River flood disaster may increase due to climate change. To evaluate the impact of climate change on river floods, in this study, pseudo-warming ensemble experiments under +2 and + 4 degree scenarios were conducted using numerical simulation for not only rainfall and runoff, but also river flow and flooding, in which there have been few previous studies. Study target was the flood in the upstream reach of the Arakawa River due to typhoon Hagibis in 2019. Comparison of ensemble average of the +2 (+4) degree scenario and the current case showed that total rainfall, peak discharge and water level increased by 7.1 (33.8)%, 12.2 (35.8)%, and 130 (286)%, respectively. In the current case, no inundation occurred, but the inundation occurred in 6 or 18 of 20 members of the +2 and + 4 degree scenarios, respectively. These facts suggested that floods that exceeded H.W.L. but did not cause flooding would increase the possibility of flooding due to the progress of climate change.

 Recently, large scale inundation has occurred frequently. In addition, in the study of flood in consideration of climate change, the occurrence of extremely large floods has also been calculated. If large flood and large scale urban inundation occur, the flood water flows to the lower part of the urban area. Then amount of flood water may flow into the underground spaces, human damages and economic damages may be extremely large. In this study, inundation analysis in Shonai River basin will be performed using the future forecast flood discharge obtained by Tachikawa et al., and its inundation characteristics will be clarified. As a result, there was time delay on the temporal change of flood in the case of inundation consideration in upstream region, but the peak discharge did not change. In addition, on the inundation of the downstream region, it was shown that the underground inundation increases, although there is no big difference in the state of inundation on the land in the analysis conditions considering levee breach.

 Railway operators enforce train operation control such as suspended operations based on precipitation observed by rain gauges, to ensure the safety of trains during heavy rains. Rainfall intensity and frequency are expected to increase in the future with the increase of greenhouse gas emissions, and it is expected that stability in railway transportation will face a decline due to such safety procedures. This study evaluated regional differences in the stability in railway transport due to such precipitation change along four railway lines by calculating the changes in frequency and duration of suspended operations at present and at the end of the 21st century. Results showed that at the end of the 21st century, frequency of suspended operations on lines in the Tōhoku region would be 3.1 times the current average, and 1.4 times the current average on lines in the Kantō region, and duration of operation suspensions being 4.2 times in the Tōhoku region and 2.1 times in the Kantō region. In this way, it was shown that the frequency and duration of suspended operations increased, and that the degree of increase was greater in the Tōhoku region than in the Kantō region.

 A drought occurs due to long-term continuation of light rain as the trigger. In addition, time varied conditions for water infiltration, storage and runoff process may become an important factor. Applied the MCMC method as data assimilation techniques to a simple runoff model and long-term hydrological observed data including a huge drought event in 1994, time variation characteristics for water movement at the drought occurences are investigated. As the results, it is found that increased vertical infiltration and decreased horizontal runoff rather than decreased storage height under the ground surface make the potential of a drought occurence increase.

 This study aimed to estimate crop damage during floodwater inundation considering the effects of climate change on rivers in lowland areas. The Chitose River, a tributary of the Ishikari River, flows through low-lying land and is a topographically vulnerable area that is strongly influenced by backwater from the main river. In addition, Hokkaido produces several crops along the river basin, including wheat and soybeans, that account for a large share of the market, which in turn has a large impact on the national supply chain. In this study, we used rainfall information obtained from d4PDF downscaling data and Thiessen polygon data on farmlands to estimate crop damage in the Chitose River basin. Based on the amount of present and future crop damage estimated, the likelihood of damage exceeding 1.2 times that of the greatest damage ever recorded hitherto occurring in the future was shown.

 We analyzed the characteristics of changes in rainfall and river discharge due to global warming in four first-class river basins in Kyushu, focusing on the 48-hour annual maximum precipitation events in each year from the d4PDF past experiment and the 4°C rise experiment. It was found that the frequency of the summer season was higher in southern Kyushu than in northern Kyushu. Next, in addition to the increase in the frequency of major floods in all basins due to global warming, the peak number of hyetograph showed a decreasing trend in all basins, and the number of heavy rainfall events causing one- and two-peaks floods increased. This study indicates that global warming may cause not only an increasing precipitation, but also a decrease in the peak number of hyetograph, resulting in an increase in peak flood discharge.

 This study aims to estimate the impact of climate change on the probability of flood control operation during an extreme flood (hereafter, extreme flood operation) using simple runoff model and dam operation model which is developed based on actual dam operation rule with d4PDF in Shimouke Dam, in which the operation has been carried out in July 2020, and Matsubara Dam, which is located directly below the Shimouke Dam as research fields. As the result, it was revealed that the frequency of extreme flood operation by the Shimouke Dam in the future might increase relative to the present with the increase of flood discharge. On the other hand, the frequency of the operation of the Matsubara Dam in the future might be almost same as the present. Therefore, it was revealed that the connected two dams were able to operate effectively in order to reduce damage to the downstream area of the dams.

