A Spatial Analysis of Casualties Sustained During the Chi-Chi Earthquake and its Applicability to Earthquake Casualty Modeling

Earthquakes are by nature a spatial phenomenon. Seismic intensity and subsequent damage are directly related to the geographic properties of the affected area. This dissertation analyzes the spatial pattern of fatalities and hospitalized injuries sustained during the 1999 Chi- Chi earthquake in Taiwan. Seismic intensity, geographic properties such as soil type and distance to the epicenter, population density, building collapse, and demographic characteristics will be analyzed with respect to casualties in order to determine the spatial distribution of casualty risk. This research can be used to develop simple casualty models that predict the location of casualties to facilitate prevention of morbidity and mortality post-event and to assist in pre-event mitigation in order to mitigate the casualty risk from future earthquakes. This dissertation develops earthquake casualty models centered on three different parameters and evaluates the predictive ability of these models. The maximum coseismic slip provides a better framework than either of the two commonly used spatial centering parameters: epicenter and surface rupture. The strengths and weaknesses of the coseismic slip methodology are further evaluated and additional predictive variables are incorporated into the model, including construction class and additional geologic data. This results in the development of construction class specific casualty vulnerability functions. The relationship between specific ground motion parameters and earthquake casualties is quantified and found to be in accordance with what would be expected from structural engineering research. Finally, a method for translating existing curves from past events to future earthquakes is introduced when the vulnerability functions derived from Chi-Chi are used to model a historical event, the 1976 Tangshan earthquake.