Understanding of representative permeability of large region groundwater flow system is important for accurate calculation on an alluvial fan. I applied a method to estimate permeability from seepage geometry by using complex analysis, Schwarz-Christoffel transformation or Joukowsky function. This analysis plays a role to associate physical space with complex potential space, which is easy to describe groundwater flow. Inputting computer-simulated hydraulics data, large-region permeability is the value from 10^-3 to 10^-4 order. This result is 1/10 for permeability under Dupuit-Forchheimer assumption, and it is almost as same as or ten times the permeability of drilling survey. The estimation by complex analysis delivers adequate permeability value without using the unknown aquifer stratum thickness.

 Coastal subsurface dams have been playing a vital role of the stable water supply by storing fresh groundwater in the upstream side of the cutoff wall and by preventing the saltwater intrusion in the downstream side. In subsurface dam sites, a method for evaluating the soundness of the cutoff wall is required for proper dam management. Here we investigate the relationship between the soundness of the cutoff wall and the saltwater distribution in the reservoir through experimental and numerical studies. The results show that the saltwater intrusion phenomena vary greatly according to the soundness of the wall, suggesting that the saltwater distribution in the vertical direction measured in subsurface dam sites may be an index to assess the soundness of the cutoff wall associated with the weak area having a relatively high hydraulic conductivity.

 It is necessary to clarify the mechanism of the concentration and development of underseepage flow around the landside levee toe for estimation of the levee failure risk due to underseepage flow. In this paper, the underseepage flow in a ground foundation under a levee was modeled by considering the anisotropy of permeability coefficient. The developed model was applied to the underseepage flow in a ground foundation with a weak point. The influence of the thickness of cover soil and expansion of weak points on the piezometric head change in the foundation layer and the concentration of the flow toward the weak points was clarified. In addition, it was shown that the risk of levee failure may increase due to the development of the large velocity region near the levee toe when the weak point is formed on the thin cover soil.

 In recent years, there has been increased risk of piping failure due to water leakage and sand volcano in the levee due to the infiltration of river water into the permeable foundation ground. Elucidation of the piping mechanism of river levees and examination of important monitoring points are important issues in the future. In this paper, three-dimensional saturated-unsaturated flow analysis and experiments were carried out focusing on the generation of sand volcano and piping hole formation under the levee for the piping mechanism. As a result, the state of the propagation of the decrease in water pressure in the foundation ground due to the generation of sand volcano was reproduced by analysis. In addition, as the piping hole develops, at the tip of piping hole, the flow velocity increases due to three-dimensional water collection from the range of about 10 times the diameter of piping hole, and in particular in the case of multiple layers, seepage acts on the entire piping hole . It became clear that it becomes a factor to promote.

 Since the seepage flow around river levees is one of the causes of bank breach and an important engineering problem, many studies have been conducted in the field of geotechnical engineering and river engineering since long time ago. As the system of equations of seepage flow is relatively simple, and numerical simulation is not so difficult, most problems can be numerically solved. The present study uses the Dupuit-Forchheimer type simple model to analyze the seepage flow around river levees with double-layer foundation in order to elucidate the basic physical mechanism. With two-layer foundations, the piezometric head in protected lowlands and the risk of piping increases with decreasing the ratio of the permeability of the upper layer to that in the lower layer and the extent of the permeable layer. It is theoretically found that the permeability of river levees increases the piezometric head in protected lowlands.

 It is difficult to make one-month weather forecasts using physics-based numerical models. Therefore, the Meteorological Agency announces probability forecasts for temperature and precipitation in three classes rather than numerical forecasts. In this study, we examine the possibility of numerical prediction of average temperature and precipitation using machine learning. We used LSTM suitable for learning time series data and CNN suitable for image learning in the middle layer of the model using ground meteorological observation data and sea surface temperature data as training, validation and evaluation data. Although the average temperature was predicted to rise and fall, there was also a tendency to underestimate summer temperature and overestimate summer and winter temperatures. In the forecast of precipitation, although the trend of increase and decrease can be generally predicted, it was not possible to predict the sudden increase of precipitation. It is also suggested that the sea surface temperature data may deteriorate the prediction accuracy.

 Frequency of heavy rains is increasing in Japan and therefore it is important to improve the accuracy of precipitation prediction for dam management and appropriate provision of information. In this study, we tried deep learning models using MSMGPV and in-situ precipitation data to predict the precipitation several hours later and studied the possibility of prediction in terms of spatial range of MSMGPV and leading time. The followings are obtained. 1) The accuracy of prediction was generally improved, except for some precipitation patterns, compared to Fujimori et al. (2019) who constructed a deep learning model using only AMeDAS data around the target point. 2) The accuracy of precipitation prediction did not drop with leading time for prediction. 3) The accuracy can be more improved by learning both AMeDAS data and GPV together.

