Toshihiro Osaragi

Simulation Model of Evacuation Behavior Following a Large-Scale Earthquake that Takes into Account Various Attributes of Residents and Transient Occupants

Understanding human behavior, such as waiting, returning home, and evacuation, after a great earthquake is very critical in establishing detailed disaster prevention planning. In the present paper, we construct a simulation model to describe how human behavior varies as a function of physical damage, such as the spread of urban fires and street blockage due to collapsed buildings. The proposed model is applied to a densely built-up area of Tokyo using a database of the spatiotemporal distribution of railroad passengers, automobile users, and pedestrians. Using the model, we attempt to demonstrate some new findings that can be applied to disaster prevention planning by examining evacuation plans and various settings in earthquake simulations.

Effects of Ground Surface Relief in 3D Spatial Analysis on Residential Environment

The inherent worth of visibility from high-rise housing has been actively discussed in recent years. Securing sunshine is also a fundamental and important issue in residential areas. Since visibility and sunshine vary according to various local factors, it is difficult to extract and analyze them in urban areas covered by a number of buildings. In previous research, visibility and sunshine analysis have been carried out under the assumption that the ground surface is a flat plane. However, it is necessary to incorporate geographical relief when conducting 3D urban spatial analysis. In this article, we construct a visibility analysis system, adapt it to a 3D urban surface model, and examine the effects of geographical relief on urban spatial analysis.

Map element extraction model for pedestrian route guidance map

A model that can generate compact and intelligible route guidance maps is essential for distributing efficient information on pedestrian route guidance through mobile terminals. In this research, we focus on the fact that the existing route guidance maps have been prepared by considering the characteristics of peoples spatial cognition. By analyzing which roads and buildings are represented in existing maps, i.e., which map elements are important and necessary for pedestrian route guidance, we construct a model that can extract key map elements from the geographical database. The route guidance maps generated by the proposed model are shown and evaluated in comparison with the existing maps.

Estimating Spatio-Temporal Distribution of Railroad Users and Its Application to Disaster Prevention Planning

To develop measures for minimizing human damage from a devastating earthquake, it is important to understand the characteristics of the population and its spatio-temporal distribution in an urban area. In the present paper, a model is constructed that simulates the route selection behavior and the transfer choices of railroad users using a geographic information system. The spatio-temporal distribution of users is estimated by applying the model to the Tokyo metropolitan area, using data collected in a person-trip survey. Some numerical examples using the proposed model are shown for detailed disaster prevention planning. In particular, the number and the spatio-temporal distribution of people with difficulty returning home are discussed.

Modeling of Land Use Transition and its Application

A model for describing the spatial distribution of land use utility is proposed. The utility function used consists of the benefit and the cost in the process of land use transition. The benefit is described as the positive utility obtained through the land use activity. The cost is described as the negative utility necessary for changing the state of land use. Using the model, the effects of socioeconomic factors on utility can also be evaluated. As numerical examples, the proposed model is applied to the actual land use data and the effectiveness of the model is assessed.

Predicting Spatiotemporal Distribution of Transient Occupants in Urban Areas

In order to discuss in detail the environment and urban systems, it is necessary to consider not only static physical objects like buildings, but also the spatiotemporal aspects like the distribution of population.This paper aims to construct models that describe the spatiotemporal distribution of population in urban areas. The models are composed of parameters describing the number of persons per unit floor-area of buildings, which varies according to the time fluctuation factors and the location factors, and are calibrated using a person trip survey data and GIS data.We discuss the characteristics of the spatiotemporal distribution of population and the accuracy of the models, and demonstrate that the proposed models can benefit all phases of urban planning, which include risk assessment and disaster management.

