Solitary steps (isolated waterfalls) caused by tectonic movements such as faulting on exposed bedrock, are said to migrate upstream over time due to increased flow velocities and concentrated erosion of the bedrock at the steps. In the case of bedrock, erosion is caused by sediment transported by flowing water rather than by flowing water, and the erosion process due to sediment transport has recently been formulated¹⁾. However, numerical calculations using that equation have shown that the steps disappear over time. This study theoretically clarified whether solitary steps (isolated waterfalls) can form upstream on mild-slope riverbeds while maintaining their shape.

 Sediment discharge from mountainous areas causes severe sediment and flood disasters. Therefore, it is important to predict the sediment discharge from mountainous areas in order to prevent increased damage due to climate change. To improve the accuracy of sediment transport model for mountainous areas, we included the effects of the expansion of the river width and hiding effect due to boulder protrusion in the numerical model “SiMHiS” which integrates sediment production, sediment supply, and sediment transport. As a result, we succeeded to reproduce the amount of sediment discharged from the upper reaches of the Pekerebetsu River and the expansion of the river width during the 2016 heavy rainfall in Hokkaido, Japan. The original model involved extreme riverbed change. However, because the new model is able to consider the decrease in shear stress due to the widening of the river, extreme erosion is suppressed. Furthermore, because it considers the hiding effect of fine sediment appropriately, extreme deposition is suppressed.

 The behavior of large wood (LW) during flood events has been an important issue in a river management filed, and in recent years there has been growing interest in the relationship between LW and sed-iment transport mechanisms. In this study, a simple PSM-type LW model was modified to calculate sweeping transport of LW pieces and applied to reproduce their behavior in an existing large-scale movable bed flume experiment. Relatively good reproducibility was obtained for the location of LW accumulation, and a high contribution from the drag term, which evaluates the orientation of LW pieces, was suggested.

 In this study, we developed a basin-wide river temperature prediction model for five first-class rivers in the Chugoku district, and investigated the change in model accuracy when rainfall was also used as an input layer, in order to realize highly accurate basin-wide river temperature prediction. Specifically, the model was compared under two conditions: one in which only temperature was input, and the other in which rainfall was input in addition to temperature. These were conducted under the condition that water temperatures at all locations in the watershed were predicted simultaneously using meteorological data at all locations in the watershed as input. As a result, the input of air temperature alone overestimated the water temperature by more than 5°C during the outflow period from June to August. Especially at the main river sites with large catchment areas, the input of rainfall also reduced the temperature to a maximum of 2°C and the mean absolute error rate (MAPE) was less than 10% at most of the target sites. These results show that using not only air temperature, but also rainfall in the input layer enables highly accurate river water temperature prediction on a basin-wide scale.

 This study tried to develop a new model that predicts the distribution of river vegetation by applying a machine-learning model in multiple steps and analyzing the factors behind vegetation recruitment and disappearance. This study used Random Forest (RF) as the machine-learning model. The multiple-step RF model was optimized for each river segment. Based on the input dataset, the optimized multiple-step RF model predicts the river vegetation distribution for the following year. As a result of applying the multiple-step RF model to several river channels of the Kinu River, the F1 scores improved by approximately 10% compared to the simple RF model, and they had over 0.8 in all river segments. The model could predict the distribution of vegetation on sandbars along the inner banks of meandering channels and the distribution of grass plants that frequently recruit and disappear. Furthermore, SHAP analysis of the model results successfully detected the main factors for vegetation recruitment and disappearance in each river segment. The results suggest that the multiple-step RF model in this study could help predict river vegetation distribution.

 This paper tried to improve the versatility of flood flow analysis by using the river land covers classified by machine learning with normalized indices of a new satellite image at different times. Here, we chose random forests (RF) for machine learning. The target river was a forested gravel-bed channel in Kinu River, Japan. For river land cover classification, RF was learned using normalized indices from the satel-lite image taken in October 2016, and the trained RF was applied to new satellite images taken at differ-ent times. The river land cover classification results indicated an effective classification with an accuracy over 0.76 in F-measure, even for untrained images. Then, a flood flow analysis was performed using the roughness coefficients concerted from the RF's land covers. The flood water levels were compared with those of a correct data analysis, resulting in a difference of less than 0.1 m. The average F-measure cal-culated for each river cross-section indicated over 0.8, which noted that the proposed condition in the previous research for ensuring accuracy in flood flow analysis was also verified for the gravel river of Kinu River. These results strongly supported that the normalized indices of satellite images could be helpful for high-frequency quantitative monitoring of the impacts of river land cover changes in flood risk management.