 High-frequency and high-accuracy spatial distribution information of precipitation is required for disaster prevention and water resource management especially in small river basins due to its short runoff response time in hydrology. However, many small reservoirs are suffering the lack of such information because of poor surface gauging system and it seemed hard to grasp precipitation over the basin area. C-band radar rainfall gauging system provided by the Ministry of Land, Infrastructure, Transport and Tourism (MLIT), which is delivered in short intervals of five minutes, might be a solution for it because of its spatial and temporal resolution, but there is still room for improvement of accuracy. On the other hand, Japan Meteorological Agency (JMA) radar-raingauge analyzed precipitation data is more accurate but less frequent for example once in thirty minutes. In this study, the deep learning method was tested for improving accuracy of C-band radar-based rainfall by using JMA radar-raingauge analyzed precipitation data as supervised information. It was shown that driving a distributed hydrological model with the corrected rainfall reproduced well the reservoir discharge.

 The purpose of our study is to create a deep-learing model for water levels that can support appropriate operations of drainage facilities in an agricultural lowland. This model may help make a guidline that describe an effective use of a deep learning approach with artificial datasets for water level during heavy rainfall events even if insufficient observed data are avariable in the lowland. The study has two steps. First, the artificial data of water levels were made by a hy-drolological model using three cases of rainfall amounts (100, 300, and 500 mm/h) during three days, eeach of whose has 1, 000 different patterns. Secondtly, our deep-learning model that employs a LSTM model utilized the artificial data as observed data to predict water levels. LSTM model predicted water levels in an hour in the lowland after an appropriate learning process. The results show that the predic-tions of three cases have good aggreement with the observed data except for un-learned events like ex-tremely heavy rainfalls. As a future work, a lead-tme extension of the model and a comaprison among other deep-learning architectures are required.

 Urban floods occur frequently, and there are great expectations for the application of machine learning models that can be easily constructed to flood forecasting fields in urban medium and small size basin, which have complicated runoff mechanisms. Therefore, in this paper, we constructed a deep learning model using an actual river basin dataset of urban small and medium rivers, which was created at a short observation interval of 1 minute, which is different from that of large river basins. Benchmark tests were performed when the parameters were changed. An ANN model was also constructed for comparison with the deep learning model, and the performance was evaluated by a new index called the PD ratio, which is the number of parameters of the deep learning model and the number of observed data. As a result, it was found that the deep learning model is superior to the ANN model for the learning/verification flood at the same PD ratio, and in particular, the ANN model shows rapid fluctuations in the verification flood that do not follow the actual results.

 This paper presents a study on the performance of the Ensemble Kalman Filter with different system error settings when multi-site river flow observations are assimilated into a distributed hydrological model based on kinematic wave theory. We varied the scaling factor of the model error term and its correlation structure for a fixed observation error scaling and found the filter behaviour and the prediction accuracy to be highly dependent on the model error properties.While using standard normal model errors without scaling produced inferior results, with scaled errors, the effect of the scaling diminished when the model errors had higher correlation lengths. Identifying the proper model error correlation structure is therefore more important than determining the magnitude of the scaling for improving the filter performance.

 Recent progress in high-performance computing have enabled meteorological and climate communities to perform large-ensemble weather/climate predictions. To maximize the values of these large ensemble weather/climate prediction data, this study aims to develop a computationally-inexpensive machine that emulates a physics-based rainfall-runoff-inundation model. This emulator predicts maximum inundation depth from the spatial and temporal rainfall data for individual events. We first developed three types of regularized regressors, and then used the Random Forest to conduct ensemble learning of those regressors. This machine structure aims to have two characteristics: preventing over-learning for the underdetermined problem, and stacking multiple weak regressors for the non-linear transformation. There was almost no difference in the predicted accuracy of the maximum inundation depth between three regressors. Regressors tended to underestimate deep maximum inundation depths. The Random Forest significantly improved the prediction accuracy by stacking the weak regressors. In particular, the underestimated inundation depth seen in regularized regressors was greatly improved in the Random Forest.

