DEVELOPMENT AND APPLICATION OF SPATIOTEMPORAL ANALYSIS METHODS IN SPATIAL INFORMATION SCIENCE

这项研究开发了一套方法，用于分析空间对象的时空域，不能完全分析现有的空间或时间的方法。本研究选择了四种常用于表示空间现象的空间对象：1）点，2）线，3）多边形和4）表面。沿着方法开发的实证应用也是并行进行的。点的分析基于其对应的表面的分析：点通过空间平滑转换为表面，并通过使用峰、坑和锯齿的表面分析方法进行分析。这种方法可以检测点分布的局部和全局模式，如果将其与GIS中的图形分析相结合，则可以更有效地工作。每个环节的时间变化，这导致了网络的整体变化的评估方面进行定量评估。该方法特别适用于分析诸如道路网变化之类的线的离散变化，与线一样，多边形也可以用基元分解方法分析，即多边形分布的变化被分解为单个多边形的变化。将该方法应用于便利店的市场区域分析，得到了一些有用的实证结果，证明了该方法的有效性。分析的表面是基于自然定义的两种类型的时空奇点的时空镶嵌。这种方法给出了一个共同的时空现象的两种不同的观点，这表明了一种可能的方式结合空间和时间的分析，已分别在不同的领域研究。

This study developed a set of methods for analyzing spatial objects in the spatiotemporal domain that cannot be fully analyzed by existing spatial or temporal methods. This study chose four types of spatial objects frequently used in representing spatial phenomena: l) points, 2) lines, 3) polygons, and 4) surfaces. Along method development empirical applications were also done in parallel.Analysis of points is based on analysis of their corresponding surface: points are converted into a surface by spatial smoothing and analyzed by a surface analysis method which uses peaks, pits, and cols. This method can detect both local and global patterns in point distributions and works more efficiently if it is linked to graphical analysis in GIS.In analysis of lines links are used as primitive elements to describe change over time. Each link is quantitatively evaluated in terms of temporal change, which leads to evaluation of the whole change of a network. This method is appropriate especially for analysis of discrete change of lines such as change of road networks.As well as lines, polygons are analyzed by the primitive decomposition approach- change of a polygon distribution is decomposed into a set of changes of individual polygons. Application of this method to market area analysis of convenience stores yielded some useful empirical findings which demonstrated the validity of the method.Analysis of surfaces is based on spatiotemporal tessellations naturally defined from two types of spatiotemporal singularities. This method gives us two different viewpoints on a common spatiotemporal phenomenon, which shows one possible way of combining spatial and temporal analyzes that have been studied separately in different fields.