 We evaluated the applicability of a deep learning model that quantifies scum coverage from fixed-point camera images of river surfaces across multiple locations. Furthermore, we developed a system capable of continuously tracking the spatiotemporal behavior of scum by sequentially assimilating the output of the detection model into Lagrangian particle tracking simulations. This approach compensates for data gaps during nighttime and between locations. The results demonstrated that fine-tuning the detection model, initially trained on dummy images, with a small number of original images improved scum detection accuracy and reduced false detection rates. While the detection model alone could only capture discrete scum behaviors, the assimilated simulations visualized the transport of scum repeatedly moving downstream and upstream with the tides. These findings demonstrated the utility of the developed behavior tracking system in capturing the complex spatiotemporal behavior of scum in urban tidal rivers.

 To clarify the behavior of floating debris in the Neya River system in Osaka, we conducted continuous detection of floating debris using the object detection model YOLOv5 applied to water surface images captured by fixed-point cameras. We then analyzed the influence of wind and tides on this behavior. The results revealed that in the Neya River system, the density of floating debris was higher near the tidal limit, suggesting accumulation due to the cessation of flow during high tide, which allows debris transported from upstream and downstream to accumulate. Long-term observations focused on Suminodo, where a high number of floating debris were detected, showed that both wind and tides significantly contributed to the abundance of floating debris. Despite low precipitation periods, a considerable amount of floating debris was detected, indicating transportation upstream by westerly winds and prolonged drift in the tidal area. Moreover, the amount of floating debris varied according to the lunar cycle, with an increase during spring tide compared to neap tide. This study's findings on wind and tide effects on floating debris behavior are expected to contribute to effective measures to reduce floating debris.

 In order to clarify the impact of climate change on water supply from Anou dam, we used data from long-term continuous experiments under the RCP8.5 scenario of the MRI-AGCM3.2S of the Meteorological Research Institute of the Japan Meteorological Agency. Then, a water supply model simulation was performed using a runoff analysis model. As a result, it was found that in the future the number of days that water supply will be restricted will increase approximately 1.2 to 1.8 times compared to the present climate condition, and the number of days that water supply will be cut off will also increase approximately 1.5 to 2.0 times than the present climate conditions.

 The impact of the heavy flooding caused by the historical heavy rain in July of 2020 on the Kuma River was evaluated in terms of the transition probability of “operation method during an extraordinary flood for disaster prevention” due to climate change, focusing on the existing storage-type dam, Ichifusa Dam, and the gate-controlled flood mitigation dam under construction, Kawabegawa Dam. The evaluation was conducted using a dam model incorporating an outflow model and dam operation rules, targeting the inflow volume to simulate the outflow discharge and water level in the dam. Additionally, the probability of excess floods at the downstream point of both dams, Hitoyoshi, was also evaluated. Measures to prevent abnormal floods and excess floods were verified, including the redevelopment of Ichifusa Dam, increasing the constant outflow discharge of Kawabegawa Dam during flood control operation, and creating reservoirs within the watershed areas. As a result, the quantitative evaluation of the changes in the transition probability of the operation for both dams and the probability of excess floods at Hitoyoshi due to climate change was possible. Furthermore, it was demonstrated that the redevelopment of Ichifusa Dam has a significant effect and, in combination with reservoirs, there is a possibility of achieving the target safety level under all climate conditions.

 In order to climate change, it become increasingly important to enhance the operation of existing dams. By operating dams more flexibly, we can maximize their functions, improve flood control capabilities and increase hydropower generation. However, there is currently a lack of consideration for advanced dam operation, including pre-release, post- flood operation and flexible dam operation. This study used long-term ensemble rainfall prediction in multipurpose dams to determine target drawdown water level for prerelease operation, and flexible dam operation for hydropower production. We investigated how these new rules improve overall dam operation including flood control safety, water usage and hydropower generation. By adjusting the reference water level based on ensemble rainfall prediction, we can effectively manage the dam water level while monitoring flood risk and water usage capacity. Additionally, maintaining a high water level in the dam when no flood is expected ensures the efficient operation of the dam-type power plant, leading to increased hydropower generation. If the maximum hydropower intake discharge is small comparing to reservoir storage, the minimum discharge can be selected as the hydropower intake rate to maximize hydropower without affecting water usage.

 In order to reduce risks caused by large-scale floods and to effectively utilize for hydropower production, advanced dam operation using ensemble rainfall prediction is anticipated. This study investigates methods to optimize pre-release and flood control operations using ensemble rainfall prediction, and applies dam operation methods that provide benefits for both flood control and water use in practical scenarios. Flood control operations aimed at minimizing the maximum outflow from the dam as well as reduced the energy loss from hydropower plants. By combining pre-release with optimized flood control operations, both flood control and water use benefits were further enhanced. Additionally, by preparing 'template operations, ' where pre-determined operations are selected based on optimization results, it became easy to understand the process for optimization that provide benefits for both flood control and water use from multiple operation options which increase receptivity for dam operators.