 The objective of this study is to investigate characteristics of precipitation downscaling by means of a deep learning method, which have not been revealed well so far. This study utilized atmospheric variabes as input, and watershed-scale precipitation as the targeted data. CNN was seleted as a deep learing method for precipitation downscaling. Then, this study analyzed impacts of the spatial range obtaining input variables and the selection of combination of input variables on the accuracy of estimated precipitation. The results indicates several characteristics of CNN precipitation downscaling (CNN-DS) : (1) A wide horizontal range of the input variables are required compared to the target area. (2) CNN-DS requires physical information to some extent. (3) Depending on a variable, it may decrease the accuracy. (4) The use of more input variables related to precipitation does not always improves the accuracy.

 A rainfall occurrence prediction model was developed using a convolutional neural network (CNN), a representative machine learning algorithm in image recognition. A spatiotemporal data array was created from the time series of related atmospheric variables from multiple ground gauge observation sites and used as the image data set. By feeding the atmospheric data array into the CNN algorithm as an input, the algorithm was trained to classify whether there will be rain in the next 30 min. The trained models demonstrate promising results for three different cities in Japan, with a 64 – 76 % detection ratio for a 30 min prediction lead time. The high false alarm ratio is an issue that should be addressed in further research, with additional input data. This paper presents the basic concept of the developed model and the results from modeling tests, with various model structures and input data combinations.

 To detect the flowing groundwater with high accuracy using 1 m Depth Temprature, skilled technology based on many years of experience is required. Therefore, in this study, we developed a Convolutional Neural Network model that estimates the planar position of the flowing groundwater from the results of 1 m Depth Temprature survey. Numerical simulation was performed to make typical thermal condition under the ground. And a data set used for supervised learning was created from results of the numerical simulation. By learning of the Convolutional Neural Network model with the result of the numerical simulation, The Convolutional Neural Network model showed 99.7 % accuracy. Additionally, in order to confirm the Convolutional Neural Network system, the estimated route by the Convolutional Neural Network model was compared with the estimated route by the engineer. Results of the comparison showed that the Convolutional Neural Network has good reproducibility.

 We developed and validated artificial neural networks (ANN) to map flow characteristics in the most upstream basins throughout Japan. The ANN output mean annual runoff height (QMEAN) and flow percentiles of daily runoff height including 9 different groups, and input basin characteristics including climate, land use, soils, geology and topography. The generalization performances of ANN showed R² = 0.70 in QMEAN and R² = 0.20~0.74 in the flow percentiles. We succeeded in mapping the flow characteristics in the most upstream basins throughout Japan, which reflected rainfall and snowfall characteristics in Japan. The flow characteristics map revealed that it is suitable to develop run-of-river hydropower stations in heavy snowfall area facing Japan Sea within Tohoku and Hokuriku region.

 This study aims to determine the most dominant meteorological factors for extreme precipitation, such as probable maximum precipitation (PMP); this is important in determining the conditions that generate extreme events to understand their magnitude. This study analyzed historical heavy rainfall events and their corresponding meteorological factors. The factors considered in this study are precipitable water (PW), surface dew point (SDT), temperature, relative humidity (RH), convective available potential energy (CAPE), and vertical wind velocity (VVEL) in eight different regions in Japan. To collect sufficient data, we utilized reanalysis data from “Japanese 55-year Reanalysis” (JRA-55), and the relation between historical rainfall events and the selected factors was analyzed via the cumulative density function (CDF) exceeding 0.95. In southern Japan, the VVEL and PW were highly correlated factors for historical events. In northern Japan, PW was the highest correlated factor for the events. Conversely, temperature had the lowest extreme ratio in all areas. Based on this analysis, we can conclude that PMP estimation via PW is reasonable for these areas.

 A heavy rainfall prediction has long been an important goal in the field of disaster prevention. Robust and accurate predictions highly contribute to evacuating from disasters like a flood. Optical flow has become one of the standard methods for rainfall predictions, e.g. High-resolution Precipitation Nowcasts and Radar nowcasts provided by Japan meteorological agency (JMA). The aim of this study is to examine the applicability of U-Net, one of deep neural network models, for the prediction of hourly 5 categorical precipitation up to 6 hours ahead, and the effectiveness of learning pseudo rainfall data created for data augmentation to the model performance. For the model development, we used the data of Radar/raingauge Analyzed Precipitation (RAP) provided by JMA and generated the pseudo data by applying random rotation and random scaling to the original data. Using these data, we trained the model and then verified the capability of the U-Net model for rain forecasting. As a result, we successfully applied the U-Net model for a rainfall prediction and found that the model catches a rainfall intensification and weakening. The model trained with the pseudo data outperforms the one without the augmented data, particularly in the predictability of heavy rainfall for long forecast time.

