Development of GIS-Based Building Damage Database for the 1995 Kobe Earthquake
2.2 Building Damage Data by BRI
This survey was carried out by professionals, researchers, and volunteer students who major in Architecture or City Planning. After the investigation, City Planning section of Hyogo prefecture conducted a further survey with the same method to supplement data in insufficient areas. The survey was conducted not only in Nishinomiya City but also in other cities: Kawanishi, Itami, Takarazuka, Amagasaki, Kobe and Awaji Island, to grasp the comprehensive damaged area due to the earthquake. The aim of the survey was for future academic contribution. Therefore, the assessment of the building damage was more strict and the method of the survey was more technical than that of local governments. The damage was classified into 5 categories, i.e., "collapse or heavy damage", "moderate damage", "slight damage", "no damage", and "burned". However, each building was evaluated by visual observation from outside of the building, so that the inventory does not contain "roof type" and " construction period", and the classification of "non-wooden buildings" in "structural type" is not divided. Figure 2 shows the polygon data of buildings in the BRI's GIS database. The coordinate system is sixth of JPN.
Figure 2: Polygon Data of Buildings by BRI
3. Matching the different Data Sets
3.1 Pretreatment
First, the authors had to make the Nishinomiya data (Data-1) and the BRI data (Data-2) with the same coordinate system. The coordinates of Data-1 were transformed from the fifth coordinate system of JPN to the sixth. Next, the regions of the both data sets were compared to extract the researched area for the database as shown in Figure 3. While the whole area of Nishinomiya City was made as the target of Data-1 because of the purpose for the tax reduction, the area for the Data-2 was limited. As a result, the south area of the City was extracted as the research object. One problem of the matching was that overlapped point data had existed for one building in Data-1 because the data were constructed based on the household unit. It was necessary to integrate these multiple point data into one.
3.2 Integration by the Positional Relationship
To integrate the both spatial data sets, only a spatial positional relationship was considered. Firstly, the point data of Data-1 inside the polygon data of Data-2 was extracted as shown in Figure 4. Secondly, in the other data except for the above, the distances between the points and polygons were calculated, and a point located within 3m from a polygon was matched with it as shown in Figure 5. The shortest distance is the perpendicular distance between a point and an edge of polygons or the distance between a point and a corner of polygons. Figure 6 shows the method to calculate the shortest distance. After that, these matched data were printed out on a paper, and they were compared with a detailed map.
Figure 3: The Area of Nishinomiya City Covered by the Two Data Sets
3.3 Processing of Overlapped Data
In case of being more than two points inside one polygon as shown in Figure 7, these multiple overlapped data were integrated into one data set considering the building use. The reason of the overlapped points was mostly due to the difference in dealing with attachment buildings, such as garages or warehouses. The cause of it was considered and these additional data were omitted from the viewpoint of the building use. In addition, this process is carried out considering the construction period, structural type, and damage classification. Figure 8 shows the flowchart of the matching process.
Figure 4: Matching of the Point Data inside Polygons