 It is essential to manage sedimentation effectively for the long-term use of dams. This study aims to improve the operation of sediment bypass tunnels (SBT) by predicting the flow discharge (Q) and suspended sediment concentration (SSC) using ensemble rainfall forecasts. Additionally, we examined the hysteresis and the behavior of SSC during various flood periods. We found that small floods exhibit high SSC in the early stages of a flood event, while large floods show more complex behavior. By predicting Q and water recovery rate, SBT can be operated without the risk of delays in start time and without the loss of water recovery.

 Purpose of this study is to propose a digital twin technology that supports evacuation actions in conjunction with water level forecasts during floods. In order for local residents to understand evacuation sites and routes appropriately and to take action before flooding occurs, it is necessary to provide contents that enable them to feel the conditions near and behind the river channel and that show the hydraulic phenomena leading to flooding in chronological order. In this study, we used 3D point cloud data obtained by laser surveying of the Chiribetsu River in Muroran City, Hokkaido, Japan, to represent the river channel topography, Fundamental Geospatial Data for the hinterland, and PLATEAU for buildings, and iRICNays2DFlood to calculate hydraulic phenomena such as river water level and flood depth to create a time series of flood damage. The hazard map image reproduced on the digital twin was proposed by adding evacuation sites and routes to the image. We believe that this kind of content, which gives a realistic impression of flood damage, will be useful for local residents to take appropriate evacuation actions.

 Road flooding prediction can help road users select rational routes and road administrators to implement efficient traffic regulations. This research proposes a method to calculate the probability of road flood occurrence in real time. The method accumulates flood data from a single flood sensor and rainfall data, then applies logistic regression analysis to these data. Since only one flood sensor is required per location, the method can be applied to locations where road flooding occurs suddenly or where multiple flood sensors cannot be installed. The usefulness of the proposed method was confirmed through experiments using road flood and rainfall at three locations over a period of 2 years and 5 months.

 Heavy rainfall disasters have become more severe and more frequent due to climate change. For mitigating the damage caused by flood inundation, real-time monitoring of flood inundation conditions is necessary for appropriate evacuation. Focusing on vehicle traffic information, the purpose is to clarify the relationship between flood inundation and traffic information and propose a method of real-time flood inundation monitoring for suitable evacuation information. The result showed that there is a correlation between depth of 50cm or more and vehicle traffic condition, and at flooded road link, relative traffic speeds decreased. Finally, from the presence or absence of vehicles passing on the link, it is possible to estimate the extent of inundation on a regional standard mesh basis.

 This study assessed the impact of inundation damage on the total population of a 500 m mesh in areas inundated by Typhoon Hagibis and the July 2020 heavy rains employing the difference-in-differences method. We also evaluated the influence of regional characteristics on population changes due to inundation damage. The results showed that the total population decreased with increasing inundation depth. The average rate of change in the total population per 500 m mesh cell due to inundation damage was estimated to be -2.9% in areas inundated by Typhoon Hagibis and -10.4% in areas inundated by the July 2020 heavy rains, compared to the case where no inundation damage had occurred. Furthermore, the impact of inundation damage on the total population varied with regional characteristics; for instance, the longer the distance to medical institutions, the greater the decrease in the total population due to inundation damage.

 In Japan, various disasters occur wholly every year. There is a growing concern about the significant damage caused by spatial/temporal chained multi-hazards, in which multiple disasters occur simultaneously, disrupting infrastructure systems. This study aims to investigate and analyze case studies of temporal chained multi-hazards as a first step in studying spatial/temporal chained multi-hazards, and to evaluate multi-hazard risks with a focus on water treatment plant (WTP). To achieve this, we conducted field surveys on the temporal chained multi-hazards related to WTP due to the heavy rains in Shizuoka Prefecture in 2022, and Akita in 2023. Additionally, we evaluated the disaster occurrence potential for WTP in Japan using various hazard maps. As a result, we confirmed the damage potential due to some form of disaster in a lot of WTPs.

