In recent years, heavy rainfall disasters have caused frequent damage to bridge piers due to scouring and the resulting fall of bridges in many areas. The purpose of this study is to develop a model for evaluating risk of disaster due to scouring based on machine learning. We developed models by four different machine learning algorithm based on the data of disasters in the past, and compared the performance of the models. As a result, we selected the neural networks(NN) algorithm as its the most performance. In addition, we developed the model to apply NN based on importive explanatory variables selected by the feature selection, the model can make possible to evaluate the risk for initiation of disaster due to the scouring.

 In July of 2018, a flood with return period of 1/15 years attacked the Yubetsu River basin in Hokkaido. The Iwane Ohashi Bridge over the Yubetsu River suffered great damage. This flood resulted in a subsidence of a bridge pier, causing a great deformation of the bridge girder. The detailed understanding of the reason of the damage will provide a useful insight into future management of the river and bridge. In this study, we investigate the main cause of this disaster by using numerical model of flow and bed evolution, iRIC-Nays2DH. As the results of numerical calculation of two-dimensional bed variation, bars are not able to form during the high discharge condition. The presence of the bridge pier greatly affects to the flow and bed evolution around the bridge pier, resulting in a deep scouring around the pier. This scouring might be the main factor causing the damage of the bridge.

 We have developed a method to estimate inundation ratio by microwave radiometers, and to downscale it into 15-second grid with the help of geospatial dataset in Japan. In this study, this method is applied to all land area in Japan (excluding some remote islands), then high-resolution surface water map (GSMaWS) is produced. GSMaWS is compared with simulated inundation area by Today’s Earth-Japan (TEJ) and it is found that GSMaWS tends to overestimate inundation ratio in urban area, in Hokkaido, and along the coast, but GSMaWS and TEJ show positive temporal correlation in many river basins in eastern part of Japan in October 2019. For the flood event caused by Typhoon Hagibis, inundation area estimated by GSMaWS and TEJ are evaluated with the reference of inundation map published by Geospatial Information Authority of Japan. Just after the peak of heavy rainfall, in early morning of October 13, GSMaWS shows similar estimates of inundation area with TEJ, while the accuracy becomes worse during heavy precipitation.

 A heavy rainfall caused by Typhoon No. 19 resulted in flood inundation at the Chikuma River in Nagano Prefecture Japan on October 13, 2019. In particular, severe flood inundation damages at Naganuma district along the Chikuma river were observed. In this paper, flood inundation situations and building damages in the district were focused and reported based on an on-site field survey and numerical simulations.

 From the results of analysis, it was found that highly damaged houses were concentrated along the two flow pathways in the inundation areas because of the local bathymetry and land surface roughness. Also, highly damaged houses were found in the zone where the maximum flood flow velocities are more than 2 m/s and inundation heights are more than 2 m. We concluded that the high resolution information would be required to represent the flow behavior and building damages near the river bank failor point.

 A disaster caused by heavy rain and storms occurred due to the effect of Typhoon No. 19 in October 2019. In Nagano City, large scale inundation damage was caused by overflow and dike break from the left bank of the Chikuma River. In this study, the inundation analysis model was developed to examine the inundation water behavior. The obtained inundation water depth was good agreement with the observed inundation water depth. Therefore, the validity of the analytical model was shown. In addition, the feature of drainage was shown from the analysis results considering drainage process. However, sewerage system, waterway and drainage by pump trucks were not taken into consideration in this analysis model, and future issues regarding model improvement were also shown. In addition, the analysis model with 2m grid was developed on the dike break site, and the inundation water flow on the road and the fluid force around the house were obtained. From these results, the state of hydraulics on the dike break site was shown in detail.

 This study aims to compare flood control effect of two retarding basin against excess flood by field surveys and flood simulation. Field sites were chosen at Ichinomiya River 2^nd retarding basin and Ara River 1^st retarding basin which were flooded due to big typhoons in 2019. The results showed that the Ichinomiya River 2^nd retarding basin showed the limitation of flood control function under huge flood like typhoon no. 21 even though the height of overflow weir was changed. The Ara River 1^st retarding basin showed the flood control function in the large flood due to typhoon no. 19. Also noted that selecting appropriate heights of the overflow weir is important to reduce flood discharge and water level in the downstream reach of rivers.

