, Korea has implemented the Total Maximum Daily Load (TMDL) management system to control pollutant discharge loads and achieve target water-quality objectives at the watershed scale. Effective management of pollutant loads requires continuous monitoring of both flow and water-quality conditions. Currently, the TMDL monitoring network operates based on manual grab sampling conducted at approximately eight-day intervals, which offers high reliability due to direct field measurements. However, the low sampling frequency, along with limitations caused by adverse weather conditions and safety concerns for field personnel, restricts its ability to capture short-term and continuous variations in river conditions. Recent advances in sensor-based automatic monitoring technologies enable the acquisition of continuous, high-frequency time-series data under diverse climatic conditions and spatial constraints. In this study, the Nakdong River was selected as the study area, and pollutant load estimation characteristics were analyzed using data from both automatic monitoring networks and the conventional TMDL monitoring network. High-frequency flow and water-quality data collected from the automatic monitoring network were used to develop optimal estimation models based on various machine-learning techniques, and the resulting pollutant loads were compared with those estimated from the TMDL monitoring data. The findings of this study are expected to provide a scientific basis for improving the efficiency of river water-quality management through near–real-time pollutant load estimation and for supporting timely decision-making processes.

This work was supported by the Ministry of Climate, Energy, Environment through the project ‘Research and Development on Technology for Securing Water Resources Stability in Response to Future Changes (grant number RS-2024-00332114)’.

ABSTRACT. Flooding causes significant loss of life and economic damage and affects healthy development of society. It is crucial to forecast the spatiotemporal evolution of floods quickly and accurately. Deep learning (DL) models demonstrate significant advantages in improving computational efficiency while maintaining accuracy. Existing research of predicting dynamic flood evolution still remains some gaps for predicting flooding process from the initial time step, weak transferability for flood scenarios from unseen inflows processes and breach locations, and potential enhancement of common neural network frameworks. This paper proposes a DL model called FloodUnet based on a U-Net architecture, improved by integrating residual module and channel attention mechanism, to achieve rapid and accurate prediction of flood evolution. Input data includes topographic elevation, water depth at the current time step, and source term at the current time step. The target is the water depth map at next time step. It follows a rolling prediction mode, which means the predicted water depth is used as the input of next time step. FloodUnet can predict a series of flooding depth maps from the initial frame and maintain high-precision prediction on unseen flood cases. It achieves an average Root Mean Square Error (RMSE) of 0.2 m and an average Nash-Sutcliffe Efficiency coefficient (NSE) of 0.9 on testing sets of unseen breaches and inflows through a 4-fold cross validation. It reaches significantly high computational efficiency which is three orders of magnitude faster than the hydrodynamic model with a 24-hour lead time, while remining a relatively low training cost. It has obvious advantage in prediction accuracy compared to an ordinary Convolutional Neural Network (CNN) and a U-Net model as benchmarks. Through an ablation study, it demonstrates that residual module and channel attention mechanism can enhance feature representation for complex flood dynamics and ensures stability during multi-step rolling prediction.

ABSTRACT. Wave propagation is a fundamental process in open-channel hydraulics and plays a crucial role in describing unsteady flow behaviors under rapidly varying conditions. Although conventional numerical solvers have achieved considerable success in hydrodynamic simulations, they still suffer from notable limitations when dealing with wetting–drying processes and complex irregular topography. Physics-informed neural networks (PINNs) have shown potential for modeling open-channel flows; however, their performance deteriorates in dam-break problems due to strong nonlinearities and flow discontinuities. This study proposes a Gradient-Weighted and Residual-based Adaptive Refinement PINN (GW-RAR-PINN), which combines gradient-weighted loss modulation with residual-based adaptive refinement to enhance shock resolution in dam-break flows. One-dimensional dry-bed and wet-bed dam-break problems were simulated to assess the performance of GW-RAR-PINN against the baseline PINN and other representative variants. Ritter’s and Stoke’s analytical solutions are employed as reference cases to identify an optimal PINN framework for dam-break flow simulation. GW-RAR-PINN demonstrated the highest accuracy, reducing the root-mean-square error (RMSE) by 32.6% and 39.9% for water depth and discharge in dry-bed cases, and by 22.0% and 18.9% for water depth and velocity in wet-bed cases, respectively. The Gradient-Weighted (GW) module reduces the contribution of regions with steep gradients in the loss function, alleviating over-enforcement of governing equations near discontinuities, while the Residual-based Adaptive Refinement (RAR) module adaptively enriches collocation points in high-gradient regions to better resolve shock structures. The combined strategy stabilizes network training and improves the representation of localized hydraulic features. These findings demonstrate that GW-RAR-PINN provides a robust and accurate framework for simulating strongly nonlinear dam-break flows.

ABSTRACT. In recent years, basin-wide flood management has been promoted throughout Japan as a countermeasure against heavy rainfall disasters caused by extreme weather events; however, the impacts of flow regime alterations associated with basin-wide flood management on riverine environments remain insufficiently understood. Therefore, this study applied multiple machine-learning–based species distribution models (e.g., LightGBM), originally developed for the Omaru River system in Miyazaki Prefecture, to the Oyodo River system, and evaluated the effects of flow regime alterations induced by basin-wide flood management on riverine biological communities using benthic macroinvertebrates as indicators. A comparison of model transferability among machine-learning algorithms revealed that LightGBM exhibited the highest predictive performance, with particularly improved accuracy for gliding and net-spinning functional groups. Under basin-wide flood management scenarios, local-scale diversity increased at certain sites, whereas diversity at the basin scale declined. In addition, the reduction in differences in community structure between upstream and downstream reaches suggests that basin-wide flood management scenarios may shift down-stream communities toward structures more similar to those found in upstream reaches.

ABSTRACT. This paper focuses on wind effects on plastic debris in rivers. The plastic debris entering ocean from rivers has been getting more serious in recent years. It is important to investigate how plastic debris is transported to ocean through rivers in order to take measures to reduce them. The plastic debris with low specific gravity is thought to drift near the free surface and its motions can be greatly affected by wind. However, the wind effects on the drifting plastic debris have not been sufficiently studied. Therefore, we conducted flume experiments to investigate the wind effects on drifting behaviors of plastic debris. We measured the drifting velocity of plastic debris mode by the high-speed camera in wind-induced open channel flows by changing the plastic model shape, thickness, specific gravity, water flow and wind speed. We prepared three shapes, i.e. cylinder, square pole, and triangle pole. The static water depth was 10 cm, mean water flow velocity Um was 30 cm/s and 0 cm/s, wind speed was varied systematically from 0 to 6.49 m/s, and the water flow and wind flow were in the same direction. Results showed that the drifting velocity increased with wind speed, and it increased by most 45 % compared with no-wind conditions. It is also suggested that although the model shape effect on the amount of drifting velocity increase was little, it increased with the lower model specific gravity or larger thickness because the project area of wind drag force increased. In addition, the drifting velocity got greatly larger than free-surface velocity in conditions where only wind blows (Um = 0). It became by most 320% of the water surface velocity. Based on these findings, it is thought that the wind could play an important role in mass transportation of plastic debris in rivers, and the wind effects depend significantly on debris specific gravity and thickness.

ABSTRACT. This study investigates the effects of an upstream submerged vegetation patch on scour–deposition patterns and three-dimensional flow structures around a spur dike under clear-water, mobile-bed conditions. Two long-duration flume experiments were conducted: (i) a baseline case without vegetation and (ii) a vegetated case with an upstream patch simulated using an array of rigid cylinders. Experiments were run until morphodynamic equilibrium was achieved, requiring approximately 24 days for the non-vegetated case and 28 days for the vegetated case. Bed topography was surveyed with an ultrasonic sensor, and velocity fields were measured using acoustic Doppler velocimetry. In the near-dike region (−1L ≤ x ≤ 1L), the vegetation patch markedly reduced dimensionless scour volume compared to the baseline. Velocity measurements revealed weakening of the horseshoe-vortex system at the dike tip and downstream displacement of maximum local scour. Conversely, over the full test section, the vegetated case showed greater maximum scour depth and total scour volume. These results reveal a spatial decoupling of scour processes: upstream submerged vegetation mitigates deep local scour near the spur dike—enhancing potential structural stability—while increasing net erosion at the reach scale, especially in downstream areas. These findings offer experimental guidance for optimizing integrated vegetation and spur-dike designs in sustainable river engineering and management.

ABSTRACT. Density currents driven by spatial variations in temperature and salinity commonly occur in lakes and coastal regions, and they play a crucial role in controlling tratification dynamics and material transport. To properly understand such flows, numerical models that account for threedimensional flow structures and non- hydrostatic effects are required. The lattice Boltzmann method (LBM) is a numerical approach that analyzes fluid motion by solving a discretized Boltzmann equation rather than directly solving the Navier–Stokes equations. Owing to its explicit treatment of non-hydrostatic effects and its algorithmic suitability for parallel computation, LBM is expected to be an effective tool for simulating density-driven flows in aquatic environments. A key feature of conventional LBM is its use of uniform square or cubic lattices, in which the advection process is designed to be free from numerical errors by applying a first-order upwind scheme aligned with the lattice structure. However, natural water bodies such as lakes are characterized by horizontally extensive and vertically shallow geometries, and the spatial resolution required in the horizontal and vertical directions is generally non-uniform. As a result, the use of uniform lattices leads to inefficient computations and limits the applicability of LBM to realistic environmental flows. In this study, we develop a density current simulation model based on LBM that incorporates nonuniform rectangular lattices suitable for environmental flow analyses. The proposed method evaluates inflow and outflow fluxes at each lattice using three-dimensional interpolation and appropriately accounts for lattice non-uniformity during the streaming process. To assess the accuracy and computational efficiency of the proposed model, simulations of internal waves in a rectangular domain and internal wave breaking over a slope are conducted. In the internal wave simulation within a rectangular domain, grid refinement is applied near the density interface, enabling a significant reduction in the total number of lattice while successfully reproducing internal wave splitting observed in previous laboratory experiments. In the simulation of internal wave breaking over a slope, local grid refinement near the sloping boundary allows the proposed model to reproduce experimentally observed breaking patterns while maintaining computational efficiency. These results demonstrate that the proposed non-uniform lattice LBM is an effective and efficient tool for simulating density-driven flows in lakes and coastal environments.

ABSTRACT. Aquatic vegetation is a key factor to characterize sediment transport processes and environment in rivers. Specifically, the sediment transport and deposition mechanisms around a vegetation patch are closely connected with three-dimensional mean and turbulent flow structures. Although plants and woods in natural rivers have various physical properties (e.g., flexural rigidity, vegetation height), previous studies have not fully revealed the effects of the vegetation flexibility on the three-dimensional flow structure around a vegetated area. In this study, we investigate the secondary flow structures and sediment transport processes around a vegetation patch by flume experiments. First, we performed velocity measurements by particle image velocimetry (PIV) for flows around a single upright rigid strip blade and around a single inclined rigid strip blade. Mean and turbulent flow structures around the simplified vegetation models are compared with each other to reveal the effect of deflection of vegetation elements. Next, we conducted vertical and horizontal PIV measurements and sediment deposition experiments for submerged, rigid and flexible vegetation patch flows. This aims to explore the effects of vegetation flexibility on the secondary flow structure and sediment deposition within and behind the vegetation patch. The velocity measurements for the single vegetation element demonstrate that the inclination of the vegetation element causes the downward flow on the side of the vegetation element and upward flows behind. For the vegetation patch flows, rotating secondary flows develop around the submerged flexible vegetation patch, but their direction is opposite to those around the submerged rigid vegetation patch. These results suggest that the deflection of the flexible vegetation elements creates the characteristic secondary flow structure and enhances the suspended load transport into the vegetation patch wake region.

ABSTRACT. Accurate monitoring of suspended sediment is critical for understanding river morphology and aquatic ecosystems; however, traditional point-measurement instruments such as physical samplers and LISST face significant constraints due to logistical difficulties and maintenance costs. Conventional remote sensing platforms, such as satellites and drones, offer broad spatial coverage but suffer from limited temporal continuity and surface-only observations. Consequently, they cannot provide the continuous and vertical monitoring required to understand sediment transport mechanisms and capture rapidly changing events. To address these gaps, this study evaluated the field applicability of the CCTV-Hyperspectral Imaging for Suspended Sediment Transport (CCTV-HISST) system, which is designed to provide high-resolution and 24/7 continuous observation of suspended sediment concentration (SSC). Sediment tracer experiments using loess at Gam Creek and Cheongmi Creek provided validation data under varying solar radiation conditions. A real-time radiometric calibration method, involving the simultaneous observation of a Spectralon reference panel and the water body, was implemented to minimize atmospheric and illumination uncertainties. For SSC conversion, we utilized the LiCAR (Light Classification and Adaptive Regression) model, and the resulting surface concentration maps were vertically extended using the Rouse-alpha model to reconstruct a continuous 3-dimensional SSC structure. Notably, the integration of the Rouse-alpha model allowed for the precise observation of complex sediment mixing patterns, providing a detailed view of the spatiotemporal evolution of the plume. The application of the LiCAR model demonstrated high predictive accuracy (R2 = 0.97, MAPE = 13.3%) with negligible bias. Across all sites, the spectral sensitivity remained stable; however, specific biases were observed due to variations in water depth, sediment mixing, and solar-camera geometry. This study confirmed that CCTV-HISST provides stable and reliable measurements even under complex hydraulic and optical conditions, effectively filling the gap in continuous temporal monitoring left by traditional remote sensing while offering deep insights into sediment dynamics.

ABSTRACT. Climate change is expected to affect water cycle through changes in precipitation, river streamflow, and soil moisture dynamics, thereby posing a threat to surface-water-fed wetland habitats and their biodiversity. This paper examines past trends and future impacts of climate change on riparian, water-dependent habitats within Poland's main river network. Changes in duration of flooding and inundation events were used to assess impact of climate change riparian wetland habitats.

One can distinguish two types of surface water-fed wetlands, which differ in their lengths of inundation. In this work, it is assumed that swamp is inundated by shallow water for relatively long period. Swamps are not always peat-accumulating wetlands, and in many cases, the major soil type in those areas is alluvium. Due to environmental conditions, swamp vegetation is adapted to growth in stagnating or slowly flowing water. Swamps are usually associated with adjacent rivers or lakes. Marsh has less open water than a swamp but is frequently inundated. The flooding phenomenon is much more dynamic, leading to the development of alluvial soils. Moreover, unlike swamp, marsh lacks woody vegetation and is dominated by grasses, rushes and reeds.

The main challenge in assessing inundation characteristics in riparian areas is estimating bankfull discharge flow along the entire river stretch where river-dependent habitats are located. The flood risk mapping is not useful for this purpose, as Q1% flow has no ecological meaning.

Therefore, national-scale, data-driven approach to bankfull discharge estimation was developed. Existing spatial and hydrological databases covering the entire territory of Poland were used, including a digital elevation model, river cross-section surveys, and discharge records from gauging stations. A total of approximately 500 sites were selected where these datasets overlapped, allowing simultaneous analysis of channel geometry, slope, and observed discharge. At these locations, Manning’s roughness coefficients were calculated and combined with cross-sectional information, including bankfull indicators, to estimate bankfull discharge values. In next step, regression models were developed to relate bankfull discharge to selected physiographic characteristics: catchment area, channel slope, and channel dimensions. These relationships were then applied to the entire river network, enabling the estimation of bankfull discharge for all river reaches across the country. Finally, a national-scale hydrological model (SWAT) was used for both baseline and future climate scenarios to analyze the duration and frequency of bankfull-flow exceedance. The resulting changes in inundation patterns were interpreted in terms of their potential impacts on riparian habitats at the national scale.

ABSTRACT. Flood water detention utilizing traditional flood mitigation systems, such as open levees(locally knowns as kasumitei) and abandoned paddy fields, is a promising nature-based solution (NbS) for small catchments. However, the effectiveness of peak discharge reduction depends on the spatial configuration of abandoned paddy fields used as storage areas. This study aims to elucidate the effects of ground elevation, areal extent, and ridge configuration on peak discharge reduction in abandoned paddy fields connected to open levees. Two-dimensional inundation simulations were conducted using iRIC (Nays2DFlood solver) for the Taigawa River basin in Toyooka City, Hyogo, Japan. The calculation area included the river channel, abandoned paddy fields, and open levees, and was divided into uniform 0.3 m grid cells. The riverbed slope was set to 1/100, and Manning’s roughness coefficients were set to 0.03 for the river channel and 0.06 for the abandoned paddy fields. The ground elevation in the abandoned paddy fields was set to 0.1 m, 0.3 m, 0.5 m, and 0.7 m. The abandoned paddy field area was varied by changing the field size, and ridges with a height of 0.4 m were installed within the fields. Simulations were conducted for combinations of these conditions. The simulation results showed a linear relationship between the ratio of peak discharge reduction and the area of abandoned paddy fields. In addition, even a slight lowering of the ground elevation contributed to peak discharge reduction. The width of the abandoned paddy fields perpendicular to the river was more associated with the reduction ratio than their longitudinal length. Under the no-ridge conditions, the ratio of peak discharge reduction increased linearly as the ground elevation of the abandoned paddy fields decreased. In contrast, under ridge scenarios, no significant effect of increasing the number of ridges was observed at lower ground elevations. These findings provide practical guidance for the management of abandoned paddy fields and open levees for flood detention in small catchments.

ABSTRACT. This paper examines the experiences of Chinese water governance institutions in advancing the United Nations 2030 Agenda SDG 6 (Clean Water and Sanitation). The purpose is to synthesize practical insights from China’s institutional innovations, policy frameworks, and implementation strategies, offering perspectives for global water governance in dynamic and changing environments. The study employs a mixed-methods approach, combining case analyses of representative water projects (e.g., drinking water safety for rural residents, promoting the rural toilet revolution, national water conservation initiative) and policy document reviews. Quantitative data on water access, pollution reduction, and infrastructure efficiency are integrated to assess progress. The results reveal that China’s multi-level governance model—emphasizing adaptive policymaking, technology-driven solutions (e.g., smart water systems), and public-private partnerships—has enhanced water security and resilience. However, challenges persist in balancing economic growth with ecological conservation. The findings align with the Congress theme “Water in the Changing World: Innovation and Adaptation” by demonstrating how institutional agility and technological innovation address climate uncertainties, urbanization pressures, and shifting resource demands. This paper underscores the value of context-specific yet replicable governance frameworks, contributing to global dialogues on equitable and sustainable water futures.

ABSTRACT. This study evaluates the potential for carbon dioxide reduction and cost savings through small-scale hydropower generation and greywater reuse utilizing urban water resources. Six building-scale scenarios were developed by combining rainwater and greywater in varying proportions for power generation and greywater reuse. Each scenario was simulated using 12 years of rainfall and sewage data, followed by comparative analysis. Rainfall data were obtained from Automated Surface Observing System (ASOS) observations. Due to the absence of long-term measured data, water usage was synthesized based on previous studies. Parameters such as catchment area and available head were configured to optimize each scenario. This study provides new guidelines for urban water resource utilization and demonstrates the potential for reducing carbon dioxide emissions through greywater reuse and eco-friendly power generation, thereby enabling proactive responses to climate change. This work was supported by Korea Environment Industry & Technology Institute(KEITI) through R&D Program for Innovative Flood Protection Technologies against Climate Crisis, funded by Korea Ministry of Environment(MOE)(RS-2023-00218973) and Korea Environment Industry & Technology Institute (KEITI) through the Aquatic Ecosystem Conservation Research Program (or Project), funded by Korea Ministry of Environment(MOE)(RS-2022-KE002214).

