在大地震發生後，經常需要進行緊急的救災或是補強弱化結構的行動，為降低救災風險且順利進行結構補強，強餘震的即時風險評估是非常重要的。針對此一研究問題，傳統上採用RJ模式(Reasenberg and Jones ,1989)研究餘震發生的時間及規模分布，因為RJ模式假設餘震的發生時間與規模獨立，Chen等人(2015)修正RJ模式提出餘震規模分布與時間相依的MRJ模式。利用最大概似法估計在主震發生後的初期因為資料不足，RJ或MRJ模式無法獲得可靠的風險評估預測，故使用貝氏方法。因此本文考慮利用貝氏方法估計RJ和MRJ模式，並據以評估和預測餘震的風險。上述兩種模式分別應用最大概似估計方法與貝氏方法，針對下列三個餘震序列2008 發生在龍門山斷層的中國汶川地震、1999 發生在車籠埔斷層的台灣集集地震以及2011 發生在日本的東北大地震，說明如何進行強餘震的即時預測。;Emergency rescue and structure reinforcement are often needed after a drastic earthquake. To reduce the hazard while doing the work, the near real-time assessment of strong aftershocks is of great urgent demand.. To do so, the RJ (Reasenberg and Jones, 1989) model is conventionally used to study the time-magnitude hazard of aftershocks. On the other hand, Chen et al. (2015) proposed a modified RJ (MRJ) model that includes a time-dependent magnitude distribution. Note that the low detection capability of aftershocks right after the main shock often leads to sparse data and hence the maximum likelihood inference or forecast may not be valid. Therefore, Bayesian methods are employed to analyze the RJ and MRJ models for the assessment of the time-magnitude hazard of aftershocks. The gridding method is further used to explore or forecast the spatial hazard of aftershocks. Analyses of relevant models are illustrated for three aftershock sequences, after 2008 7.9 Wenchuan, China, earthquake、1999 7.7 Chi-Chi, Taiwan, earthquake and 2011 9.0 Tohoku, Japan, earthquakes.