 Typhoon Hagibis 2019 recorded heavy rainfall at a lot of area in east Japan. As a result, levee breaching occurred at four places in the Tokigawa River and two places in the Oppegawa River. When the river channel management is planned, the inundation between the river and the flood plain should be understood to discuss the change of the high-risk area. Therefore, the effects of the levee breaching of the Tokigawa River and the Oppegawa River on flood volume and discharge-hydrograph of downstream were considered. The levee-breach timing and time it took for the levee to be washed away on the 0.0k right levee of the Oppegawa River greatly affected the inundated volume. Considering the flood volume from the Otanigawa River, which is a branch of the Oppegawa River, the flood volume in some simulation case was similar to that of the actual case. Condition of the simulation case was that the levee-breach time was 24:00 on October 12, and it took 3 hours for the levee to be wash away. On the other hand, the effect of the levee breach on the downstream peak discharge was small. In Typhoon Hagibis, the duration of rainfall was longer than the time until flood reached downstream; therefore peak discharge continued long after when levee breaching occurred.

 Continuous disasters occurred during short terms would cause more severe flood damage. In the present study, to consider how we should make disaster recovery, we investigated the disaster damages by the floods of Natsui River occurred due to the Typhoon No. 19 and No. 21 in 2019 based on the field surveys and numerical simulation. The field surveys clarified that the inundation area due to the Typhoon No. 19 was approximately 20 times of that due to the Typhoon No. 21 and the overflow due to the Typhoon No. 21 were confirmed at the same point where the levee collapsed due to the Typhoon No. 19. On the other hand, the numerical simulation indicates that it could be possible to occur the comparable inundation disaster with when flooding due to the Typhoon No. 19. Our results suggest that it is important to establish a strategy to evaluate flood risks and to develop a rapid restoration method of collapsed instructure in the future.

 Typhoon No. 19 which occurred in October 2019 caused a lot of damage in wide range of Kanto region in Japan due to levee breaks, river floods, and urban inundation. As global climate change becomes more extreme, typhoons and localized torrential rain, which cannot be handled by the existing infrastructure currently, are increasing. This is no exception in any cities in Japan, and once unprecedented rain falls, river floods and urban inundation damage will occur. In this study, we used S-uiPS, which is a sophisticated inundation prediction method, to reproduce the inundation cases in the downstream area of the Tama River. For the input rainfall at that time, the XRAIN actual rainfall data of Typhoon No. 19 in 2019 is used. From this reproduction calculation, it is clarified that even if the amount of rainfall in the target area is small, if there is high-intensity rainfall upstream and the water level in the river rises, significant flooding may occur in the target area.

 Evacuation is promoted by using hazard maps, but there is a problem of delayed evacuation due to the unexpected flood inundation by a levee breaching arong the rivers. In this study, rainfall-runoff simulation was conducted on all rivers in Toyama Prefecture to evaluate flood hazard areas in the prefecture by evaluating their overflow, erosion and infiltration potentials of the river levees.The results of the evaluation of each potential indicate the risk of overtopping due to water level rise at the confluence and the risk of erosion at the bend. Evaluation of river flood risk areas by combining the three potentials. Erosion was highly rated in the overall evaluation, and the estimation of risk areas reflects the erosive effect of rapids, which is a characteristic of rivers in Toyama Prefecture.

 The oil spill to inland inundation area requires works such as collection and removal of spread oil and countermeasures to avoid the outflow to adjacent water area, besides the drainage of inundation water. Thus, it is considered that the time to recover the devastated area will be becoming longer.

 In this study, first, for predicting the oil spreading in water area, a numerical model was developed. This model was constructed by flood model and oil spill model. From the comparison between computed results and literature, although there are some rooms for tuning the model parameters, it was shown that the processes of oil spreading were appropriately computed by the developed model. Next, numerical simulation were carried out with assuming the oil spill accident at inland inundation area. Calculated results showed that oil was widely spread by the effects caused from oil characteristics as well as flood flow.

 In this study, we developed a numerical simulation model for predicting advection/diffusion of oil spills due to flooding and implemented the model to analyze the oil spills generated by the heavy rainfall in Saga prefecture in August 2019. The simulation results revealed that it is necessary to include the influence of the water surface gradient in addition to the influence of the flood flow for accurately reproducing the advection of oil. Also, the results showed that oil gradually spread due to oil film diffusion even after the flow velocity decreased.