ABSTRACT. Fecal coliform (FC) is widely used as a representative sanitary indicator for indirectly indicating fecal contamination and the potential presence of associated pathogens (e.g., E. coli O157:H7 and Salmonella). FC can enter aquatic systems through the combined effects of diverse point and nonpoint sources, including effluents from wastewater treatment plants, livestock activities, and manure/compost application. FC concentrations are characterized by substantial variability driven by rainfall runoff events and hydrologic and hydraulic processes. Because the causes of fecal contamination are difficult to interpret based solely on observations at individual monitoring sites, watershed-scale spatial analyses that comprehensively consider land-use structure, source characteristics, and hydrologic and topographic conditions are essential. This study aims to quantify the spatial distribution of FC concentrations at the mid-watershed scale within the Han River basin and to identify dominant factors and contamination patterns through linkage with watershed characteristics. Monthly streamflow and water-quality data (pH, TN, TP, water temperature, and fecal coliform, among others) were compiled for outlet monitoring stations of each mid-watershed for 2015–2024, retrieved from the Water Environment Information System. In addition, a suite of explanatory variables was developed, including population density, land-use composition (forest, agricultural land, and urban areas), livestock manure generation, mean slope, precipitation, and sewerage coverage rate. Seasonal stratification, as well as wet/dry and growing/non-growing period conditions, was incorporated to examine the relationships between FC concentrations and watershed factors and to characterize spatial differences in contamination patterns across mid-watersheds. The findings of this study are expected to provide scientific evidence to support the development of policy and management guidelines for reducing fecal microbial contamination in the Han River basin. Acknowledgements: This research was supported by Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education (No. 2022R1C1C1010804)

ABSTRACT. Asian cities are facing increasing complex climate challenges, including extreme rainfall, flooding, and rising carbon emissions. In response, urban wetlands have gained attention as Nature-based Solutions (NbS) for climate risk reduction. Existing studies, however, have largely focused on the ecological value of wetland or the functions such as water purification, and floodwater storage. While few studies have suggested that urban wetlands have increasing potential in reducing climate risks by understanding their functional synergies and trade-offs, a limited attention was given to how policy design and governance shape their multifunctionality and climate risk orientation. This study examines how urban climate governance and wetland management policies in five Asian cities leverage the environmental, social, and economic functions of urban wetlands to enhance climate resilience. To this end, we develop policy-based indicators: the Functional Diversity Index (FDI), which captures the breadth of wetland multifunctionality, and the Climate Risk Orientation Index (CROI), which measures the extent to which policies prioritize climate risk reduction. The analysis draws on 25 urban wetland and climate governance policy documents published between 2010 and 2025 across selected Asian countries, including South Korea, Japan, Singapore, Indonesia, and the Philippines. Study cities were selected based on document availability, language of publication, and policy maturity. A structured codebook was developed to classify wetland’s environmental, social, and economic functions (e.g., flood mitigation, recreation, and tourism) as policy design elements, such as regulatory instruments, citizen participation, and biophysical intervention. Content coding analysis was conducted in the following sequence: (i) sentence-level extraction of wetland-related policy statements, (ii) inductive and deductive coding using the predefined codebook, and (iii) synthesis and cross-case comparison to identify dominant policy functions and design strategies, as well as key differences across cities. The results demonstrate that the urban wetland multifunctionality is driven primarily by policy design and governance arrangements, rather than by prevailing ecological conditions alone in achieving climate risk reduction. Substantial cross-city variation is observed in both FDI, and CROI. The Philippines exhibits disaster response-oriented governance structure with a strong emphasis on flood and disaster mitigation, resulting in a high CROI but a moderate FDI. In contrast, South Korea adopts a conservation- and capacity-building-oriented approach that integrates biodiversity protection, ecotourism, and environmental education, yielding a high FDI but a relatively lower CROI. These contrasting patterns highlight the critical role of governance - particularly public participation, and legal institutionalization - in shaping how urban wetlands are mobilized for climate resilience.

ABSTRACT. Phosphorus loss from agricultural activities is a major non-point source pollutant that degrades adjacent aquatic ecosystems. Process-based models such as APEX are widely used to evaluate phosphorus reduction policies for sustainable agriculture. However, their high computational demand makes them inefficient for policy analyses requiring thousands of scenario simulations. This study aimed to develop a machine-learning meta-model that approximates APEX to rapidly predict phosphorus loss under diverse upland crop cultivation conditions in Korea. APEX model configurations were established using the 「STATISTICAL YEARBOOK OF AGRICULTURE, FOOD AND RURAL AFFAIRS (2024)」 selecting representative crops with large cultivation areas for each crop group: corn (cereals), soybean (legumes), sweet potato (tubers), Kimchi-cabbage (vegetables), and perilla (specialty crops). In addition, highland fields are regarded as representative vulnerable environments for phosphorus loss from croplands and were included in the APEX model, with kimchi cabbage selected as the representative crop. Input datasets integrated topographic information (coordinates, field area, slope, and field length), meteorological data from the Korea Meteorological Administration (precipitation, wind speed, temperature, solar radiation, and humidity), and soil data and cropping schedules from the Rural Development Administration. Model calibration was conducted using runoff volume (m³) and total phosphorus (TP) loads (kg/ha) compiled from previous monitoring studies. Sensitive parameters affecting runoff and TP loads were first identified through Sobol sensitivity analysis, followed by automated calibration for the selected parameters. Subsequently, the calibrated APEX model was driven by observed meteorological data from weather stations and future climate scenarios based on the Shared Socioeconomic Pathways (SSPs) provided by the National Institute of Agricultural Sciences to generate training datasets across management options, including no-tillage, cover crops, fertilizer-rate adjustments, and vegetative buffer strips. The meta-model was developed by applying multiple machine-learning algorithms and comparing their predictive performance using scenario inputs (climate and management variables) and APEX outputs (runoff and TP loads). The developed meta-model is expected to improve the efficiency and scalability of policy assessments, supporting phosphorus mitigation strategies for upland fields and the evaluation of climate-change vulnerability. Acknowledgements: “Development of Adaptation Technologies and Vulnerability Assessment of Agricultural Runoff and Irrigation Water Quality under Changing Climate and Cropping Systems” (Project No. 00396736), funded by the National Institute of Agricultural Sciences, Rural Development Administration, Republic of Korea.

ABSTRACT. The Nakdong River Basin is increasingly vulnerable to ecological drought as flow conditions are affected by multi-purpose weirs and hydrologic variability intensifies under climate change. Traditional drought assessments—meteorological, agricultural, hydrological, and socioeconomic—primarily quantify deficits in precipitation, soil moisture, water resources, or water supply–demand balance, and therefore do not necessarily indicate whether ecological flow requirements are satisfied. This study is motivated by the need to translate hydrologic deficits into an ecologically meaningful diagnosis that identifies when and where river ecosystems face water shortages. Accordingly, this study aims to (1) estimate site-specific environmental flow requirements using habitat-based modeling, (2) quantify ecological drought events as times when observed discharge falls below those requirements, and (3) evaluate the spatio-temporal vulnerability of ecological drought across the Nakdong River Basin. Seven ecologically vulnerable sites (Gudamgyo, Gimcheongyo, Seonjugyo, Hwangganggyo, Jeongamgyo, Samsanggyo, and Gangdongdaegyo) were analyzed using 10 years (2015–2024) of daily discharge and accumulated precipitation data. Environmental flows were estimated with PHABSIM based on habitat suitability indices (HSI) for representative fish species at each site. The Environmental Ecological Flow Index (EI) was defined as the ratio of observed discharge to the estimated environmental flow. Ecological drought events were identified as EI < 1 using run theory, and event-based duration and severity were derived. Long-term tendencies were explored using the Mann–Kendall test. Ecological droughts clustered in two periods—late 2015 to early 2018 and late 2021 to early 2023—during which EI < 1 persisted at multiple sites. Gangdongdaegyo showed recurrent, high-severity events with long durations, indicating pronounced vulnerability, whereas Samsanggyo remained relatively stable with EI generally ≥ 1. Seasonally, ecological drought risk was higher in winter than in summer, reflecting increased vulnerability during the dry season. Overall trend significance was limited, but site-specific tendencies were observed. By spatio-temporally identifying the sites and periods when ecological drought is most severe in the Nakdong River, this study can serve as baseline information for sustainable river management policies and ecology-oriented flow management in the Nakdong River Basin.

ABSTRACT. Intensity–duration–frequency (IDF) curves are essential tools for quantifying the probability of extreme rainfall events and form the basis for designing urban stormwater management systems. However, under accelerating climate change and rapid urban expansion, rainfall regimes have become increasingly variable, prompting the need for more robust and accurate IDF estimation methods. Conventional frequency analysis used to construct IDF curves relies on the full set of empirical frequency samples, among which data points with frequencies greater than 50% (corresponding to return periods shorter than 2 years) account for half of the dataset. Although these high-frequency samples strongly influence curve fitting, they contribute little to engineering design, which typically focuses on return periods ranging from 2 to 100 years. Their disproportionate influence may therefore distort the IDF curve shape and reduce the reliability of design rainfall estimates.To address this issue, this study introduces a sparsification framework that selectively reduces the density of high-frequency samples prior to curve fitting. Several sparsification strategies are developed and applied to empirical rainfall datasets, followed by a comprehensive sensitivity analysis to evaluate their effects on IDF curve parameters, curve smoothness, and design rainfall estimates. The analysis examines how different sparsification intensities alter the relative weight of low-frequency samples and improve the stability of fitted IDF relationships. The results show that appropriate sparsification substantially mitigates the overrepresentation of high-frequency data, enhances the influence of low-frequency (i.e., return period > 2 years) samples, and leads to more physically meaningful and engineering-relevant IDF curves. The improved curve fitting yields more reliable design rainfall estimates, particularly for longer return periods that are critical for modern stormwater infrastructure planning. This study demonstrates that sparsification is a promising enhancement to traditional frequency analysis and provides a practical pathway for improving IDF curve robustness under changing climate conditions.

ABSTRACT. Transboundary river basins frequently operate under severe information asymmetry: upstream infrastructure can strongly influence downstream hydrologic hazards, yet operational data (e.g., reservoir releases, rule curves, gate operations) are often undisclosed or inaccessible. This lack of transparency undermines flood preparedness and integrated water resources management, particularly during extreme events. This study develops and demonstrates a satellite-only hydroinformatics framework to infer upstream reservoir operations in a data-restricted transboundary basin. Three challenges are addressed: temporally sparse reservoir water-level observations from satellite altimetry, difficulty in inflow reconstruction without observed streamflow, and the ill-posed nature of release inference without independent downstream constraints. We propose a modular workflow that integrates multi-source satellite observations with modeling and optimization. Reservoir water level is reconstructed by integrating satellite altimetry, optical imagery, and synthetic aperture radar (SAR). Reservoir inflow is simulated using a conceptual rainfall–runoff framework forced by satellite climate inputs and constrained by surrogate discharge derived from L-band microwave signals. An independent downstream proxy is constructed from Sentinel-1 SAR by translating surface-water dynamics into a unitless surrogate discharge signal, which is used to constrain a parsimonious rule-based operations model parameterized by a constant maximum discharge capacity and a flood-release ratio. The optimized model reproduces reservoir water-level dynamics with strong skill (R² = 0.89; RMSE = 1.88 m) and captures the 2020 flood dynamics, including near pass-through behavior during peak inflow (estimated peak inflow and release of 1,165 m³/s and 1,147 m³/s). This distinction is critical for disaster resilience, as downstream hazard depends strongly on whether reservoirs attenuate inflow or pass it through during extreme events. The proposed satellite-only framework provides actionable and physically interpretable information to support integrated water resources management and improve downstream preparedness in shared river basins.

ABSTRACT. Monitoring inland water quality (WQ) remains a challenge due to the limited spatial coverage and sparse in situ observations, which are insufficient to capture the heterogeneity of WQ conditions across large regions. Although recent studies have started to apply remote sensing and artificial intelligence (AI) based approaches to estimate inland WQ variables, most existing efforts have been confined to local or basin-scale applications. As a result, nationwide assessments that consistently characterize long-term WQ dynamics across diverse hydrologic settings remain limited, despite their importance for integrated water resource management, environmental policy, and climate impact assessments.

This study presents a comprehensive assessment of long-term inland WQ trends across all surface water bodies in South Korea from 2015 to 2024, based on a deep learning reconstruction framework. Our data-driven models were employed to estimate multiple WQ indicators, including water temperature, chlorophyll-a concentration (Chl-a), suspended solids (SS), total nitrogen (TN), and total phosphorus (TP), by integrating multispectral optical observations from the Harmonized Landsat–Sentinel-2 (HLS) S30 product with meteorological forcing datasets. The framework produces spatially and temporally consistent 30-m resolution WQ reconstructions for all detectable inland water bodies, which were subsequently aggregated to a 1-km grid to produce national-scale trend analysis. This spatial aggregation strategy enables consistent comparison of long-term hydrologic trends across spatial scales.

The reconstructed dataset enables spatially explicit trend analysis of WQ at the national scale, revealing heterogeneous temporal patterns across rivers, reservoirs, and lakes. Initial trend analysis focusing on Chl-a reveals pronounced regional variability, with statistically significant increasing trends identified using Sen’s slope estimator, particularly in basins containing major industrial regions.

By providing a spatially consistent, decade-scale national reconstruction of inland WQ dynamics in South Korea, this study enables robust trend analysis across diverse surface water bodies.. The proposed framework supports national-scale water quality monitoring and surface water management under changing climatic and land-use conditions.

ABSTRACT. Taiwan’s mountainous watersheds are frequently impacted by typhoons that trigger highly nonlinear rainfall–runoff–sediment interactions, posing critical challenges to reservoir sedimentation management and flood control. The Shihmen Reservoir, one of northern Taiwan’s major multipurpose reservoirs, has suffered rapid capacity loss and turbidity issues due to extreme sediment inflows. Although the physics-based SRH-W (Sedimentation and River Hydraulics–Watershed) model can accurately simulate rainfall, runoff, and sediment dynamics, its high computational demand limits its use in real-time applications for flood forecasting and reservoir operation. To overcome this challenge, this study developed an artificial intelligence (AI)-based surrogate modeling framework that integrates SRH-W simulations with deep learning to achieve both physical reliability and computational efficiency. A total of 25 typhoon events from 2015 to 2023 were simulated using SRH-W to generate high-resolution datasets covering rainfall intensity, infiltration, surface roughness, soil erodibility, and sediment concentration. These data, distributed across 13,697 unstructured mesh cells, were processed into time-series and spatial image sequences for training the AI model. Several deep learning architectures were tested, including LSTM, GRU, BiLSTM, BiGRU, and Transformer networks. Among them, the Deep BiLSTM achieved the highest accuracy, with R² values of 0.84 for flux-based predictions and 0.66 for grid-based sediment concentration and flow fields. The model successfully reconstructed hydrographs and sediment dynamics during major typhoons such as Nisha (2017) and Soudelor (2015), reproducing both flood peaks and sediment response patterns. Compared with SRH-W, the AI surrogate accelerated simulations by approximately 10⁵ times while maintaining high predictive fidelity. The results demonstrate that the proposed hybrid framework can effectively simulate complex hydrosediment processes for large-scale watersheds, providing a powerful tool for near-real-time flood and sediment forecasting. This approach establishes a foundation for integrating digital twins of watersheds, supporting adaptive reservoir operation, sediment management, and sustainable flood risk governance in Taiwan and other mountainous regions.

ABSTRACT. Deep learning-based flood surrogate models have shown promise in accelerating spatiotemporal flood simulations, yet their cross-regional transferability remains a significant challenge, limiting widespread application in data-scarce catchments. This study proposes a transfer learning framework to address the transferability of flood surrogate models across catchments with inconsistent base resolutions and diverse terrain features. Within this framework, a Residual SwinUNet model is developed to fuse multi channel coarse grid flood maps with fine‑grid topographic features for high‑resolution flood map reconstruction. The framework is applied to the Shenzhen River catchment in China and Richmond River catchment in Australia, using various transfer learning strategies. Results demonstrate that under data-abundant conditions, the proposed model accurately reconstructs high-resolution flood maps for both uniform and spatiotemporally distributed rainfall events. In data-scarce scenarios, full fine-tuning strategy recovers over 90% of baseline accuracy using less than 3% of the pre-training events, while Low-Rank Adaptation matches or even surpasses scratch training performance with only 10–35% of trainable parameters. Furthermore, cross-scale experiments reveal that the framework is effective across a broad range of scale factors. Pretraining on smaller source scale factors enhances transferability, while larger target scale factors amplify the benefits of pretraining. This study establishes a new paradigm for enhancing cross-regional adaptation of flood surrogate models for data-scarce catchments.

ABSTRACT. Recent advances in artificial intelligence (AI) have enabled foundation models to directly predict precipitation from large-scale Earth system datasets, offering new opportunities for subseasonal-to-seasonal (S2S) precipitation prediction. These models range from data-driven emulators of conventional numerical weather prediction systems to end-to-end foundation models that infer precipitation directly from atmospheric states. AI-based precipitation prediction has demonstrated performance comparable to that of physics-based models. However, the hydrologic value of AI-based precipitation prediction depends not only on overall forecast accuracy but also on its ability to represent extreme precipitation characteristics, including intensity distribution, duration, and spatial coherence. Extreme events (i.e., flood and drought) are rare and highly nonlinear, making it difficult for data-driven models to learn. This hampers the direct use of AI-based precipitation forecasts in hydrologic applications, highlighting the need for systematic evaluation and comprehensive analysis of extreme precipitation characteristics in AI-driven S2S predictions.

This study compares precipitation predictions of a data-driven foundation model with those from a conventional physics-based model from 2017 to 2025. We further analyzed flood events during this period and focused on the predictability of extreme precipitation events for these two types of models. Results show that while foundation model-based precipitation captures large-scale variability, notable discrepancies remain in extreme rainfall, event duration, and regional consistency. In particular, smoothing and underestimation of short-lived intense events are evident in some regions and seasons, which may limit the hydrological application of the foundation model at S2S lead times. We will discuss the strengths and limitations of foundation model-based precipitation for S2S-scale studies.

ABSTRACT. Dense water level monitoring in river networks is often constrained by installation and maintenance limitations, despite the critical need for reliable real-time information for flood response and river management. To address spatial observation gaps, this study proposes a machine learning–based virtual sensing framework for multi-site water level estimation that explicitly incorporates river network connectivity. The framework represents the river basin as a directed graph in which hydrologically connected components—including dams, water level stations, and rainfall gauges—are treated as nodes linked according to upstream–downstream topology. Spatial dependencies within the river network are learned through a graph-structured virtual sensing model (Graph-LSTM), while temporal dynamics are captured using a recurrent neural network structure. To enhance predictive performance during flood conditions, a weighted dual-loss function is introduced to emphasize high water level events, forcing the model to focus on peak dynamics that are critical for flood monitoring. The proposed framework is applied to a dam-regulated reach of the Seomjin River in South Korea, where water levels are virtually estimated at three key bridge sites. Comparative experiments demonstrate that the proposed virtual sensing framework consistently outperforms conventional single-site and multi-site LSTM-based models, achieving Nash–Sutcliffe efficiency (NSE) values exceeding 0.965 at all target locations. In particular, the framework significantly improves peak water level estimation, substantially reducing quantitative peak error ratios at downstream sites where cumulative hydrological effects are more pronounced. These results indicate that integrating river network structure into machine learning–based virtual sensing frameworks can significantly enhance the accuracy and robustness of real-time multi-site water level estimation in complex river systems.

ABSTRACT. Flood control in cascading reservoir systems is increasingly challenging under climate change. Deep learning approaches demonstrated strong potential for modeling inflows and outflows in regulated river systems, yet significant gaps remained because hydrological processes and reservoir operations were typically modeled separately. This limitation was especially pronounced for cascading dam systems and for assessments of future flood changes using large ensemble climate datasets. This study addressed these gaps by developing a DL-based model cascading dams and applying it to the Shimouke and Matsubara dams on Kyushu Island, southwestern Japan. First, the proposed model employed a customized Bidirectional Long Short-Term Memory network and was trained, validated, and tested using 40 years of hourly data (1986–2025). The model was then applied using 720 ensemble years of rainfall from the Database for Policy Decision-Making for Future Climate Change (d4PDF) to simulate cascading dam performance under the future climate scenarios (2031-2090). Analysis of 4,495 simulated flood events indicated substantial increases in both flood frequency and flood magnitude. For the 100-year return period, flood peaks were projected to increase, based on the average of 12 d4PDF ensemble members, by +25% (from 2,448 to 3,068 m³/s) at Shimouke and by +85% (from 2,955 to 5,456 m³/s) at Matsubara. Flood volumes were also projected to increase substantially, by +27% (from 38.54 million m³ to 52.98 million m³) at Shimouke and +19% (from 45.79 million to 56.66 million m³) at Matsubara. These projected values may exceed both the spillway design capacity and the designated flood control storage at the two reservoirs: 2,300 m³/s and 53.30 million m³ at Shimouke, and 5,000 m³/s and 45.80 million m³ at Matsubara. Overall, the results indicate that future floods are likely to become more severe and may exceed the design limits of the existing reservoir system, highlighting the need to strengthen preparedness and implement adaptive reservoir management strategies.