 This study aimed to evaluate the flood control functions of multi-purpose dams in snow-covered areas, considering increased flooding risks associated with climate change. In snow-covered areas such as Hokkaido, heavy rainfall during the snowmelt period makes the prevention of extreme floods difficult around multi-purpose dams due to their high water levels. This study estimated the frequency of disaster prevention operation due to severe flooding based on the climate change ensemble data for the whole year, including heavy rain in summers and snowmelt combined with heavy rainfall. The results indicate that the frequency of disaster prevention operations due to extreme flooding will increase in the future, especially during winter and snowmelt. The results obtained will be useful for planning flood control measures that can adapt to climate change, such as pre-flood dam release, dam regeneration, and construction of new dams.

 Future changes of flood control effect (FCE) by major existing dams were analyzed for 59 river systems all over Japan using d4PDF rainfall data and the rainfall-runoff model 1K-DHM. The rate of flood peak discharge with/without dams increased for floods over a particular magnitude (FCE was reduced) for some river systems near the Pacific Ocean and not (FCE was remained) for ones near the Seto Inland and Japan Seas. The former has smaller dam capacity despite receiving heavier rainfall, where larger future changes of extreme discharge are projected but its impact was not significant. Similarly, the frequency of emergency operation is also related to the flood storage, indicating that flexible dam operation such as preliminary release will make a significant contribution to flood control in a changing climate in Japan.

 In recent years, floods on a scale larger than planned level have frequently occurred in Japan, which needs urgent actions to mitigate these impacts regarding future increasing impacts under climate change. In this study, we focused on the flood control function of existing dams. We examined the flood control function in major river basins nationwide, the effect of preliminary discharge, and the priority evaluation method for rationally upgrading the functions of existing dams. As a result, we tried to clarify a relative comparison of flood safety for each river basin using the total equivalent rainfall volume in the basin (= total flood control capacity of dams in the basin / total river basin area). Specifically, we clarified the differences such as the installation level of flood control dams and the possibility of participation in the flood control function of other reservoirs such as hydropower dams. In addition, we showed that there is a possibility to select high priority dams which should enhance flood control capacities by using the relationship between the comparative constant proposed in previous studies and the basin area control ratio of dams.

 This research conducted ensemble inflow simulations for electric power generation dams in the Sai River in Japan. After developing the Water and Energy Based Distributed Hydrological Model 4 components for Snow (WEB-DHM-4cS), we validated the model by conducting hydrological simulations for three consecutives years from August 2015 to July 2018 and confirmed that the model could properly express peak discharges, snow-melt discharges, and base flows in the basin. Then, we performed ensemble inflow simulations of the flood event at July to September 2018 in the Sai River by inputting an ensemble 39-hour rainfall dataset to WEB-DHM-4cS. As a result, we predicted the flood discharge (800m3/s) at Ikusaka Dam with lead times of 7 to 31 hours and the accumulated inflow volume at Takase Dam with lead times of 6, 12 and 24 hours highly accurately with an error rate being less than 5%.

 The current study aimed to propose a method for the prediction of inflow, storage levels, and discharge flow rates from dams for extreme flood disaster prevention management. In recent years, owing to the damage from frequent large floods nationwide, the prediction of water storage levels and discharge flow rate to be utilized for effective dam management has become critical. In this study, using Elastic Net, a sparse modeling method capable of identifying relationships between data even from small amounts of information, we predicted the inflow volume for dams that have experienced cases in the past wherein engaging disaster prevention management was required during extreme flooding. Subsequently, the water storage level was estimated based on predicted inflow and discharge based on operational regulations. Furthermore, the predicted discharge flow rate was shown to be effective for predicting the water level of downstream rivers. In summary, it is considered that the proposed method can be utilized in judging pre-emptive discharge and recognizing the effect of the discharge on the downstream area.

 Flood disasters often occur worldwide as heavy rainfall events have become increasingly frequent due to global warming. In these circumstances, it is important to evaluate flood risk accurately and thus essential to develop a highly accurate runoff-inundation model and prepare detailed hydrological data. However, such data are rarely available in areas where an observation network is not well developed. In this study, by taking a dam failure event in Myanmar as an example, we developed a runoff-inundation model based on detailed spatiotemporal precipitation data and extracted inundation areas using satellite data. The results show that it is possible to prepare necessary data, develop a reliable model, and conduct risk assessment even for areas with insufficient hydrological data.

