位於多板塊交界帶的台灣與日本，斷層多而複雜，因此在強主震後經常發生集群式的有感餘震。為能在強震後進行即時的短期餘震風險評估，本文利用混合二維常態分布描述餘震的空間分布，結合Reasenberg-Jones(簡記為 RJ)模型，建立餘震時空風險模式，簡記為SRJ模式。本文分別分析 1999年 9月 21日台灣集集規模 7.3 地震，以及2011 年 3 月 11 日發生在日本東部海域規模9.0地震後的餘震資料，藉SRJ模式或傳統的點格--RJ方法建立相對餘震風險圖用以預警強餘震，並且根據各種準則，評估上述相對餘震風險圖在即期預警短期強餘震的效果。在綜合評估之下，SRJ模式對於上述兩筆餘震序列的風險評估優於點格--RJ方法。;Taiwan and Japan are located in the junction of multiple tectonic plates where involve complex fault zones. Therefore, there are clustered aftershocks of large earthquakes. In order to assess the risk of aftershock in near-real-time after the main shock, we propose a model with a mixture of bivariate normal distribution to describe the spatial hazard of aftershocks. The spatial hazard function joint with the Reasenberg-Jone (RJ) model then gives a new spatial-temporal-magnitude hazard of aftershocks, denoted by SRJ model. We employ the SRJ model to illustrate two aftershock sequences of, namely, the Mw 7.7 earthquake occurred on September 21, 1999, in Chi-Chi, Taiwan, and Mw 9.0 earthquake occurred on March 11, 2011, in Tohoku-Oki, Japan. The relative aftershock hazard (RAH) maps are constructed based on SRJ model or gridding-RJ method. The effect of RAH maps on depicting possible rupture area of forthcoming large aftershocks is evaluated according to a variety of criteria based on ROC curve and Molchan error diagram. The results finally demostrate that the RAH map based on the SRJ model provides an efficent and better forecast of spatial-temporal magnitude aftershock hazard than the one hazard on gridding--RJ model.