ABSTRACT. Wetlands play a critical role in hydrologic and ecological systems; however, accurate wetland water level prediction remains challenging due to data scarcity and complex hydrological interactions. This study evaluated multiple data-driven models for daily wetland water level prediction, including decision tree (DT), random forest (RF), support vector machine (SVM), artificial neural networks with varying hidden node configurations (ANN1–ANN10), long short-term memory (LSTM), temporal convolutional network (TCN), autoregressive integrated moving average (ARIMA), and Kalman filter model. The Upo Wetland, the largest inland wetland in South Korea, was selected as the study area. Daily water level data from 2009 to 2015 were used as the target variable, with meteorological variables and upstream water levels as inputs for the machine learning models. Model performance was assessed using the correlation coefficient (CC), Nash–Sutcliffe efficiency (NSE), and root mean square error (RMSE). ANN7 achieved the highest predictive accuracy (CC = 0.974, NSE = 0.942, RMSE = 0.082), followed by ANN6 (CC = 0.970, NSE = 0.938, RMSE = 0.085). The random forest model also demonstrated strong accuracy (CC = 0.965, NSE = 0.926, RMSE = 0.092). Deep learning models (LSTM and TCN) also captured temporal dynamics reasonably, whereas ARIMA and the UCM Kalman filter showed lower predictive skill. Overall, machine learning and deep learning methods outperformed conventional statistical models, providing valuable baseline information for data driven wetland water level forecasting, particularly in data limited or ungauged systems.

ABSTRACT. This study investigates the use of the “Informer” based deep learning architecture for streamflow forecasting at the Shivade station in the Upper Krishna Basin, Maharashtra, India. Utilizing 17 years of daily rainfall and runoff data, the research aims to forecast streamflow 24 to 120 hours ahead, offering a robust alternative to traditional Rainfall-Runoff modeling approaches in a very complex basin wherein “data scarcity” is the vital issue. 305-km-long Shivade Basin, situated in the Upper Krishna Basin of Maharashtra, India, has three tributaries of the Krishna River (Tarali, Urmodi, and Venna rivers) which contribute the flow in the basin at Shivade station (as an outlet station). The basin consists of seven rain gauge stations and one discharge measurement station, namely Upshinge, Thoseghar, Targaon, Sandavali, Nagthane, Marali, Jawalwadi, and Shivade(73°17’ to 81°9’ East and 13°10’ to 19°22’ North). Surprisingly all these rain gauge stations are situated at the downstream extreme region of this basin and the entire upstream part of this basin is ungauged. Consequently, different kind of “scarcity of the rainfall data” is there in the basin and making it very difficult, complex hydrological and hydraulic case of “Rainfall- Runoff” model. Conventional hydrologic models often rely heavily on extensive basin parameters, which are not always available or easy to obtain, especially in data-scarce regions. To overcome this limitation, data-driven models such as deep learning have been emerged as powerful tools in last decade. The Informer model, known for its efficiency in handling long-sequence time series data for forecasting any random variable, is applied to capture the complex nonlinear relationships between rainfall and runoff in this basin. That too, as Informer is an advancement on “transformer” technique which is invented to forecast the parameter over long range of forecasting intervals, present study demonstrates its application for 24 hr (small range) to 120 hrs ahead (long range -5days ahead) forecasting of runoff in this basin. All the developed models performed well with good prediction accuracy, with a correlation coefficient (r) of 0.89 and a reduced Root Mean Square Error (RMSE). Performance remained acceptable even at longer lead times (up to 120 hours), although prediction of extreme events remains a challenge. The novelty of this work lies in the application of a transformer-based deep learning model with limited basin-specific input, demonstrating its potential as a scalable, data-efficient solution for short- to long-range streamflow forecasting in ungauged or poorly gauged basin.

ABSTRACT. The objective of this study is to develop and validate a two-dimensional depth-averaged numerical model for simulating mudflow propagation in the district of Baling, Kedah, Malaysia. The model solves the depth-averaged mass and momentum equations, with advection terms discretized using the Constrained Interpolation Profile (CIP) scheme to minimize numerical diffusion and preserve sharp flow fronts in rapidly varying flows. Mudflow rheology is represented using a Herschel–Bulkley formulation to account for yield stress and shear-dependent viscosity associated with high fine-sediment concentration. Turbulent effects and internal collisional resistance from coarse debris are incorporated through a pseudo-Manning parameter. Mudflow initiation is represented using a dam-break approach to mimic sudden slope failure and rapid release of material. Due to limited information and large uncertainties, small bridges and minor hydraulic structures along the flow path are not explicitly represented in the model. Model calibration is performed by adjusting Manning’s roughness coefficient to account for unresolved surface resistance and structural effects. Model validation is conducted using post-event aerial photographs, with particular emphasis on comparing the simulated and observed downstream flow extent. The simulated inundation area shows good agreement with aerial imagery, indicating that the model can reasonably reproduce overall propagation patterns and run-out behaviour. Remaining discrepancies are attributed to uncertainties in rheological parameters, surface roughness, and terrain resolution. The results demonstrated that the proposed CIP-based depth-averaged model provides a practical and reliable tool for rapid mudflow hazard assessment in mountainous regions where detailed field data are limited.

ABSTRACT. According to the National Fishery Information System, the number of weirs installed on rivers in Korea as of 2025 is approximately 34,000. Each year, approximately 150 weirs are newly constructed, and 100 weirs are decommissioned due to urbanization and declining agricultural water demand. Many of the abandoned weirs remain instream, blocking sediment transport to the downstream reach, disrupting ecological connectivity, and deteriorating upstream water quality. Over 90% of weirs are more than 30 years old, and 5,900 are structurally damaged, underscoring the need to remove abandoned, aged, and damaged weirs. However, previous studies on morphological changes after dam or weir removal are rare. This study presents numerical simulations for morphological changes following the removal of Gongneung Weir-2 using a quasi-steady flow model. The quasi-steady model assumes steady-flow conditions over short time intervals, based on the premise that flow changes much faster than stream morphology. The quasi-normal flow assumption by Cui et al. (2006) is adopted to compute post-removal bed evolution, particularly to address discontinuities formed by upstream sediment deposits. Total sediment load was estimated using the Engelund-Hansen’s formula, and morphological changes were calculated using the Exner equation. Gongneung Weir-2 is a 1.5 m high instream installation constructed in the 1970s on the Gongneung-cheon Stream, a tributary of the Han-gang River. The weir was removed on April 14, 2006, due to a land-use change in the nearby area. The study reach is a 1.0 km section that includes the location of the removed Gongneung Weir-2. Topographic survey data are used for initial conditions and model validation. The simulation was conducted to evaluate the time required for the bed to reach morphological equilibrium. After the weir removal, the bed slopes in both upstream and downstream reaches changed abruptly at the beginning. A comparison with measured data reveals that the numerical model successfully simulates the morphological change after weir removal. Even considering field conditions, it takes less than a month to reach equilibrium, in which the slope upstream of the dam equals the slope downstream. The quasi-steady model successfully reproduced morphological changes following weir removal, indicating its utility for river restoration and for predicting sediment transport.

This work was supported by Korea Environment Industry & Technology Institute(KEITI) through 「Research and Development on the Technology for Securing the Water Resouces Stability in Response to Future Change(RS-2024-00335281)」 Program, funded by Korea Ministry of Climate, Energy, Environment(MCEE)

ABSTRACT. Compound channels are characterized by complex flow structures driven by momentum exchanges between the main channel and floodplains that become even more complex in the presence of riverine vegetation. Understanding the flow-vegetation interaction in these kinds of channels, therefore, is important to accurately describe sediment transport patterns for effective river management. Numerical models are required to be validated with experimental data for applications in describing such complex phenomena. Three-dimensional (3D) calculation models excel in capturing flow characteristics, but applying them to large-scale sites or experiments becomes a challenge because of their large computational requirements, especially for shallow open channel flows. This gave way to the introduction of quasi-3D models capable of representing 3D flow characteristics but with less computational cost, such as the Bottom Velocity Computation (BVC) method that was used in this study. It evaluates bottom velocity and vertical velocity distributions from a set of depth-integrated momentum and vorticity equations. The numerical model was validated using experimental data from the large-scale straight compound channel experiment conducted at the River Experiment Center of the Korea Institute of Civil Engineering and Building Technology (KICT). The effect of vegetation on the flow was represented by a multidimensional drag force term that considers flow direction. A comparison of the calculation results for the two-dimensional (2D) model and a quasi-3D model without shallow water assumption, known as the General Bottom Velocity (GBVC), was made to describe the influence of floodplain vegetation on the flow. When compared to experimental results, the quasi-3D calculation model introduced in this paper outperformed the 2DC model, and it was able to simulate the flow characteristics in both the vegetated and non-vegetated sections of the compound channel.

ABSTRACT. The safety and resilience of riverine infrastructure have become critical concerns as hydrological regimes continue to evolve under changing climatic conditions. Among these concerns, pier scour remains one of the leading causes of bridge failure worldwide, highlighting the need for improved physical understanding and predictive capability. Conventional pier scour formulations have been developed primarily under free-surface flow assumptions and therefore exhibit inherent limitations when applied to highly complex erosive flow regimes induced by deck submergence associated with extreme flood events.

In recent years, the Phenomenological Theory of Turbulence (PTT) has gained significant attention as a promising framework for elucidating the physical mechanism inherent in various scour processes. Concurrently, global climate change has increased the frequency of extreme flooding events, which often cause riverine free-surface levels to rise until bridge decks become partially submerged. Such condition induce pressure-flow through vertical contraction, wherein the nonlinear interaction between submerged deck flow and pier obstruction triggers complex scour mechanism that can substantially intensify pier scour. Despite the growing practical importance of these occurrences, research on this topic remains in short supply compared to the rising urgency of the issue. Motivated by this necessity and the proven utility of the PTT in other scour processes, this study attempts to apply the PTT to this problem to explore its potential for analyzing pier scour under pressure-flow.

By extending the PTT framework to pressure-flow conditions at bridges, this study seeks to establish a unified physical interpretation linking pier obstruction, vertical contraction effect by submerged deck, and sediment response mechanisms. Such an interpretation is expected to enhance the robustness of pier scour prediction under extreme flooding scenarios and to support the development of physics-based methodologies for bridge safety evaluation and design.

ABSTRACT. Open-channel flows over a drop structure with a trench show different flow structures depending on the trench geometry and hydraulic conditions, and may accompany oscillating hydraulic jumps under certain conditions [Fujita J. Vis. 5 335 (2002)]. Therefore, this flow can be a good benchmark problem to elucidate the essential aspects in numerical modeling for accurate reproduction of unsteady turbulent flows with large free-surface deformation ubiquitous in environmental and geophysical applications. The present study is an extension of Yoshimura et al. [KSCE J. Civ. Eng. 28 1041 (2024)], who examined the effect of turbulence models and computational grid spacings on the prediction results by using OpenFOAM and found that only the SST k-omega model using a fine grid can reproduce the oscillating hydraulic jumps. Here we examine (i) which elements of the SST model enable it to capture the hydraulic jump phenomena and (ii) the impact of the wall boundary condition (i.e., using off-the-wall empirical functions or imposing the no-slip/impermeable conditions directly) on the prediction accuracy, by using an originally developed in-house code. It is clearly shown that the eddy-viscosity limiter of the SST model activated near-wall regions to suppress its overprediction, plays an important role in reproducing the unsteady hydraulic jumps. It is also revealed that the impact of the wall boundary condition is substantial. When the no-slip/impermeable conditions are imposed directly on the wall, as demonstrated by Yoshimura et al., only the SST model with a fine grid can capture the oscillating phenomena. When the empirical wall functions are used, on the other hand, computations not only on a fine grid but on a coarse grid can reproduce the oscillation, although qualitatively.

ABSTRACT. In conventional numerical simulations of river flows based on the shallow-water equations, it is common practice to prescribe discharge at the upstream boundary and water level at the downstream boundary for subcritical flows. However, from the viewpoint of the characteristic structure of the shallow-water equations, only two boundary conditions are physically admissible: velocity at the upstream boundary and water depth (or water level) at the downstream boundary.

When discharge is prescribed at the upstream boundary, it must be converted into velocity, which requires an additional assumption for water depth (or water level). As a result, three boundary conditions (velocity and depth at the upstream boundary, and depth at the downstream boundary) are effectively imposed, leading to an over-specified and potentially unstable boundary value problem.

A commonly used remedy is to assume uniform flow at the upstream boundary and determine velocity and depth from the prescribed discharge. However, this approach requires an additional specification of slope, which is not unique and may induce numerical instability depending on its choice.

In this study, we propose a boundary-condition formulation that avoids over-specification by inheriting one of the boundary conditions from the internal characteristic variable, namely a Riemann invariant, computed inside the computational domain. Using this approach, only the physically required number of external boundary conditions is imposed.

The method is implemented in the two-dimensional shallow-water solver iRIC-Nays2dH for two configurations: (1) prescribed discharge at the upstream boundary and prescribed water level at the downstream boundary, and (2) prescribed water levels at both upstream and downstream boundaries.

The proposed schemes are validated using steady-flow experiments conducted in a 1/50-scale physical model of the upper Ishikari River (river kilometer 156–157). The results demonstrate that the Riemann-invariant-based boundary conditions provide significantly more stable and accurate reproductions than conventional methods.

In particular, the water-level-based formulation at both boundaries enables discharge to be determined internally by the model without prescribing it externally, which is highly attractive for practical river engineering applications.

ABSTRACT. This study integrates Detached Eddy Simulation (DES) and Dynamic Mode Decomposition (DMD) to systematically investigate the complex turbulent structures and modal characteristics induced by set-back side obstructions in erodible compound river channels. By varying the obstruction lengths and their corresponding local scour conditions, the evolution of shear layers, vortex dynamics, pressure fluctuations, and turbulent kinetic energy (TKE) distributions is analyzed in detail, along with dynamic mode decomposition conducted on typical cross-sections. Research shows that the retreat distance of obstacles plays a key role in the evolution of the shear layer and vortex structures. Obstacles with a short retreat distance (SSO) generate a thin and high-energy shear layer, accompanied by high-frequency vortex shedding and concentrated turbulent kinetic energy. As the retreat distance increases, the shear layer gradually thickens and diffuses, leading to a decrease in vortex shedding frequency and a stabilization of the flow pattern. Local scour alters the near-bed flow field, resulting in an expanded flow cross-section, downward displacement of the shear layer, and a reduction in main channel velocity, which in turn increases near-bed instability and localizes the vortex structure. DMD analysis shows that high-frequency modes dominate the flow characteristics in all cases; however, under scouring conditions, the modal frequency decreases, and energy shifts toward low-frequency modes, indicating an enhanced coupling between bed morphology and flow dynamics. The findings provide insights into the design and maintenance of typical side obstructions, such as spur dikes or bridge abutments, and their potential short-and long-term influence on the local morphology of riverbed.

ABSTRACT. The runoff angle of the water jet from a lateral overflow weir in both straight and curved channels (with curvature radii of 0.50 m, 0.70 m and 0.90 m) was experimentally investigated using the Froude number, relative water depth, and curvature radius as parameters. The main results are as follows; 1) The upstream runoff angles θ1 increase with increasing the relative water depths and Froude numbers, while the downstream runoff angles θ2 follow the same trend. θ1 decreases as the curvature radius increases. However, the dependence of θ2 on the curvature radius is more complex. 2) In subcritical flow over lateral overflow weirs, regions with reduced water depth are observed. These regions form a semicircular shape in straight channels and an elliptical shape in curved channels. As the curvature radius decreases, the shape becomes a distorted ellipse skewed upstream. These patterns are related to the runoff angle. In supercritical flow, such regions are not observed, and the runoff angle increases due to the inertial force in the flow direction. 3) The formula for estimating θ1 was developed as a function of the Froude number, relative water depth, and relative curvature radius, and it showed good agreements with the experimental results.

ABSTRACT. Most countries in the world have significant investments in urban water infrastructures (e.g., storm drainage and flood management systems). Every day, people rely on these systems to protect lives, property, and natural water environment. These infrastructures have reduced the vulnerability of the cities, but at the same time could make them more vulnerable to climate extremes, due to the lack of consideration of what might occur when the design criteria are exceeded. Furthermore, recent assessment reports on climate change have indicated for the late 20th century a worldwide increase in the frequency of extreme weather events because of global warming, and this trend would be very likely to continue in the 21st century. Consequently, research on developing innovative approaches for limiting and adapting climate change impacts on urban water infrastructures is highly critical due to the substantial investments involved. However, it has been widely recognized that the main difficulty in dealing with climate change impacts for urban water infrastructure design is “how to estimate accurately the changes in the extreme rainfall and temperature processes at the urban basin scale projected by global/regional climate models because these models do not contain an adequate description of the hydrologic governing processes at relevant high temporal and spatial resolutions as required by the impact and adaptation studies”. This necessitates some form of downscaling of the climate model simulations from a coarse spatial resolution (20 – 250 km) down to much finer spatial grids and even point values if changes in local extreme hydrologic processes are to be assessed. In addition, the required time scales for assessing the climate change impacts on the urban hydrologic processes are usually less than one day. Therefore, the overall objective of the present paper is to provide an overview of recent advances and shortcomings in the development of downscaling methods for estimating design storms in the climate change context from both theoretical and practical viewpoints. In particular, another focus of this paper is on the recently published technical guide by IAHR to provide some guidance to water professionals and engineers on how to consider the climate change information in the design of urban water infrastructures.

ABSTRACT. Due to climate change and irrational human activities, the hydrological regime of the Yellow River Basin has undergone significant changes since 1950s, which has restricted the sustainable utilization of water resources in the basin. The purpose of this study is to investigate the intra-annual runoff distribution characteristics in the Taohe River Basin, as the second largest primary tributary in the upper Yellow River, which accounts for approximately 10% of the basin’s total surface water resources. The historical monthly runoff datasets during 1980~2016 from the Luqu hydrological station and Hongqi hydrological station were used in the investigation, which serve as representative stations of the headwater region and the basin outlet respectively. Three statistical indices, including the extreme values ratio (EVR), runoff concentration degree (RCD) and runoff concentration period (RCP), were employed to analyze the spatio-temporal variations in runoff concentration. The results indicate the following: (1) the annual runoff distribution curve of the Luqu hydrological station exhibits a unimodal pattern peaking in September; whereas that of the Hongqi hydrological station presents a weakly bimodal pattern, with peaks in July and September. (2) The EVR at the Hongqi hydrological station is significantly higher than that at the Luqu hydrological station, with values of 7.178 and 4.758 respectively. That indicates that the intra-annual runoff distribution at the Hongqi hydrological station is more pronounced than that at the Luqu hydrological station. (3) The RCD at the Hongqi hydrological station is 0.38, with a slight decreasing trend, while that at the Luqu hydrological station is 0.08, with no significant trend. That illustrates that the intra-annual runoff distribution is relatively uniform in the headwaters and a trend toward homogenization at the basin outlet. (4) The RCP at both the Hongqi and Luqu hydrological stations occurs in early August, while that of the Hongqi hydrological station lags by approximately 8 days on average. However, the RCP at the Hongqi hydrological station has alternated with the Luqu hydrological station since 2009, primarily attributable to the regulation function of the Jiudianxia Hydraulic Project commissioned in 2008.

ABSTRACT. In recent years, global warming has led to an increase in atmospheric temperature, which has caused a rise in water temperature in both riverine and marine environments. These thermal changes significantly impact aquatic ecosystems and water quality. The Ariake Sea, located in Kyushu Island, southwestern part of Japan, is an enclosed sea with vast tidal flats and a large tidal range of up to 6 meters. In its inner part, a long-term trend of increasing sea water temperature and decreasing salinity has been observed. It is suggested that riverine discharge plays a crucial role in these long-term changes in the marine environment. Therefore, the objective of this study is to evaluate the long-term changes in the thermal contribution provided by major rivers flowing into the inner part of the sea. 　This study examined the long-term changes in water temperature for major rivers flowing into the inner part of Ariake Sea from 1973 to 2022. For the temperature analysis, representative values for each year were determined by fitting a cosine function to the observation data. This method allowed for the appropriate evaluation of long-term trends despite irregular intervals of the observation. In addition, the study examined the long-term changes in the impact on the thermal environment of the inner part of the sea by calculating the thermal flux, which combines the river water temperature with the river discharge. 　The results of this study showed that the rate of water temperature increase in rivers is significant, particularly during the summer season. In some observation points, the rate of increase in river water temperature was more than double that of the sea water temperature. Furthermore, the thermal flux from rivers flowing into the inner part of the sea has increased during the past fifty years. This trend was especially clear during summers as well. These results suggest that the heat supply from rivers has been increasing, and that the cooling effect of inflowing river water on a coastal sea temperature has been decreasing. Therefore, understanding the long-term changes in thermal transport from river systems is important for predicting future environmental transitions and ecological shift in the Ariake Sea under climate change conditions.