 Railway operators enforce train operation control based on actual precipitation observed by rain gauges to ensure safe train operation in the times of heavy rainfall. Now by utilizing rainfall forecasting information they may make train operation safer. We studied about several methods for correcting prediction rainfall values by the translation model and the high-resolution precipitation nowcast distributed by Japan Meteorological Agency. Assuming that real-time precipitation values recorded by rain gauges are true values, we evaluated the accuracy improvement of each method by verifying accuracy of quantitative precipitation and predicting the time of the issues of the train operation control with 10 minutes ahead prediction. As a result, it was found that the prediction accuracy was improved by the correction with the error ensemble prediction in both prediction methods, and the accuracy after the correction was higher in the translation model than in the high-resolution precipitation nowcast.

 Small and medium-sized rivers, managed by local governments, have the problem that they do not have observation and real-time data transmission facilities. In this study, we discussed a system of observation facilities driven by a solar system and LPWA communication in radio dead zone and verified the feasibility and problems. In addition, by combining the observed water surface level with the results of flood analysis (conducted in advance), expecting to improve disaster prevention capabilities in radio dead zones.

 A flood forecasting system using physical models (such as rainfall runoff model, indefinite flow model) and data assimilation has been introduced in rivers in Japan. On the other hand, the application of DNN which empirically predicts unknown quantity from known quantity without depending on physical model has been observed. By using data assimilation based on physical interpretation and combined use of DNN to ensure accuracy from correlation between various quantities, it can be expected to simultaneously improve explainability and accuracy.

 In this paper, first, the results of prediction calculation by assimilation variables were compared and analyzed, and it was grasped that assimilation variables which affect the prediction accuracy most change even during flood. Based on this, a model to select an optimum assimilation variable by DNN was constructed, and a combination of assimilation variables selected successively was applied to carry out a water level and flow rate prediction simulation by data assimilation. As a result, the prediction accuracy improvement by the combined use of DNN and data assimilation was confirmed.

 Deep neural networks (DNN) with a deep learning approach have recently been applied to riverine flood models. We have developed the DNN-based flood model, in which a convolutional neural network (CNN) is embedded, to predict water levels during flood events with image classification techniques. Our model needed to have two major improvements for practical uses. First, lead time should be longer because our previous model provided only an-hour lead time prediction. Secondly, an effective creation of image-based inputs for CNN may have the appropriate spatial information optimized for watershed observatory networks. The former improvement provided appropriate model predictions of water level up to 3-h lead time. In addition, the simulation of a smooth flood wave was proper up to 6-h lead time. For the latter one, to reasonably predict water levels, the arrangements of stations for rainfalls and water levels based on the disctance from the predicted location to each station was better than the other arrangement such as the arrangement only of a few stations.

 There are various types of filters used for flood forecasting, including particle filters, but there has been no case in which the performance of filters is evaluated by employing an information criteria WAIC. Since particle degeneration is a problem in flood forecasting, five types of filters from the fields of moving object tracking, signal processing, image recognition, weather forecasting, that have the potential to work with degeneration were selected and used to evaluate the performance with WAIC. The filter performance was sufficient for all five types, but there was a slight difference in WAIC. In order to evaluate the significance of the WAIC difference, simulating the predicted rainfall error, analyzing the magnitude of the water level error, and estimating the ratio of the difference between filters to the water level error were conducted. As a result, it was found that the error between filters in the predicted water level was small, and the difference between the filters was also extremely small. Therefore, it was concluded that there was almost no difference in the performance between the five types of filters.

 Targeting the Lake Toyanogata basin in the Kamedago district of Niigata City, which is located in a low-lying agricultural area, we developed a machine learning model that outputs the inflow to the Lake Toyanogata by inputting rainfall data and drainage data from the Lake Toyanogata. The applicability to a short-term water level prediction model was then examined. For the input learning data of the machine learning model, we attempted to use the artificially generated mock data acquired by using the drainage analysis simulation as well as the observed data in the past rainfall event. As a result, it became possible to perform a simulation in a short time even for a big rainfall event that was not included in the measured data, suggesting the effectiveness of the method of complementing the learning data with the artificially generated data. In addition, it was shown that the proposed short-term water level prediction model could be applied to a tool that supports the decision of the operation of drainage pump stations.

 There are several types of open levees, which are traditional river technologies. In the Matsuura River, it is common to find open levees with retarding zones in the downstream region.

 This study focuses on the Okawano area of the Matsuura River, where the form of an open levee is still evident. The area was flooded in August 2019, after which the function of the open levee was evaluated by two-dimensional plane calculations.