 We developed an operation support system for a single hydroelectric dam using ensemble flood prediction to reduce water-related disaster risks intensified by climate change and increase renewable energy generation promoted by the Carbon Neutral policy. The system was experimentally applied to dam operation simulations for warm seasons of two consecutive years. The results showed that the system successfully contributed to controlling flood risk while increasing hydropower generation.

 We assessed how accurate the current rainfall forecast was in decision making for pre-release at multipurpose dams in Japan, as the use of numerical rainfall forecast for pre-release is becoming more active at dams nationwide. For this analysis, inflow was calculated by using a tank model whose parameters were proposed based on geological features, and the regular controls and quantities of each dam were also incorporated based on our survey. After that, we evaluated the capacity to be secured and time required for prerelease which were predicted at different forecast initial times in the most recent disasters. As a result, there were uncertainties such as false alarms of the capacity to be secured and large range among forecast updates. On the other hand, under the flood control capacities and downstream flow capacities of target dams, there were few cases where it is difficult to recover the water use capacity, and the forecasts were effective information for reducing the degree of emergency spillway gate operation.

 A method for prior release of a multi-purpose reservoir was developed considering operational medium-range ensemble rainfall forecast and its error characteristics. Ensemble members which to be focused in decision making for prior release were determined based on an accuracy analysis on each order of ensemble member so as to balance effects in flood management and storage recovery. The proposed method was applied to a hypotethical reservoir in Japan, demonstrating its potential effectiveness to improve prior release operation.

 In recent years, large-scale floods such as heavy rains in July 2018 have occurred frequently. On the other hand, due to the delay in river improvement works, there are many dams where provisional operations are being carried out by devaluing the flood control start flow rate, and there are some cases leading to the emergency spillway operation during extreme floods. The purpose of this study is to propose a method to determine the dam flood control operation method that can reduce the damage not only in the frequent small and medium-sized floods but also in the large-scale floods. In this paper, the flood control operation method of Hiyoshi Dam, which is located in the upstream of Katsura River, was studied. Firstly, return periods of large ensemble rainfall events (d4PDF) were evaluated and then we conducted rainfall runoff inundation analysis to know how the dam discharge, flooded area, economic damage in the downstream area, etc. change depending on the flood control operation method. As a result, we proposed a method to quantitatively compare the effects of dam flood control operation methods, including those without dams, and to select the most effective dam flood control operation method under various scales of rainfall groups.

 Numerical flood flow simulation was conducted to investigate the effectiveness of HAMAO Flood Control Basin located in the upper reach of ABUKUMA RIVER against an extraordinary flood in 2019. Because of the uncertainty of calculation condition due to the data missing and the bank overflow, an iterative process of calculation was employed in order to obtain the simulation results consistent with available field data. According to the numerical simulation results, the flood control basin was not effective for the extraordinary flood because its capacity was used up before the flood peak. The numerical results also suggested that, on the contrary, the bank overflow to the floodplain in the reach upstream from the flood control basin effectively reduced the flood peak.

 Dams located in heavy snowfall regions are required to restore their water utilization capacity during snow melting period because the inflow volume into reservoir decreases during the snowfall period. In this study, a dam operation model using deep reinforcement learning was trained by inputting snowmelt and temperature data of the catchment area to achieve both control of snowmelt floods and operation to restore the water utilization capacity of dams in heavy snowfall areas. From the results, this model can be adopted to both flood control and restoreing water utilization capacity during snowmelt period. We also obtain the results that this model can control discharge rate according to mass of snowmelt or weather conditions.

 In this study, saturated and unsaturated seepage simulations were carried out using 2D and 3D nonuniform field model for fingered flow in a two-layer unsaturated zone near the ground surface, and the differences between the 2D and 3D fields were discussed. In addition, a continuous/quantitative evaluation of the occurrence of fingered flow is attempted based on the statistical properties of the velocity distribution obtained from the simulation. The results showed that the upper layer, which has low permeability and high water retention, produced a generally uniform flow field, while the lower layer, which has high permeability and low water retention, produced a distinct finger flow. Although it was difficult to visually compare the 2D and 3D flow fields, it was shown that the differences between the two fields could be quantitatively evaluated based on their statistical properties.

 This paper classified the levee foundation leakages in the Chikuma River and the Kakehashi River using the levee foundation vulnerability index t_b^* and soil characteristics of the toe of landside levee. We found: 1) underground leakages occur when 0.3 ≤ t_b^* ≤ 1.4; 2) sand boilings occur when t_b^* ≥ 1.4; 3) boilings and sliding failures following a sand boiling occur depending on the thickness and grain size of the levee foundation soil layer at the toe of landside levee; and 4) a leakage causing from the regional underground water in the alluvial fan and tributaries and a leakage seen frequently during floods occur when t_b^* < 0.3.

