Integrating the geographic information system and predictive data mining techniques to model effects of compound disasters in Taipei

Abstract

Located in the circum-Pacific seismic zone, Taiwan is continually threatened by such natural disasters as typhoons, floods, landslides, and earthquakes, which increase risk of property loss and severely endanger public safety. Taipei City, the political and economic capital of Taiwan, must address disaster prevention and relief operations for compound disasters and extreme climatic events in addition to existing metropolitan disaster prevention operations. This study formulates 48 compound-disaster scenarios based on threats to Taipei City due to heavy rainfall and surrounding faults. Hydrology and flood analysis results and the Taiwan earthquake loss estimation system are utilized to assess the potential for compound disasters and the number of people they would displace in Taipei’s administrative districts. Analytical results can be used to create a pre-disaster static potential diagram and a refuge or shelter capacity assessment table. The disaster potential diagram is adopted to conduct geographic information system spatial and data analysis, and temporary refuges or shelters planned by the city government are integrated for shelter capacity comparison. Furthermore, a dynamic assessment curve for the number of displaced people during a disaster is plotted using data mining and attribute filtering. Subsequently, a cross table is obtained and employed to predict the number of refugees in the various administrative districts. Finally, conclusions and recommendations are provided for making disaster prevention and relief decisions simultaneously concerning earthquakes and flooding.