This paper describes a numerical model developed to simulate the flow and scour around a bridge pier. The flow is modeled by large eddy simulation. The morphological change of riverbeds is calculated by coupling sediment transport with models for sediment pick-up and deposition. The sediment transport is calculated by 3D momentum equations of sediment particle. Sediment pick-up and deposition are mod-eled by stochastic models for the purpose of reducing computational demand. The numerical model was applied to the local scour around a vertical cylinder pile under clear water condition. The results agreed well with experimental data. It is concluded that this model can evaluate a local scour considering 3D movements of sediment particles.

 The spacing between two spur dikes and the local scour depth are two important considerations in spur dike design. The spacing is related to the downstream separation zone of a spur dike. In meandering channels, the flow field and sediment transport around a spur dike are affected by both the spur dike location and channel sinuosity. The paper employs a 3D model to simulate the flow and sediment transport around a spur dike with different location in meandering channels under different sinuosity. The results show that the downstream separation zone of a spur dike shortens as the spur dike moves downstream or the channel sinuosity decreases. Therefore, the spacing can be adjusted with the location and channel sinuosity for costeffectiveness. The local scour depth around the spur dike tip increases as the spur dike moves downstream and reaches the maximum when the spur dike is placed near the crossover section of a meander, where measures may need to be taken to mitigate the local scour.

 The problem of large amounts of sand deposition at movable weir often occurs both upstream and downstream of weir at the time of flood, and investigation of the cause is an important issue for river management. Sand deposition at movable weir at the time of flood is especially remarkable in downstream area, and is often considered to be a problem. However, there are few studies on sand deposition at upstream side, and there are many unclear points. In this study, a hydraulic model experiment using a model weir in an experimental channel was conducted with reference to the Heisei Grand Weir.The results showed that increase tractive force in upstream of the weir promoted sand movement and deposition. In addition, the water surface dropped below the lower edge of weir at high flow rates, and suggested that this is caused by the formation and movement of dune.

 Recently, levee breaching due to overtopping flows which causes severe damage in residensial area is reported, and for disaster mitigation it is of great importance to understand the mechanism of breaching process in terms of flows and sediment transport, and consider countermeasures. In this study, a numerical model is refined by considering the effect of pressure gradient near a bed in the sediment transport model, and applied to the hydraulic experiment of levee breaching. Comparing the numerical results, it is shown that erosion at the toe of embamkment is slightly simulated by considering the effect of pressure gradient.

 Stones and gravels can cause serious abrasion at a dam. This study attempts to clarify the mechanical characteristics of transport of a stone and gravel along a dam through the experiment and computational fluid dynamics (CFD). The CFD is used to evaluate the fluid force acting on the particle that has been tracked in the experiment. The momentum equation of the particle is verified from the both side of the experiment and CFD. The results show that pressure gradient force has dominant influence on jumping at the top of the dam. The pressure gradient force relative to the particle mass does not depend on the particle size, which impies that even large particle can jump at the spillway and collide with downstream bed.

 There is concern that the risk of riverbank erosion will increase as sandbars develop. Therefore, in this study, we conducted laboratory-scale experiments. A set of experiments were conducted where we formed sandbars, and changed the sandbar height and flow discharge in each case. As a result, it is verified that the riverbank erosion rate increases as the sandbar height or flow discharge increases under the tractive force condition that some sand on the sandbars is transported. On the other hand, our experiments show that when the flow discharge increases to the extent that most of the sand in sandbars is transported, the erosion rate under high flow discharge become smaller than that under low flow discharge.

 We organized the final level and goal of rainfall forecast accuracy required for dam pre-release by targeting at multi-purpose dams in Japan and rainfall for the past 16 years, as the use of numerical rainfall forecast for it is becoming more active at dams nationwide. Specifically, we assessed the impact on the index directly related to the judgement of pre-release calculated by inputting observed rainfall as it is or with processing to runoff-dam model. As a result, just several events requiring pre-release happened nationwide per year. They were caused mainly by typhoon body and stationary front in descending order. Required lead time is within a day and a half in Typhoon cases, and ranges to 2-3 days in front cases. This indicates the whole lead time of current Meso-Scale Model should be high accurate and be also extended further. About spatial distribution, it is indicated that current model’s resolution is generally sufficient, and the level of position accuracy should be finer than the one of current forecasting by one digit.