 From the evaluations, it was apparent that the flow velocity of the flood flowing into the retarding zone was extremely slow in the open levee of the Okawano area. In addition, it was verified that the water level inside the open levee at the time of the flood inflow was the same at all points. As a result, the paddy fields near the open levee were not seriously damaged by the inflow of the flood.

 Although a significant peak cut function could not be determined, the results of delaying the peak was obtained.

 According to the concern about increase in heavy rainstorms due to global climate change, it is required to enhance the flood retention function in the upper river reach in order to improve the safety level against flood disaster. This paper discusses the possibility of introducing a new type flood control facility, called "running water type retarding basin (RWRB hereafter)", on narrow floodplains in a valley (FPV hereafter). A model of FPV having a meandering river channel and remain of old river channel periodically was constructed based on topographic maps and various materials for river planning. An overflow weir and a levee opening was installed at the upstream and downstream ends of RWRB for each floodplain along the meandering channel, respectively. Numerical flood flow simulation on the FPV equipped with RERB showed that the RWRM decreased the peak flood discharge with 5 % in a distance of 3.7 km without causing a severe inundation on the floodplain.

 This paper discusses the possibility of introducing a new type flood control facility, called "running water type retarding basin (RWRB hereafter)", in the upper Abukuma River where the largest record flood occurred in 2019. Unlike a usual retarding basin that is installed in the plains to store flood water horizontally, RWRB is designed to flow excess flood water over narrow floodplain along a steep upper river channel by premeditated river overflow. In this study, adjoining two narrow paddy, having the total length of 4000 m and the total area of 100ha, were selected for a numerical model simulation of RWRB. Three structures were installed in the model; a culvert connecting the two areas to enlarge the distance of water running; a side-channel spillway for inflow at the upper end, the crown of which is HWL+0.5 m; and a drainage gate for outflow at the lower end. Flood control capability of the model RWRB obtained from a numerical flow simulation under the hydrological condition of 2019 flood suggetsed that a group of RWRB installed on the narrow floodplains along the upper Abukuma River would effectively decrease the peak of flood hydrograph.

 In this paper, flood control function of agricultural land and wetlands were considered using an inundation analysis model, expecting to work as green infrastructure against flood. The study area is located in Zenzo River basin, a tributary of the Kaifu River, Tokushima Prefecture, where population decline is severe in recent years. Based on the rainfall by Typhoon 1412 several probable rainfalls were evaluated, and the flood control function was examined with the different rainfalls, as well as the difference of land use. The results show that, since the flood damage due to inundation above floor level remarkably increase at and over 20-year period probable rainfall, the flood control function has a limit at some stage against extreme rainfall events. On the other hand, in the past land use where more farmlands existed, temporal variation of inundation volume reduces more rapidly compared with the current land use. The results indicate it will be necessary to suppress development in high risk areas, and the past land use will be a good reference for a wise land use.

 It occurred widespread and simultaneous flooding and debris flow by the typhoon No. 19 in September 2019. Because heavy rainfall disasters occur frequently and heavy rain damage has become severe, we have to make effective use of various hydraulic structures. In the present research, the flood control functions of open levee in Kinugawa River were examined by 2D flow numerical analysis. The flood control functions were classified into two functions, that is, reservoir function and flood return function. As a result, it was suggested that the reservoir function does not work in steeply river such as Kinugawa River. It was also clarified that an inundated current with sediment deposit spreads wide in inland area in spite of the flood return function, because of the obstacles such as road embankment constructed at the open levees.

 A new flood control concept called ‘Cascade method’ which permits multiple dams around an upstream region in a river basin to overflow from their emergency spillways has been proposed. For the sake of simplicity, we assume a river basin in which multiple dry dams are constructed in a series where no water flows in from rain or tributaries other than main river upstream. In this study, under the condition of common reservoir capacities, the new flood control method permitting upstream dry dams to overflow except for a downstream dry dam is compared with the conventional one not to overflow in each dam. In addition, each optimum design high water discharge for two or three dry dams constructed in a series are investigated. As a result of this study, it is made clear that the new method is more effective than the conventional one, and uniform design high water discharge for each dry dam is practically appropriate in the Cascade method.

 R1 Saga floods caused serius damage in the Rokkaku River basin. We carried out field surveys in the flood area and numerical simulation. The field surveys indicated that the flooding whose depth was higher than 2m spreaded in the left side of Takeo River, and hence the flooded area and water volume were 56.8km² and 40.9 million m³. In addition, the numerical simulation demonstrates that the maximum flood volume of inland waters is seven times greater than the maximum capacity of drainage pumps, indicating that it is difficult to drain the inland flood waters using these pumps. We need to develop a flood control strategy considering both of flooding from riverine and inland waters.