 The purpose of this study is to develop a numerical inundation model considering the effects of inundated buildings. Flood experiments considering inundation inside of the buildings were carried out, and their measurement results were precisely reproduced by a numerical model which applies the Torricelli’s theorem to estimate the inflow discharge at the building boundaries. The inundation model was applied to a real-scale urban area to find the significant difference from the simulation results of the conventional model of the MLIT guideline. Consequently, the effects of inundated buildings are not negligible, and the Torricelli’s theorem is applicable to estimate the inflow discharge at the building boundaries. In addition, the treatment of the buildings with more comprecated configuration and how to determine the area and the height of small inlet of the building walls should be more discussed.

 To conduct the numerical simulation accurately in urban area including houses or buildings, research on increasing the accuracy of two-dimensional (2D) flood inundation model is necessary. The objective of this study is to grasp the effect of anisotropy of porosity in calculation grid on a flood inundation pattern by comparing a hydraulic model experiment with numerical simulations by a 2D flood inundation model which includes an anisotropic porosity in each grid. In numerical simulations, anisotropy was represented by considering an anisotropic porosity model comparing with experiments. It is expected to improve the calculation accuracy by giving an appropriate anisotropic porosity to the model developed in this study.

 This study attempts to conduct a monitoring & evaluation (M&E) of the integrated flood risk management (IFRM) in Metro Manila. In this study, we monitored the performance level at each city in Metro Manila by carrying out a series of interviews and surveys. The performance for IFRM based on the 15 indicators we utilized in this study was appraised qualitatively by the practitioners from each city in Metro Manila. Then, the monitored qualitative appraisals were evaluated using the fuzzy set theory, which was a suitable method for handling imprecise or qualitative data. The indicators and qualitative appraisals were given fuzzyweights, and these were aggregated to evaluate the overall performance for IFRM for each city. The results show that majority of the cities in Metro Manila are performing above the Good level, but this level still suggests that more work and attention are needed to attain substantial achievements for IFRM. The approach in this study is suitable for M&E activities that heavily depended on qualitative data or information.

 According to the River Law in Japan, municipalities are to manage Class C rivers and Non Statutory rivers. Basically, the Class C rivers require river management mutatis mutandis to the River Law. In this study, we investigated the current situation and issues of local river management focusing on the independent water systems in Nagasaki. We conducted statistical analysis, interviews and field surveys. As a result, we found that the management and the manager's burden are different between rivers connected to the Class A or B water systems and that of the independent water systems. Though municipalities are addressing such difficulties as the lack of personnel and river inspections, it may be difficult to achieve the management required by the River Law only by the river managers. On the other hand, the active maintenance by residents played an important role to keep rivers in good status. Considering current situation of rural small rivers, we propose to introduce a river management system that incorporates residents as well as to establish a river management archive to share information among river managers, residents and stakeholders.

 To rebuild robust infrastructures against severe natural disasters caused by climate change, our study extracted crucial data from the dam-construction guideline that was submitted to the expert panel on future flood control measures by Japan Ministry of Land, Infrastructure and Transport, and then estimated the costs for the seven flood control countermeasures that consist of dam construction, river channel dredging, set-back levees, levee heightening, detention basin, cut-off channel and circle levee. We proposed a simple estimation method that can obtain the feasible business costs related to flood controls. As a case study, future flood risks under the impacts of climate change and costs of flood damages were predicted for Naka River flowing through Fukuoka City. The most economical and effective combination for the implementation of adaptation measures to the target river was examined based on the estimated cost of flood damages.

 Studies on the change of real estate prices pre- and post-floods are conducted to understand residents’ flood risk perceptions. Earlier studies suggest, in flood-prone areas, real estate value will be minimally affected by flood occurrences. At the same time, real estates are valued less in the high-risk areas within the flood-prone areas. However, studies have suggested that information on the degree of flood risk, correlated with the price decline, has changed due to improved flood control performance. Thus, the impact of flood risk information on real estate prices is expected to be increasing, given the fact that residents are increasingly accessing flood risk information. In this study, we analyzed a relationship between land price and flood risk information targeting Nagoya City, a region renowned for its flood-proneness. The analysis reconfirmed that the areas previously flooded are priced low at regular times and do not change after floods. The study also confirmed the information on the estimated flooding area has a larger impact on price than the actual flooded area, and the price drop of real estates in the area with no recent flood experience is larger within the estimated flooding areas.

