台灣地區地震延時模型之建立

Abstract

地震的延時在地震工程中是一項重要的參數，它可以評估建築物的表現、潛在的山崩和土壤液化。雖然目前已有許多類型的地震延時參數被提出，但先前的學者仍然偏好使用「顯著地震延時」來建立地震延時的預測模型。由於一個地區受震的反應主要受到震源、距離以及場址本身的條件影響，因此有許多學者將地震延時的預測方程式假設為地震規模、距離參數和場址參數的函數。 台灣強地動觀測網蒐集大量的地震記錄，本研究將從該資料庫選擇近二十年地震規模大於5.5的記錄，並為台灣地區建立地震延時的預測模型。本研究的過程包括判斷先前學者所提出的預測模型是否適用於台灣地區，以及對各個參考模型進行比較，並觀察其優缺點。最後將分別採取傳統迴歸分析以及混合效應回歸分析，為台灣地區建立適合的地震延時預測模型。

Duration is an important ground motion parameter. It can be used for assessing building performance, potential landslides and liquefaction. Many types of duration parameters have been proposed, but significant duration is the most widely used duration parameter. The ground motion is controlled by the source, path and site conditions. For this reason, the predictive models for the duration parameter are usually functions of earthquake magnitude, distance parameter and site parameter. In this study, the earthquake recordings from the Taiwan Strong Motion Instrumentation Program (TSMIP) are utilized. The earthquake events with local magnitude larger than 5.5 in recent years are selected. The objectives of this study are to identify if there are any shortcomings in the previous predictive models for the duration parameters, and to compare the performance of different models. In addition, new predictive models for the duration parameters for use in Taiwan are developed by the classical regression method and the mixed-effects regression method.