ABSTRACT. Rote Ndao Regency is one of the regencies in East Nusa Tenggara Province, Indonesia, covering an area of approximately 1,280.10 km². The region exhibits a seasonal climate pattern characterized by a relatively short rainy season, occurring from December to April, and a prolonged dry season lasting approximately seven months. These conditions classify Rote Ndao Regency as a semi-arid region. Recurrent droughts have a significant impact on groundwater availability, which is crucial for sustaining agricultural activities and meeting local food needs. Rote Ndao Regency is also recognized as one of the areas experiencing the most severe drought conditions in East Nusa Tenggara Province. This study aims to identify and analyze drought severity in Rote Ndao Regency using the Palmer Drought Severity Index (PDSI). The PDSI method is capable of identifying drought potential and severity based on key parameters, including precipitation, potential evapotranspiration, air temperature, and soil moisture. The data used in this study consist of secondary data, including rainfall records for the period 2014–2023 obtained from the Keka, Pantai Baru, Danau Tua, and Eahun rainfall stations, as well as climatological data covering the period 1994–2023. Potential evapotranspiration was calculated using the Modified Penman method, while spatial analysis and drought mapping were conducted using ArcGIS 10.8 software. The results indicate that all rainfall stations in Rote Ndao Regency experienced drought conditions classified as extremely dry. The most severe drought occurred in 2016, with a PDSI value of −3.95, indicating extreme drought conditions in the study area. These findings underscore the necessity for adaptive and sustainable water resource management strategies, including optimizing rainwater harvesting, integrated groundwater management, and enhancing irrigation efficiency. Furthermore, the results provide a scientific basis for dryland agricultural planning, the selection of drought-tolerant crops, and the formulation of drought mitigation and adaptation policies in semi-arid regions such as Rote Ndao Regency.

ABSTRACT. The Tienshan Mountains of Central Asia, a key region in global arid and semi-arid zones, faces highly uneven precipitation distribution due to its unique topography and climate. While extreme heavy precipitation has been widely studied, research on extreme light precipitation is limited. Additionally, spatial distribution patterns and driving mechanisms of extreme events under varying climatic and geomorphic conditions remain underexplored. This study systematically examines the spatial-temporal trends of extreme hydro-climatic events, focusing on both extreme heavy and light precipitation, to provide insights for water resource management and disaster prevention. A distinct hydrological regime shift has occurred since 2000. The frequency anomaly of extreme light precipitation events (R1p) plunged from positive to negative, indicating a marked decline, whereas extreme heavy precipitation events (R99p) surged, reflecting a substantial increase in frequency. Spatially, a prominent dipole pattern is identified around 80°E, where extreme heavy precipitation frequency increases eastward and decreases westward. Vertically, the mid-altitude zone acts as an amplification center, exhibiting the sharpest intensification of heavy precipitation and the steepest decline in light precipitation frequency. These patterns result from the combined effects of Tibetan Plateau thermal dynamics and monsoon-driven moisture transport, creating distinct differences in extreme precipitation between the eastern and western Tienshan. Future studies should explore the interactions between the plateau and atmospheric circulation to improve the prediction and mitigation of extreme events, aiding water resource management and disaster preparedness.

ABSTRACT. Glacier- and snow-dominated river basins in arid Central Asia are highly sensitive to climate change, yet large uncertainties remain in simulating cryospheric processes and projecting future runoff. This study integrates process-based hydrological modeling and data-driven forecasting to improve the understanding of glacier–snow–runoff dynamics in the Tarim River Basin, the largest inland river systems in China. First, a distributed hydrological model, SWAT–Glacier–SRD, was developed and applied to the Kumarik River catchment, the largest tributary of the Tarim River. An empirical snow redistribution module was incorporated into the conventional SWAT–Glacier framework to explicitly represent terrain-controlled snow transport processes. Snow redistribution was parameterized as a function of slope and snow density under threshold conditions. The model was calibrated using multiple observational constraints, including streamflow, glacier area change, glacier mass balance, and snow-related indices. The enhanced model substantially improved the simulation of spring streamflow and cryospheric components, achieving a Kling–Gupta Efficiency of 0.82 during calibration and 0.57–0.91 during validation. Long-term simulations indicated a mean glacier accumulation rate of 0.573 m w.e. and an ablation rate of 0.850 m w.e. over 1975–2019, with a noticeable slowdown in glacier mass loss after the late 1990s. Second, future runoff variations for seven glacier-fed headwaters of the Tarim River were projected using an attention-enhanced Long Short-Term Memory (LSTM) model. The deep learning framework was designed to provide robust long-term runoff projections under multiple climate scenarios. Results indicate sustained increases in annual and seasonal runoff throughout the 21st century under medium- and high-emission scenarios, while changes remain limited under low-emission conditions.

ABSTRACT. This study explores the application of a bias-aware adjustment framework to enhance the usability of Singapore’s Third National Climate Change Study (V3) dataset, which delivers dynamically downscaled projections with a spatial resolution of 2 km. To support the transition of high-resolution climate projections into local impact assessments, five CMIP6 global climate models (GCMs) were considered and subsequently adjusted through the Quantile Delta Mapping (QDM) approach. The analysis focused on four critical variables, temperature, precipitation, relative humidity, and wind speed, at both daily and hourly temporal scales across five meteorological stations. This study applies site-level (station-based) bias correction for direct comparison with historical observations. Initial comparisons indicate systematic differences between model outputs and observations, especially for precipitation (at hourly scale) and wind speed (at both scales), with Mean Absolute Percentage Error (MAPE) values exceeding approximately 29%. The application of the QDM method effectively reduced these differences, bringing key statistical indicators, including mean, standard deviation, and skewness, to closely align with observed characteristics across all months and stations. Furthermore, the QDM approach maintains the underlying climate change signal. By aligning high-resolution projections with site-specific observational benchmarks, this work provides a consistent framework for supporting local-scale climate impact studies, including applications in hydrological analysis, extreme event assessment, and sectoral risk evaluation.

ABSTRACT. The Water-Energy Nexus (WEN) has been widely employed as a resource management framework to address sustainability challenges across various water and energy sectors from generation to distribution. While effective for analyzing their synergies and trade-offs as integrated systems, the nexus perspective is limited in explaining how the systems may fail under extreme weather events due to their interdependency. Service disruptions during disasters are rarely driven by resource shortages but often by infrastructure interdependency, where the operational failure of one system directly constrains the functionality of another. In urban environments, water supply systems are especially vulnerable because their core operations such as abstraction, treatment, pumping, and distribution are highly dependent on electricity supply. This distinction highlights a fundamental gap between nexus perspective aiming resource coordination efficiency and interdependency perspective aiming systems operational resilience. This study advances an infrastructure interdependency framework and examines possible cascading failures between power distribution networks and water supply systems under extreme climate conditions. Focusing on the Metropolitan Area, coupled synthetic water–power network is developed using publicly available geographic and statistical data, addressing data availability constraints while preserving the topological characteristics of real-world infrastructure systems. The power grid vulnerability is evaluated using graph-theoretic indicators, including articulation points, betweenness centrality, and clustering coefficients, which capture structural mechanisms that amplify cascading disruptions. To enhance resilience within this interdependent system, a hierarchical multi-objective greedy algorithm is proposed to optimize the spatial deployment of Distributed Generation (DG). The algorithm prioritizes maintaining power supply to critical water infrastructure while simultaneously mitigating structural vulnerabilities in the power network. Simulation results show that deploying 92 DER units eliminates 11 critical articulation points, reduces average betweenness centrality by approximately 20%, and increases clustering coefficients by 5.9%. Under an extreme disaster scenario involving a 30% power grid disruption, the operational continuity of water facilities improves by 7.7% compared to the baseline system. By explicitly distinguishing infrastructure interdependency from the water-energy nexus, this study translates interdependency theory into a quantitative, network-based resilience strategy. The findings demonstrate that decentralized, safe-to-fail energy configurations are essential for securing water services under climate uncertainty, offering actionable insights for interdependency-aware infrastructure planning.

ABSTRACT. Droughts that develop gradually are among the most critical climate-induced hazards, characterized by strong spatial and temporal heterogeneity and persistent impacts on agricultural production and water availability. Concurrently, the increasing frequency of short-duration, high-intensity rainfall events has amplified variability in agricultural water supply, underscoring the need to enhance the performance and adaptability of irrigation systems in rural regions. In reservoir-based paddy irrigation districts, a portion of delivered water re-enters the canal network through surface runoff and subsurface percolation. Although this return flow represents a potentially valuable secondary water source, it has received limited quantitative attention in conventional irrigation assessments. This study quantifies return flow generation and evaluates water reuse potential through an integrated water cycle analysis using the U.S. Environmental Protection Agency’s Storm Water Management Model (EPA-SWMM). A detailed hydraulic representation of an agricultural irrigation network was developed, explicitly incorporating control structures, channel geometry, longitudinal slopes, and drainage facilities. The model links water sources, conveyance canals, and irrigated areas to simulate the spatial routing and redistribution of irrigation water. Daily water balance components—including irrigation supply, precipitation, drainage discharge, evapotranspiration, infiltration, and storage variation—were analyzed to estimate return flow dynamics and associated reuse volumes. The analysis further examines how network configuration and canal connectivity influence the magnitude and temporal patterns of reusable water within the irrigation district. The results provide essential baseline information for managing agricultural water resources from a system-wide water cycle perspective. Moreover, the proposed modeling framework supports the development of effective water allocation strategies and operational guidelines under drought and water-stressed conditions, thereby contributing to the long-term sustainability and resilience of irrigation systems.

This work was supported by Korea Institute of Planning and Evaluation for Technology in Food, Agriculture, and Forestry (IPET) through Intelligent Agricultural Infra Management for Climate Change Development Program, funded by Ministry of Agriculture, Food and Rural Affairs (MAFRA) (RS-2025-02219764)

ABSTRACT. Dam construction as a National Strategic Project involves a high level of complexity, making it vulnerable to corruption risks throughout all stages of the project lifecycle. Although various control measures have been implemented, there remain limitations in systematic approaches to identifying, predicting, and mitigating corruption risks in an integrated manner within the existing risk management framework. Therefore, this study aims to identify, predict, and mitigate corruption risks in dam construction projects using a Corruption Risk Assessment (CRA) approach integrated with the risk management framework in accordance with Circular Letter of the Minister of Public Works and Housing No. 12/SE/M/2024.

This study employs a quantitative event-based analysis by examining project documents, including supervision reports, internal audit findings, cost and schedule deviation data, and records of contract variations (variation orders). Corruption risk identification is conducted across all project phases, namely planning, procurement, construction implementation, and supervision. Risk assessment is carried out using a risk scoring model that combines the probability of occurrence with financial and operational impacts, resulting in measurable risk levels. Mitigation strategies are formulated through strengthening the role of the Risk Management Unit, implementing a Loss Event Database, applying tiered risk escalation mechanisms, and adopting risk-based supervision.

The results indicate that the highest levels of corruption risk occur during the procurement of goods and services, contract variations, and quality control in on-site construction activities. The dominant factors influencing these risks include weak internal controls, low accountability, and pressure to achieve physical project completion targets. The application of an integrated Corruption Risk Assessment (CRA) approach within the risk management framework has proven effective in reducing the risk level of most critical events from high to moderate. Consequently, this study provides both practical and theoretical contributions to the development of a regulation-based corruption risk mitigation model applicable to strategic infrastructure projects

ABSTRACT. Water security is a critical component in achieving sustainable development and supporting the national vision Indonesia Emas 2045. The Opak River Basin is highly vulnerable to volcanic disturbances, as most of its rivers originate from Mount Merapi in Yogyakarta, Indonesia. These rivers play a strategic role in supplying water for irrigation, domestic use, industry, and environmental flows. However, the basin has experienced severe water stress since 2019, and according to Indonesia's Vision 2045 Towards Water Security projection this trend is expected to be worsen by 2045 due to population growth, economic expansion, and climate change. This study aims to assess the carrying capacity of water resources in the Opak River Basin by 2045 in relation to projected demand, ensuring equal distribution of water allocation and sustainable management through water availability and demand balance. In accordance with Indonesian Law Number 17 of 2019 on Water Resources explain that water allocation prioritises. In situations of scarcity, these legal rules establish governing strategies. Key challenges to be anticipated include water shortages, baseline availability and consumption, and projected sectoral demands for irrigation, industry, domestic use, and environmental flow. The methodological approach will be combining the water availability projection, water balance analysis, and climate change scenarios to evaluate resources capacity. Current water availability is estimated at 40.05 m³/s (3.46 million m³/year), with demands distributed between irrigation (3.66 m³/s across 18 weirs) and non-irrigation users (0.83 m³/s across 6 intakes). The increasing stress indicates the need for comprehensive conservation techniques and adaptive water allocation to be implemented, especially during dry seasons, despite a largely balanced supply-demand ratio. Projections suggest that by 2045, climate change will substantially modify local hydrological patterns, leading to increased variability and potential reductions in reliable water availability. Moreover, volcanic eruptions from active volcano introduce extreme disruptions which lahars can block river channels and damage weirs for up to a year, halting water allocation. Along with variably conditions happen when these factors are synthesized through the water balance analysis, the findings demonstrate a clear and critical imbalance, revealing that the water availability in 2045 will be insufficient to meet the total projected demand. Ultimately, the study concludes that the water resource carrying capacity of the Opak River Basin is under severe strain, highlighting an urgent need for adaptive management strategies, infrastructure optimization, and integrated conservation policies to mitigate future deficits and ensure sustainable water security for the region.

ABSTRACT. Global irrigation models (GIM) estimate irrigation water withdrawals (IWW) based on a series of assumptions on crop distribution, water availability and irrigation practices. These estimates inform assessments of global water scarcity, agricultural production and future water demand, making it crucial to characterise their uncertainty. While uncertainty analyses typically focus on parameters and numerical inputs, many influential assumptions and modelling choices are qualitative and pragmatic in nature, resisting numerification, and therefore they escape conventional uncertainty and sensitivity analyses.

We address this gap using a post-normal science framework that evaluates the epistemic quality and influence of modelling assumptions. We conducted an elicitation workshop with an extended peer community comprising eleven irrigation scientists and five irrigators from traditional irrigation systems in Spain. Participants ranked the most influential assumptions from an inventory of 100 assumptions identified across nine GIMs. The ten highest-ranked assumptions were then assessed using a pedigree matrix that evaluates situational limitations, plausibility, availability of alternatives, peer agreement and perceived influence on model outputs.

Participants identified assumptions related to irrigated area datasets, irrigation efficiency and water availability as dominant drivers of global IWW estimates. These assumptions are implemented using single datasets or uniform representations despite the existence of credible alternatives. Pedigree analysis reveals that most highly influential assumptions rest on weak epistemic foundations characterised by limited empirical support, strong simplifications, low agreement or derivation under practical constraints. When mapped in a diagnostic diagram, these assumptions cluster in a "danger zone" defined by high influence and low pedigree (i.e., weak knowledge base).

Our findings show that global irrigation estimates are shaped by influential assumptions that remain largely unexamined. We demonstrate that involving an extended peer community provides a holistic means to identify and appraise consequential modelling choices. This approach complements stochastic uncertainty and sensitivity analyses and supports more transparent and policy-relevant use of global irrigation models under conditions of deep uncertainty.

ABSTRACT. Hydrological variability in river catchments is significantly influenced by land-use change, especially in areas experiencing rapid urbanisation and agricultural development. The Tukituki Catchment is one of the region's largest catchments and has significant changes in land-use transformation over recent decades, most notably the growth of urban areas, dairy farming, plantation forestry, and urban areas. This study proposes to evaluate the impacts of land-use change on river flow regimes in the catchment using the Soil and Water Assessment Tool (SWAT), a semi-distributed, process-based hydrological model.

The Soil and Water Assessment Tool (SWAT) model will be set up using high-resolution climate data obtained from NIWA, land-use information from the New Zealand Land Cover Database (LCDB), soil data from S-map, and topographic data derived from a digital elevation model. The model will be calibrated and validated using observed streamflow records at selected gauging stations within the catchment. The impacts of Land-use change scenarios on important hydrological components such as surface runoff, baseflow, peak discharge, and seasonal flow patterns, will be measured by stimulating historical and contemporary recent land-cover conditions.

The proposed analysis is expected to reveal that agricultural intensification and urban expansion increase surface runoff and peak flows, while reducing baseflow during dry periods. In contrast, afforestation may reduce mean annual flow due to increased evapotranspiration. In general, it is predicted that changes in land use will increase flow variability and modify the Tukituki River system's natural flow regime.

This study will provide how the SWAT model may be used as a decision-support tool for sustainable land and water resource management in New Zealand and offer insights information on the hydrological impact land-use change.

ABSTRACT. Integrated water resources management (IWRM) is increasingly recognized as a critical framework for balancing water use and ecosystem protection in arid inland river basins. This study focuses on the Haba River Basin, an arid inland river system, to investigate the impacts of IWRM on ecosystem services (ESs) and to elucidate the spatial heterogeneity mechanisms underlying shifts in ES trade-offs and synergies. By analyzing land-use change characteristics and the evolution of ES interactions, we assess the role of basin-scale management policies in coordinating ecological conservation and water resource utilization.

The results reveal distinct developmental stages associated with changing water and land management priorities. During the initial land development stage (1990–2000), 10.65% of the basin experienced land-use change. In the intensified development period (2000–2010), driven largely by agricultural expansion and water reallocation, 30.29% of the land underwent substantial transformation, with approximately 78% of grassland, sparse grassland, forest land, and desert converted into arable land, leading to pronounced trade-offs among ecosystem services. From 2010 to 2020, as water resource management shifted toward ecological restoration and vegetation recovery under the IWRM framework, land development intensity declined markedly, and only 3.65% of the total area exhibited land-use change.

Spatial heterogeneity analysis covering 1990–2020 indicates a gradual transition in ecosystem service relationships from trade-offs to synergies between the periods 2000–2010 and 2010–2020. The years 2010 and 2020 emerged as critical temporal nodes for achieving synergy enhancement and trade-off reduction. Net primary productivity (NPP) was identified as a key driving factor of comprehensive ecosystem service (CES) performance, reflecting the integrated effects of water availability, vegetation restoration, and land management regulation.

By integrating ES trade-off–synergy relationships with the spatial identification of ecosystem function hotspots and embedding them within a unified water resources management policy framework, this study proposes zoned and differentiated ecological management measures tailored to regional conditions. These measures have facilitated the effective implementation of ecological protection policies in arid environments, promoting a shift in ecosystem service development from trade-offs to synergies and enhancing overall ES functionality in the Haba River Basin. The findings provide robust scientific evidence and practical guidance for sustainable ecosystem management under integrated water resources management, particularly in arid and semi-arid regions of Central Asia.

ABSTRACT. In recent years, balancing flood risk reduction and the conservation/restoration of river ecosystems has become a critical challenge in river management worldwide under accelerating climate change. In Japan, a series of large-scale flood events since the late 2010s has exposed the limitations of conventional flood control approaches that relied primarily on river channel improvements, serving as a catalyst for a policy shift toward integrated, basin-scale flood management. At the same time, given the limited availability of land space, large-scale river improvement works remain unavoidable, raising concerns about the degradation of river-specific habitats such as floodplain vegetation and water body, gravel bars, and riparian woodlands. Under the circumstances , Japan has begun to incorporate the concept of Nature Positive into river policy following the adoption of the Kunming–Montreal Global Biodiversity Framework in 2022 and the positioning of the “30 by 30” target as a national goal. A key challenge in implementing this policy is the establishment of quantitative river environmental targets that can be embedded in statutory river planning processes. This paper reports on the framework and current practices for setting quantitative environmental targets being implemented across 109 nationally managed river systems in Japan. The approach is based on two nationwide and long-standing monitoring and evaluation schemes: the Nationwide Census on River Environments, conducted continuously for nearly 40 years, and the River Environment Management Sheet, which organizes habitat area and relative environmental conditions at 1-km river segment scales. By integrating these datasets, declining habitat types and species requiring conservation and recovery are identified, including not only legally designated threatened species but also formerly common species that have shown long-term population declines. Quantitative environmental targets are primarily defined in terms of habitat extent. In addition, spatial connectivity is considered as a key indicator, including longitudinal connectivity of aquatic habitats important for migratory species and lateral connectivity between in-channel and floodplain water bodies important for floodplain-dependent species. Target values are established by referencing past favorable environmental conditions as well as the potential habitat areas that can be created through river improvement works. Progress toward nature-positive outcomes is evaluated through comparison with defined baseline conditions. These efforts are advanced within Japan’s multi-layered river governance structure. This paper synthesizes institutional framework features, technical challenges, and insights gained from early implementation, and discusses how quantitative environmental targets can function as a practical tool to bridge biodiversity policy and large-scale flood management projects.