 The non-life insurance industry requires sophisticated storm surge risk assessment for risk management. In the United States, a stochastic storm surge risk assessment has been performed using a method with low calculation load called JPM-OS, but the calculation accuracy is lower than that of conventional method JPM. In the storm surge risk assessment procedure with low calculation load proposed by past study, which can consider the uncertainty of astronomical tide level, the consideration on the computational accuracy is insufficient.

 Therefore, in this study, utility of the proposed procedure in previous study is confirmed by applying it to the Tokyo Bay where assets are accumulated and considering whether the calculation accuracy can be maintained by reducing the calculation load.

 Some place names that have been handed down since ancient times often capture the characteristics of the land and imply disaster risk. In this study, we proposed a method to easily evaluate flood inundation risk using flood damage names and open data. According to this method, it was found that the flood damage risk in the inundated areas is scattered throughout the country. In addition, we were able to extract a potential flooded space that is not displayed on the hazard map.

 In low-lying areas, global issues of water use and disaster risk management are especially important for sustainability.

 The purpose of this study is to expand disaster prevention awareness through production based on the collective experience of all the junior high school students who experienced the heavy rain disaster in Saga prefecture in August 2019 and presentation activities to the region. These efforts were activities of the author’s previous public junior high schools such as the 2019 culture presentation, and were an extension of the basin learning that we continued.

 From the questionnaire survey, it was confirmed that junior high school students' awareness of disaster prevention increased through production and presentation activities. In addition, by interacting with residents, administrative officials, and experts in the presentation activities, it became an opportunity to develop human resources for sustainable development for their town.

 This study is considered to be important as the introduction of the concept that connects schools and regions with heavy rain disasters common to low-lying land in Japan and overseas.

 In Japan, evacuation advisories are simultaneously issued for a wide area when there is a risk of river flooding. But in the U.S., if there is not enough time to evacuate, they will not issue evacuation orders in some cases. This study focused on the middle basin of the Chikugo River and analyzed inundation situations and lead times to determine the evacuation considering the risk of a horizontal evacuation to a evacuation site versus the risk of remaining in a vertical evacuation at home. Then, we examined the appropriateness of making a "recommendation for vertical evacuation" rather than a uniform evacuation advisory. As a result, the risks of horizontal and vertical evacuation vary greatly in each area, and if there is not enough time to evacuate, it is safer to stay home instead of evacuating to a evacuation site. It is necessary to select appropriate evacuation actions for each area.This study shows that vertical evacuation advisory should be issued in some areas instead of uniformly encouraging horizontal evacuation.

 In recent years, river floods due to torrential rains have frequently occurred under the influence of climate change, while typhoon floods have been increasing. It’s still fresh in memory that Typhoon No. 19 in 2019 broke 140 dikes in 71 rivers nationwide and caused serious damage. In this study, numerical prediction calculations were performed for the 23 wards of Tokyo, assuming the levee on the right bank of the Arakawa river breaks. This elucidated the generation and expansion process of inundation due to Arakawa flooding. Furthermore, based on these results, the following two points were examined, which are considered to be seriously damaged when the levee of the Arakawa river were breached. (1) Focusing on K subway station, we analyzed the inundation process inside the station and in the tunnel, elucidated the inundation process, and verified the damage mitigation effect of the subway. (2) Focusing on Adachi Ward and Arakawa Ward, elucidating the flooding process up to the main facilities requiring special attention, and examined evacuation measures for hospitals and welfare facilities for the elderly.

 Using multiple rainfalls from numerical weather simulations considering the global warming, inundation simulations were implemented for the Kakehashi River basin, Ishikawa, Japan. Then, economic losses were estimated based on the maximum inundation depth estimated by the simulations. In the evaluation of economic loss, future variations of population distribution caused by depopulation were considered. At the same time, variations of economic losses were investigated by considering possible changes in urban structure caused by the depopulation in the target area. Under depopulation condition, the economic losses diminishes in residential district. When people relocated from the higher inundation risk areas to the lower risk area, 40 % of the original economic losses would be diminished at the maximum relocation ratio.