 In this research, model experiment and seepage flow analysis were carried out to clarify the mechanisms of piping by sheet-pile installation. As a result, from the model experiment, sheet-pile installation had the effect of delaying piping hole penetration by suppressing the deformation of the foundation ground, although it was difficult to prevent the occurrence of sand volcano. From the seepage flow analysis, it was found that the rapid increase of seepage flow velocity to the tip of piping hole was one of the triggers of the levee breach due to piping hole penetration, and the sheet-pile installation prevented this effect.

 In this study, in order to evaluate the impact of the 2016 Kumamoto earthquake on the hydrological environment in Aso Nangou-dani, a water budget model was developed based on the hydrological data obtained from the field survey. The analysis was performed for the period from 2015 to 2019. The reproducibility of the rapid rise in groundwater level immediately after the earthquake, the subsequent water level change, and the depletion and recovery process of Shioisha spring suggest that the hydraulic parameters such as hydraulic conductivity rose due to the earthquake and have not returned to their original state since then. In the area where groundwater inflow from upstream increased due to the earthquake, it is indicated that the spring volume and base runoff increased.

 We investigated the physical parameters that suitable for Shishamo smelt spawning in the lower reaches of the Azuma River and the impacts of high suspended sediment concentration caused by the Hokkaido Eastern Iburi Earthquake by means of a statistic model and numerical simulation. The results of the statistical model showed that the most suitable spawning places were mainly determined by bottom velocity of 0.4 m/s. In the numerical simulation, the appropriate place (bottom velocity range 0.3 to 0.6 m/s) moved in the river according to the flow rate (about 5 m³/s), so there was no suitable place to cover from high to low flow rate. Silt, increased after the earthquake, was mainly deposited along riverbanks and less deposited in the low flow channel, suggesting that the effects of high silt concentration on the spawning sites in the river would be limited.

 This paper analyzes the relationship between the parameters and the surficial geology that greatly affects the runoff in mountainous watersheds in 32 dam watersheds consisting of a single geology to elucidate the physical characteristics for improving the accuracy of rainfall-runoff models. In addition, the applicability of the rainfall-runoff model to large-scale floods was examined by simulating an unprecedented large-scale flood in a dam basin with different surficial geology. The results show that the relationship between surficial geology and parameters is dominated by parameters related to intermediate runoff, and that the variation of parameters is larger in granitic and volcanic rocks. As for the applicability of the model to large-scale floods, the results of parameter estimation by large-scale floods, regardless of the surficial geology, were found to represent the peak flow rate the best, indicating the applicability of the model to large-scale floods.

 This study analyzed the correlation coefficient (R) between the monthly average flow (MF) and the total precipitation (TP) of various periods in several mountainous watershed in Japan, discussing the difference between watersheds mainly focusing on the difference of geology. As a result, in low-water periods, TPs showing strong correlations with MFs differed by watersheds. The dominance of volcanic rock formed in the Quaternary period seemed to be related to higher Rs between the MFs and TPs of longer periods because of its high permeability. In addition, Mann-Kendall test was conducted for MFs, TP of various periods and the annual precipitation to verify the validity of comparing the trend of the flow and the total precipitation of influential periods rather that the annual precipitation. No trend was seen in the annual precipitations of all watersheds, although MFs showed increasing trends in some months and watersheds during low-water periods and some of the TPs whose Rs were higher than 0.6 with that flow also showed increasing trends.

 In this study, we examined the effects of the enhanced interception evaporation rate and surface roughness expected from changing needleleaf forest to needleleaf-broadleaf on flood mitigation functions. By conducting hydrological observations in the Hashimoto forestry area, which is a mixed forest of needle and broadleaf, the obtained interception evaporation rate and surface roughness were adapted to Nagayasuguchi dam and runoff simulation was conducted. We used a tank model for the spill simulation, which was evaluated in terms of peak flow rate and generation time. As a result, increasing the surface roughness reduced the peak flow rate by 6.1% relative to the 100 year Probabilistic rainfall, and increasing the interception evaporation rate reduced it by 3.6%. Therefore It was confirmed that this is an important function in flood reduction.

 This study proposes a method to quantify the degree of freedom of the annual maximum rainfall given by the climate system as entropy and to predict the future change of the probable rainfall based on the entropy. In particular, we have developed a method for calculating the probable rainfall in the future period by calculating the process of increasing entropy of the annual maximum rainfall from observed information and constructing a Bayesian predictive distribution that preserves the process of increasing entropy. The time series of sequential probable rainfall obtained by the proposed method can be a useful indicator for considering project options based on a real time axis when considering the staged development of flood control management.