ABSTRACT. Water ecological compensation is an important instrument for compensating the conservation cost in water source basin of water diversion projects, and for promoting watershed and regional sustainable development. Under the construction of networked diversion system in China, difficulties are brought to define compensation relationships and to calculate compensation standards by the increasingly complicated relationships of import and export water diversion projects and the overlapping of the water diversion source area and the impacted area. The theory of water value flow under water diversion network is studied by analysis on water balance and water value transfer. Based on the principle of internalization of external effects, an ecological compensation standard calculation model for networked water diversion based on water resource value flow is proposed. The model includes the calculation of conservation cost and losses in water source basin, construction of coefficient matrix for water quantity allocation for water diversion from upstream source area, and for impact contribution on lower river reaches by diversion projects, based on which the compensation cost by each diversion can be calculated. Taking the Han River Basin as the case study, the water receiving areas of the Yinhanjiwei Water Diversion Project, the first phase of the South to North Water Diversion Project, and the North Hubei Water Allocation Project respectively bear annual compensation amounts of 659 million yuan, 6.577 billion yuan, and 533 million yuan. Meanwhile, the areas of Hanzhong, Ankang, Shangluo, Shiyan, Nanyang, and the area below Danjiangkou Reservoir should receive compensation of 1.072 billion yuan, 1.286 billion yuan, 876 million yuan, 1.882 billion yuan, 946 million yuan, and 1.707 billion yuan respectively annually. This study can provide theoretical and technical support for the determination of compensation standard under the water diversion network.

ABSTRACT. Vegetation in river channels plays a critical role in controlling flow resistance, sediment transport, and habitat structure, and therefore constitutes a key factor in achieving ecologically healthy and sustainable river systems. However, vegetation characteristics vary dynamically with flow conditions, and neglecting these variations in hydraulic modeling may lead to significant uncertainty in water-level prediction and flood risk assessment. This study evaluates the applicability of dynamic roughness coefficients that account for vegetation-state transitions to improve hydraulic simulation reliability for sustainable river management.

Three roughness estimation approaches were compared in the downstream reach of Senaegyo Bridge within the Samcheon Basin: (CASE 1) roughness coefficients adopted from the river master plan, (CASE 2) roughness derived from the stage–discharge relationship, and (CASE 3) roughness calculated using a vegetation resistance formula that explicitly incorporates vegetation-state transitions (erect, waving, and prone). Probable flood discharges were generated using the HEC-HMS hydrologic model and applied to unsteady flow simulations in HEC-RAS under two hydrologic scenarios: a historically extreme rainfall event and an average rainfall condition. Water-level variations associated with vegetation-state changes were analyzed for each case.

Results indicate that under the extreme rainfall scenario, CASE 2 (second prone stage) and CASE 3 (prone stage) exhibited the highest agreement with observed water levels, reflecting the importance of accounting for vegetation lodging under high-flow conditions. Under the average rainfall condition, CASE 1 and CASE 2 demonstrated stable performance across erect, waving, and prone vegetation states. These findings confirm that applying dynamic roughness coefficients reflecting actual vegetation behavior significantly enhances the reliability of hydraulic simulations.

By improving the representation of vegetation–flow interactions, the proposed approach supports more accurate flood prediction and environmentally sound river design. Ultimately, this study provides a practical basis for integrating ecological characteristics into hydraulic modeling, contributing to sustainable river management and the long-term maintenance of ecologically healthy river systems.

ABSTRACT. This study aims to implement “Collaborative Nature Restoration in rivers” in society. This approach involves various stakeholders who, through their own ideas and collaboration, engage in hands-on, nature restoration activities at familiar waterside locations while enjoying the process. It emphasizes practical actions that can be initiated with limited resources, shared responsibility, and flexible methods that allow modification or removal when necessary.

Such collaborative practices have become increasingly relevant in recent years in Japan, where river environments face growing challenges, including biodiversity loss, climate change impacts, aging infrastructure, and constraints in administrative capacity. At the same time, national policies have begun to emphasize nature-positive outcomes and the use of green infrastructure as part of river and watershed management. These trends highlight the need for restoration approaches that can be implemented locally and continuously by diverse actors.

To support this, this study collected and analyzed 69 cases of collaborative nature restoration from across Japan, using published casebooks, records of field training seminars, and additional interviews with practitioners. In addition, on-site investigations were conducted at ten locations, including both advanced practices and newly planned initiatives, with the cooperation of local implementing groups. Through these analyses and field surveys, practical knowledge on goal setting, site selection, applicable methods, implementation frameworks, stakeholder coordination, and monitoring was systematically organized and compiled.

Based on the findings, the study compiled the accumulated experiences into a structured set of practical knowledge and reference materials, rather than prescriptive guidelines, to support future practitioners.

ABSTRACT. A multi-scenario simulation and optimization framework for recycled water distribution systems serving river–lake ecological replenishment was developed using EPANET based on the operational conditions of the Qinglonghu Recycled Water Plant. The hydraulic performance and residual chlorine decay characteristics of the network were systematically analyzed under different ecological water supply scenarios. By coupling EPANET with the Non-Dominated Sorting Genetic Algorithm II (NSGA-II), key operational parameters, including pump station pressure and valve opening, were optimized with residual chlorine concentration at ecological replenishment nodes as the primary control objective. The results demonstrate that the proposed approach effectively coordinates water quantity regulation and water quality control under ecological constraints and significantly improves the compliance of residual chlorine concentrations with ecological safety thresholds while satisfying water supply demands across multiple scenarios. This study provides a quantitative and transferable methodological basis for the ecological operation and management of recycled water distribution networks and supports the safe application of recycled water in river and lake restoration and environmental flow enhancement projects.

ABSTRACT. Water supply networks are a critical urban infrastructure that directly influences public health, economic stability, and the sustainability of natural water systems. In South Korea, a large portion of these networks was constructed during the 1970s and 1980s and is now rapidly aging. As a result, pipeline leakage and corrosion have become major sources of non-revenue water, generating substantial economic losses and increasing unnecessary abstraction of raw water from rivers, which in turn places additional stress on aquatic ecosystems. Improving the efficiency and reliability of water distribution systems is therefore essential for achieving ecologically healthy rivers and sustainable social development.

This study proposes an AIoT-based Buried Pipelines Integrated Management System (ATIMS) that integrates multi-purpose Internet of Things (IoT) sensors and machine learning techniques to enable real-time detection, prediction, and localization of anomalies in underground water supply networks. The proposed framework represents an end-to-end integrated system in which sensing, data processing, model inference, and visualization are seamlessly connected, supporting a transition from reactive maintenance toward predictive and preventive management.

For fundamental data acquisition, four key soil parameters governing the corrosion environment around buried pipes—soil resistivity, water content, pH, and oxidation–reduction potential (ORP)—were selected. A durable multi-purpose IoT sensor capable of continuously measuring these parameters under underground conditions was developed and equipped with an LTE communication module for real-time data transmission. To construct a high-quality training dataset, a hybrid data strategy combining field-measured and statistically processed data was adopted. Preprocessing procedures, including Box–Cox transformation, were applied to reduce sensor bias and skewness and to enhance statistical robustness.

Based on the refined dataset, a deep multi-layer perceptron (DMLP) model was trained to learn the complex non-linear relationships between soil characteristics and pipeline risk states. Verification results demonstrate excellent predictive performance, with a coefficient of determination (R²) of 0.99 for leak status prediction, 0.82 for leak localization, 0.92 for corrosion thickness prediction, and 0.93 for corrosion rate prediction. These results confirm the effectiveness of the proposed algorithms for early-stage anomaly diagnosis.

By minimizing water losses and preventing unexpected failures, the proposed ATIMS contributes to reducing unnecessary water withdrawal from rivers, thereby supporting ecological river health while ensuring a stable and resilient urban water supply. Ultimately, this study presents a practical technological pathway toward sustainable water infrastructure management that aligns engineering innovation with environmental protection and long-term societal sustainability.

ABSTRACT. South Korea has experienced pronounced hydro-climatic shifts over the past five decades, including a 1.5°C increase in mean temperature since 1970 and a growing frequency of extreme rainfall events exceeding 30 mm hr⁻¹. These intensifying climate extremes have exposed structural and operational limitations in conventional gray infrastructure, particularly in urban flood control and water quality management. In response, this study evaluated the effectiveness and policy integration of Nature-based Solutions (NbS) as adaptive strategies to enhance water circulation and climate resilience in urban areas. A multi-layered methodological framework was employed, combining long-term climate trend analysis (1970–2024) with empirical evaluation of representative NbS implementation cases across South Korea. Selected case studies included ecological stream restoration, constructed wetlands integrated with wastewater treatment plants, agricultural non-point source management systems, and urban Low Impact Development (LID) technologies. Performance was assessed across four domains: (1) water quality improvement, (2) flood mitigation capacity, (3) carbon sequestration potential, and (4) economic feasibility relative to gray infrastructure alternatives. Results demonstrate substantial multi-functional benefits of NbS. Constructed wetlands integrated with secondary wastewater treatment achieved 67% biochemical oxygen demand (BOD) removal and 62.8% total phosphorus (T-P) reduction, while reducing energy consumption by more than 90% compared to membrane-based systems. Urban LID facilities, including bioretention cells, tree box filters, infiltration trenches, and hybrid wetlands, achieved peak flow reductions of 75–80% during extreme rainfall events, demonstrating strong hydraulic buffering capacity. River-scale NbS interventions implementing “Room for the River” strategies reduced flood velocities by up to 2 m s⁻¹ in critical basins. Carbon accumulation rates ranged from 0.15 to 2.2 kg C m⁻² yr⁻¹ across wetland types, indicating meaningful climate mitigation co-benefits. Economic analysis further revealed that NbS-based flood management approaches were 76% more cost-effective than conventional pipe replacement strategies. These findings confirm that NbS function not merely as supplementary environmental measures but as integrated, cost-efficient, and climate-adaptive infrastructure systems. Embedding NbS within national water management policy, through performance standardization, cross-sectoral coordination, and hybrid gray-green design frameworks, was found essential for advancing long-term resilience under accelerating climate extremes.

This research was supported by the National Research Foundation of Korea(NRF) funded by the Ministry of Science and ICT (No.RS-2023-00277793).

ABSTRACT. Climate change is intensifying extreme rainfall events in tropical regions, challenging conventional water management approaches that have primarily focused on urban drainage while overlooking catchment-scale processes. In the Philippines, rainfall variability has been governed by four distinct climate types, ranging from pronounced wet–dry seasonal patterns to persistently high rainfall regimes. When combined with heterogeneous land-use transitions from upstream natural and agricultural areas to downstream urbanized environments, these climatic conditions have created complex challenges in managing both the quantity and quality of stormwater runoff during extreme weather events.

The Rainfall and Stormwater Runoff Management Technology for Tropical Catchments (RainS-TECH) project was presented as an integrated, nature-based water circulation framework that strengthened hydrologic connectivity and climate resilience from upstream to downstream catchment systems. Rather than treating flooding and water quality degradation as isolated urban problems, a holistic, watershed-scale approach was adopted, linking land-use change, sediment mobilization, and pollutant transport across natural, agricultural, and built environments under extreme rainfall conditions.

The methodological framework was structured using a multi-line-of-defense strategy across pre-developed, actively developing, and post-developed catchment conditions. Continuous dry- and wet-weather monitoring of rainfall, runoff, streamflow, sediment accumulation, and water quality was conducted and combined with empirical analysis and hydrologic modeling to characterize runoff generation, sediment transport, and pollutant build-up and wash-off processes under varying climate and land-use settings. These analyses were used to inform resilience-oriented design and performance evaluation under extreme rainfall scenarios.

Nature-based stormwater management was operationalized through the implementation of three complementary decentralized technologies across varying land-use contexts: rainwater harvesting systems with controlled reuse, bioretention systems enhanced with sediment and litter interception, and infiltration-based systems incorporating detention and regulated discharge. These blue–green infrastructure and low-impact development solutions were designed to promote stormwater retention, infiltration, temporary storage, and delayed release, thereby restoring disrupted hydrologic pathways while mitigating both peak flows and pollutant loads during high-intensity storm events. Lab-scale development and pilot-scale implementation were supported by smart monitoring and a web-based data platform, enabling performance evaluation, adaptive management, and resilience assessment.

By translating sponge city principles into a catchment-based, climate-responsive methodology, transferable design and policy insights were generated to support the development of climate-resilient water circulation strategies in Southeast Asia and other monsoon-influenced regions.

ABSTRACT. As a nature-based technique, floating constructed wetland has been widely used for the treatment of various wastewater. The pollutants removal in the floating wetlands had been frequently investigated, while the microbial community structure responsible for the function was rarely studied. This study built a pilot-scale floating wetland used for the purification of a waterlogged area caused by coal mining. The wetland was assembled with artificial fiber and ceramsite as substrate to promote the treatment. With the operation of one year, the wetland performed average removal rates of 32.4%, 25.5%, 23.6%, and 37.1%, for turbidity, BOD₅, TN, and TP, respectively. In terms of microbial community structure, the water under the floating island exhibited a high similarity to the water outside, except for the slightly higher Alpha diversity index. However, the substrate showed significantly higher OTU richness, Shannon diversity and phylogenetic diversity (PD) than the water. Bray-Curtis and UniFrac analyses showed that the microorganisms on the substrate were clearly different from the water, indicating the substrate harbored a "hotspot" responsible for the biological processes. The results also indicated that a stable nitrification function represented by Nitrospira could been quickly established with the wetland operation. As a typical nitrite-oxidizing bacterium, the enrichment of Nitrospira indicates that the substrate provided a stable environment for the growth of nitrifying bacteria, which facilitated the conversion of ammonia nitrogen and nitrite. However, the microorganism related to denitrification, organic carbon utilization, and methyl metabolism reflected a gradual succession. The species initially appeared were common organic-degrading bacteria, such as Flavobacterium, Chryseobacterium, and Cloacibacterium. Over time, genera such as Hyphomicrobium, Rhodobacter, and Roseiflexus began to increase. They are typically associated with the metabolism of low-molecular organic compounds like methanol, denitrification, and photosynthetic processes. This succession of microbial community structure suggests that, with the wetland operation, the substrate not only maintained nitrification functions but also gradually formed a microenvironment conducive to denitrification. The OTU number, Shannon index, and PD of ceramsite were slightly higher than those of fiber ropes, reflecting its advantages in supporting a diversified microbial community. In addition, the ceramsite showed higher abundance of Actinobacteria and Planctomycetes, which are beneficial for phosphorus mineralization, polyphosphate, and sedimentation. The high diversity and PD values of ceramsite indicate that it harbored various microbial groups that potentially participate in phosphorus cycling.

ABSTRACT. Urban flooding aggravated by climate change calls for more adaptive and cost-efficient stormwater management strategies. While green‑grey infrastructure (GGI) and real‑time control (RTC) are established structural and non‑structural measures for flood mitigation, their combined effect under future climate uncertainty remains underexplored. This study aims at evaluating the effect of integrating RTC with GGI optimization on the cost‑effectiveness and climate adaptability of urban stormwater systems, addressing whether RTC can enhance the environmental‑economic performance of GGI under extreme rainfalls and how it reshapes the cost‑benefit contribution across infrastructure types. A simulation‑optimization framework integrating multi‑climate scenario analysis (SSP126, SSP245, SSP585 for 2030–2050) with GGI layout and RTC operation design was developed. This framework couples the Storm Water Management Model (SWMM) with different control strategies to simulate system performance, while the NSGA‑II algorithm generates Pareto optimal trade‑offs between minimized life cycle cost and maximized flood peak reduction. A case study in Shenzhen, China demonstrates that RTC significantly improves the cost‑benefit relationship of GGI. Under baseline climate, RTC reduced life cycle costs by 32%~47% compared to static control for equivalent peak flow reduction (20%~60%). In future climates, RTC lowered the additional opportunity cost of maintaining baseline performance by 50%~100%, and even under 100‑year extreme storms it provided 14%~18% higher benefits without extra cost. RTC also shifted optimal layouts toward greater use of grey storage, alleviating reliance on scarce urban space for green infrastructure while preserving system‑wide benefits. These findings highlight that integrating real-time control with coordinated green-grey infrastructure planning provides a more adaptive, cost‑effective, and spatially feasible pathway for urban flood risk mitigation under climate uncertainty. The proposed framework offers a practical tool for designing resilient stormwater systems that balance economic efficiency with environmental performance.

ABSTRACT. In response to the intensification of climate change and extreme weather events, sponge city strategies and nature-based urban water circulation systems have emerged as key approaches for strengthening urban resilience. However, despite the expanding deployment of green infrastructure (GI) products, standardized Measurement, Reporting, and Verification (MRV) frameworks capable of scientifically validating carbon mitigation and climate adaptation performance—and translating these outcomes into quantifiable metrics—remain insufficiently developed. This study proposes an integrated MRV framework for moss-based green infrastructure using a highly controlled and reproducible U-Ecotron (Urban Ecotron) chamber. The Urban Ecotron serves as an experimental platform capable of repeatedly simulating complex urban climate scenarios, including heatwaves, elevated CO₂ concentrations, and varying rainfall intensities. Such controlled environmental conditions ensure reliability, comparability, and reproducibility in evaluating product performance under realistic urban climate stresses. Through controlled experiments, we quantified: (1) direct carbon removal based on net CO₂ uptake and carbon flux dynamics; (2) emission reductions resulting from decreased cooling energy demand driven by evapotranspiration and surface temperature mitigation; and (3) potential emission avoidance associated with stormwater runoff reduction and alleviation of urban infrastructure burdens. Notably, climate adaptation effects were converted into carbon-equivalent values using energy consumption models and emission factors, enabling an integrated assessment of both mitigation and adaptation performance. Based on these results, we propose an MRV framework aligned with ISO 14068-1, categorizing quantified outcomes into three carbon credit types: Avoidance, Reduction, and Removal. This research establishes an experimental and methodological foundation for carbon-based verification of nature-based green infrastructure and contributes to the scientific advancement of sponge city strategies and performance-based climate finance mechanisms.

Acknowledgement: This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT) (No. RS-2023-00259403).

ABSTRACT. Climate change has increased the frequency and intensity of flash floods, highlighting the need for real-time flood monitoring systems. CCTV-based visual monitoring, combined with deep learning image segmentation, has emerged as a promising approach for automated flood management. However, most studies validate segmentation models on curated datasets or footage from limited sites (1-2 locations), leaving model transferability across diverse real-world monitoring conditions underexplored. Furthermore, while task-specific architectures such as U-Net have traditionally been used for water segmentation, foundation models now offer an alternative with broader generalization potential, yet comparative evaluation of their cross-site transferability remains limited. To address this gap, this study evaluates model transferability across multiple monitoring sites through a two-stage validation framework. First, models trained on static benchmark images via Bayesian optimization were evaluated on four pixel-wise annotated video sites representing diverse monitoring conditions to quantify transferability from static training data to real-time video streams. Second, we validated performance on 30 videos equally distributed across flood warning levels, leveraging monocular depth estimation to establish geometric correspondence between segmentation masks and recorded water levels, enabling statistical assessment of flood state classification against official warnings. Results quantify foundation model advantages in cross-site transferability and demonstrate depth-calibrated segmentation as a viable approach for automated flood warning systems. Performance evaluation across diverse monitoring conditions reveals key factors affecting model reliability, including illumination variability and site-specific geometric characteristics. By validating against flood warning levels with statistically sufficient samples, this study provides a replicable framework for assessing deep learning methods in real-world water monitoring applications, enabling informed model selection for operational deployment.

This work was supported by Korea Environment Industry & Technology Institute(KEITI) through R&D Program for Innovative Flood Protection Technologies against Climate Crisis, funded by Korea Ministry of Climate, Energy, Environment(MCEE)(RS-2023-00218873).

