本計畫旨在發展以歷史地震實際的受損數據（人員或建築物），建構經驗的易損曲線。易損曲線是地震損害評估模式中的基本骨幹，通常是以解析法或經驗法來建立。傳統上，實際地震的受損資料通常以行政區作為統計基準，據此獲得的罹難率或建築物倒塌率與地動參數的關係通常都相當散亂，有別於以行政區基準的傳統作法，本計畫主持人以「等震度等母體數」之概念，發展出一套處理實際地震災損資料的新方法。初步的分析成果顯示，若以傳統方法行政區為基準的方式處理集集地震的災損數據，將呈現相當散亂的現象；利用新方法處理相同的災損資料，則可使數據集中化地分佈於有限的帶寬上。本計畫將進一步針對集集地震的災損資料，考慮地質條件、建築物構造別、建立各類建築物及人員罹難之易損曲線，同時探討各類建築物及人員罹難之關連性，並與解析法所使用之參數進行比較及驗證，並與解析法所使用之參數進行比較及驗證，以提升現有災損推估系統之本土適用性。由於地震數據的不確定性，發展易損曲線必須考量其不確定性。故評估模式的不確定性，即知識不確定性(epistemic uncertainty)在建立易損曲線模式時有其重要性。本計畫中將應用貝式修正(Bayesian updating techniques)，配合集集地震所調查及記錄之資料，進行模式不確定性的率定。經由率定及統計特徵化處理的易損曲線可能具有相當的潛力，可以應用於地震災損推估，對於地震災害防治工作將有顯著的助益。 ; Earthquake loss estimation models are used to quantify potential social and economic losses after occurrence of a disastrous earthquake. Reliable and fast estimation of spatial distribution of losses provides needed information to the government agencies to prioritize actions that are required for emergency responses and disaster mitigation. Human injury and fatality is one of the most important losses in a seismic event. However, it is usually caused by the collapse of buildings and infrastructures and conflagrations during and shortly after the earthquake. Human fatality rate depends on spatial distribution of people at the time of the event, and thus is a dynamic process. The number of injuries and fatalities may be dramatically different even the same earthquake attacks the same area at different time. Despite the complexity of the causes, the estimation model for human injury and fatality is an important module in an earthquake loss estimation system and is essential for developing emergency preparedness plans. In this project, the focus is placed on the development of the empirical fragilities curves based on damage data of historic seismic event. These curves are the backbones of any earthquake losses estimation model and may be developed through analytical approach or empirical approach. The latter will be employed in this study. Empirical fragility curves are often determined based on historic seismicity at a locality. The fragility data collected from historic events are district-based and often very scatter. Rather than using district-based data, the principal investigator (PI) has developed a new method that models fragility data with the concept of “iso-seismic intensity”and “iso-population”. Preliminary analysis showed that this new approach can transform the fragility data from the 1999 Chi-Chi earthquake into a contracted band, which is very encouraging. Thus, the main goal of the proposed project is to develop and/or refine empirical fragility models that consider local geologic conditions, varying building types and human injury and mortality based on the Chi-Chi earthquake data. Furthermore, relationships between different types of buildings and the induced mortality ratio would be established for analytical-based earthquake loss estimation models. For the vast uncertainty in the data, there will be considerable uncertainty in the developed fragility models. Thus, assessing the model uncertainty (epistemic uncertainty) would be essential. In the present study, the well-established reliability theory with Bayesian updating techniques will be used to calibrate model uncertainty based on observed and surveyed data from the Chi-Chi earthquake. The fully calibrated and statistically characterized fragility models will have a great potential to be an indispensable tool for estimating earthquake losses, which contributes in a significant way to the earthquake hazard mitigation effort. ; 研究期間 9708 ~ 9807