 This study revealed the issues related to the use of large-scale ensemble climate projection data, which is necessary to consider flood control planning in light of climate change. The analysis is based on the case studies of Shonai River and Kano River. The future change multipliers of the annual maximum rainfall at the planned rainfall duration under the global mean temperature increase of 2 and 4 degrees Celsius were estimated from the climate projection data with a spatial resolution of 5 km. The results under the 4 degrees Celsius temperature increase were generally consistent with the future change multipliers presented by the Ministry of Land, Infrastructure, Transport and Tourism. On the other hand, the results for the Shonai River are larger than those of the MLIT when the temperature rises by 2 degrees Celsius, indicating that precipitation will be larger when the temperature rises by 2 degrees Celsius than when it rises by 4 degrees Celsius. The possibility that the number of ensemble experiments may affect the characteristics of such precipitation change multipliers is clarified by validation using 20 km spatial resolution climate projection data with a large number of ensembles.

 Future changes in river discharge were calculated for the Omono River from the d4PDF rainfall data, which consists of 1500 years of past experiments and 5400 years of future experiments, and a rainfall-runoff inundation model. The results were generally consistent with the estimation by the Ministry. The range of the projections is analyzed in terms of three factors: the rainfall event extraction method based on the rainfall duration, the SST pattern, and the upstream/downstream. This study shows that it is important to consider the difference between the SST results and the upstream/downstream results when examining the magnitude of future changes in river discharge based on large ensemble climate projections.

 The effect of the climate change on the discharge of the Sayogawa river was investitaged using two rainfall-runoff models and d4PDF. A storage function model and a distributed rainfall-runoff/flood-inundation model were applied as the runoff models. Fistrly, the flooding in August 8-9 2009 by Typhoon 9, the historically largest event in the region, was reproduced by the runoff models. Then, the yearly maximum 24 hour rainfalls of d4PDF for the Sayogawa river was analyzed and the 24 hour rainfall - return period relation was summarized. Finally, the peak discharges driven by the hourly rainfall time series of the yearly marimum 24 hour rainfall are analyzed. As the result, it was observed that the peak discharge value of a return period is different according to the runoff models. The discharge with the very intensitve rainfall tends to be higher by DRRFI, a physically based model, which is in general more sensitive to the rapid change of the rainfall. In other word, DRRFI may provide the safer side consideration when applied to river planning.

 In recent years, flood disasters have become more frequent and their scale has been increasing. One of the reasons for this is the increase in rainfall due to global warming, and it is necessary to evaluate future disaster under changing climate. In particular, the rivers in Toyama are known rapid flow and are often erosion, important to predict erosion areas. In this study, rainfall-runoff calculations were carried out for river basins in Toyama using d4PDF to reproduce past conditions, and temperature rise experiments of 2°C and 4°C were conducted to evaluate the effects of inflow, overflow, and possible erosion. As a result of comparison with non-exceedance probability, the flow rate increased by about 1.1-1.7 times when the temperature increased by 2°C and by about 1.3-2.2 times when the temperature increased by 4°C with 150-year probability. In the evaluation of water overflow potential, the number of dangerous areas in the Oyabe River, which has a relatively gentle gradient, increased. The erosion potential of the Joganji River, which has a steep gradient, increased by more than 60% when the temperature was raised by 2°C. the Shogawa and Kurobe Rivers also increased by more than 60% when the temperature was raised by 4°C.

 This study aims to elucidate the future change of the patterns of snake line in a changing climate. Constituted by two parameters, hourly rainfall intensity and soil-water index, a snake line is a curve for judging early-warning on rainfall-triggered sediment disasters. To quantify snake line pattern, we focused on analyzing the geometric range of snake line in terms of the maximal hourly rainfall and soil-water index. We adopted high-resolution future climate projections of 2-km and 5-km Non-Hydrostatic Regional Climate Models under the scenarios of RCP2.6 and RCP8.5. Snake lines obtained by climate projections were verified by the reanalyzed precipitation of Radar-AMeDAS. Obvious changing trends for the two parameters and corresponding spatial distributions are revealed on each mesh. Based on six types of snake line stretching, we examined the spatial distributions and statistics of the future changes of snake line patterns as well as their relation to rainfall prone to sediment disasters under the two climate scenarios.

 This study aimed to analyze external forces and to estimate high erodible sites under the current and future climate scenarios using d4PDF downscaling data. The Toyohira River is a rapidly flowing river that flows through the urban Sapporo area; therefore, high-speed currents develop, leading to conditions that have traditionally made the area vulnerable to flooding, from which erosion is likely to occur. To estimate high erodible sites, we first calculated the outflow of 3000 and 5400 cases of d4PDF data based on current and future climate forecasts, respectively . Next, the current and future climate scenarios were classified according to the rainfall distribution using cluster analysis, and external forces were analyzed. In addition, the high erodible sites were evaluated from the calculation of the hydraulics on the Toyohira River channel using the results of outflow calculations in the present climate scenario as an external force. Thus, it is estimated that greater flooding than has ever been previously experienced will occur owing to climate uncertainty, even in the present day. It is further estimated that some areas are highly erodible, and care should be taken when planning river management strategies.