 In recent years, water-related disasters have become more severe and frequent due to climate change. Additionally, there is concern about the occurrence of a massive tsunami caused by the Nankai Trough earthquake, which has a 70-80% probability of occurring within the next 30 years. Therefore, it is necessary to consider the combined disasters of tsunamis and floods. In this study, we used a comprehensive flood analysis model to examine the characteristics of tsunami run-up during compound disasters for the Asahi River, Hyakken River, and Yoshii River, which flow into Kojima Bay. The results confirmed that tsunamis run up the rivers for several kilometers more significantly when the scale of the flood is small. Conversely, when the scale of the flood increases, the run-up distance of the tsunami decreases, but the river water level rises due to the flood itself. Additionally, rivers with a larger incidence angle of the river mouth relative to the tsunami propagation direction in the sea tend to show a more pronounced effect of flood flow, leading to a narrower range of water level rise.

 Using the storage function runoff model, as a rainfall-runoff system, we proposed a differential equation that can theoretically estimate the stochastic characteristics of discharge with rainfall intensity, two storage coefficients, and initial discharge as known random variables, and its validity was verified through simulation methods. The results show that the variance of parameters related to the topography and geology of the watershed affects the variance of discharge around the flood arrival time, and the variance of rainfall intensity affects the variance of discharge in proportion to its value. The stochastic properties of the runoff model obtained in this study, through comparison with the results of stochastic response analysis of the physical model, lead to the solution of problems related to the lumping process of the runoff model.

 This paper presents the numerical study of the wind effect on flood propagation as well as discharge by rainfall-runoff-inundation simulation. A shallow water equation and a distributed rainfall-runoff-inundation model with the shallow water equation were used. First, the shape, velocity and range of the flood propagation with and without winds are investigated. As a result, the speed of the propagation increases with the wind in the moving direction, but if the depth increases or decreses depends on the locations of the entire wave. Likewise, the wind movement changes clearly by giving the observed wind. Finally, the changes of the discharge at Arakawa River catchments by runoff simulations are investigated targeting the real flood event on 2019. As a result, the changes of the discharge with the actual observed wind was not so significant. When the observed wind is doubled or tripled in magnitude, the changes of the discharge become more apparent. The paper presents these results.

 This study proposes a runoff forecasting method that considers snowmelt in addition to rainfall, which will remain applicable throughout the year in snowy and cold regions. Climate change has complicated considering heavy rainfall combined with snowmelt and not many studies and practical examples of runoff models that account for snowmelt exist. In this study, we propose a distributed runoff modeling method that consists of snowmelt estimation, slope runoff considering snow cover, and river channel tracking. This method enables estimation of the amount of snowmelt, without the need for snow accumulation models by using the snow depth analysis of JMA. We first confirmed the validity of the water balance using long-term slope runoff calculations and used past cases to perform predictive calculations. The results indicate that accurate calculations are possible for both the snowmelt season and the summer season, demonstrating that it is possible to use public meteorological information to calculate highly practical runoff predictions that can be applied to drainage basins including snowy and cold regions throughout the year.

 This paper analyzed high-altitude snow depth data and AMeDAS data in the Yoneshiro River basin and proposed a method for predicting snowmelt runoff using methods other than numerical models. High-risk snowmelt runoff was predicted by setting thresholds for snow depth, precipitation, and temperature data from AMeDAS observations. This prediction method was verified using 2022 data and its practicality was demonstrated. The effectiveness of high-altitude snow depth observations was demonstrated for snowmelt runoff after snow melts at AMeDAS points. In the case of the Yoneshiro River basin, high-risk snowmelt runoff was observed under conditions of a sudden decrease in snow depth of approximately 30 cm or more per three days at high altitudes, temperatures of approximately 10°C or more, and rainfall of approximately 5 mm per day. This result can be used for disaster prevention information and flood control systems during snowmelt runoff in municipalities that do not have flood prediction systems using numerical models, and demonstrated the importance of high-altitude snow observation.

 For a wide range of Reynolds numbers, the velocity and turbulence intensity distributions in classical hydraulic jumps have been systematically investigated under a given inflow Froude number with an undeveloped inflow condition below a gate. Effects of the Reynolds number on the magnitude and distribution of the velocity and those of the turbulence intensity in the jumps are clarified. The energy balance in the jump zone are obtained, demonstrating that the energy dissipation in the jump is not accomplished at the end of the surface roller of the jump. Based on the energy dissipation in the jump, it is shown the jump length which can be useful for the determination of the apron length.

 In consequence of progress of urbanization or mixed settlement in the rural area, falling water sound on drops has been recognized as a noise by people who came to reside newly around there. Therefore, this study conducted the comparison of relationship between sound power level calculated from sound pressure level measured on site and turbulent energy dissipation rate simulated by CFD, for the purpose of predicting the falling water sound on water-cushion-type drop. The results showed that turbulent energy dissipation rate had higher correlation with sound power level than water flow energy per unit time and unit width focused on in previous studies, and that turbulent energy dissipation rate had possibility to predict the effect of discharge or head drop on sound power level.

