GIS Based Studies in the Humanities and Social Sciences - Chpater 10

10 A Toolbox for Spatial Analysis on a Network Atsuyuki Okabe, Kei-ichi Okunuki, and Shino Shiode CONTENTS 10.1 Introduction................................................................................................139 10.2 Tools in SANET..........................................................................................141 10.3 Software and Data Setting .......................................................................142 10.4 Network K Function Method ..................................................................144 10.5 Network Variable-Clumping Method....................................................146 10.6 Network Cross K Function Method.......................................................148 10.7 Network Voronoi Diagram ......................................................................148 10.8 Network Huff Model................................................................................149 10.9 Conclusion ..................................................................................................151 Acknowledgments..............................................................................................151 References.............................................................................................................151 10.1 Introduction In the real world, we notice many events and situations that locate at specific points on a network. These are referred to as network spatial events. Some typical examples relevant to studies in the humanities and social sciences are as follows: Homeless people living on the streets (Arapoglou, 2004). Street crime (Harries, 1999; Painter, 1994; Ratcliffe, 2002). Graffiti sites along streets (Bandaranaike, 2003). Urban cholera transmission (Snow, 1855). Traffic accidents (Yamada and Thill, 2004). Illegal parking (Cope, 1990). Street food stalls (Stavric, 1995; Tinker, 1997). 139 Copyright © 2006 Taylor & Francis Group, LLC 140 GIS-based Studies in the Humanities and Social Sciences FIGURE 10.1 Churches alongside the streets in Shibuya-Shinjuku, Tokyo. In addition to the types of events listed above, there is another large class also representing network spatial events, but these occur alongside a net-work. A typical example is shown in Figure 10.1, where the circles indicate the locations of churches in Shibuya-Shinjuku, Tokyo. It can be seen that these are not freely situated over the region, since their positions are strongly constrained by their location along the streets. Not only churches, but also almost all facilities in an urbanized area, are located at the side of streets, and it is actually the gates or entrances of these facilities that lie adjacent to the thoroughfare. This chapter focuses on the analysis of events and facilities that are placed at specific locations on and alongside a network, and are called network spatial events. A decade ago, analysis of network spatial events was very difficult, because network data were poor and there were few tools for their analysis, such that researchers had to assemble data and develop methods them-selves. This task demanded much time and effort. The modern advent of geographical information systems (GIS) and the abundance of network data that are accessible today have, fortunately, made matters easier, and many GIS-based tools are available. In this chapter, we introduce a user-friendly toolbox, called SANET, which is the abbreviated name for Spatial Analysis on a Network. This tool is useful for answering, for instance, the following questions: Does illegal parking tend to occur uniformly in no-parking streets? Are street crime locations clustered in “hot spots”? Do fast-food shops tend to contend with each other? Copyright © 2006 Taylor & Francis Group, LLC A Toolbox for Spatial Analysis on a Network 141 How extensive is the service area of a post office? What is the probability of consumers choosing a particular down-town store? In the subsequent sections, we show how to answer these questions using SANET. 10.2 Tools in SANET SANET was released in November 2001, and it has been evolving ever since (Okabe, Okunuki, and Shiode, 2004). The current 2005 edition of SANET is the third version, and it provides the following 15 tools: 1. Construction of a node-adjacency data set. 2. Assignment of a data point to the nearest point on a network. 3. Aggregation of attribute values belonging to the same item. 4. Generation of a network Voronoi diagram. 5. Generation of random points on a network. 6. Enactment of the network cross K function method. 7. Enactment of the network K function method. 8. Partition of a polyline into constituent line segments. 9. Assignment of polygon attributes to the nearest line segment. 10. Enactment of the nearest-neighbor distance method. 11. Enactment of the conditional nearest-neighbor distance method. 12. Calculation of polygon centroids. 13. Enactment of the network Huff model. 14. Enactment of the variable clumping method. 15. Comparison of two networks. In the subsequent sections, the procedure for spatial analysis on a net-work using these tools is outlined. First, in Section 10.3, SANET and datasets set up on the computer are described. Second, in Sections 10.4–10.8, we show how to achieve spatial analysis with the network K function method (Tool 7) using an illustrative example in Figure 10.1; also shown are the network variable-clumping method (Tool 14), the network cross K function method (Tool 6), the network Voronoi diagram (Tool 4), and the network Huff model (Tool 13). Copyright © 2006 Taylor & Francis Group, LLC 142 GIS-based Studies in the Humanities and Social Sciences 10.3 Software and Data Setting The software SANET consists of two components: the main program, and the interface between this and a GIS viewer. The main program performs the geometric and algebraic computation needed for running the tools mentioned in Section 2. This program works independently, and can, in theory, be interfaced with any GIS viewer. The interface between the main program and a viewer will clearly depend on the choice made from the many viewers available. SANET currently adopts ArcView, which is one of the most popular GIS viewers. The main program and the interface can be downloaded from the SANET Web site: http:// okabe.t.u-tokyo.ac.jp/okabelab/atsu/sanet/sanet-index.html. This download can be made without charge for nonprofit-making uses.Also posted on this Web site is the detailed manual of SANET and information about the most recent version. The GIS viewer ArcView is obtainable at a reasonable price from Environmental Systems Research Institute, Inc. (ESRI). After installing both SANET and ArcView on a personal computer, the computer-readable digital data of a street network and churches has to be obtained. There are many ways of recording and managing the digital data of a street network. The main program of SANET employs adjacent-node tables that are commonly used in computational geometry. The adjacent-node tables for the street network of Figure 10.2 are shown in Table 10.1. This illustration consists of straight-line segments whose end points (called nodes) are labeled by numbers. Table 10.1(a), called a header table, shows that node i, say node 0, is headed to the ID = 0 in Table 10.1(b). Table 10.1(b) shows that the nodes adjacent to the node corresponding to ID = 0 (i.e., node 0) are nodes 1 and 5 (reading downwards). 0 5 4 1 491 2 9 10 FIGURE 10.2 Nodes of a street network. Copyright © 2006 Taylor & Francis Group, LLC A Toolbox for Spatial Analysis on a Network 143 TABLE 10.1 Adjacent Node Tables (a) Header table Node ID Head (b) Adjacent node table ID Adjacent Node 0 0 0 1 1 2 1 5 2 5 2 0 3 8 3 2 4 11 4 491 The structure of street data varies in differing GIS software packages. ArcView uses Polyline, which is not compatible with the adjacent-node tables. Therefore, when SANET is used, we have to transform Polyline to enable it to function. This transformation is made by using Tool 1. The digital data for churches may be given either as the coordinates of their representative centroid points or as polygons representing the areas occupied by the buildings. SANET assumes that features are represented by points. With data given in the latter form, the centroids of the polygons are easily located by using Tool 12. For SANET, all network spatial events are precisely on a network. As is seen in Figure 10.1, churches are not exactly located on streets, because a point does not indicate the gate of a church but the centroid of its buildings. In practice, these entrance data are difficult to obtain, and, hence, we have to estimate them from the centroids. SANET assumes that the nearest point on a street from the centroid of a facility is its gate. The location of these access points is derived by using Tool 2. An example is given in Figure 10.3, which shows the access points of the churches plotted in Figure 10.1. FIGURE 10.3 The access points of churches in Shibuya-Shinjuku, Tokyo, obtained using Tool 2. Copyright © 2006 Taylor & Francis Group, LLC ... - tailieumienphi.vn