 Global hydrological models are continuously being developed with the objective of evaluating more precisely water scarcity risk at finer spatial resolutions. Acomplishing this objective, however, can be hindered by the difficulty in representing changes that are driven by urban infrastructure and may result in misleading risk evaluations. In this study, we simulated the hydrological response of the Tone River basin and Arakawa River basin in Japan by employing the H08 global hydrological model. We were able to reproduce the observed river discharge flows. Then, we implemented a new method that integrates water intake and drainage system to improve the evaluation of water scarcity risk. To this end, we compared four different scenarios varying the way in which the urban water supply and the drainage system can be integrated in the H08 model. The results showed that integrating both urban water supply infrastructure and the drainage system gives a more realistic view of the degree of water scarcity in urban areas.

 This study projects long-term changes in available water resources over Japan using the 150-year continuous run of the Meteorological Research Institute's super-high resolution climate model (MRI-AGCM3.2S). Available water resources was evaluated for multiple land use scenarios to isolate the impact of climate change and land use change. The difference in evapotranspiration is as large as 500mm in the area where population change is large. In response to the regional distribution of changes in climatic conditions, the amount of available water resources will increase significantly in north part of Japan, while the amount of evapotranspiration will also increase nationwide. It was projected that the amount of available water resources will decrease significantly in the central mountainous area. Time series of climatological value of available water resource was analyzed for each river basin. The long-term change patterns vary depending on the basin, and the climatological values do not always change unidirectionally from the present to the future. Available water resources will become more severe before the end of the century in some basins.

 In recent years, the development of global hydrological models and their higher resolution have been progressing, but with the higher resolution problem, and water stress tends to be overestimated in grid-scale assessment. In this study, we tried the evaluation at the city scale, and estimated the water demand and the supply in each city. We evaluated the water stress and solved the problem. Monthly water stress evaluations were conducted for 20 cities, taking into account the location of the actual water intake points and the amount of water transported from urban water infrastructures. The monthly water stress evaluation allowed us to verify the seasonal change of water stress and the effect of the urban water infrastructure on water supply.

 Recently, it has been suggested that saturated lateral flow on a local scale, such as hillslope hydrodynamics, affect the terrestrial water heat balance such as evapotranspiration and soil moisture, and that these areas are not uncommon globally. However, no study expresses the contrast of hydrological quantities on the hillslope at a fine resolution on the global scale with reasonable computational cost. Here, we implemented a subgrid saturated lateral flow scheme in the integrated land simulator (ILS), and carried out a global simulation for 163 years. Then we expressed the hill-valley moisture contrast by diagnostically downscaling the simulation results using subgrid topography information. The results suggested that by considering the saturated lateral flow, the soil moisture and the evaporation tend to decrease in a wide range of the global average, and the runoff amount increases in the area where the evaporation decreases, especially in the alpine area. We re-evaluated the land surface water balance from the viewpoint of water redistribution by the hillslope dynamics, and experimentally explained the possibility of the existence of vegetation in the valley in the arid area. In addition, we compared the downscaled result with high-resolution soil moisture satellite observation data (SMAP) and expand the possibility of verification of a global land model by high-resolution satellite data. Although it is necessary to improve the method of applying land cover in the subgrid and climate forcing dataset, it was suggested that the hill-valley moisture contrast can be represented in the global land model with the saturated lateral flow.

 The purpose of this study is to clarify whether terrestrial water cycle variables can be better represented by improving water retention curve and considering horizontal and vertical distribution of soil hydraulic parameters in a land surface model. The hydraulic function of a land surface model ILS was changed from Clapp and Hornberger (CH) model to observation-based van Genuchten (VG) model. Horizontal and vertical variability of soil hydraulic parameters were estimated globally at 0.5 degree resolution using the latest soil dataset and a machine learning-based pedotransfer function. The improvement of the soil scheme resulted in an increase in the coefficient of determination of estimated soil moisture from 0.45 to 0.54, as well as an improvement in runoff and river discharge. The change in the hydraulic function to VG model improved the accuracy of estimation, especially for sandy soils, due to the increased water holding capacity of the soil. Other hydraulic variables such as runoff and river discharge are better represented as well, which is expected to contribute to the improvement of accuracy of terrestrial hydrodynamic modeling.