 A multi-agent evacuation simulation model based on C ++ language was constructed for Taehwa-dong, Jung-gu, Ulsan, Korea. Taehwa-dong currently does not have quantitative criteria on the location and size of shelters during inland inundation so that the evacuation time by age group was reviewed for the necessary evacuation strategies in the area. In addition, a temporal evacuation shelter was designated to improve the long evacuation time in the region. This study is useful for the preliminary planning of the road maintenance and evacuation facilities as well as the suggestion of optimal evacuation routes.

 The purpose of this study is to improve the presentation of hazard maps and promote evacuations in areas where residents typically underestimate flood risks. This study proposes a new risk rank evaluation method using flood inundation simulation results. Two rivers, the Jinzu and the Jouganji River, were chosen, and Toyama City was selected as the target area. We simulated the worst-case flood inundation to demonstrate its effects to the local people. In addition, a new risk rank evaluation of the flood inundation that identified safe and unsafe areas was proposed to formulate a new flood hazard map in this study. Based on the results of the flood simulations, hazard maps consisting of only two colors (safe and unsafe zones) were created based on the new risk rank proposed in this study. We believe that the hazard maps that uses only two colors can have a strong impact on the residents, who underestimate the risks, and can promote evacuation.

 The possibility of monitoring inundation process during floods using ETC2.0 probe data was investigated by comparing between the probe data and numercal flood simulation during three severe flood events in Japan. The comparison indicates that the traffic volume significantly reduced during the floods. In addition, the behavior of a single car during the flood corresponded to the inundation process. To further understand inundation process, “Dimensionless vehicle traffic volume” defined by the ratio of the traffic volume during flood stage to mean traffic volume during normal stage was computed. This was significantly reflected the progress of inundation on a main road during the flood. Consequently, our results demonstrate that ETC 2.0 probe data can applicable to monitor the inundation process during floods

 Recently, heavy rainfall is increasing in urban areas and pluvial flood caused damage such as flooding above floor level. Therefore, it is an urgent task to develop a flood control plan that take into consideration the impact of climate change. In this study, we calculated 10-year precipitation for future climate in Osaka's Umeda district from out-put rainfall of climate model using d4PDF which is massive ensemble climate prediction data. As the result, it became clear that the rainfall intensity and total rainfall both increase compared with the present planned rainfall. Further-more, it was indicated that the reduction of inflow water by installing a water stop board at the entrance is effective for the future hourly maximum rainfall event.

 First, an identification method based on identifying the basic flow equations of open-channel unsteady flows in sewer pipe systems having Preissmann Slots with the basic flow equations of closed-conduit unsteady flows in urban sewer pipe systems, is proposed.

 Second, a practical identification method based on identifying the basic flow equations of open-channel steady flows in the sewer pipe systems having the widened Preissmann Slots with the basic flow equations of closed-conduit steady flows in urban sewer pipe systems, is proposed.

 There is an increasing possibility that rainfall patterns will change due to the effects of climate change associated with global warming. In addition, the use of underground spaces such as underground mall and subway is progressing in Osaka, and the subway network is expanding. In this paper, subway model considering the detailed shape of the tunnel based on the construction records shown in the published books is built, and conducted pluvial flood simulation.

 The results show that the detailed tunnel shape affects the propagation of flood water. In addition, the subway stations at high risk for flood can be pointed out by the improving subway model.

 In recent years, it has been pointed out that the number of local torrential downpours and the amount of rainfall in Japan have increased, and that they will continue to increase in the future. In urban areas, underground spaces such as underground shopping malls have been used. For this reason, when pluvial flood occurs in urban areas with underground space, there is an inundation risk not only above the ground but also underground. In this study, pluvial flood analysis was done using assumed the maximum rainfall, and the inundation risk of an underground mall was evaluated. Then, using the data of flow velocity and water depth obtained from the analysis results, the safety evacuation of the underground mall was examined by calculating the specific force per unit width. In addition, as a countermeasure of underground inundation, the effect of disaster prevention by installing flood boards was examined. From the research results, it has been shown that installation of waterproof boards was effective in reducing the damage caused by underground inundation.