 The flood water level was reproduced where breach of river dike occurred at the confluence of Oppe River and Tokigawa River in the flood event of Typhoon Hagibis. We compared the estimination of flood water level using a two-dimentional numerical model with roughness coefficient, and tree height and thick growth of vegetation added. The estimated flood water levels were substantially coincident with the maximum flood trace heights along the river. We showed the rise of flood water level in Oppe River and Tokigawa River with vegetation affected by Typhoon Hagibis. We also evaluated the lowering effect of flood water level due to vegetation management that was demonstrated by changing the vegetation location and the percentage of vegetation. 40% tree removal near the breach location of Oppe River is found to be the most effective way in the condition of the least tree removal area of this study.

 This study demonstrates the accuracy of vegetated lower Asahi River surveying using both the improved airborne LiDAR topo-bathymetry (ALB) and newly launched unmanned aerial vehicle (UAV)-borne green LiDAR system (GLS), with comparisons among systems-derived, depth-averaged numerically simulated, field- and sensor-based data. This report also examined the applicability of both systems data for low-water flow simulations considering vegetation effects, which enables us to select the most feasible LiDAR-based technique for developing a novel hydrodynamic flow model. Results on topo-bathymetry and vegetation assessment revealed that both systems data were almost identical to the field observations, with an outperformed accuracy of a few dozen centimeters. Based on overall aspects taking account of high spatial resolution aerial image and shortcoming to deep-water, it can be inferred that the compact GLS would be a cost-effective tool worldwide for shallow vegetated rivers management measures after further studies.

 In a disaster management, a SAR satellite is an effective tool to extract inundation areas since it enables remote observation over wide areas even at night or in a cloudy situation. At present observation opportunities of SAR satellites are limited to about twice a day. However, thanks to the recent technological innovations in miniaturization of SAR satellites, high-frequency observations by small SAR constellations could be possible in the future. In this research, a constellation of small SAR satellites was designed with Walker Constellation method, and it was demonstrated that 40 satellites can accomplish 1 hourly observation. Combining this assumed condition with hourly flood plain water depth of CaMa-Flood, the cover ratio of flood area by the observation was analyzed and the constraints were indicated as a specific case.

 In recent years, cities are facing pluvial floods more frequently due to global climate change and rapid urbanization. This problem causes more severe damages in the cities with old aged drainage infrastructures. In these cases, the urban drainage systems cannot meet the original design requirements, which causes traffic jams and hygienic problems in the downtown areas. This study adopts InfoWorks Integrated Catchment Modeling (ICM) to evaluate the effects of design storms and channel sediment depths for the old storm drainage system in Yangon, Myanmar on the urban flood severities. The simulations with different design storms clarify that they have larger effect on flood discharges but smaller effect on flood durations, relatively. Whereas, those with the varied sediment depositions suggest that they could strongly influence both of the flood discharges and durations. The study could find that the maintenance work was essential for the reduction of such the urban flood severity.

 A rainfall-runoff-inundation model emulator was developed by machine larning using outputs from large ensemble climate simulations as a training dataset. The prediction of flood inundation area by the emulator was evaluated in the Omono river basin, which has been freaquently flooded in the past. The experimental results show that the inundation area near the river channel can be reproduced by the enulator with about 80 to 90% accuracy. We applied the machine to the actual flood inundation case in July 2017, and found that the result of prediction by the emulator is comparable to that by the results using rainfall-runoff-inundation model although the reproducibility differs according to the training data of machine learning.

 In this study, we developed a real-time river flood forecasting method about 1 day ahead utilizing basin-average rainfall in order to use for making decision to refuge a railway vehicle detained at a rolling stock depot before flooding occurs. In this method, the basin-average rainfall at the nearest water level observation station installed in the river that causes flooding to the rolling stock depot is used. It is evaluated that river flooding may occur when the basin-average rainfall exceeds the design rainfall. In 3 out of 5 cases in which the river water level exceeded the designed high water level, this method could be predicted more than one day before the river water level reached the designed high water level. Furthermore, the river water level exceeded the designed high water level in 4 of the 11 cases in which the predicted value of the basin-average rainfall exceeded the design rainfall. Therefore, it was decided that this method would be used as a decision criterion for refuging a railway vehicle from a rolling stock depot.

 In this study, an optimization method using a genetic algorithm was developed for rainfall-runoff and flood inundation analysis, and the effectiveness of the optimization method was examined by applying it to the Kagetsu river during North Kyushu heavy rain in 2017 and comparing the results with existing analyses. It shows that the method (1) enables the estimation of flow discharge and optimization of basin parameters based on rainfall and water level information, (2) improves the reproducibility of inundated areas, although the reproducibility of water level and discharge hydrographs is comparable, and (3) estimates optimal parameters more efficiently than the empirical method.