 The maintenance of aging water infrastructure facilities is highly important, thus a simple inspection method is desired. This paper proposes a quick and simple inspection method utilizing time-reversal waves along with a pair of omni-directional transmitter/receiver devices. This method is to evaluate the maximum amplitude of the converged time-reversed pulses in an attempt to determine the presence or absence of deformation of pipes. When the difference in amplitude between the two is small and difficult to identify, the spectrogram of the received waveform and convolutional neural network (CNN) have been successfully employed to improve the accuracy.

 Freshwater carbon, which utilizes the absorption of carbon dioxide through photosynthesis by aquatic plants and phytoplankton, is attracting attention as a measure to reduce carbon dioxide. Previous studies have shown that stratification in freshwater lakes can significantly change carbon dioxide fluxes. However, more research is needed regarding the stratification formed by the presence of aquatic plants. Therefore, in this study, we conducted a field survey at Lake Akan and Chuurui Bay in 2022 and analyzed water temperature data, with the aim of evaluating the influence of aquatic plants on the formation of the surface mixed layer. The results showed that the presence of aquatic plants promoted the formation of a surface mixed layer. In addition, we introduced a novel formula, taking wind energy into account, to estimate the surface mixed layer thickness. This innovative approach demonstrated good reproducibility, sparking new avenues for research in this field.

 In Lake Biwa, stratification is generated due to water temperature differences excluding well-mixed winter season. The change in water density induces the formation of internal waves when wind stress is applied to the lake surface.The study aims to elucidate the internal wave mode in Lake Biwa, theoretically and numerically, with the Coriolis, using the vertical water temperature profile obtained by field observations. The results of the field observations revealed that the Kelvin and Poincaré internal waves were induced based on the topographical features under the Coriolis effect. Theoretical solutions also showed internal wave modes comparable to the numerical simulations. Estimated internal wave modes numerically and theoretically agreed with the field observations. This suggests that defining the internal wave periods in Lake Biwa using theoretical solutions enables easy estimation of the internal wave modes.

 It has recently been found that submarine turbidity currents travel very long distances due to the extremely small entrainment or diffusion between turbid water and fresh water at the density interface. In this study, we analyzed numerically the motion of submarine turbidity currents neglecting the entrainment of seawater from the top of the layer, and examined the behavior of long-runout submarine turbidity currents in the downstream direction. As a result, It turned out that the profiles of density interfaces of the submarine turbidity currents resemble the S₂ and S₃ curves of open channel flow depending on the conditions, and there are cases where the flow converges to a normal flow condition (in which flow does not slow down). Although there are cases where the flow disappears, there is a precise critical layer thickness (thickness of submarine turbidity currents), like the critical flow depth, between such cases and cases where the flow converges to the normal flow condition. The layer thickness must be somewhat small to converge to a normal flow condition.

 There are concerns about MPs that have flowed into the ocean. In this study, we attempted to elucidate the process by which MPs flow from land into Tokyo Bay based on regular surveys of the river sediment in the Ebi River. The results showed that the number and types of MPs fluctuate greatly over time, but that seasonality is small, and that when a typhoon passes, the number of MPs increases to 75 times the normal amount. We also found that more than 90% of the MPs in the Ebi River are plastic fragments, with few artificial turf or fibers.

 We investigate the characteristics and amount of floating microplastics and macro plastics in rivers, coasts, and estuary by direct collection. The results shows that as for microplastics, PE and PP are dominant, and the number density is also high in rivers and coasts that flow around urban areas. As for the macro-drifting garbage, plastics are the most dominant, especially bags, fragment debris, and cigarette butts. It was also confirmed that when the wind and waves are strong, streaks do not occur, and the amount of garbage decreases. Furthermore, the results of the expiration dates printed in the garbage show that illegally dumped or improperly disposed garbage moves quickly to estuary.

 The fluctuations in the vertical distribution of microplastics (MPs) in the tidal reach of the Oita River were investigated. The greater the river depth was, the greater the MP mass concentration was; and hence, MPs were more abundant in the bottom layer. Interestingly, MPs accumulated on the salt water wedge in the middle layer. In addition, the highest mass concentration was found in the flood tide, followed by the ebb tide, first low tide, high tide and second low tide. This suggests that MPs are deposited on the bottom and that the deposited MPs are rolled up during flood and ebb tides. Furthermore, the concentrations of MPs smaller than 1 mm in size (SMPs) were one or two orders of magnitude greater than those of MPs larger than 1 mm (LMPs), indicating that evaluating the SMP concentration is important for assessing plastic emissions from land to the ocean.