 The Cabinet Office of Japan issued that the Nankai Trough Earthquake will occur in the next 30 years with a high probability of occurrence. In urban areas, people use subways and underground spaces. These tend to be flooded by the tsunami. Therefore, people have to evacuate from the subway stations after the tsunami warning will be issued. The study area is Osaka City, which has many subways and underground mall. InfoworksICM (sewer/river integrted simulation software) was used to calculate the flood depth and the start time of the flooding subway stations. From the results, over 80% of all subway stations are flooded. All stations in study area were classified into stations directly inundated from the ground, stations indirectly inundated through subway tracks from other stations and non-inundated stations. It was revealed that the number of subway stations indirectly inundated through subway tracks from other stations was more than that of stations directly inundated from the ground. In addition, since water overflowed from the entrance of the subway stations and the tunnel gates, it is necessary to take measures to prevent secondary damage.

 To advance the flood control in the Rokkaku River basin, we examined the effectiveness of irrigation ponds as a measure against flood inundation inside the levee. The flood control function of irrigation ponds was evaluated by applying a distributed rainfall-runoff-inundation model (RRI model) and adopting data of the heavy rainfall in August 2019. The result found that the existing irrigation ponds reduced the floodwater volume by 150,000 m³. Furthermore, we evaluated the flood control function of the ponds, considering the pre-release of water from the ponds in the same rainfall condition. As a result, we found that the flood water volume decreased by 383,000 m³ compared with the case of no irrigation ponds and its prerelease. Besides, the pre-release of water from ponds reduced the inundation depth in the area many houses exist. In conclusion, irrigation ponds have a flood control function that mitigates the flood inundation inside levees.

 Under the global warming condition, it is indispensable to consider flood control measures on the premise of huge inundation along with bank overtopping or levee breach caused by record-breaking heavy rainfalls. At the same time, risk reduction ability of such measures have to be evaluated. In this study, multiple rainfall patterns were obtained by pseudo-global-warming numerical weather simulations, and runoff analyses and inundation simulations were implemented for Kakehashi River basin, Ishikawa, Japan. Inundation simulations were conducted under conditions with various types of makeshift flood protection walls which were assumed to be installed on existing roads. Then, inundation control ablity of the makeshift protection walls were evaluated. In addition, economic loss were estimated for each inundation simulation results. The results indicate that appropriate installation of makeshift flood protection walls could diminish inundation extent and depth in the residence and commercial areas and greatly reduce economic loss up to about 20%.

 The typhoon No. 19 in 2019 caused severe damages in large area of Japan. In the Ara River basin system, a broad inundation was caused due to the collapse of 7 points of levee along tributaries, and a flood risk was also extremely high on the main stream of the Ara river. Here, we estimated that the inundation area is 25.9 km² and the amount of flood water is about 31 million m³ based on the field surveys. In addition, we examined the possibility of flooding in the Ara River where flooding did not occur depending on the difference in rainfall distributions. We conducted runoff analysis by using RRI model and river flow analysis calculating 1-D unsteady flow using hindcast rainfall and ensemble forecast rainfall data. We calculated the river discharge and river water level by these analysis. As a result, we showed that the river flow increased due to the difference in therainfall distribution, and that there were some places where the flood overflow could occur in the Ara River.

 A flood forecasting method that can analyze the longitudinal water surface profiles time series of ongoing floods with high accuracy is required to ensure the timely evacuation of residents. In this paper, the water level prediction method using flood flow propagation and water level prediction method with modified runoff of tank model is examined based on the hydraulic quantities obtained from the assimilation analysis.

 As a result, the water level propagation time of the flood in September 2013 is clarified, and the predicted water level approximately agrees with the observed water level up to 3 to 5 hours ahead. The analysis method based on the assimilation flow rate can predict the water level by flood flow analysis using the flow rate with the modified state quantity of the tank model. This predicted water level approximately agrees with the observed water level over a long lead time. Therefore, it is shown that this prediction method is highly applicable to disaster prevention information.