A Method for Estimating Land Use Transition Probability Using Raster Data

In the field of urban and regional planning, several Markov chain models for land use conversion have been proposed. However, some problems have been encountered when estimating land use transition probabilities. In this paper, a by taking into account spatial units of land use transition, fixed state of locations, and some new findings on land use conversion are presented using numerical Timmermans, 2003; Saarloos, Arentze, et al. 2005). The strengths and also indicated (Parker, Manson, et al. 2003). Most MAS/CA models are, by of locations, and time-varying probability Systems in Architecture and Urban Planning, 69-84. Markov chain, Land use, Transition probability, Building lot, Fixed state, Jos P. van Leeuwen and Harry J.P. Timmermans (eds.), Innovations in Design & Decision Support 70 nature, interdisciplinary. Therefore, it is obvious that we should draw on and combine knowledge from many disciplines in order to develop creative new tools for empirical analysis. A large body of basic research using Markov chain models exists in LUCC studies (Drewett, 1969; Bourne, 1971; Bell, 1974; Bell and Hinojosa, 1977; Robinson, 1978; Jahan, 1986; Ishizaka, 1992; Muller and Middleton, 1994; Theobald and Hobbs, 1998; Qihao, 2000). Markov models may be combined with CA for LUCC modeling, as evidenced by joint CA-Markov models (Li addresses fundamental problems hidden in the Markov chain models, the challenges shown in this paper are common and partly overlap those of MAS/CA models. In particular, the question has evolved from “Which is the best?” to “What are the conditions under which it is the best, and how can we flexibly combine the appropriate approaches on a case-by-case basis?” Employing Markov chain models to predict the distribution of land use is always plagued with several types of errors. One type stems from the uncertainty, which is inevitably inherent in the transition matrix. Since each coefficient in the matrix refers to one sub-area, an error of this type is directly related to the manner in which the sub-areas are formulated. Therefore, the authors traced the manner in which such error was generated and developed methods for estimating and decreasing it. The latter involved the division of a total area into appropriate sub-areas by using a simulated neural network Furthermore, Markov chain models have predominantly focused on providing information regarding the amount, location, and type of LUCC that has occurred. Only a few models have been developed to address the manner in which and the reasons for change (Qihao, 2002). Then, the authors described land use transition probabilities by using the concept of land use utility, and measured and evaluated the effects of socioeconomic factors on land use utility (Osaragi and Kurisaki, 2000). Thus, several methods have been proposed to sharpen Markov chain models. However, there still remain several types of errors in estimating transition probabilities. In the previously proposed models, land use transition probability was typically estimated using time series raster data by counting the number of cells. Although this method is very simple, there are several problems associated with it, as described below. First, it has been widely recognized in geography that there exist common scale-related problems related to the verification and validation of MAS/CA models. Nevertheless, we should take into consideration that the basic unit of land use change is not a cell but a building lot. Adopting the conventional method entails the risk of overlooking the true transition structure (Yoshikawa, 1994). Toshihiro Osaragi and Yoshitsugu Aoki and Reynolds, 1997; Balzter, Braun, et al. 1998). Although this paper (Aoki, Osaragi, et al. 1993, 1996). A Method for Estimating Land Use Transition Probability 71 Second, in CA, the system is homogeneous such that the set of possible states is the same for each cell and the same transition rule applies to each cell underlying Markov chain models (Stewart, 1994). One basic assumption is the consideration of LUCC as a stochastic process. However, there are some locations where land use is in a fixed state. In other words, land use transition at such locations should not be considered as a stochastic process. Hence, the conventional model, which assumes that land use transition at all locations will vary stochastically, might yield an incorrect estimate. Finally, while cellular modeling techniques offer greater flexibility for representing spatial and temporal dynamics, these dynamics are based on the future, except in a few instances where it has been tested (Bourne, 1971; Bell, 1974; Bell and Hinojosa, 1977). Therefore, it is necessary to consider the variations in the transition probability in terms of time. Hence, this study addresses the abovementioned issues and improves upon previous models by proposing the following methods: (1) A method for estimating land use transition probabilities based on building lots using raster data. (2) A method that takes into account the existence of building lots in a fixed state. (3) A method for expressing changes in transition probability based on the concept of land use utility. 2. LAND USE TRANSITION PROBABILITY BASED ON BUILDING LOTS 2.1 The order of the Markov chain has only been formally tested in a few studies (Bell, 1974; Robinson, 1978). In several Markov chain models, the “first-order Markov chain” property is simply assumed to simplify the Markov chain property, an investigation was carried out using the actual land use raster data that has been used in this study (Osaragi, 2005). Since the result showed that the land use transition process could be adequately described by a first-order Markov model, this simple property is assumed in the following discussion. (Parker, Manson, et al. 2003). This feature corresponds to the assumption stationary transition probabilities (Parker, Manson, et al. 2003). The models The Order of the Markov Chain 1” theoretical discussion. In other words, it is assumed that “the state at time t + is dependent only on “the state at time t.” In order to ensure the first-order proposed thus far assume that the transition probability will remains Table in 72 2.2 In conventional land use transition models, the probability Pij for a land use change from j to i is estimated using the following equation: ij ij kj k m P m , (1) where mij is the total area of all locations where the land use category changes from j to i during a certain time interval. In other words, the transition probability was estimated with the assumption that all the spots (cells of raster data) varied independently. However, the actual spots do not change independently; rather, all the spots included within the same building lot generally changed together. This implies that the spatial unit of land use change is a building lot. A simple example of the difference between the transition probability matrices based on cells and building lots is shown in Figure 1. The transition probability matrices obtained are evidently distinct. Therefore, in order to accurately ascertain the structure of the transition, it is necessary to estimate the transition probability by a method based on building Toshihiro Osaragi and Yoshitsugu Aoki Transition Probability Based on Building Lots Figure 1. Example of difference in transition probability estimated by counting cells and building lots. A Method for Estimating Land Use Transition Probability 73 lots. Thus, the above-mentioned conventional model (equation (1)) should be transformed into the following enhanced equation. ij ij kj k n p n , (2) where nij denotes the total number of building lots where the land use category changes from j to i. In this study, a transition matrix composed of nij, which is called the “transition matrix of building lots,” is utilized, and its probabilistic expression is referred to as the “transition probability matrix of building lots.” 2.3 The transition matrix of building lots should be estimated using vector data of building lots. However, it is relatively simple to synthesize raster data, and large amounts of such data have been collected to date. Assuming efficient and effective use of existing raster data, a method for estimating the transition probabilities of building lots using raster data is proposed. The number of building lots where the land use category changes from j to i, denoted by nij, can be estimated using the following equation. ij ij ij m n a , (3) where mij denotes the total area of the building lots where the land use category changes from j to i, and aij denotes the average area. The average area aij of building lots in the transition from j to i can be estimated by the method shown in Figure 2. If adjacent cells of the raster data display identical transitions, they are considered to constitute one building lot. Therefore, the average area aij of the building lots in transition from j to i is estimated by the following equation. ij ij ij m a n , (4) where n’ij is the number of building lots estimated by the method shown in Figure 2. However, the number of building lots where the land use category Estimation Method of Transition Probability of Building Lots