 Pre-release is the effective measure against large-scale floods that have frequently occurred in recent years, whereas its implementation period based on ‘Pre-release guidelineis’ is limited. In addition, there are few considerations of pre-release in tandem series of dams including hydropower ones. This study investigated the effect on the maximum discharge of each dam which shows the flood control effects and the waste discharge of hydropower dams, which is directly linked to the amount of hydropower generation, by changing the start time and target water level of pre-release in tandem series of dams. At the time of large-scale flooding, the maximum discharge of each dam and the waste discharge were reduced by the pre-release. Furthermore, by advancing the start time of the pre-release from 3 days to a week and increasing the target water level drawdown, the pre-release effects on both flood control and water utilization has increased.

 As the frequency of droughts increases with climate change, more effective use of existing dams is required. In this study, we developed a model to support the optimal operation to effectively use the capacity of dams using reinforcement learning and verified its usefulness. The developed model predicts the optimum replenishment quantity based on various conditions such as the water storage quantity and inflow quantity of each dam at the present time. In the reinforcement learning model, it was possible to predict the shortage of the reference point and the operation with less invalid discharge compared with the operation in which the replenishment quantity is set according to the water storage ratio of each dam. In addition, as a method for effectively utilizing the capacity of a group of dams, efficient operation in accordance with the water storage conditions is required based on the characteristics of individual dams, such as the inflow volume and storage volume. The predicted amount of supplementation by the reinforcement learning model was almost consistent with this trend.

 Improving the flood control function of dams is being considered to help mitigate the damage caused by intensifying floods. In this study, with the aim of improving the flood control function of dams in the upper Tone River basin, deep reinforcement learning of artificial intelligence (AI) was applied to the integrated flood control operation of dams to reduce the water level downstream. By applying the Deep-Q-Network model to AI reinforcement learning, a model was constructed to appropriately limit the discharge of the dams. In addition, using the discharge of each dam together with water level prediction model downstream of the dams, reinforcement learning was conducted to reduce the water level at the downstream points. Using unseen floods to validate the model, it was shown that many of the validation floods achieved lowerwater levels compared to the normal dam operation. Furthermore, it was demonstrated that using this model, the dams did not need to perform an emergency water discharge, and the validity of the model was thus confirmed.

 To observe of seepage rate at rockfill dams is important for dam safety management, but the observed volume includes that from the reservoir through the dam body and foundation, as well as that from meteorological effects such as precipitation. In order to therefore grasp the state of the impermeable function of the dam body daily, the observation results such as the seepage rate, reservoir level, and meteorological conditions at the managed dams have been organized and analyzed, and established general-purpose model to evaluate how much percolate, considering the relation between the percolation volume and the reservoir level and rainfall response characteristics. As a result, the validity of the model was confirmed by applying it to dams with different meteorological conditions and measurement status of seepage rate. In addition, since the base seepage rate of the dam body can be separated from the calculated value of the general-purpose model, it is inferred to be applicable for the evaluation of dam body safety such as the abnormality judgment of the seepage rate.

 In this study, a new flood-control concept "Cascade" method was considered in which the upstream Shimouke dam can overflow from its emergency spillway while the downstream Matsubara dam is required only to use its normal spillway and never its emergency spillway in the upper reaches of the Chikugo river basin under the heavy rain in July 2020. As a result of this study, it is made clear that the flood control capability in the basin can strengthen by changing the design high water discharge of the Shimouke dam. In addition, the current design high water discharge of the Shimouke dam, 350 m³/s, is appropriate for cases under uniform precipitation condition in the basin. Moreover, flood control capability with Cascade method compared with a conventional method can be estimated in cases where there is an inflow from a tributary without a dam in the basin where two dams are constructed in series.

 The objective of this study is to propose flood risk reduction measures involving dam linkages, with consideration of the flood characteristics of snowy cold regions due to climate change. In the case of multipurpose dams in snowy cold regions, disaster prevention operations may be carried out during abnormal floods due not only to heavy rainfall in summer but also to snowmelt and rainfall in early spring. In this study, the frequency of disaster prevention operations during abnormal floods caused by rainfall and snowmelt in the present and future climates was estimated for two dams in the upper reaches of the Toyohira River in Hokkaido, using d4PDF ensemble data of rainfall and temperature. In addition, we examined whether flood control safety can be improved by transferring the flood control capacities of the two dams, taking into account their functions and hydrological characteristics. As a result, it was shown that dam linkage is effective as an adaptation measure against climate change.