 While urban areas are the main source of plastic waste, almost no study has made on road pollution. High production rates of plastic waste are expected near train stations. In this study, we observed density of plastic and tyre particles on urban roads (particles / m², mg / m²) in Kawasaki City and Noda City, Japan. (1) The plastic density was reduced to one-fifth in areas with periodic road cleanings by the city government. (2) The plastic density was about 10 times higher near large, highly developped stations than in small stations. (3) In Noda City, where the main transportation is automobile, the proportion of tyre fragments was about 3 times higher than near train stations of Kawasaki City.

 A degradation model based on accelerated degradation tests of drink bottle libs under dry and wet conditions was applied to a high-resolution grid-based map considering population distribution and meteorogical conditions (annual cumulative UV radiation and annual rainfall days) to assess the potential for microand nanoplastic (MNP) generation within the class A river basins in Japan. The MNP generation potentials for each grid were aggregated for each class A river basin boundary, and the values for the Tone and Yodo Rivers had large catchment populations and areas. On the other hand, the MNP generation potential per catchment area was found to be extremely high in the Tsurumi River, which has a high population density. The highest generation potential per catchment population was found for the Ota River, which has a Seto Inland climate with low rainfall. In the future, the MNP generation potential should be clarified by taking into account the generation of MNPs due to physical external forces in the basin and the surface runoff effect induced by rainfall.

 In order to continuously measure sediment dynamics during flooding in the observation box installed in the Nenotani River of the Ota River system, we developed a new measuring device using a hard material that do not deform easily against stress, compared and investigated different three measuring patterns. The measuring principle of the device is based on a method in which pressure gauges are installed inside and outside a water-filled container. After sediment deposition, the pressure gauge inside the container measures the total stress of the deposited sediment, the pressure gauge outside the container measures the pore water pressure. The difference between the two is used to derive the sediment deposition height. The results of the application of this method to the observation box showed that it was not possible to properly evaluate the sediment accumulation height for the plastic bag type and the polyethylene tank type. For 18-liter square can type, it is considered that the sediment accumulation height can be evaluated to some extent by applying a correction formula based on field experiments and by considering the stress acting on the wall surface.

 This study proposes a generalized method for predicting Japanese dam inflow with various basin characteristics. Climate change has several consequences, including frequent major flooding. Accordingly, the Japanese government has announced a policy against floods, such as pre-flood discharges, even at wateruse dams without flood control measures. Proper dam functioning requires establishing an inflow prediction method. However, creating prediction models individually for the numerous dams that exist throughout Japan is complicated and requires intensive labor and time; thus, establishing a generalized prediction method that can be applied to multiple dams would be advantageous. In this study, we attempted to generalize the inflow prediction method using Elastic Net, a sparse modeling method. We clustered dams based on the geographical information of the basin and selected a model representative for each cluster to generalize the prediction method.

 Objective of this study is to propose year-round dam inflow forecasting method by introducing snowmelt into the Elastic Net, in addition to taking into account the uncertainty of the forecast by using ensemble rainfall forecasts. In recent years, climate change has caused more frequent and severe flooding not only during the summer due to heavy rainfall, but also during the snowmelt season.With this in mind, there is a need for inflow forecasting that can be used for effective dam operation. In this study, we used Elastic Net, a sparse modeling method, to predict inflow at a dam in Hokkaido, Japan. The ensemble rainfall forecast was introduced as an explanation variable to the model. In addition, the snowmelt calculated by the Degree Hour method using the Analytical Snow Depth and air temperature in MSM was used as an explanation variable, and it was shown that the inflow rate could be predicted with good accuracy throughout the year.

 This study reports on the applicability of the UAV green laser, which can simultaneously measure land and sub-surface areas, to the measurement of dam sediment topography and the utilization of the 3D data obtained. The study targets three hydroelectric dams where sedimentation has been progressing in recent years. The main results of the study showed that (1) the acquisition rate of 3D point cloud data is generally more than 85% of the measurement area, (2) the measurement accuracy error satisfied the required accuracy, and (3) in areas where there was a possibility of missing measurements, it was effective to operate the reservoir level up or down according to the measurement method, and the combined use of narrow multibeam and green laser is effective. Furthermore, a detailed topographic mesh was created from the high-density three-dimensional point cloud data acquired by the green laser, and a reproducible simulation of flushing sand discharge was performed to construct an analytical model that can be used to predict sand discharge in the future. In conclusion, this study demonstrates the applicability of the UAV green laser to the measurement of dam sedimentation and the usefulness of the acquired data.