ABSTRACT. Hydrological data are essential for flood and drought forecasting, water resources management, and climate change adaptation. However, outliers frequently occur due to sensor malfunctions, data transmission errors, and external environmental influences. These anomalies can significantly degrade the reliability and accuracy of hydrological models. Conventional outlier detection methods rely heavily on statistical analysis and expert judgment, which are often time-consuming and costly. To overcome these limitations, this study aims to enhance hydrological data quality management by developing AI-based automated outlier detection techniques. In this study, both supervised and unsupervised learning approaches were employed to develop outlier detection models. For supervised learning, an XGBoost-based gradient boosting model was trained using historical labeled data containing normal and anomalous observations. The supervised model achieved an outlier detection accuracy exceeding 95% compared to existing rule-based screening criteria. For unsupervised learning, an Isolation Forest algorithm was applied to detect anomalies by defining confidence intervals based on multiple criteria, including maximum rainfall thresholds, comparisons with neighboring stations, water level variations, and repeated identical water level values. Unlike supervised approaches, the unsupervised method does not require prior labeling of anomalies, making it readily applicable in data-scarce environments, particularly in developing countries. In addition, the detected outliers were corrected using an XGBoost regressor for rainfall data and an LSTM autoencoder–based deep learning model for water level data. A PC-based desktop application implementing hydrological data quality control using the aforementioned AI techniques was successfully developed. This system was validated using rainfall and water level datasets collected from the Yeoju Bridge site in the Republic of Korea and the Nam Ngum River Basin in Lao PDR. This study demonstrates that AI-based automated outlier detection and correction is an effective tool for improving hydrological data quality management. Furthermore, the proposed framework shows strong potential for future extension toward real-time data processing and anomaly cause analysis systems.

ABSTRACT. Accurate container throughput forecasting is essential for port planning, berth allocation, and climate-resilient infrastructure investment. Although deep learning has progressively advanced this domain, from early hybrid Convolutional Neural Network-Long Short-Term Memory (CNN-LSTM) architectures to attention-based and multi-scale mixing models, two significant gaps remain: no systematic evaluation of state-of-the-art neural forecasting architectures has been conducted on Asia-Pacific port throughput data, and existing methods model only historical numerical signals while ignoring the contextual information embedded in port operational announcements, trade policy documents, and climate disruption alerts. This study addresses both gaps through a two-stage framework. The first stage benchmarks five leading neural architectures, including TimeMixer, iTransformer, PatchTST, N-HiTS, and CNN-LSTM, on monthly container throughput records from five major Taiwan ports (Kaohsiung, Taichung, Hualien, Anping, and Suao), evaluated across multiple forecasting horizons using Mean Absolute Percentage Error (MAPE), Root Mean Square Error (RMSE), and Mean Absolute Scaled Error (MASE), to identify the strongest numerical baseline. The second stage augments the best-performing architecture with a Large Language Model (LLM) exogenous semantic encoder that transforms heterogeneous textual signals, including port authority announcements, typhoon disruption alerts, seasonal shipping schedules, and international trade policy updates, into contextual embedding vectors fused with the numerical forecast stream via a cross-modal attention module. To isolate the contribution of semantic context, ablation studies compare forecasting performance during stable operating periods versus climate-disrupted periods. Results demonstrate that the consistent improvement over the best standalone numerical baseline, with semantic fusion contributing most significantly during climate-driven disruption episodes. These findings suggest that integrating contextual semantic intelligence into numerical forecasting offers a practical pathway for building forecast-resilient port infrastructure across the Asia-Pacific region.

ABSTRACT. As part of natural systems, rivers provide water resources for productive life and support regional development. Rapid urbanization has increased pressures on rivers, leading to highly uncertain changes in water quality. Accurate prediction of water quality is critical for early warning and effective management. Current methods are difficult to capture long-term time dependencies and have limited interpretability, reducing practical application reliability. This study proposed an interpretable deep learning framework, i.e., TCN-Attention, for river water quality prediction using temporal convolutional network (TCN) with attention mechanisms. A comprehensive dataset encompassing water quality, meteorological, socio-economic and hydro-topographic features was constructed to describe river conditions. TCN-Attention model was designed to predict six water quality indicators, with Bayesian optimization to effectively improve the model robustness. Compared with TCN, TCN-Attention achieved better results by adaptively emphasizing critical time steps, which reduced information loss and improved sensitivity to water quality mutations. The proposed model had optimal prediction accuracy in Xinwu District case study, with mean absolute error (MAE) and root mean squared error (RMSE) of 0.5155 and 0.6782, respectively. To enhance model interpretability, Shapley additive explanation (SHAP) technique was applied to quantify the contribution of each driving factor. Results indicate that domestic water use, turbidity and water body have greatest contribution to model predictions. Domestic water use reflected the intensity of urban wastewater discharge, turbidity captured suspended sediment changes that influenced light penetration and pollutant transport, and water body represented the regulation of surface water extent and connectivity, all of which affected dilution capacity and material exchange of river systems. The findings provided a physical perspective for understanding main drivers of water quality dynamics. The proposed framework offered an effective and interpretable deep learning solution for river water quality prediction to support informed decision-making in urban river management.

ABSTRACT. Accurate prediction of urban flooding is increasingly critical as cities face more frequent and intense rainfall events. However, traditional physics-based hydrodynamic models capable of producing detailed inundation maps demand substantial computational resources, making real-time forecasting impractical for operational use. While deep learning-based super-resolution methods have emerged as promising alternatives to accelerate flood modeling, current approaches primarily enhance spatial resolution without addressing temporal evolution, limiting their utility for dynamic flood forecasting applications. This study presents a data-driven hydroinformatics framework that integrates physics-based hydrodynamic modeling with deep learning to enable rapid, high-resolution spatiotemporal prediction of urban flooding. The framework employs a spatiotemporal downscaling approach that transforms coarse-resolution 1D–2D coupled flood simulations into detailed inundation maps while simultaneously forecasting future flood progression at multiple time steps. By training on sequences of physics-based hydrodynamic simulations, the model learns physically consistent spatiotemporal patterns and propagation characteristics of urban flooding processes. The framework leverages convolutional neural networks to integrate multi-source geospatial data including time-series flood depth outputs from coarse simulations, digital elevation models, and land cover classifications, explicitly capturing complex spatiotemporal dependencies in urban flood dynamics. We demonstrate the framework's applicability and assess its predictive performance through comprehensive case studies in the Oncheon Stream watershed, a highly urbanized catchment in Busan, South Korea. The model is trained and tested across various synthetic and historical rainfall scenarios representing different flood magnitudes and durations. Model outputs are rigorously validated against benchmark high-resolution hydrodynamic simulations using multiple quantitative performance metrics including root mean square error, critical success index, and structural similarity index. Results indicate that the proposed framework generates accurate high-resolution flood predictions, reducing computational time by orders of magnitude compared to traditional physics-based models, while maintaining physical consistency and capturing essential spatiotemporal flood characteristics. This hydroinformatics framework offers a practical and efficient tool for operational flood forecasting systems, effectively bridging the gap between computationally intensive detailed hydrodynamic modeling and real-time urban water management requirements.

ABSTRACT. Using a long-term dataset collected in Minamata Bay, Japan, from 2006 to 2023, this study investigated the environmental patterns associated with variability in the MeHg/THg ratio using an interpretable machine-learning framework integrated with statistical modeling. An XGBoost model was applied to quantify the relative contributions of environmental variables, with model behavior interpreted using SHAP analysis, and generalized additive models (GAM) were used to characterize nonlinear response patterns. After removing pronounced interannual drift through within-year standardization, the model explained a moderate fraction of the variability in the MeHg/THg ratio (R² = 0.43, r = 0.66), indicating that a substantial proportion of variability remains associated with unresolved environmental and biogeochemical processes. Temperature and other hydrographic variables were associated with major response patterns, whereas salinity showed a variable nonlinear response, likely reflecting hydrographic structure. SHAP–GAM analysis identified consistently positive contributions at intermediate temperatures of approximately 22–28 °C. Dissolved oxygen (DO) showed a nonlinear relationship, with negative contributions at moderate levels (~4–7 mg L⁻¹). Total mercury (THg) exhibited a nonlinear response, with higher MeHg/THg ratios occurring primarily at elevated THg levels. Overall, although the model captures only part of the observed variability, the consistency between SHAP-based feature attributions and GAM-derived response curves supports the identification of reproducible environmental response ranges associated with variability in the MeHg/THg ratio.

ABSTRACT. Urban pluvial flood prediction requires fast, street-scale information while remaining physically credible amid strong urban heterogeneity and uncertain forcings. We present a suite of AI-informed hydroinformatics methods that integrate physics-based modeling, real-time observations, and machine learning into a deployable prediction pipeline. Multi-GPU-accelerated hydrodynamic simulation reduces computational latency, enabling high-resolution, city-scale scenario exploration and supporting time-critical decision-making during extreme events. To better leverage sparse and heterogeneous observations (e.g., water-level gauges and camera-derived depth estimates), the workflow incorporates sequential data assimilation (e.g., particle filtering) and multivariate geostatistical data fusion (co-kriging) to translate limited measurements and auxiliary covariates into spatially distributed inundation updates with quantified uncertainty. Data-driven surrogates further complement the physics core by enabling rapid emulation of high-fidelity simulations and super-resolution mapping that bridges scale mismatches between coarse inputs and street-level impact fields. Overall, the framework provides a practical hydroinformatics architecture that balances computational speed, physical consistency, observational influence, and robustness under extreme events, and supports operational modes ranging from rapid what-if screening to near-real-time forecasting.

ABSTRACT. Reliable and timely urban flood forecasting is critical for mitigating flood damage and supporting effective emergency response. However, physics-based hydrodynamic models are inherently subject to uncertainties in initial conditions, model parameters, and meteorological forcings. While ensemble-based data assimilation (DA) offers a robust approach to reducing these uncertainties through real-time observation integration, its application to high-resolution urban hydrodynamic models is often limited by prohibitive computational demands. These challenges are particularly pronounced in complex urban environments, where fine-scale topographic variability and rapidly evolving surface–drainage interactions necessitate both high spatial resolution and frequent model updates. To overcome this limitation, we present an integrated forecasting framework that couples a physics-guided artificial intelligence (AI) emulator with ensemble data assimilation for efficient and reliable spatiotemporal inundation prediction. The physics-guided AI emulator is trained on high-resolution hydrodynamic simulation outputs to reproduce flood dynamics while preserving physical consistency, achieving a significant reduction in computational cost compared to conventional numerical solvers. This computational efficiency enables the operational generation of large ensembles through systematic perturbation of initial conditions and meteorological forcings, facilitating uncertainty quantification. Real-time inundation depth observations from heterogeneous sources (e.g., surveillance cameras, synthetic datasets) are assimilated to continuously update ensemble states and reduce forecast uncertainty under evolving conditions. Additionally, we investigate how data assimilation effectiveness depends on observation availability and spatial configuration, analyzing the impact of sensor density and placement on constraining ensemble spread. The framework is demonstrated for an urban drainage basin in Seoul, South Korea, using historical extreme rainfall events and synthetic precipitation scenarios to evaluate performance under diverse flood conditions. Overall, the results highlight the potential of integrating physics-guided AI emulation with ensemble data assimilation to address the trade-off between accuracy and computational speed in operational high-resolution urban flood forecasting.

ABSTRACT. Moving boundaries (often called free boundaries) in shallow-water equations mark the time-dependent water line between '(flooded) wet' and 'dry' regions and this feature is of utmost practical importance and should be adequately incorporated in the simulation. An efficient and conceptually simple implicit scheme is proposed for the one-dimensional shallow water equation systems with moving boundary treatment (i.e., wet-dry scheme). For the implementation of a wet-dry scheme in which very small water depth presents, a fully implicit backward-Euler scheme was developed for the fixed Eulerian mesh domain and a positivity-preserving limiter was employed. For the approximate Riemann problem, the Local Lax-Friedirichs flux is used. The Jacobians are estimated by centered differences in an approximate way rather than exact formulas, which, we believe, greatly reduced the coding efforts. Some benchmark problems with exact solutions and/or reference, experimental data are presented for the domain where the initially dry bed presents. Overall, satisfactory results are obtained.

ABSTRACT. Urban flood control in low-lying and coastal cities increasingly relies on polder systems, in which water levels are actively managed through pumping stations. The reliability of these pumping stations is therefore essential, particularly during extreme flood events that require continuous and high-intensity pump operation. Under such demanding conditions, unfavorable hydraulic behavior at pump intake structures can significantly degrade performance and accelerate mechanical wear. Among the most critical hydraulic phenomena is the formation of intake vortices, which may induce air entrainment, pressure fluctuations, increased vibration, and non-uniform flow distribution, ultimately reducing pump efficiency, shortening service life, and compromising overall system reliability.

This study investigates the influence of intake vortices on pump performance and operational reliability at the Ancol–Sentiong Pumping Station, a key component of Jakarta’s urban polder-based flood control system, using a Computational Fluid Dynamics (CFD) approach. The primary objective is to assess whether the existing intake configuration and pump operation patterns are capable of sustaining reliable performance during extreme flood conditions without compromising pump integrity, thereby supporting long-term operational sustainability. A three-dimensional numerical model was developed based on the actual geometry of the intake and pump bays, incorporating transient free-surface flow and gravitational effects. Multiple pump operation scenarios were simulated to represent realistic operational conditions, and mesh refinement was applied in critical regions to ensure accurate resolution of vortex structures and flow characteristics.

The simulation results reveal that localized vortices form near the operating intakes, particularly under high-discharge conditions and simultaneous pump operation. These vortices generate local flow acceleration and pressure reduction; however, their intensity remains within acceptable hydraulic limits and does not result in significant degradation of pump performance. The maximum observed water surface drawdown ranges from 0.13 m to 0.14 m, satisfying the established hydraulic design criteria for pump intake structures. Overall, the flow distribution remains sufficiently uniform to support stable and reliable pump operation.

The findings demonstrate that CFD-based analysis provides a robust framework for evaluating existing intake structures that may not fully comply with pump hydraulic standards, as well as for supporting the design of new intake configurations with improved reliability. This approach contributes to enhancing the resilience of pumping stations during extreme flood events while maintaining pump condition and extending service life, thereby strengthening the long-term performance of urban polder flood control systems.

ABSTRACT. Turbulence in narrow open-channel flows exhibits strong three-dimensionality due to enhanced interaction among sidewall boundary layers, free-surface constraints, and confinement-induced secondary currents. Such flows commonly occur in steep mountain streams, gorge sections, and river-training structures, where accurate prediction of anisotropy, Reynolds stress redistribution, and coherent-structure dynamics is essential for understanding sediment entrainment, bank stability, and scour processes. Despite their hydraulic relevance, the turbulence structure in narrow channels remains insufficiently resolved experimentally. Acoustic Doppler Velocimetry (ADV), the standard laboratory tool, is unable to reliably measure velocity fluctuations in the near-surface region and in close proximity to sidewalls because of free-surface interference, acoustic contamination, and probe-access limitations. As a result, the upper-layer turbulence dynamics where pressure strain redistribution and anisotropy evolution are most pronounced remain poorly quantified. To address this limitation, the present study investigated a high-resolution three-dimensional Reynolds-averaged Navier–Stokes (RANS) simulations of narrow open-channel flow with aspect ratio 3 using FLOW-3D HYDRO. The numerical setup reproduces controlled laboratory conditions reported by Mahananda et al. (2018), enabling systematic validation of mean velocity profiles. The predictive performance of three commonly employed eddy-viscosity closures Standard k–ε, RNG k–ε, and k–ω SST has been evaluated. The RNG k–ε model provides improved representation of vertical velocity gradients and Reynolds stress redistribution in the outer layer, while the Standard k–ε model tends to underpredict anisotropy under confinement. The k–ω SST model demonstrates enhanced near-wall stress resolution but exhibits sensitivity to free-surface boundary treatment. None of the models fully reproduces the experimentally inferred anisotropy distribution in the upper layer, highlighting inherent limitations of linear eddy-viscosity assumptions in strongly confined free-surface turbulence.

ABSTRACT. Local scour often occurs around obstacles and vegetation patches in rivers, and it can undermine bank protection and change channel stability. Understanding flow–sediment interaction around vegetation patches is essential for predicting localized scour and deposition in rivers and for designing nature-based bank protection. In particular, how an upstream submerged vegetation patch modifies leading-edge scour and concentrated scour around the downstream emergent patch remains insufficiently studied. This study reports flume experiments in uniform sand (d50=1.55 mm) compare an emergent patch alone (Case 1) with a submerged–emergent patch sequence separated by a 0.2 m gap (Case 2). Discharge is Q=0.012 m3/s and water depth is maintained at h=0.08 m using a downstream weir. Measurements over x=0–10 m include time-evolving bed-level-change maps, water-surface profiles, and equilibrium velocity profiles (t=30 h). Results show that Case 2 produces deeper and more localized scour around the emergent patch, with maximum scour concentrated in the adjacent free-flow corridor rather than inside the patch. Compared with Case 1, the upstream submerged patch weakens scour at the emergent patch leading edge, but it promotes stronger scour in the bypass-flow corridor where the flow accelerates. The 0.2 m gap is short enough that the upstream wake does not fully recover before reaching the emergent patch. As a result, more flow is routed through the free-flow corridor, where the bypass flow accelerates into a jet-like stream and strengthens the shear layer along the patch boundary. This elevates near-bed shear stress and shifts deposition farther downstream toward x=9 m. The main contribution of this study is controlled evidence that an upstream submerged patch (with a 0.2 m gap) redistributes scour by concentrating bypass flow, intensifying corridor scour alongside of the emergent patch. This dataset can provide a novel benchmark problem for interactions among 3D flow-vegetation-sediment transport to validate 3D models such as OpenFOAM and enhanced depth-integrated models such as BVC to analyze the resulting flow structures with sediment transport.

ABSTRACT. In recent years, river management has required consideration of nature-positive approaches while ensuring flood control safety. Methods such as widening the channel, lowering the riverbed, and creating three-dimensional riverbed profiles have been proposed in long-term river management plans spanning 20 years. These measures have been implemented and reported nationwide. A key aspect lacking in these approaches is environmental improvement utilizing the seepage occurring inside boulders and gravels. According to Navaratnam et al. (2017, 2018), even with the same riverbed topography, the presence of seepage between gravel and sand can suppress flow velocity near the riverbed. Recent research by the author using stone materials has focused on seepage flow through boulders, crushed stones, and pebbles. By defining stone size as the average of the long side, short side, and height, the author investigated experimentally how void space between stones changes. Here, the author presents results examining how void space between gravels changes with gravel size. Using plastic containers to displace the volume of inter-gravel voids with water weight and calculate void ratio, the author showed that the trend of void ratio variation with size differs between sizes suitable for stone assembling (≥ 0.04 m) and sizes unsuitable for stone assembling where gravel is simply stacked (≤ 0.03 m). The relationship between void ratio and the seepage flow velocity inside stones differs depending on stone size. Furthermore, comparing the variation in seepage flow velocity with gravel sizes through flow over a gravel mount, it was shown that the difference in flow velocity by gravel size is like the difference in void ratio between sizes suitable for assembly and sizes unsuitable for assembly. When comparing a horizontal channel section constructed with assembled boulders averaging 0.20 m in size (installation length 5 m in 0.80 m wide) to a section constructed with stones averaging 0.10 m in size (2.5 m in 0.40 m wide), the degree of blockage due to the assembly structure is affected, but the time-averaged flow velocity is equivalent. However, turbulence between stones depends on spatial dimensions. The seepage flow velocity between boulders obtained in the model experiment is difficult to reproduce as a scaled-up velocity converted using Froude and Reynolds similarities. Considering the size of the void volume per gravel particle, it is important to adjust the void volume between the gravel particles before applying the gravel size used between the large boulders to the prototype.