 Embankment failure due to overtopping flow takes place by headcut formation and its propagation toward upstream. Vertical, nearly vertical or cliff-shaped terrain called headcut forming at landside toe of the embankment is key to cohesive embankment failure during overflow. This steep terrain or headcut triggers further erosion and propagates upstream direction until crest of the embankment is reached and breached. In this study, experiments were conducted to understand the effect of the seepage state on initial position and migration of the headcut. For this purpose, three experiment cases firstly, no seepage into the embankment, secondly and thirdly with seepage and phreatic line exit point on downstream/landside slope at 1/3 and 1/6 of embankment height, h, from toe level respectively were conducted in laboratory. From the experimental results, it was found that the higher the exit point of phreatic line on the downstream slope, the higher the initial occurrence position of the headcut and higher the migration speed. When the experimental results were applied to a previous predictive formula regarding the headcut location, a large error occurred between the measured value and the predicted value for the case where initial phreatic line exit point was higher. In order to solve this problem, modification was made to address the effect of the seepage state in its predictive formula. As a result, for 1/6h permeation case, the predicted value for the position of the headcut almost agreed with the measured value. However, for 1/3h permeation case, it was confirmed that the predicted value of the headcut position was greatly underestimated compared to the measured one.

 The pooled water depth on the landside of a river embankment can affect the embankment’s erosion and breaching process. This phenomenon is understood as ‘water cushion effect’ that decreases the energy of overflowing water and therefore suppresses erosion. The 2019 Typhoon Hagibis, Japan, such landside inundation conditions may have influenced the behavior of levee failure. This study aimed to clarify the effects of this inundation on the erosion by overflowing water on an embankment. A laboratory experiment using a sand embankment model and simulated result showed that the higher pooled water depth, the lower the erosion due to overflow was observed. Reproduction calculation under the conditions at the embankment of the Tokigawa River where the levee failure occurred in the Typhoon Hagibis also showed a similar tendency.

 When a landslide dam collapses due to flooding, severe damage occurs at the downstream. For this reason, the measures for collapse of landslide dam are being implemented immediately. However, the difference of the destruction phenomena caused by the infiltration situation of the reservoir water impregnated upstream of the dam into the soil mass of the landslide dam has not been clarified. In the Hidakahoronai river, the largest landslide dam at the 2018 Hokkaido Eastern Iburi Earthquake was formed, and a detailed investigation was conducted by the Hokkaido Development Bureau. In this study, hydraulic experiments, in which the experimental conditions, were set based on the result of this investigation were conducted and the influence of the infiltration condition into landslide dam on the erosion phenomena of landslide dams due to overflows was discussed. As a result, it was confirmed that the situation of collapse changes greatly due to the difference of penetration, and the progress speed of the phenomenon also changes greatly depending on the way of collapse.

 The drag force and lift force acting on a structure are generally evaluated by using the average flow velocity. It has been pointed out that a surface wave called a triangle-shape water surface wave is formed and the internal flow is locally changed under the supercritical flow, but its influence has not been evaluated. In this study, we evaluated the stability of blocks using different shape blocks under the supercritical flow. Also, the stability of blocks was examined using the equilibrium of force and moment considering local and upward flow. As a result, the model considering the local flow and upward flow was consistent with the experimental result. Therefore, it was clarified that the local flow and the upward flow greatly affect the stability of blocks.

 In alluvial rivers, revetments have an important role in protecting banks from lateral erosion. In 2016 August due to heavy rain, numerous flooding events and countless revetment failures were observed in tributaries of the Tokachi River in Hokkaido, Japan. One of noticing points with those revetment failures was that they were caused with large erosion on high-water channels, nearly inducing a levee breach. In this study authors focused on the mechanism of erosion along a revetment on high-water channel and carried out a flume experiment with a meandering channel. The experiment’s result and a reproduction computation showed some viewpoints regarding high-water channel erosion. In a meandering channel there is a tendency of fluctuation of bedloads and local bedload imbalance on high-water channel. This tendency leads to high-water channel erosion along revetment as observed in tributaries of the Tokachi River.

 Owing to the recent flood scale expansion due to climate change, erosion and washout of road and railway bridge abutment back embankments are occurring frequently. A wing walls and a small retaining wall are constructed on the embankment slope of the back embankment of the railway bridge. However, the disaster mechanism is not clarified. In this study, we conducted a basic study on displacement/collapse behaviour from an open channel test on a model embankment that reproduced a wing wall and small earth retaining wall. As a result, the depression of the embankment slope occurred at the boundary between the wing walls and the small retaining wall. The collapse of the top of the embankment progressed and the wing wall and a small retaining wall washout from this depression.