地震引致人員罹難之變異性分析─以集集地震為例; The variability analysis of the human- fatalities were caused by Chi-Chi Earthquake.

NCU Institutional Repository > 工學院 > 土木工程研究所 > 博碩士論文 > Item 987654321/1356

jsp.display-item.identifier=請使用永久網址來引用或連結此文件: http://ir.lib.ncu.edu.tw/handle/987654321/1356

题名:	地震引致人員罹難之變異性分析─以集集地震為例;The variability analysis of the human- fatalities were caused by Chi-Chi Earthquake.
作者:	張詩平;Shih-ping Chang
贡献者:	土木工程研究所
关键词:	預測區間;罹難曲線;變異係數;prediction interval;mortality curve;coefficient of variation
日期:	2009-07-10
上传时间:	2009-09-18 17:28:14 (UTC+8)
出版者:	國立中央大學圖書館
摘要:	本研究利用「等震度」與「等人口母體數」的概念，透過GIS軟體的運算與分析，以集集地震為模擬地震建立強地動參數PGA與人員罹難率之關係曲線，並建立罹難曲線誤差區間，進行罹難人數之推估及驗證。研究過程中，發現等母體數為20萬人、10萬人、5萬人、2.5萬人與1.25萬人之罹難曲線相似，僅有些微差異，故可決定以一罹難曲線代表各母體數的迴歸曲線，減低工作量。且因資料點數超過50% 集中在PGA小於200 gals，罹難率大都為0的部分，不考慮這些點數加以迴歸，其迴歸曲線與考慮整體點數曲線一致。而為使罹難曲線更為合理，不因罹難率落在曲線外而表示此曲線不正確，故建立一誤差區間使罹難曲線更為可靠。採用統計上之預測區間模式，其區間與既有條件不符而不適用，因而使用變異係數建立之誤差區間，在罹難人數推估上皆較預測區間準確。且為求可信度與精度，建議採用±70% 的誤差區間。因母體數愈小，更可表現出PGA對應的實際罹難率，在母體數為1.25 萬人的點數分佈上，有許多奇異點產生，進一步探討皆為土塊厝和集合住宅倒塌引致的高罹難率。 In this study, I use the concepts of “iso- seismic intensity and equal- population intervals to establish the mortality curve of Chi- Chi earthquake. From the curve, we can find the relation between the human- fatalities and strong motion data (PGA). Then I establish an error interval of the curve, making the mortality curve be more reasonable, and proceed to estimate and verify the death population of towns. During the research process, finding no matter which equal-population interval is, the mortality curves are similar such as 200 thousand, 100 thousand, 50 thousand, 25 thousand and 12.5 thousand. So we can use the equal- population interval is 12.5 thousand to represent the mortality curve of Chi- Chi earthquake. And because the more than 50% data points concentrate on PGA be smaller than 200 gals. If we omit this part data, the regression curve will be still the same. For making the regression curve is more reliable, I use the prediction model of statistics. The prediction band includes a large percent of data, nevertheless, the band will make negative mortalities, it’s not reasonable. For this reason, I calculate the coefficient of variation of curve to establish an error interval. The error interval is more reliable and accurate than prediction interval. From the distribution of points, there are some freak points. Therefore I do some discussion about these strange data. And find the main reason is the collapse of mud- bricks and apartment buildings caused the high human- fatalities.
显示于类别:	[土木工程研究所] 博碩士論文

文件中的档案:

档案	大小	格式	浏览次数

在NCUIR中所有的数据项都受到原著作权保护.

社群 sharing

数据加载中.....