ABSTRACT. The use of deep underground storage and drainage tunnels as a countermeasure against extreme urban flooding is becoming increasingly common. However, the transition from open-channel to pressurized flow in these systems generates complex air–water interactions that remain insufficiently understood, particularly regarding the onset of hazardous geysering. This study addresses this knowledge gap by systematically evaluating how hydraulic parameters, including inflow rate, roughness height, and tunnel slope, influence flow-regime transitions and trigger geysering events in partially filled tunnels. In this study, a three-dimensional computational fluid dynamics (CFD) model was employed to analyze the geysering phenomenon observed in tunnels, and the model was rigorously validated against laboratory experiments that reproduced the complete transition from gravity-driven to pressurized flow. The simulation successfully reproduced observed hydraulic jumps, air-pocket entrapment, pressure fluctuations, and vent-pipe eruptions both qualitatively and quantitatively, demonstrating the model’s capability to resolve key interaction mechanisms relevant to geysering. Three primary parameters were considered, namely main pipe slope, roughness height and inflow rate, which reflect representative ranges observed in laboratory-scale deep tunnel systems. By systematically combining these conditions, the model captured a wide spectrum of hydraulic behaviors, enabling the classification of flow patterns ranging from typical hydraulic jumps to full-pipe inflow, and the identification of parameter thresholds associated with geysering onset. The parametric simulations revealed that increasing inflow and roughness, while decreasing slope, resulted in earlier transitions from supercritical to subcritical flow, enhanced air-pocket compression, and a higher likelihood of geysering near vent pipes. These findings provide a foundation for developing design and operational guidelines that mitigate instability caused by trapped air, thereby supporting the safe and reliable implementation of large-scale underground drainage infrastructure in regions experiencing intensifying rainfall.

ABSTRACT. During high dam’s discharge, a mass of air is entrained in the nappe and dissolves in the plunge pool at an extremely high pressure, leading to the supersaturation of total dissolved gas (TDG). Fishes live downstream of dams would suffer from gas bubbles disease even death if they were prolonged exposure to supersaturated TDG water. It is necessary to explore measures to mitigate the adverse effect caused by TDG supersaturation to fishes. Sometimes, due to the sediment deposition, floodplains form on the riverbank, and vegetation usually grows in the floodplain area. During the flood discharge period, floodwater leads to an increase in water level, forming a floodplain flow. Due to its unique hydraulic, substratum, and bait conditions, the vegetation area on the riverbank floodplain is usually a more concentrated area for fish distribution. Previous studies have shown that the presence of vegetation can effectively promote the dissipation of supersaturated TDG. Additional, compound river channels, the influence of vegetation patches on the flow structure and momentum, mass exchange will further affect the distribution of supersaturated TDG. So, how to predict the transport and dissipation process of supersaturated TDG in vegetated compound channel is significant for aquatic conservation. Here, a three-dimension supersaturated TDG transport model under the effect of vegetated floodplain was built. The Reynolds stress turbulence model was applied to obtain the flow filed condition. The effect of vegetation on the flow field in the model was realized by adding a drag term into the momentum equation. In the supersaturated TDG transport equation, the TDG dissipation term, which includes the liquid-gas interfacial mass transfer, solid wall adsorption and internal dissipation is added; in the diffusion tensor, turbulent diffusion and dispersion caused by vegetation were considered. Our three-dimension supersaturated TDG transport model under the effect of vegetated floodplain was verified with the published experiment data. The verification results show that the model established in this paper can truly reflect the distribution of supersaturated TDG under the effect of vegetated floodplain, and can be used to do relevant simulation studies.

ABSTRACT. Violent free-surface flows accompanied by large interface deformation and breakup are commonly observed in hydraulic phenomena such as dam-break flows. Accurate and stable numerical simulation of such flows remains challenging, particularly under high Reynolds number conditions on the order of 10⁵ and with a large density ratio between air and water. The lattice Boltzmann method (LBM) is a fully explicit numerical scheme, which avoids the solution of a pressure Poisson equation and is therefore expected to achieve high computational efficiency. In addition, recent developments have enabled stable simulations under high Reynolds number and large density ratio conditions, leading to an increasing number of applications. In this study, a three-dimensional dam-break problem involving violent air–water two-phase flows was numerically investigated to quantitatively assess both the predictive accuracy and the computational efficiency of two different numerical approaches. Specifically, a three-dimensional lattice Boltzmann method employing a multiple-relaxation-time collision model coupled with a phase-field formulation (3D MRT-LBM) was compared with a three-dimensional finite difference method combined with the volume-of-fluid (VOF) approach, in which the pressure Poisson equation was solved using the successive over-relaxation (SOR) method. The simulations were conducted in a three-dimensional computational domain containing an initially confined water column, and identical spatial resolutions were adopted for both numerical methods to ensure a fair comparison. The predictive accuracy was evaluated by tracking the temporal evolution of the water-front position along the bottom wall and comparing the numerical results with experimental data reported by Martin and Moyce. In addition, the computational efficiency was assessed by measuring the elapsed computation time required to simulate the flow evolution up to 0.3 s, corresponding to the time just before the water front reached the downstream wall, without outputting intermediate results. The results demonstrated that the 3D MRT-LBM reproduced the time evolution of the water-front position in close agreement with the experimental measurements, whereas the 3D FDM exhibited a delay of approximately 0.02 s in the arrival time relative to the 3D MRT-LBM. This indicates that, for the spatial resolution considered in this study, the 3D MRT-LBM provides higher accuracy in predicting the water-front propagation. Furthermore, the elapsed computation time was approximately 1 h for the 3D MRT-LBM, compared with approximately 26 h for the 3D FDM. These findings clearly show that the 3D MRT-LBM can substantially reduce computational cost while maintaining high predictive accuracy in three-dimensional simulations of violent free-surface flows such as dam-break problems.

ABSTRACT. Climate change is deeply affecting numerous regions around the world and especially in the southeast Asia region. Among these countries that are already facing climate related challenges the strategy to follow for economic development is highly questioned. How to design the needed infrastructures that could mitigate the future climate conditions? How to choose the agricultural strategy to ensure food security under the pressure of longer droughts? How to design and develop the water services that will match the demands of the fast-developing urban environments? are some of the frequently asked questions by numerous governments and municipal decision makers. The task to elaborate is complex and requires mobilizing a combination of climate forecasts addressing various scenarios with hydrologic models delivering finally an assessment of the water resources and of the magnitude of the extreme processes. This process requires local data to be able to deliver meaningful results that can exploited through strategic development plans. Unfortunately, in numerous developing countries, the availability of local data – hydro-climatic data, accurate DEM, hydrographic data and hydrometric data – is extremely limited. To overcome this challenge and to support the economic development, it’s essential to elaborate robust approaches that could provide sound results to support cation plans. Cambodia is one of the countries most exposed to climate change impacts. Climate change is projected to increase the frequency and intensity of flooding, sea level rise, and heat stress. Cambodia is investing massively to develop water services all around the country. Obviously, the future conditions must be taken for the design of the new services. This task is complicated due to the poor level of availability for quality local data. To support a massive development strategy endorsed by the Asian Development Bank over 12 years, a robust climate risk assessment methodology has been elaborated. The method mobilizes CMIP6 dataset for SSP2-4.5 and SSP5-8.5.scenarios and produce forecast for 2050 horizon. The climate conditions are used to simulate hydrologic conditions (HEC-RAS models) and to simulate extension of inundated areas and duration of droughts. Impact of inundations is assessed by counting the number of buildings affected. The freely available world building database provides a sufficient accuracy. The results obtained with the various simulations allow to assess the future conditions and elaborate criteria to establish comparison among locations and to identify priorities for actions. The presentation will present the methodology and its application to Cambodia context.

ABSTRACT. Climate change is expected to increase the frequency and intensity of extreme events worldwide. Although there is a broad consensus that impacts will worsen, they will not occur uniformly, being dependent on local climatic conditions and vulnerability. Therefore, assessment of climate change impacts at finer spatial scales, particularly at local and basin levels, is essential to support effective adaptation planning. In this context, the study evaluates the impacts of climate change on droughts, wildfires and floods events in the Ceira River basin. Based on the results achieved, local climate adaptation measures were defined. The Ceira river basin is a natural runoff river basin, part of the Mondego river basin located in the centre of Portugal, being affected in the last years by more frequent and intense extreme events. Among these, floods are particularly relevant, as the Ceira River has a significant contribution to floods in Coimbra, a major urban center located downstream. To assess future climate change impacts, an ensemble of 13 regional climate models was used, considering RCP4.5 and RCP8.5 and three future periods (2011–2040, 2041–2070, 2071–2100). Drought impacts were evaluated using the SPI and SPEI, while wildfire risk was assessed using the FWI. Flood impacts were analysed through the variation of annual maximum daily peak flow for different return periods. To this, hydrological modeling and extreme value analysis using the GEV distribution was applied. The results show that the frequency, duration and intensity of drought events will increase (up to more 8 events and 28 months in extreme drought by 2071-2100). Regarding wildfire risk, it is also expectable a significant increase (up to 20 more days per year with elevate wildfire risk). Despite overall reductions in mean flows, extreme peak flows will tend to increase (around 22% in 2011-2040 for 10, 20 and 50-year return periods), leading to more impacting floods. These findings highlight the need for integrated basin-scale climate adaptation and risk management strategies. As such, a guide with several measures for adapting the Ceira river basin to climate change was developed, contemplating the required and important holistic approach, as different thematic areas were considered: forests; natural areas and biodiversity; wate resources; river Infrastructure; tourism; human health. This work was developed in close collaboration with the Portuguese Mondego RBD authority (APA - Portuguese Environmental Agency) and municipalities, under the scope of the EEA Grants “Ceira River Basin Management under Climate Change” project.

ABSTRACT. Urban nature-based stormwater systems such as low impact development (LID) technologies are widely promoted as climate-adaptive stormwater solutions, yet their soil-level resilience to extreme rainfall and heatwave events remains insufficiently understood. This study quantified coupled hydrological and thermal soil responses across multiple LID facilities to identify design-relevant mechanisms governing resistance, recovery, and long-term functional stability under extreme rainfall and heatwave stress. The analysis was conducted at the Kongju National University (KNU) Cheonan campus (Republic of Korea) and included five LID facilities: bioretention (BR1), rain garden (RG1), infiltration garden (RG2), surface constructed wetland (SCW), and tree box filter (TBF). High-resolution soil moisture, electrical conductivity (EC), and temperature data collected through IoT-based sensors were aligned with identified extreme rainfall and heatwave events to derive event-based response metrics, including soil moisture change (Δθ), EC dilution magnitude (ΔEC), thermal stress intensity, time to peak response, and recovery time.

Extreme rainfall responses revealed pronounced design-dependent contrasts. Median Δθ ranged from <0.05 in SCW outlets to >0.35 in TBF soils, with maximum values exceeding ~0.6, indicating susceptibility to prolonged saturation in highly confined systems. Recovery times varied by more than an order of magnitude, with SCW and BR1 typically recovering within ~5–10 h, compared to >30–40 h in TBF. EC dynamics showed consistent dilution during rainfall (ΔEC ~10–40%), with delayed recovery in infiltration-dominated facilities, suggesting prolonged solute redistribution and altered soil chemical conditions following large events. Heatwave events induced substantial soil drying (Δθheat up to 0.12) and temperature increases of 5–10 °C, with recovery strongly controlled by antecedent moisture storage and subsurface connectivity. Machine-learning analysis using random forest and XGBoost achieved strong predictive performance for Δθ (5-fold cross-validated for random forest of R² ≈ 0.78; RMSE ≈ 0.06), identifying baseline soil moisture, LID type, antecedent dry days, and rainfall magnitude as dominant controls. SHAP analysis further revealed nonlinear thresholds and contrasting sensitivities across LID designs.

These findings highlight critical design trade-offs in LID systems: designs that maximize infiltration and storage enhance peak attenuation but may experience prolonged recovery and reduced resilience under repeated extremes, whereas hydraulically buffered systems exhibit faster recovery and greater functional stability. The results provide quantitative guidance for climate-adaptive LID design, emphasizing the need to balance infiltration capacity, internal drainage, and thermal buffering to enhance long-term soil resilience under intensifying extreme weather.

Acknowledgement: This research was supported by the National Research Foundation of Korea(NRF) funded by the Ministry of Science and ICT (No.RS-2023-00277793).

ABSTRACT. Urbanization and the increasing frequency of short-duration, high-intensity rainfall have highlighted the limitations of conventional urban drainage systems (CUDs). This study investigates infiltration-type rainwater drainage systems (IRDs) to evaluate their hydraulic behavior and runoff attenuation performance in comparison with CUDs. Full-scale rainfall experiments were conducted using IRD and O-shape channel systems under identical initial and boundary conditions, focusing on runoff initiation, peak discharge, and post-rainfall drainage characteristics.

The experimental results show that IRDs effectively delay runoff initiation, reduce peak discharge, and enhance hydrologic attenuation through infiltration and temporary storage. To reproduce these processes, a three-dimensional numerical model was developed. Direct application of experimentally measured hydraulic properties resulted in limited agreement with rainfall observations due to the heterogeneous and anisotropic structure of the IRD layers. To address this limitation, effective permeability and porosity were introduced to represent the averaged hydraulic response of the system.

The numerical model incorporating effective parameters achieved improved agreement with experimental results, and sensitivity analysis confirmed the dominant influence of permeability and porosity on discharge behavior. The findings demonstrate the hydrologic advantages of IRDs over CUDs and provide a practical modeling framework for evaluating and designing infiltration-based urban drainage systems.

ABSTRACT. The Philippines is one of the most typhoon-prone regions in the world, making an understanding of typhoon characteristics critical for disaster preparedness, particularly under changing climate conditions. The Database for Policy Decision-Making for Future Climate Change (d4PDF) has been widely applied in various regions to support policymakers in anticipating potential climate. However, studies assessing future changes in typhoon characteristics affecting the Philippines using d4PDF remain limited. This study addresses this gap by applying a deep learning-based cluster-matching model to analyze future changes in typhoon characteristics affecting the Philippines. The proposed model was developed using an encoder–decoder neural network, which effectively overcomes limitations of conventional cluster-matching approaches by handling high-dimensional features, large datasets, and the complex dynamics of typhoons. Analysis of typhoons simulated in the d4PDF historical period (1951–2010) and the +4K future scenario (2051–2110) revealed significant changes in typhoon characteristics. Typhoons affecting the Philippines are projected to become more frequent, with an increase of approximately 15% in the annual number of events. Notably, future typhoons are projected to become more severe, with increases in maximum wind speed of 35%, decreases in central pressure of 28%, and longer lifetimes by about 1.5 days. These findings highlight the urgent need to strengthen preparedness and implement adaptive disaster management strategies in the Philippines.

ABSTRACT. Irrigation system performance is a crucial factor in supporting the success of the global cropping plan (GRP), particularly in island regions. East Nusa Tenggara (NTT) Province has diverse geographical conditions, uneven rainfall, and limited water resources, making irrigation management a crucial aspect in increasing agricultural productivity. This study aims to analyze the performance model of irrigation systems against GRP in three different island irrigation areas in NTT (Timor, Flores, and Sumba). The research method uses a quantitative approach with a correlation design and multiple linear regression modeling. Data were collected through field observations, technical documentation including rainfall data, land cover and climatology data, and questionnaires to irrigation officers and farmers. Irrigation system performance variables include aspects of physical infrastructure, crop productivity, supporting facilities, personnel organization, documentation, and the Water User Farmers Association (P3A). Meanwhile, the GRP variable is measured based on the suitability of the planting schedule, cropping pattern, and the level of planting realization. Data analysis uses Pearson Correlation and multiple linear regression to determine the effect of irrigation system performance on the GRP. The correlation test results show that irrigation system performance significantly influences the success of the RTTG (p-value 0.001) with r=0.890, indicating a very strong relationship between irrigation system performance and RTTG. The model analysis results show that all irrigation performance indicators significantly influence RTTG (p-value <0.05) and an R-square of 0.76. This means that 76% of the variation in RTTG can be explained by physical infrastructure, crop productivity, supporting facilities, personnel organization, documentation, and the Water User Farmers Association (P3A), with the remaining 24% explained by factors outside the model. Physical infrastructure indicators are the most dominant factor determining the suitability of planting schedules, cropping patterns, and the level of planting realization (β=0.45). The results of this study indicate that improving irrigation system performance, particularly in physical infrastructure, is essential to support the success of RTTG in island regions such as NTT. These findings are expected to serve as a basis for formulating more effective and sustainable irrigation management policies.

ABSTRACT. Although non-turf extensive terrestrial green roofs contribute to cooling down building in daytime and keep building warm in nighttime during cold season in subtropical climate, their disadvantages include thick substrate, heavy weight bearing, as well as high maintenance and low resiliency during extreme climate. This study attempts to develop ultra-thin wetland green roof which has compatible thermal performance, and yet light, resilient and low maintenance. The effect of type and depth of growth substrate, and different species of wetland plants on the thermal performance of the ultra-thin wetland green roofs were explored, and then compared with the conventional extensive terrestrial green roofs and extensive wetland green roofs. The research adopts one three-stage field experiment conducted on the flat rooftop of one concrete building in Taichung, Taiwan. Results from the 1st stage demonstrated that the extensive field soil wetland roof without plant outperformed the extensive coconut fiber board and rock wool wetland roofs without plant by keeping the buildings warmer during daytime, and by keeping the building warmer during nighttime. Results from the 2nd stage presented that the ultra-thin 10cm-field soil wetland roof without plant outperformed the extensive 5cm-, 15cm- and 20cm-field soil wetland roofs without plant by highest passive cooling effect during daytime, and highest marginal increase in thermal insulation performance during nighttime. Results from the 3rd stage demonstrated that the ultra-thin 10cm-field soil wetland roofs with water dropwart, waterhyssop and roundleaf rotala outperformed tall scouring rush by higher passive cooling effect during daytime, and weaker thermal insulation performance during nighttime. Moreover, the ultrathin wetland roofs had similar thermal performance as the extensive terrestrial as well as wetland green roofs with roundleaf rotala, yet their advantages also include 1/2 substrate in thickness, lighter, resilient and low maintenance. In sum, the thermal performance of the innovative ultra-thin 10cm-wetland green roofs with water dropwart, waterhyssop and roundleaf rotala are excellent extensive green roofs to adopt which are mosquito larvae free, no need for periodical soil replacement and frequent weeding, no risk for clogging of drainage pipe, as well as beneficial for urban flooding mitigation.

ABSTRACT. Climate change poses universal challenges to all coastlines and waterfront environments due to rising sea levels and intensifying storms. While many coastal areas are more visibly affected by increased extreme wave conditions or storm surges, others encounter subtler concurrent challenges. Urbanised coastlines are of particular concern as they typically contain extensive waterfront infrastructure, with significant industrial and commercial value. Furthermore, they are often densely populated, with established communities whose well-being can be impacted both directly and indirectly. It is imperative that these values are safeguarded from both tangible and intangible losses. The necessity of developing effective adaptation solutions for existing waterfront infrastructure, while maintaining prior investments, presents a principal challenge for most urbanised coastlines. Singapore’s shoreline serves as an example of such challenges.

In the face of this challenge, the concept of Smart Waterfront is introduced. This concept integrates two systems: a predictive structural system with configuration flexibility and an active feedback system. These systems can be represented by one or multiple components. When combined, they form a smart waterfront system capable of completing the cycle of active feedback, predictive response, and adaptable flexibility. The carrying actors may consist of various combinations of structural and feedback options. This research will focus on two main components: XblocPlus and SMRS.

XblocPlus is a single-layer concrete armour unit developed by Delta Marine Consultants (DMC) in 2017 and first deployed in Afsluitdijk, Netherlands in 2019. It is a new addition to the Xbloc family, which has been in use worldwide since 2003. In 2021, through the Global Innovation Challenge (GIC) by PUB, Singapore’s National Water Agency, a collaborative research effort between DMC and PUB led to the development of a new type of XblocPlus unit, which is better suited to Singapore’s wave climate. The XblocPlus system employs a pattern placement technique, and its patterned interlocking nature allows for the calculability of its force of connection, which is directly related to the predictability of its interlocking connections. The XblocPlus system will be integrated with a cost-effective monitoring solution, Smart Remote Monitoring System (SRMS).

The primary objective of this research is to investigate the integration of a predictive system with advanced sensing technology that provides high-frequency feedback for effective monitoring. Throughout the process, we will explore various aspects, including hydraulic and structural performance, construction and installation efficiency, environmental benefits, monitoring techniques, and the evaluation of the combined system of predictive and feedback technologies.

