Presentation Information
[U15-P05]Utilization of the Integrated Earthquake Simulator for Local Government Disaster Prevention: Accuracy Verification and Web Application Development
*SHO ITO1, Shinpei Ito1, Takayuki Kawajiri2, Hideyuki O-tani3, Yoshihide Sekimoto4 (1.Nikken Sekkei Research Institute, 2.MIERUNE Inc., 3.Japan Agency for Marine-Earth Science and Technology, 4.Center for Spatial Information Science, The University of Tokyo)
Keywords:
Integrated Earthquake Simulator,Web Application,Local Government Disaster Prevention,Accuracy Verification,Road Blockage
Traditionally, earthquake damage predictions by organizations such as the Cabinet Office of Japan have primarily relied on macroscopic analyses using damage functions for each mesh area. However, to promote effective disaster prevention measures, it is necessary to identify damage to individual buildings and road conditions. Conventional area-based analyses have limitations in enabling local governments to grasp critical locations to protect or to formulate specific evacuation routes. In this study, we developed a web-based system that allows local government officials to easily perform high-precision damage simulations. This system utilizes the "Integrated Earthquake Simulator (IES)" developed by the Earthquake Research Institute at the University of Tokyo and the 3D city model "PLATEAU." We conducted accuracy verification and hearings with local governments to evaluate the system's practical utility.
First, to ensure the reliability of the simulator, we verified its accuracy focusing on Mashiki City, which suffered severe damage during the 2016 Kumamoto Earthquake. Specifically, we compared the simulation results by IES with actual building damage data and estimates using the conventional damage functions based on seismic intensity provided by the Cabinet Office. The results confirmed that the system secures accuracy equivalent to or greater than that of the Cabinet Office's method. Notably, unlike conventional mesh-based analysis, this system enables behavior analysis at the "individual building level," making it possible to evaluate the risk of "specific road blockages" caused by collapsed houses.
Demonstration experiments and hearings conducted with local government officials using the developed system yielded positive feedback, particularly regarding road blockage prediction, which was described as "useful for planning routes to transport stockpiles to evacuation centers immediately after a disaster." On the other hand, precise simulation with IES requires detailed building attribute data (construction year, structure, number of stories, etc.), highlighting the challenge of conducting analyses in areas where such data is missing. Moving forward, we plan to examine methods to derive reliable simulation results even when building information is insufficient, such as by using statistical imputation or estimation from similar data, thereby enhancing versatility for social implementation.
First, to ensure the reliability of the simulator, we verified its accuracy focusing on Mashiki City, which suffered severe damage during the 2016 Kumamoto Earthquake. Specifically, we compared the simulation results by IES with actual building damage data and estimates using the conventional damage functions based on seismic intensity provided by the Cabinet Office. The results confirmed that the system secures accuracy equivalent to or greater than that of the Cabinet Office's method. Notably, unlike conventional mesh-based analysis, this system enables behavior analysis at the "individual building level," making it possible to evaluate the risk of "specific road blockages" caused by collapsed houses.
Demonstration experiments and hearings conducted with local government officials using the developed system yielded positive feedback, particularly regarding road blockage prediction, which was described as "useful for planning routes to transport stockpiles to evacuation centers immediately after a disaster." On the other hand, precise simulation with IES requires detailed building attribute data (construction year, structure, number of stories, etc.), highlighting the challenge of conducting analyses in areas where such data is missing. Moving forward, we plan to examine methods to derive reliable simulation results even when building information is insufficient, such as by using statistical imputation or estimation from similar data, thereby enhancing versatility for social implementation.