 Soil moisture through soil water potential controls the transpiration from vegetation. This study investigates how vegetation impacts soil moisture bimodality in coupled analytical land-vegetation-atmosphere model for mid latitude region. The influence of precipitation on soil moisture is direct, but soil moisture through evapotranspiration controls mass and energy transfer between land and atmosphere, thus influencing the precipitation. In the current study, we partitioned evapotranspiration into transpiration and soil evaporation and considered three different vegetation schemes, i.e., static Leaf Area Index (LAI) for barren land and dense evergreen forest and dynamic LAI for the cropland. The investigation found that cropland showed soil moisture bimodality while barren land and dense evergreen forest had unimodal soil moisture distribution. The dominant mode of soil moisture on barren land is on the dry side, while for the dense evergreen forest has a mode on wetter side. This result indicates the importance of dense vegetation in the land-atmosphere coupled system to reduce the probability to dry and hot season. This information can further be utilized to include the role of vegetation activities in land-atmosphere model development.

 Evaporation from natural lakes and artificial reservoirs is a loss of water resources, but existing estimations of it are based on simplified methods. The evaporation is a phenomenon related not only to in-lake thermodynamics but to change in water surface area driven by riverine in- and outflows, so this study reestimates it on a global scale with a coupled model framework of hydro- and thermodynamics in rivers and lakes, of which estimate is consistent with previous ones. It was found that evaporation per area from reservoirs is bigger than that from natural lakes due to the difference of spatial distribution, and reservoir construction increases evaporative loss due to warm-up of water surface temperature, in addition, to increase in surface area. The estimation of storages in each water body also showed that most of the residence time is caused by lakes. These results suggest that it is important to consider the thermodynamics of lakes and the interactions with rivers for the estimation of water resources and their quality.

 A recent study released global maps of streamflow characteristics with high accuracy, which are expected to calibrate hydrological models without gauging streamflow. We evaluated the availability of mean annual runoff height (Q_MEAN) and baseflow index (BFI) in the global maps when calibrating a global hydrological model, respectively. We note that the calibration parameters were composed of snowmelt coefficient (C_s) and saturated permeability coefficient in soil (K_sat). Although it was difficult to calibrate the model on the basis of the Q_MEAN, the calibration with the BFI performed well, which was comparable to that of observed discharge data. In addition, K_sat controlled the accuracy of streamflow in this study. The calculation process of BFI unlike Q_MEAN depends on K_sat, which can contribute the success of calibration with BFI.

 In this study, the impacts of time integration methods on runoff estimation were investigated in two land surface models (LSMs): MRI-SiB and SiBUC. Both models were forced by the same atmospheric data from MRI-AGCM 3.2 with 20-km spatial resolution. For updating soil moisture, SiBUC implements an explicit midpoint method, while MRI-SiB applies a semi-implicit method. In this research, two simulation cases were performed by SiBUC by changing different time integration methods. It was found that applying different time integration methods affected the runoff characteristics. The simulation by the explicit method resulted in higher surface runoff than subsurface runoff. On the other hand, the results by the semi-implicit method showed the opposite characteristics. Analysis of the simulated discharge show the different responses of the estimated flow to the rainfall due to the change in the ratio between surface and subsurface runoff. It is thought that a semi-implicit scheme is one of the key factors of low surface runoff in MRI-SiB.

 Hydropower is an important development opportunity for the Mekong River Basin (MRB) and its member countries. However, the rapid development of hydropower projects in the last decades has threatened the natural flow regimes of the basin. When all hydropower projects are completed, they will put more pressure on the hydrology of the basin. Due to the unavailability of the detailed hydropower operation rules, especially for the future development, we first estimated the general patterns of dam operations by maximizing the hydropower generation based on the proposed and existing storage volumes and turbine flow capacities. Then we propose a simple storage model representing the hydropower dam operations in this river basin. For the hydrological assessment, we used a distributed hydrologic model, Rainfall-Runoff-Inundation (RRI) model. The simulation was conducted from 2010 to 2016. The results suggest that the RRI model incorporated with our proposed model can reproduce the discharge hydrograph of the MRB. The hydropower developments would increase the dry seasonal flow while decreasing the wet seasonal flow. The impact can be seen as far as downstream of the Lower Mekong Basin (LMB). The impacts of future hydropower development on the LMB are significantly larger than the present hydropower development. On the monthly scale, under the future hydropower development scenario, the discharge changes range from -17% to +40% at Chiang Saen, and -16% to +100% at Kratie. In the dry season, the discharge at Kratie was significantly increased up to 40%. While in the wet season, the discharge was decreased by 15%. This study provides important information for reservoir operators and policymakers for the sustainable development of the basin.