ABSTRACT. Climate change alters precipitation patterns and recharge processes, leading to temporal variability in groundwater flow conditions. Such climate-driven changes are increasingly associated with non-stationary flow conditions in groundwater systems, which can influence groundwater velocities and subsurface environmental processes relevant to long-term remediation performance. Permeable reactive barriers (PRBs) are widely used for sustainable groundwater remediation, in which contaminant transport and reaction within reactive media are key determinants of system behavior. The purpose of this work is to investigate transport characteristics within PRB systems by focusing on flow-driven transport behavior across pore and Darcy scales. The key issue is how transport parameters, commonly used to evaluate PRB performance, can be examined under variable-flow conditions when pore-scale structures and flow pathways are subject to change. Granular activated carbon (GAC) was considered as a representative reactive medium, and total petroleum hydrocarbons (TPH) were used as a model contaminant to examine transport behavior in a flow-through PRB system. In this study, transport behavior within a GAC-based PRB system was examined using repeated tracer-based transport characterization under controlled flow conditions, while pore-scale structural features of the reactive media were observed using X-ray computed tomography. Transport parameters were evaluated at the Darcy scale, while pore-scale observations provided complementary insight into transport processes. The results indicate that transport behavior within PRB systems varies over time under sustained flow and can be consistently described by combining pore-scale observations with Darcy-scale transport interpretation. At a Darcy velocity of 1 m/d, TPH breakthrough occurred at approximately 2,000 pore volumes (PV), substantially earlier than the theoretical breakthrough range estimated from the adsorption capacity of the GAC. Concurrently, effective porosity decreased from 28.1% at the initial condition to 23.7% after 7,000 PV of operation, while pore connectivity declined from 186.7 to 161.5 mm⁻³, indicating progressive changes in flow pathways within the reactive media. The accelerated breakthrough was primarily due to the evolution of dispersivity in both spatial and temporal dimensions, which could be incorporated into an empirical scaling factor. By focusing on transport parameter characterization within a PRB under system dynamics, this work contributes to a better understanding of groundwater remediation processes and supports efforts toward sustainable groundwater management under climate change.

ABSTRACT. Groundwater is a valuable water resource that can be used for various purposes in human life. On Jeju Island, the southernmost island in South Korea, residents rely heavily on groundwater for their water needs, reaching 82%. Therefore, to ensure sustainable groundwater use, accurate groundwater level forecasts for the distant future are essential, enabling proactive management to prevent groundwater depletion. This study aims to accurately predict long-term (1-3 months) future monthly groundwater levels using data-driven models targeting two groundwater level observation wells located in the mid-mountainous region of Jeju Island. To improve groundwater level predictions, three artificial intelligence models (LSTM, GRU, and ANN), including deep learning, were used to predict long-term future groundwater levels. Afterward, the prediction results of the artificial intelligence models were used as input data for an ensemble model to predict long-term future groundwater levels. The improvement effect of the ensemble model on groundwater level prediction was analyzed for the entire data period and the low groundwater level period (November to May), which is a period of interest in groundwater depletion. As a result, all AI models and the ensemble model adequately predicted long-term future groundwater levels for the entire data period, and in particular, the groundwater level prediction performance of the ensemble model was superior to that of the individual AI models. It is noteworthy that the performance of groundwater level prediction among AI models varied across observation wells and across forecast periods, and no single AI model consistently demonstrated the highest performance. Therefore, an ensemble model leveraging the results of different AI models is necessary to improve groundwater level prediction performance. Data-driven models performed better in predicting groundwater levels during the period of low groundwater levels than during the entire data period. This period corresponds largely to a period of gradual decline in groundwater levels due to lack of rainfall. Therefore, data-driven models can relatively easily learn these predictable groundwater level decline patterns and more accurately predict future groundwater levels. In particular, the ensemble model demonstrated excellent predictive performance, with NSE values of over 0.72 for three-month forecasts, and produced predictions with NSE values up to 0.14 better than those of individual AI models. This supports the need for long-term future groundwater level prediction and groundwater level management using ensemble models to ensure sustainable groundwater use.

ABSTRACT. Groundwater systems play a critical role in transporting land-derived nutrients to coastal waters, particularly in volcanic island environments where subsurface flow dominates hydrological processes. In such settings, nutrient inputs occur not only through diffuse submarine groundwater discharge (SGD) but also via groundwater-fed springs and anthropogenic effluents linked to groundwater use. Understanding the integrated magnitude of these fluxes across spatial scales is essential for sustainable groundwater and coastal water management under changing environmental conditions. This study presents a quantitative assessment of terrestrially derived nutrient inputs to coastal waters at three sites along the eastern coast of Jeju Island, Korea: Gimnyeong harbor, Shinyang beach, and Pyoseon Beach. Submarine groundwater discharge was estimated using a numerical approach based on the Darcy flow equation, incorporating site-specific hydrogeological parameters. In addition to diffuse SGD, nutrient inputs from surrounding groundwater-fed springs and aquaculture effluents were quantified and integrated to derive total land-derived nutrient loads to the coastal zone. Nutrient concentrations measured in each discharge pathway were combined with corresponding flow estimates to calculate nutrient mass fluxes, enabling a comprehensive assessment of groundwater-mediated nutrient delivery. The results reveal pronounced spatial variability in integrated nutrient discharge among the study sites. Shinyang beach exhibited the highest nutrient export, with estimated daily loads of 151.23 kg day⁻¹ for total nitrogen (TN) and 14.67 kg day⁻¹ for total phosphorus (TP), reflecting the combined influence of diffuse SGD, spring discharge, and aquaculture-related groundwater effluents. In contrast, Gimnyeong harbor and Pyoseon beach showed substantially lower nutrient fluxes, indicating site-specific differences in groundwater flow regimes, land use, and discharge pathways. These findings demonstrate that land-derived nutrient impacts on coastal waters cannot be adequately evaluated by considering diffuse SGD alone. The integrated framework presented here highlights the importance of accounting for multiple groundwater discharge pathways when assessing cross-scale groundwater–coastal interactions and provides a practical basis for developing targeted and sustainable coastal water management strategies.

ABSTRACT. Sand dams are small-scale water storage structures typically constructed in ephemeral streams in arid and semi-arid regions, where flood flows are intercepted and water is stored together with transported sediments over extended periods. By retaining water within the sand layer, evaporation losses are reduced, and freezing during winter is limited, allowing the stored water to be utilized over relatively long periods. As a result, sand dams have been increasingly considered as decentralized water-supply facilities in regions experiencing seasonal water scarcity.

In this study, the water-supply reliability of sand dams installed within river channels was evaluated under drought conditions using a model-based analytical approach. To quantitatively assess water-supply capacity, an integrated framework combining the Soil and Water Assessment Tool (SWAT) model and a reservoir outflow estimation model was applied to simulate inflow, storage behavior, and available supply. Water-supply characteristics were examined across multiple design scenarios by systematically varying sand-dam parameters, including dam height, storage capacity, and intake structure configurations. Simulations were conducted at a daily time step to capture short-term hydrological variability.

The results indicate that water-supply reliability exceeding 95% can be achieved under specific design and operational conditions. These findings suggest that sand-dam performance is sensitive to structural design and intake settings. Overall, the analysis indicates that sand dams can provide a degree of flexibility in water supply under climate variability, while further investigation under diverse hydrological and site conditions is required to support broader application.

Acknowledgements This research was conducted as part of the 2026 Purpose-Oriented R&R Long-Term Research Program funded by the Korea Institute of Civil Engineering and Building Technology (KICT).

ABSTRACT. Groundwater recharge is a key component of watershed hydrologic processes and plays a critical role in sustaining water resources under changing climate conditions. This study analyzes the impacts of climate change on groundwater recharge in the An-Seong watershed, South Korea, using the Soil and Water Assessment Tool (SWAT) with a CMIP6 multi-model ensemble approach. The SWAT was calibrated and validated to simulate major hydrologic components, including surface runoff, evapotranspiration, and groundwater recharge. Future climate projections were derived from downscaled General Circulation Models (GCMs) participating in CMIP6, and a multi-model ensemble was developed using the Taylor skill score to minimize uncertainties among individual models. The results indicate that during the winter dry season, increased evapotranspiration driven by rising air temperatures may lead to a reduction in groundwater recharge despite an increase in precipitation. Future groundwater recharge is projected to decrease by approximately 10–20% compared to historical conditions, implying an increased risk of seasonal water scarcity. This study demonstrates that groundwater recharge responses to climate change depend not only on changes in precipitation but also on rising temperatures and evapotranspiration regimes, and that the integration of a multi-model ensemble with a hydrologic model provides a valuable scientific basis for developing climate change–adaptive groundwater management strategies.

Acknowledge: Research for this paper was carried out under the KICT Research Program (project no.20250258-001, Development of Elemental Technologies for River Management based on The New Normal in Response to Water Issues) funded by the Ministry of Science and ICT.

ABSTRACT. Climate change poses escalating threats to coastal groundwater systems through sea-level rise, altered precipitation regimes, and intensified pumping demands. As seawater intrusion can lead to irreversible aquifer degradation, anticipating critical transitions before they occur is essential. However, conventional monitoring often detects changes only after severe damage has occurred. This study develops a multi-indicator framework for assessing coastal groundwater resilience by integrating Critical Slowing Down (CSD) theory with information-theoretic metrics. We applied Early Warning Signal (EWS) analysis to long-term electrical conductivity (EC) time-series data from South Korean coastal monitoring wells, representing diverse hydrogeological settings. The methodology employs rigorous preprocessing (outlier removal and LOWESS detrending) followed by the moving-window computation of four complementary indicators: variance and lag-1 autocorrelation (capturing recovery dynamics), skewness (detecting asymmetry), and Fisher Information (quantifying systemic order). These are aggregated into a composite resilience index to robustly detect declining. Theoretical validation using multiple bistable models confirms the framework's ability to distinguish approaching regime shifts from random fluctuations, supporting the application of CSD theory to multi-stable coastal aquifers. Field applications reveal heterogeneous resilience patterns. Sites in Incheon—highly vulnerable to intrusion—exhibited strong destabilization signals (concurrent increases in variance/autocorrelation and declining Fisher Information) prior to observed EC transitions. Conversely, Jeju Island sites demonstrated transient warning signals followed by recovery, suggesting inherent resilience. Notably, the framework detected subtle system stress during periods when raw EC data showed no distinct anomalies, demonstrating its capacity for early detection. The proposed framework offers three key advances: (1) physically interpretable indicators grounded in dynamical systems theory, (2) a composite index balancing sensitivity with noise filtering, and (3) transferability across diverse hydrogeological settings. By providing probabilistic early warnings, this approach supports the risk-informed decision-making essential for sustaining coastal freshwater resources under climate uncertainty. Acknowledgement: This work was supported by the National Research Foundation of Korea(NRF) grant funded by the Ministry of Science and Technology (RS-2024-00356786)

ABSTRACT. With the intensification of global climate change and rapid urbanization, extreme excessive flood disasters occur frequently and their destructiveness intensifies, posing a severe threat to urban development, people's lives and property safety, and urban resilience. As the core support for urban flood control and drainage, emergency flood control facilities are also key carriers for improving flood resilience, mainly including temporary water-retaining facilities for urban underground spaces, emergency drainage pump trucks for waterlogging, and dike breach closure equipment. Their application efficiency directly determines cities' capacity to respond to extreme floods and resilience recovery level: temporary water-retaining facilities for underground spaces can block water intrusion into subways, utility tunnels and other key areas; emergency drainage pump trucks can efficiently drain waterlogging and accelerate post-disaster recovery; dike breach closure equipment can curb the expansion of breaches and reduce disaster impacts. Therefore, precise simulation is urgently needed to clarify the operation mechanism, coordinated efficiency and resilience improvement paths of various facilities in excessive flood scenarios. Focusing on urban excessive flood risks and closely centering on the goals of flood risk modeling and resilience improvement, this study proposes a coupled simulation method for excessive flood evolution and emergency flood control facilities. Taking excessive floods in Xi'an and the application of relevant facilities as a case, it verifies the model accuracy, clarifies the applicable scenarios, effects, coordination logic and resilience contribution of various facilities, so as to provide scientific support for urban flood risk management, flood control and disaster reduction decisions, optimal facility allocation and resilience improvement.

ABSTRACT. Storm surges cause coastal roads to flood during typhoons, with rising sea levels amplifying the flooding damage. This study simulates the flooding of Hong Kong’s urban road network triggered by storm surges during Typhoon Mangkhut, and compares the flooding scenarios after a one-meter sea-level rise, so as to assess the road flooding risk. A storm surge model, integrated with a cyclone wind field, was employed to simulate tidal changes in the Pearl River Estuary (PRE). Meanwhile, a high-resolution urban hydrodynamic model was developed to evaluate the road inundation, taking into account buildings and drainage systems. Through analyzing the flooding characteristics from the hydrodynamic model alongside vehicle stability, this study estimates the potential losses to Hong Kong’s urban roads due to storm surges following a one-meter rise in sea level. The findings may contribute to urban road flood risk management and resilience assessments in the context of climate change.

ABSTRACT. Global climate change has led to an increase in the frequency and severity of extreme events. Heavy rainfall and storm surges associated with tropical cyclones have substantially increased the risk of compound flooding in low-lying coastal cities, necessitating the development of accurate and integrated modeling approaches. Urban drainage systems play a critical role in flood dynamics in coastal cities. Although underground pipelines effectively drain stormwater and reduce surface flooding during heavy rainfall, elevated storm surge levels can block drainage outfalls, weakening drainage capacity and triggering backwater effects. This study proposes an urban compound flood simulation framework that accounts for the bidirectional coupling between urban surface runoff and subsurface drainage flows under storm surges and rainfall induced by tropical cyclones. The proposed model is applied to the Macau Peninsula, a low-lying coastal city that is affected by typhoons every year. Compound flood events triggered by Typhoons Hato and Mangkhut are used for validation. The model successfully reproduces the spatiotemporal evolution of inundation across the Macau Peninsula. The results highlight that compared with simulations that neglect drainage networks, the coupled scenarios reduce areas with inundation depths below 0.5 m by 6.97%–7.59%, while the total inundated area decreases by 3.93%–4.32%. Sensitivity analyses indicate that maximum inundation depth in the BiH2C model is more sensitive to the Manning coefficients in the channel than to those on the floodplain, whereas inland flood dynamics are primarily controlled by floodplain roughness. Topographic resolution significantly influences inundation depth, with grid sizes finer than 10 m potentially reducing the reliability of simulated inundation extents. Overall, the bidirectional coupling scheme outperforms unidirectional approaches by more effectively capturing surface–subsurface interactions, thereby enhancing simulation accuracy. The proposed framework is expected to be applied to other coastal cities facing similar compound flood risks.

ABSTRACT. Motivation Compound flooding arises from the interaction of multiple drivers, such as rainfall, river discharge, and storm tides, and often leads to severe and widespread damage. Timely forecasting of compound flood events is therefore critical for effective emergency preparedness, particularly for extreme events such as cyclones. However, large-scale compound flood modelling at fine spatial resolution is computationally prohibitive, involving multiple upstream processes and long runtimes that severely limit forecast lead time. This creates a fundamental trade-off between spatial resolution and operational timeliness.

Methodology We address this challenge by combining physics-based flood modelling with flood map super-resolution using deep learning. The two-dimensional hydrodynamic model BG-Flood was used to generate both coarse- and fine-resolution compound flood simulations. A progressive-focused Transformer architecture was trained to learn the mapping from coarse-grid to fine-grid compound flood maps, explicitly capturing spatial dependencies and fine-scale inundation patterns. Once trained, the system enables rapid inference of fine-resolution flood maps from fast coarse-grid simulations.

Results The proposed framework produces high-quality fine-resolution compound flood maps within less than 10 minutes, compared to approximately 20 hours required for conventional fine-grid simulations. The results demonstrate that flood map super-resolution can retain essential spatial detail while dramatically reducing computational cost. This approach enables near–real-time large-scale compound flood forecasting and provides a practical pathway toward operational early warning systems for extreme compound flood events.

ABSTRACT. China is one of the countries most severely affected by flood disasters worldwide. According to a United Nations report, under a once-in-a-century flood scenario, China ranks first globally in terms of the population exposed to such hazards. In response to this critical challenge, this presentation, titled “Extreme Floods and Risk Adaptation” systematically addresses the following three core research areas: (1) the development and application of a global hydrological model; (2) the localization and improvement of a global hydrological model tailored to regional characteristics in China; and (3) the construction of a national-scale real-time flood risk forecasting system based on meteorological-hydrological-hydraulic coupling. Through these studies, we aim to reveal the spatiotemporal evolution patterns of extreme flood disasters in China under climate change, advance methodological frameworks for flood risk modeling and forecasting, and propose targeted risk adaptation strategies. This work is intended to provide a theoretical foundation for deepening the understanding of the mechanisms through which climate change influences extreme flood risks, while also offering scientific support and technical tools for major decision-making in flood risk adaptation management in China.

ABSTRACT. In meteorology, "training rainstorms" refer to a phenomenon in which successive convective cells or thunderstorms move over the same region in a linear sequence, akin to cars on a train track. These storms repeatedly generate rainfall over prior locations, producing prolonged and concentrated precipitation. In regions with saturated soils or topographical depressions, such rainfall can trigger severe flooding. During the summer monsoon in China, training rainstorms are frequent, as exemplified by the unprecedented rainfall in Hong Kong on September 7, 2023. On this day, multiple air currents supplied abundant moisture, fostering intense raincloud clusters and a distinct training rainstorm in the Pearl River Delta, with peak hourly precipitation reaching 150 mm and causing widespread urban flooding. This study examines the impact of the rainstorm ‘train effect’ on the spatiotemporal dynamics of urban inundation, using the 7 September 2023 Hong Kong event as a case study. A spatially distributed hydrodynamic model is employed to simulate rainfall-runoff interactions, with model performance validated against observations at 80 inundation locations. The influence of the train effect on flood response is assessed considering catchment slope, percentage of impervious area, spatial aggregation, and rainfall spatial variability indices. Special attention is given to the severely flooded Kowloon and Sha Tin areas, representing both urban and rural–urban catchments. Diverse training rainstorm scenarios are constructed by combining extreme storms with varying return periods to investigate their effects on runoff and inundation patterns. The analysis focuses on the temporal and spatial characteristics of rainfall critical for the initiation and intensification of urban flooding. This study aims to elucidate how catchment-specific urban features modulate flood responses under training rainstorms, providing insights for flood risk management and urban planning.

ABSTRACT. Coastal urban areas are increasingly vulnerable to compound flooding, where the simultaneous occurrence of rainfall and storm surge produces cascading impacts far exceeding those of individual flood drivers. This growing threat necessitates high-fidelity flood models capable of accurately capturing rainfall-surge interactions for effective risk assessment and mitigation planning. However, traditional physics-based hydrodynamic models face a fundamental trade-off between spatial resolution and computational cost, limiting their applicability for real-time forecasting and large-scale risk analysis. Recent deep learning-based super-resolution methods offer a promising alternative by leveraging coarse-grid hydrodynamic models to capture complex spatiotemporal dynamics while employing neural networks for efficient spatial refinement. Nevertheless, existing CNN-based approaches suffer from blurring artifacts and limited generalization capacity, struggling to resolve fine-scale features such as flood texture details and sharp inundation boundaries. Moreover, most approaches focus on single-driver flood events, failing to capture the unique dynamics of compound flooding in coastal urban environments. To address these limitations, this study proposes DiffFlood, a novel super-resolution framework based on conditional denoising diffusion probabilistic models. The framework employs an iterative stochastic denoising process to progressively reconstruct high-resolution inundation maps from coarse-grid hydrodynamic outputs. A U-Net backbone enhanced with self-attention modules is adopted to capture long-range spatial dependencies critical for resolving complex urban flood patterns, with high-resolution terrain features and geographic attributes incorporated as conditional inputs. The framework is validated in Kowloon, Hong Kong using 80 hypothetical compound flood scenarios encompassing different return periods, rainfall temporal patterns, and sea level rise projections. A coupled 1D-2D hydrodynamic model generates paired low- and high-resolution inundation maps for training and evaluation. Results demonstrate that DiffFlood effectively reproduces test flood scenarios (8 hypothetical scenarios and 1 historical event), achieving a root mean square error as low as 0.028 m, with Nash-Sutcliffe efficiency and R² exceeding 0.96. Prediction time for a 24-hour rainfall event is reduced to 1770 seconds, representing a 12-fold speedup compared to fine-grid hydrodynamic simulation with GPU acceleration. Compared to baseline methods including UNet and SN-CGAN, DiffFlood preserves comparable regression accuracy and outperforms baseline methods in classification metrics, achieving CSI and F1 scores of 0.90 and 0.91, respectively. Visual inspection confirms that DiffFlood reconstructs sharper inundation boundaries and more realistic texture details. These findings highlight the potential of diffusion-based super-resolution as an efficient and accurate tool for compound flood modeling, supporting real-time forecasting and climate-adaptive urban flood management.