Bus crews in Japan have problems such as the severity of their working conditions and the low level of income. Therefore, few young people work for bus crews. In addition, bus operators are strictly managing their operations based on the “working hours improvement standards of motorists” developed by the Ministry of Health, Labor and Welfare due to traffic accidents caused by overwork of the drivers in high speed bus services.

 In preparing a work schedule for bus crews, it is necessary to create a work schedule that can satisfy various conditions among limited crews, which is a heavy load on work schedule preparers. In this research, Airline crew scheduling problem is applied to the creation of the work schedule of bus crews by adding restrictions specific, and its usefulness is shown, and systemization and problems are examined.

 Some firms have moved out from the Pacific coastal area where severe tsunami damage is expected. The construction of seawalls reduces the risk of inundation and affects their relocation. In order to capture the benefit of the seawalls, it is necessary to grasp the change in the number of firms associated with the construction of the seawalls. The risk of tsunami inundation varies spatially de-pending on the distance from the coast and the topographic conditions. In addition, there are regional location factors by industry. Therefore, we use the Difference-in-Differences method to capture the influence of the publication of inundation risk information on the location of firms. Specifically, we analyzed the change in the number of firms by industry using location data before and after the publication of the inundation risks. As a result, we found that industries with local demand, such as education, do not change their locations even after they know that their locations are in inundation areas, while industries with wide demand areas, such as manufacturing and wholesale, move out from the inundation areas.

 A timetable that trains depart from the station at the same time every hour is convenient. This type of timetable is called a “Cyclic Timetable”. We defined the indicator that represents the level of cyclic timetable called cyclic rate and analyzed a time-series change of cyclic rate. We also analyzed a time-series change of relationship between cyclic rate and passengers per distance in regional railway of Japan. As a result, we showed that 51.7% of private railways have cyclic rate of 50% or more. On the other hand, only 10.9% of JR regional lines and 12.5% of former JNR lines have cyclic rate of 50% or more.. In addition, we indicated that most cyclic timetable lines have high service frequency, short distance and many transit line. Moreover, we showed that the lines that have been cyclic timetable since 2005 had 10.7% higher passengers per distance than no cyclic timetable lines.

 This study investigates the theoretical properties of dynamic system optimal (DSO) and dynamic user equilibrium (DUE) assignment considering the heterogeneity of users’ value of schedule delay in tandem bottlenecks. First, we formulate the DSO assignment as an infinite-dimensional linear programming problem. To analytically derive the DSO solution and show its essential properties, we apply the dynamic programming principle and optimal transport theory. Second, we formulate the DUE assignment as an infinite-dimensional linear complementary problem and develop its analytical solution method by utilizing the analytical DSO solution. Specifically, we prove that the queuing delay at a bottleneck in a DUE solution is equal to an optimal toll that eliminates the queue in a DSO solution under certain conditions of a schedule delay function. This fact enables us to derive a closed-form DUE solution using the DSO solution. Finally, we discuss the theoretical relationships between the DUE and DSO states based on the closed-form equation.

 In various wide-area plans and operations including national spatial strategies, regional blocks were granted as higher-level plans after various controversies and conflicts. As the number of people in charge of local area activities decreases, it is unavoidable that local blocks should be reproduced from a bottom-up perspective based on a real connection (contributing to the region through visitation) between each individual and the block. This study applied an inverse estimation of regional blocks through principal component analysis based on a nationwide survey of regional experience seekers. As a result, 11 principal component axes which define the basic national structure were extracted from the relevant population. Furthermore, results showed quantitatively that some regions are involved in some blocks in terms of relation. Some local blocks have no mutual relation. New guidelines were inferred for the formation of practical policies for use in future societies.

 The purpose of this study was to clarify the evaluation method of the historical value of the historic environments related to Japanese Americans in Hawaiʻi. In this study, first, historic environments related to Japanese Americans registered in the National Register of Historic Places and the Hawaiʻi Register of Historic Places have been extracted. After that, based on a detailed analysis of the descriptions of the Registration Forms prepared for nominating these historic environments, evaluation perspectives and evaluation axes have been categorized. In addition, the characteristics of the evaluation methods were clarified based on the analysis of the interrelationship of each type. I believe that the results of this research will contribute to the promotion of historic preservation not only for Japanese Americans in Hawaiʻi but also for other ethnic groups.

 Recently, appropriate planning for urban shrinkage with the viewpoint of mitigating disaster risks has become more important under situations such as climate changes and depopulations. This study aims to examine the future population of 2040 that will be exposed to landslide prone areas, to design the future urban structure scenarios including the mitigation of landslide disaster risks and to evaluate them from multiple perspectives to assess the impact of urban shrinkage on each evaluation. Major findings are as follows: (1) It is assumed that 17.6% of population in 2040 will live in landslide prone areas. (2) Comparing BAU, urban structure scenarios with mitigating landslide disaster risks have higher evaluation in all indices selected from the “Handbook on Evaluation of Urban Structure.” (3) Restricting the renewal of houses that exceed the useful life in 2040 in landslide prone areas and migrating the people who will live there to inducing areas for dwelling may be possible to construct a sustainable future urban structure.

 To design maintenance policies for public facilities, the minimization of life cycle costs has been adopted. However, a maintenance agency has a budget constraint. Accordingly, if the maintenance cost has a larger share of the budget, it is necessary to reduce the expenditures for other items, or to increase the revenue by increases in public charge or tax rates. Any increase in public charge or tax rates (except for lump-sum tax) generates deadweight losses. These additional costs are called the marginal cost of public funds (or MCF). Our study takes account of the MCF endogenously to design maintenance policies. As a result, we show that the endogenous change in the MCF should be taken into account because it affects the optimal maintenance policy, and how the optimal maintenance policy changes due to the change in the MCF theoretically as well as quantitatively.

 Although there are various investigations of land use classification by deep learning from satellite images or topographical maps, it has not yet reached an operational level. This study proposes the operational level land use classification method, which indicates more than 85% of accuracy in OA (Overall Accuracy), by deep learning. CNN and Land Use Subdivision Mesh System of National Land Numerical Information as training data are applied to the proposed classification method. Consequently, the following findings are obvious: 1) the number of epochs should be about five, 2) the number of training samples for each class should be more than 10, 000. In addition, the method to integrate multiple training results is developed by using the fact that the training results are slightly different each time. This study proposes the new land use classification method that can show an operational level classification accuracy of about 90% in OA on a regular basis.