 This study proposes a workflow to support decision making by providing materials for decision making, such as grasping the effective sediment deposition situation using from ALB measurement and objectively determining the priority of countermeas-ures mainly based on the evaluation of the sufficiency of flow capacity in small and medium rivers managed by prefectures with long management extension. This paper carries out the analysis by this method in 4 rivers of prefectural management of different river scales, and examines the effectiveness of the method, while constructing a mechanism for taking measures mainly in neces-sary places considering the degree of safety and importance in flood control by effectively utilizing limited resources of river management. In addition, this paper discusses the actual condition of flood control safety of prefectural controlled rivers, and discusses the criteria of future external force considering climate change as an application of this method.

 Due to climate change and severe fiscal conditions, there is a demand for the effective utilization and efficiency improvement of existing dams in dam management practices. In response, this study constructed an AI model capable of predicting appropriate discharge amounts during low water management by applying reinforcement learning techniques to three dams in the Dozan River of the Yoshino River Basin. The reinforcement learning method utilized was Twin Delayed Deep Deterministic Policy Gradient(TD3), which we believe to be more suitable for predicting dam operations compared to Deep Q-Network(DQN) method frequently used in previous research. The results showed that, compared to operations conducted in accordance with operational rules, it was possible to predict operations that could increase power generation without significantly increasing the number of drought adjustment days. This confirms the effectiveness of this method as a support for future river flow management of dams.

 In recent years, a lot of river bridge decks have been washed away due to heavy rain. Our purpose is to investigate the velocity and pressure distributions characteristics around a bridge deck during overflow based on the experiments. The results showed that the velocity distributions are an asymmetric between the upper and bottom of a bridge deck. It is suggested that the flow velocity particularly has been developed on the bottom side. On the other hand, it has occurred the negative pressure on the bottom side during overflow caused by separation flow, and then, it is suggested that negative lift force affected on a bridge deck.

 Dry masonry revetment is one of Japan's traditional revetment construction methods, and its value is being reevaluated in recent years. However, the lack of knowledge on the strength of dry masonry revetments and the lack of clear design criteria have prevented their adoption as a method of revetment construction. This study focused on the mechanism of backfill material suction out from the inside of revetments, which is one of the causes of damage to dry masonry revetments, and aimed to clarify how the mechanism is affected by different structures of drop-off structures. The results showed that backfill material is sucked out by the pressure difference in the longitudinal direction, and that the tendency of suction differs depending on the structure of the dropout structure, that suction is active near the point of maximum flow velocity near the revetment, and that the rock weir is a structure that is less likely to cause suction.

 In recent years, cases of water leakage and sand erosion caused by levee seepage have been reported in many parts of Japan. Since it is difficult to deal with such deformations only by flood control activities during outflow of water, it is important to identify dangerous areas in advance and take countermeasures. In this paper, the conditions that affect the safety of seepage are summarized from various literatures, and a case study is conducted focusing on the flat shape of levees, which has not been considered as a condition so far. The results of the case study showed that leakage and sand erosion were more likely to occur on the inner bank of the curved portion of the levee, and the significance of this result was proved by a hypothesis test. The reason why leakage and sand erosion are more likely to occur on the inner bank was examined using a simple flow-netting method. The results showed that the local hydrodynamic gradient tends to increase on the inner bank, and that it depends on the curve radius. The results suggest that it is necessary to consider the planform of the embankment, which has not been considered so far, as a condition that influences the safety of seepage.

 In response to the increasing frequency of flood events due to climate change, the importance of flood risk information has grown. Flood risk is composed of hazard, exposure, and vulnerability. In flood inundation analysis, which evaluates hazard, the assumption that levee breaches occur when river water levels reach the design high water level (HWL) is commonly used. However, actual flood events often show that levees do not necessarily breach at HWL, and breaches due to seepage or erosion without overtopping have also been reported. For quantitative hazard evaluation, it is essential to establish levee breach conditions that reflect actual phenomena, and probabilistic evaluations are already implemented in several countries. Therefore, this study developed a fundamental breach probability evaluation model that simplifies the erosion-induced breach process. The model was applied to the Tokachi and Satsunai Rivers in eastern Hokkaido, demonstrating that the relative breach probabilities at different evaluation points correspond to the levee and external conditions input into the model.