Development of Web Application for Disaster-Information Collection and Its Demonstration Experiment

In the event of a devastating earthquake, a large number of streets will be damaged and/or blocked by collapsed buildings, and the use of emergency vehicles is expected to be paralyzed and unavailable. It is, therefore, important to quickly collect and utilize disaster-information for mitigation. We develop a Web application for collecting and sharing disaster-information collected by users in real time. Furthermore, we demonstrate that the system can provide effective information in real time for reducing the damage of disaster, by performing a demonstration experiment and a simulation carried out by assuming a devastating earthquake in densely built-up wooden residential area in Tokyo.

Use of the Area-Dividing Method to Minimise Expected Error in Land-Use Forecasts

Employing Markov chain models to predict the distribution of land uses is always plagued by several types of error. One type of error stems from the uncertainty which always resides within the transition matrix. In this paper we therefore present a method for estimating such error and for minimising it. As each matrix coefficient refers to one subarea, error is related directly to how the subareas are formulated, and so our method involves dividing a whole region into more appropriate subareas. A simulated neural network is used to achieve this division optimally. We report how experiments were run within an actual urban area. It was found that land-use prediction error is indeed minimised whenever the area-dividing method is used.

A Decision Support System for Fighting Multiple Fires in Urban Areas Caused by Large Earthquakes

Extinguishing multiple fires resulting from large earthquakes is particularly difficult because such fires break out simultaneously in numerous locations. Therefore, effective disaster mitigation requires immediate identification of those fires that are most likely to spread widely. In this chapter, a Fire-Spread Potential (FSP) index, which defines the number of other buildings that could be expected to be destroyed by fire-spread from particular buildings, is calculated and applied to an aftermath simulation of a hypothetical scenario. We then constructed an agent-based simulation model to describe firefighter activities and used the FSP values to evaluate the decision-making support needed to fight multiple fires simultaneously. The chapter demonstrates that FSP values could be effectively used for firefighter decision-making support in order to identify high-risk buildings, thereby mitigating the disaster.