 In the recent decade, many projects of water infrastructure development namely, hydropower development and irrigation expansion have been rapidly increasing and are expected to have more in the Mekong River Basin, thus the substantial influence on the streamflow of the Mekong mainstream are unavoidable. Also, climate change is the dominant driving factor affecting the streamflow regime. For better water resource management for the Mekong River Basin, this study aims to assess the streamflow change driven by these drivers, i.e., hydropower development, irrigation expansion, and global climate change based on the SWAT model simulations. The results show that hydropower and irrigation development have limited impacts on annual flows under 2040 development scenarios compared to the early development period. However, climate change can reduce the streamflow up to 13%-36% and 14%-62% in June between 2040 to 2080 for RCP4.5 and RCP8.5 scenarios, respectively, while the streamflow is projected to increase up to 45%-85% and 65%-130% in October between 2040 to 2080 for RCP4.5 and RCP8.5 scenario, respectively. Furthermore, the streamflow shows significant increases in the long-term period during SeptemberNovember and December-February, but decreases during March-May and June-August. Whilst, hydropower decreases the streamflow in the wet season by 7%, but increases in the dry season by 29%. And, for the irrigation, there is not much impact on the streamflow, but it shows substantial flow reductions during the dry season and up to 18% in the middle part to the lower part of the basin. Furthermore, the combined impacts of all driving factors cause substantial flow reductions during the first half of the wet season (-42%), but increase the flow in the dry season by 40% and 14% compared to the early development and nearly current scenario, respectively. This result can be a reference for water management in the Mekong River Basin, particularly flood and drought management as well as agricultural production planning.

 Green infrastructure (GI) is a global, innovative stormwater runoff control measure in response to urbanization and climate change. This study quantified GI performance in improving the urban hydrologic environment. Storm Water Management Model was employed to simulate stormwater runoff in the Yazawa and Maruko River basins in Setagaya, Tokyo, Japan. Under the 75-mm/h rainfall scenario, the existing GIs were found to be incapable of controlling runoff at the government-specified rate of 10-mm/h. Furthermore, while simulating 1–10-year return periods of rainfall with 1–3-h durations, the reduction of runoff depth and peak runoff was less effective when rainfall intensified. A negative relationship was obtained using regression analysis between rainfall depth and the reduction rate of runoff depth and peak runoff. Meanwhile, rainfall duration positively and negatively correlated with the runoff depth reduction and peak runoff reduction rates, respectively. Furthermore, the reduced surface runoff considerably reduced the combined sewer overflow in the 1-year rainfall return period.

 A highly accurate and practical flood prediction model is required to carry out disaster prevention operations and evacuation actions with a margin in small watersheds with a flood arrival time of less than one hour. In this paper, we propose to build a deep learning model that predicts the water level by using prediction learning of the soil water index. The prediction accuracy was verified at the Nakatsu River water level observatory point (basin area 42.37km², flood arrival time less than 1 hour) in the upstream area of Miyagase Dam. As a result, the predicted water level up to 6 hours ahead could be predicted with high accuracy unless the current state of the soil water index deviates significantly from the predicted value.

 In recent years, with the increase in extreme rainfall, floods of small and medium-sized rivers have occurred frequently.We focused on a small and medium-sized river (155 km²) and conducted runoff simulation using four digital elevation models (Asia 30, Asia15, fifth-order mesh, and Japan Flow Direction Map (JFDM)) with different spatial resolutions. As a result, JFDM with a spatial resolution of 77 m x 92 m (three times scale-up) gives high values for both the Nash-Sutcliffe coefficient and the correlation coefficient.

 The accuracy of the meteorological input data is important for the proper reproduction of the water cycle. In this study, we conducted model simulations of river discharge for the whole of Japan using the integrated water resources model (H08) driven by three different precipitation data based on climate reanalysis, gauges and radar with gauges, and evaluated the reproducibility of river discharge. Regardless of which precipitation data were used, rivers in northern Japan tended to be underestimated their spring peak discharge. The use of precipitation data generated by densely observed gauges and gauges with radar has been confirmed to improve the reproducibility of river discharge than the use of precipitation derived from climate reanalysis. The accuracy of river discharge is relatively the same between experiments with gauges only and gauges with radar. However, in mountainous areas such as the middle basin of the Tenryu River where the precipitation observation network is insufficient, the use of radar precipitation data is expected to improve the reproducibility.

 It is necessary to develop a rainfall index that directly predict the occurrence of flooding for early evacuation of the residents. In this study, we proposed a method to directly and probabilistically predict the occurrence of inundation from a rainfall index. We calculated the 24hours rainfall averaged over the catchment area (Lv4 rainfall) corresponding to the hazardous water level at each 109 class A river systems nationwide based on past observed data of rainfall and water level. Then, we evaluate the flood risk probability that the water level reach the hazardous level, by using the Lv4 rainfall and an ensemble predicted rainfall data. This method was applied to the heavy rainfall events in July 2020 at the Kyushu region. The results indicated that the rates of concordance, oversight and missing were 27.8, 0.76 and 5.46%. The median of lead time was 19 hours. These show the fundamental validity of this method.