 Agricultural drainage systems serve to eliminate floodwater not only from farmland but also from non-farmland in agricultural basins where urbanization is remarkable. To provide a quantitative evaluation for this issue, we have developed a model to visualize the runoff components of each land category contained in drained water during heavy rainfall. A tracer sub-model was constructed that traces rainwater falling on the target landuse by giving a tracer with a concentration of 1, and incorporated in the inundation analysis model developed in the previous research. As a result of applying this model to the area where rapid urbanization is in progress in recent years, it has been confirmed that components drained by the agricultural drainage pump station during flood peaks contains a large amount of rainwater derived from non-farmland, and contribute to inundation of farmland.

 In this study, we have developed a model to evaluate the influence of runoff from paddy fields used for upland crop cultivation （rotational paddy fields）on the basin scale. For the modeling, a field survey was conducted to confirm the outflow mechanism and characteristics of the rotational paddy fields. The results showed that the runoff from the rotational fields responded sharply to the rainfall pattern, the peak runoff was large compared to the rice paddy fields. We considered that the cause of the increase in runoff was the inundation between the ridges. To reproduce the runoff, a model was constructed that represent changes in the surface inundation depth by expressing the ridge shape with a cosine function. This model reproduced the observation results well. As a result of applying this model to the case watershed, it was suggested that the inundation damage would increase greatly through increase in the area of rotational paddy fields.

 A two-dimensional flood flow analysis was conducted to investigate the characteristic of inundation and the water storage function of abandoned paddy fields along Tai River, small river, in Toyooka City, Hyogo Prefecture, Japan. The simulation results showed that 25% of total discharge was stored in abandoned paddy fields during the event of 1/30-year rainfall. Furthermore, 22% of peak discharge was cut and stored by abandoned paddy fields. These water storage functions were provided weir and water bar to reduce velocity of flood in the abandoned paddy field, set by local people.

 A rain garden, which functions as runoff control by rainwater storage and infiltration, draws attention as a Green Infrastructure technology in an urban setting. This study developed a rain garden in a Community Café and a Detached House in Fukuoka City, with the goal of 100% runoff control of rainfall exceeding the Fukuoka City sewerage plan. After implementation, we conducted on-site inflow tests and water balance monitoring to compare the estimated values of the design with the actual runoff control effects. As a result of the inflow test in the Café, the runoff per unit time was 0 mm, which means that the water could be stored for more than 130 mm against the estimated water storage height of 100 mm. At the Detached Houses, we observed runoff from the rain gardens, focusing on the daily rainfall of 297.6 mm that occurred during the monitoring period. As a result, the observed runoff was about 0.24 mm, compared to the estimated runoff of 7.9 mm. Therefore, the rain garden designed for infiltration and storage was found to be highly runoff control effects against heavy rainfall.

 We carried out flood simulations using a numerical model, and verified the adaptive limit rainfall of the flood damage mitigation effect of the Paddy Field Dam (PFD) in the Somaigi River basin in Tochigi Prefecture. As a result, it was clarified that the maximum effect of the PFD is demonstrated by the rainfall with 120-year (268 mm/day) return periods in this basin. It was also confirmed that although the amount of inundation could be reduced up to about 700 mm/day of rainfall, it was confirmed that the effect of PFD decreased as the amount of rainfall increased, since the number of paddy fields flooding above the height of levee on paddy filed increased.

 This study aims to develop a reservoir-hydropower plant model and incorporate it into a distributed hydrological model to evaluate the effect of dam operation on river discharge and power generation. The model was composed of a reservoir power generation and hydrological processes with a concept of a kinematic wave-based assumption. By using the integrated model to assess the impact of hydropower development in the Nam Ngum Basin in Lao PDR, the results indicated that the change of river discharge at the downstream of Nam Ngum 1 reservoir is +218.8% during the dry season and –28.5% during the wet season in the full development scenario from the natural condition. There is no primary effect on the inflow of the Nam Ngum 1 dam by the operation of the under construction dam. On the other hand, the annual energy product of the Nam Ngum 1 has a minor increase.

 We developed the deep neural network model of dam operation(AI model) for flood control considering the flood risk at the downstream. First, we created a regression equation that predict the downstream river-discharge using the dam discharge as input data. We made reward function of the reinforcement learning which consider the difference between capacity of river channel and estimated river discharge.

 AI model was trained through the proccess of trial and error to obtain dam operation to reduce the peak discharge at the downstream station. The result of AI model was compared with the operation rules and previous AI model which does not consider the flood risk at the downstream, and consequently validity was confirmed.