 Precipitation forecasting covering the entire rainy season is a key to water resource management in the Asian monsoon region. This study developed two models, one for 6-month rainfall prediction using convolutional neural networks and the other for predicting its reliability. 6930 samples from 42 different models participating in CMIP6 were used to predict the accumulated rainfall over the Asian monsoon region from May to October. The anomaly correlation coefficient between the predicted values and the true values was 0.89, indicating the validity of the model. On the other hand, the test results revealed that about 15% of the sample had poor prediction performance. This study proposes a “reliability index” to identify this sample group. A new model was trained with discrete values of rainfall, and the “reliability index” of each sample was calculated from the predicted probability distribution of rainfall. The results show that the reliability index can discriminate sample groups with poor prediction performance in advance.

 GraphCast, a recently developed machine learning-based weather forecasting model, can predict global weather with a spatial resolution of 0.25° for the next 10 days. This study evaluated the precipitation prediction skill of GraphCast in the Japan region by comparing it with the results of ECMWF’s deterministic model (HRES) and ensemble mean (ENS). GraphCast exhibited the lowest mean absolute error in precipitation predictions for most lead times, demonstrating superior performance over the three methods in overall precipitation predictiong. Additionally, the prediction skill for heavy rainfalls exceeding 100 (mm/24h) was evaluated using the Threat Score. The results indicated that HRES performed best up to a lead time of 3 days, after which GraphCast excelled. Furthermore, it was revealed that GraphCast's skill in predicting heavy rainfall for lead times of 4 to 10 days was comparable to that of HRES at a lead time of 4 days.

 In river construction site, it is required to assess whether there will be flooding 24 hours ahead in order to safely evacuate construction workers, heavy machinery, and construction materials.We have already developed the method for predicting water levels 24 hours ahead using Neural Network based on the spatial distribution of rainfall as input.In this method, the water level is predicted from the mean and variance of the rainfall, the center of gravity of rainfall, and the distance from the prediction point to the center of gravity, which are the parametes representing the spatial distribution of the rainfall. However, this method cannot take rainfall distribution into account in detail, resulting the predicted water level being lower than the actual water level, which is unsafe. In this study, a method was considered in which the rainfall area was divided into several parts to take into account the rainfall distribution, and was applied to water level predictions for four watersheds. As a result, it is found that underpredictions were improved and that the improvement was more effective when the rainfall distribution was uneven, i.e., when the variance values of the divided rainfall were large.

 We conducted a comparative accuracy analysis of LSTM and FFNN models, which are major models in deep learning, for prediction of dam inflows.The dam inflow prediction model aims at long-term forecasting, using only precipitation data as input.For FFNN, we prepared two deep learning models, FFNN(A) and FFNN(B), with different optimization functions. The input data, which consists of precipitation data up to 720 hours prior, was divided into several intervals and time-compressed to enhance accuracy.Among the four dams examined, three were constructed using only basin-averaged rainfall data, while one included point rainfall data to verify its impact.Among the three dams constructed using only basin-averaged rainfall data, FFNN(B) exhibited the highest accuracy. In the case where point rainfall data was added, LSTM showed the highest accuracy.However, it is likely that the model's superiority is influenced not by the presence of point rainfall data, but by its ability to learn from floods of a scale similar to the validation flood.

 In this study, we focused on vehicle drifting accidents caused by flood currents during heavy rainwater disasters and conducted vehicle drifting experiments using an actual small vehicle on an outdoor large-scale open channel. In an open channel with a gradient of 1/100 and a width of 2 meters, asphalt pavement was applied from a position 16 meters from the upstream end, and vehicles were placed facing upstream on it. We conducted three cases, two of them with the handbrake released and one case with the brake applied, gradually increasing the flow rate and measuring the flow conditions when vehicle drifting occurred. As a result, vehicle drifting occurred in the two cases where the handbrake was released, and even when the maximum flow rate was reached in the case with the brake applied, the vehicle did not move, suggesting that whether or not the brake is used greatly contributed to the occurrence of drifting. Furthermore, compared to previous model experiment results, the water depth and flow velocity when vehicle drifting occurred were slightly smaller, indicating that there is a risk of being swept away even in milder flow conditions.

 An open-channel flow experiments ware conducted on a model of a gravel bag to measure the flow field around the gravel bag and the fluid forces acting on it. The velocity and pressure distributions were measured at the upstream and downstream cross-sections of the gravel bag, revealing the flow characteristics around the gravel bag. Furthermore, using the measured velocity and pressure distributions at the upstream and downstream cross-sections, the fluid forces acting on the gravel bag were determined by considering the momentum balance. By comparing the estimated critical friction force, which considers buoyancy and lift, with the fluid forces obtained from the momentum balance, it was demonstrated the effectiveness of the measuring method of the fluid force by measuring the pressure distribution and considering the momentum balance. Additionally, it was confirmed that when measuring fluid forces based on the momentum balance, it is important to measure pressures that differ from the hydrostatic pressure distribution.