由於劇烈地震所造成的人員傷亡數量，一直都是地震災害損失研究中極為關切的議題，重大地震發生後人員傷亡推估模式及數量估計之研究就變得非常重要。本研究以集集地震人員罹難資料作為基礎，針對2,039位罹難者完成GPS衛星定位、現地補充調查及書面資料檢核等程序，建立一集集地震「罹難者屬性資料庫」。以進行 (1)人員罹難空間分佈展示、(2)鄰近斷層罹難率探討、(3)罹難率損失曲線之建立、(4)人員罹難分佈之推估與驗證。本研究結合罹難者屬性資料庫、強地動測站紀錄、地震前人口統計資料及電子地圖，利用本研究所提出的「等震度」與「等母體數」的概念，透過GIS軟體的運算與分析，以建立強地動參數與罹難率之損失曲線，並以集集地震為模擬地震，鄉鎮別為行政單位進行罹難數量之推估及驗證。在「等震度」方面，選取PGA作為地震測站強地動參數並利用Kriging內插法完成全台灣PGA的等震度分佈。在「等母體數」方面，則分別以四種不同人口數尺度進行分析區域的罹難率探討，建立罹難率與強地動參數之損失曲線。研究成果顯示以「等震度」與「等母體數」方法所獲得之罹難率與強地動參數的關係曲線，資料點關連性遠較以「行政區域」為分析區域的方法來得佳。可採用不同人口數尺度的「等母體數」進行探討，也是本研究的另一特色。當以較小的等母體數進行分析時，損失曲線的資料點較多且反應真實地震人員罹難趨勢亦較為明顯。本研究以Logistic方程式對資料點進行迴歸分析，顯示以 PGA為強地動參數所建立的人員罹難損失曲線，不僅 較高且在推估人員罹難之空間分佈亦較佳，故建議採用。 In this study, the human-fatalities and strong motion data of the Chi-Chi earthquake are collected and the concepts of “iso-seismic intensity” and “equal-population intervals” is proposed. The relationship between the strong ground motion index and human-fatality rate by means of GIS software is established. Furthermore, the Chi-Chi earthquake event is taken as a study case and reach the distribution of death tolls in different districts induced from a disastrous earthquake. The values of Peah Ground Acceleration (PGA) in different seismic stations are used as the strong ground motion indices. The Kriging method is used to finish a spatial distribution of strong ground motion in Taiwan. The scales such as (1)200,000, (2)100,000, (3)50,000 and (4)25,000 death tolls are used as equal-population number to decide the analysis regions. Finally, The relationshship of the human-fatality rate as fuction of the strong ground motion indices is finished and obtained by means of the concepts of “iso-seismic intesnsity” and “equal-population intervals”. The analysis results show that data points are more concentrated than data points obtained by means of the districts-based methods. It clearly appears that the concept we proposed is reasonable and feasible. The smaller equal-population number we take to decide the analysis regions, the more data points we get. Therefore, the close tendency between the human-fatality and the strong ground motion is apparent. In the study, the Logistic regressive equation is used to fit our data points. The curve of human-fatality rate with respect to PGA is possessed of higher value of R2 and the estimation of the distribution of death tolls induced from the earthquake is able to obtain.