 In recent years, there has been a demand for disaster countermeasures that make use of existing infrastructure. In this study, we develop dam operation models using deep reinforcement learning. To investigate the possibility of pre-discharage operation, we apply this model to pre-discharge operation by considering inflow prediction and precipitation prediction information. From the result of learning and testing with the dam operation model, we revealed that we can apply dam operation model to pre-discharge operation. On the other hand, the action that models select during testing will converge only one action when the pre-discharge operation and flood control operation are treated as a series of operation. As a result, model can fail in flood control operation.

 A simulatin-based study was conducted to analyze the effectiveness to consider the Medium Range Ensemble Rainfall Forecast of ECMWF in decision making for prior release of a hydropower dam. Prior release is initiated based on an ensemble member with the third greatest value of the maximum inflow rate, while the amount to release is determined by ensemble members with smaller values of predicted total infow volume. As a result of the application to a hypothetical target dam, it was demonstrated that the proposed method could effectively enhance its flood control capability while ensuring storage recovery.

 This study proposes dam inflow prediction methods that can be used for no-experience flood events. The frequency of large-scale flooding has been increasing in Japan, and to minimize damages, reliable dam inflow prediction is necessary to implement operations such as pre-emptive discharge.

 In this study, Elastic Net, a sparse modeling method, was used to identify intra-data relationships from limited information in order to predict dam inflow from 24 h of related climate and water data. The results showed that even in cases such as the August 2016 downpour in Hokkaido when heavy rains continued for some time and soil dampness changed, a conservative prediction is possible by adding the effective rainfall with a half-life of 720 h to the explanatory variables. Finally, the study confirmed that predictions suitable for practical use could be obtained for multiple cases of no-experience flood events.

 The authors evaluated a method using check gates to utilize in-channel storage in open-channel systems to absorb temporal mismatches between water supply and demand. The authors studied a paddy irrigation district that has had water shortage problems in its downstream area along a main irrigation canal. Field observations suggest that irrigation water is withdrawn preferentially into the upstream area, especially at peak hours for water demand. This is a possible reason for the water shortages downstream. To manage this problem, the authors present a method automating multiple check gates along the main canal so that a canal pool between an upstream control gate and a downstream control gate is used for storage. With this method, water can be stored in the pool during off-peak hours for water demand and released during peak hours. The authors tested the proposed method using numerical simulations. The simulations show that the method can moderate the decline in flow rates at canalside turnouts and the downstream end of the main canal during hours of peak water demand. With additional freeboard, the in-channel storage capacitiy can meet the simulated water demand without using reservoirs.

 Due to the increased frequency of floods caused by torrential rainfall in recent years, reservoir operation is expected to be flexible enough to mitigate damage in downstream areas during flood events, rather than following pre-defined flood control rules. This study proposes a new method to determine reservoir control options, using backward simulation of river discharge. A reverse routing method with a kinematic wave equation was performed using a time series of lateral inflows into a river channel obtained from a rainfall runoff model. Two case studies were examined for the Sameura Dam in the Yoshino River, Japan. It was confirmed that the proposed method was effective in determining the volume of water to release from the dam to control the downstream flood level, accounting for the volume and timing of inflows from tributaries and hill slopes.

 In recent years, preliminary release operation is receiving attention as one of the measures to effectively utilize the capacity of the dam. Up to now, however, There is no established method for predicting the effects of preliminary release operation on reservoir sediment characteristics and their future changes. In this study, we firstly examined the relationship between reservoir capacity and sediment progress based on sedimentation results of multipurpose dams. Secondly, we calculated the river variation analysis when the preliminary release operation was carried out to the model dam. Based on these results, we confirmed that the characteristics of sedimentation differed according to the ratio of flood control capacity to total storage capacity based on the dam sedimentation results and the river variation analysis. It was suggested that attention should be paid to changes in sediment characteristics in the case of preliminary release operation.

 In the past studies on a data assimilation, few studies have evaluated rainfall-runoff processes and hydraulic processes at the same time, and there are no studies using riverbed evolution models for real-time prediction. The objectivites of this study is to improve water level of accuracy using a paticle filter combined with the RRI model and riverbed evolutions. A particle filter is simultaneously and sequentially estimated riverbed evolution at the water level gauging and slope water depth, which is the initial conditions for the RRI model. In addition, a riverbed evolution model is predicted the riverbed according to discharge by the RRI model for the next six hours. The target river is Seri River in Shiga Prefecture, Japan. As a result, the calculated water level is assimilated to the observed one at the present time. Furthermore, the forcasted water level considering the riverbed evolution can be accurately computed for the next six hours.In conclusion, the practicality and validiry of the method we proposed in this study were verified.

 A series of physical experiments was conducted to investigate the influence of paddy ridges on the suspended load deposition under the scenario of embankment failure. A constant flow discharge and sediment was supplied at the breach as the inlet boundary, and three experimental cases were conducted by varying the downstream boundary height with or without using ridges. Experimental results show that, in the case of presence of downstream wall, the deposition occurred mainly in the vicinity of the inlet boundary and along the main flow direction. In the case of the open downstream boundary, no deposition was seen to occur near the inflow boundary owing to the high bed shear stress, most of the sediment was transported downstream and deposited there. In the vicinity of the paddy ridges, sedimentation was found at the back of the ridges but not at the front side and downstream of the ridges. A two-dimensional numerical model used here could generally simulate the deposition area and thickness, but it cannot provide sufficient explanation of the deposition near the ridges due to the three dimensional flow characteristics of the water in this area.

 Sediment Disaster Rainfall Index was calculated for sediment-related disasters caused by rainfall after the 2016 Kumamoto Earthquakes and it was confirmed that its values at actual disaster locations were below the threshold defined without considering the effect of earthquake. The threshold value considering the effect of earthquake was then examined to see how the lowering of it affects the number of disaster occurrence locations. By analyzing relationship between the calculated index value and the number of disaster locations, optimal reduced threshold value was studied. The results show that 70–80% reduction of the threshold would be suitable to consider the decrease in soil strength due to the earthquake.

 In this study, we developed a ordinal logistic regression model that predicts each probability of categories on sediment yield scale expressed in four stages by analyzing data-set, which was created based on data of July 2018 sediment disaster in the southern part of Hiroshima prefecture, topographic data and XRAIN data relating to the disaster in the same mesh size as XRAIN. In addition, we developed a simple sediment transport model to predict the sedimentation area. By using both models, a multi-case simulation was applied to predict the probability of sedimentation. The special tendency of sediment run-off was agreed with the actual disaster situation.

 Severe damage was found in the Hoyasu area, Nagano City, due to the dyke breach of the Chikuma River and its fluvial inundation and sedimentation, which deteriorated agricultural crops and the cultivated lands. Through a field survey, it is found that the deposited sediment consisted of silt and clay with fine particles, and the thickness of deposition was 20 cm or more around the breach point. Next, through numerical simulations of bed load and suspended load, the maximum flow depth and overall sediment deposition could be reproduced. However, the accuracy of the model should be more improved to predict the sediment deposition more precisely at each point. Furthermore, an assumed area map of sediment deposition was suggested by overlapping the maximum thickness of deposition calculated from several breach points.

 The prediction of sediment disasters requires an integrated approach of the rainfall infiltration, slope failure and debris flow. However, in order to predict slope failure accurately, slip surface prediction has been difficult to calculate on the complex catchment topography. Therefore, the prediction of associated debris flow and sediment runoff are also difficult. In this study, we introduce a stream-tube-based terrain partitioning method to treat rainfall infiltration, and slip surface analysis and debris flow simultaneously. The results of the application to a cathment that was damaged by the torrential rainfall in northern Kyushu in 2017, the damage around the main river was well reproduced.

 This study examined the relationship between local hydraulic characteristics and wood accumulation probability at river structure by integratedly treating bridge and a dam/weir. Specifically, after quantifying the wood accumulation probability by experiments, the dimensionless hydraulic quantities involved in the wood accumulation probability were extracted by the consideration with the aid of three-dimensional computational fluid dynamics. As a result, two dimensionless quantities representing the effects of the contraction velocity at the structure and the approach flow velocity were extracted as the dominant dimensionless hydraulic quantities. The results implied that wood accumulation probability at dams/weirs is lower than that at bridge piers because dams/weirs have comparatively large values of the involved two dimensionless quantities owing to the flow acceleration for the free surface overflow.