台灣位於菲律賓海板塊與歐亞大陸板塊交界處，地震發生頻繁，並存在著引起海嘯侵襲台灣沿岸的風險。台灣近海海嘯警報作業目前與中央氣象局地震速報系統相結合，藉由地震速報系統迅速有效的地震偵測能力，適時針對近海海嘯發出警報通知。但是現階段該系統僅能提供海嘯波浪的預估到時，為進一步強化警報資訊的內容，包括提供海嘯波浪的波形模擬與預估高度，本研究利用單位海嘯模擬方法，建立台灣近海海嘯波浪的波形模擬資料庫。 本研究是預測海嘯波到達潮位站的到時與浪高，所以我們忽略溯上的過程，因此海嘯波的模擬可以用線性淺水波方程式來表示。在線性的條件下，潮位站所收到的預估波形便可表示成單位海嘯的線性組合，因此我們以0.25°×0.25°和1公尺的起始海水面變化為一單位海嘯，並藉由美國康乃爾大學所發展的COMCOT程式，計算東經118°至125°、北緯18°至27°的範圍內，每個單位海嘯到達各潮位站的波形，並建立成資料庫。根據本研究的成果，當台灣海域發生大地震，並獲取地震速報系統所提供地震的位置與規模後，我們首先根據地震的經驗尺度(Scaling Law)，由規模推出該地震斷層的長、寬及平均滑移量，並由同區域過去的大地震預測可能的震源機制值，快速計算該地震所造成的海底地形變化，再依計算結果得到各單位海嘯的加權比重，進而與資料庫預先建置之單位海嘯波形進行合成，以產生各潮位站經數值模擬的完整海嘯波形，並得到預估的海嘯波到時與最大振幅。本研究結果可以達到下列2項優點：(1)將海嘯模擬過程中最花費時間的傳播過程事先建置成資料庫，以達到快速預警的功能。(2)海底地形變化的計算，是根據地震參數來完成，和傳統上利用地震規模和深度區段的方式相比，本方法所建置出的資料庫雖然較小，但是可以模擬出所有的結果。 Taiwan is prone to the risk of nearfield tsunamis because of the high seismic activities in the region. The Tsunami Warning System of Central Weather Bureau (CWB) currently only predicts arrival times of such tsunami events to existing tidal stations. The objective of this study is expanding the predicting information to include tsunami waveforms and maximum amplitudes of tidal stations. We fulfill this goal by building a database of tsunami waveforms which are simulated from the propagation of unit tsunamis. We ignore the run-up process and the approaching tsunami waves is governed by the linear shallow water wave equations. Accordingly, the tsunami waveforms can be expressed as a linear combination of unit tsunamis. In this study, the unit tsunami source is represented by a square box of 0.25° in dimension and 1 meter high seafloor displacement. We scan over the region bounded by 18°N, 27°N, 118°E, and 125°E to simulate the propagations of unit tsunamis of all box sources by COMCOT program. The thus calculated tsunami waves of unit tsunamis for existing tidal stations are stored in database. When a tsunami earthquake occurs in the Taiwan region, we will receive the earthquake location and magnitude from Taiwan Rapid Earthquake Information Release System (TREIRS) of the CWB. By Scaling Law, the fault length, width and average slip can be inferred from the magnitude. With empirical focal mechanism, the coefficients of unit tsunamis for each source box are derived and used to synthesize the tsunami waves of the earthquake for certain tidal stations. Merits of this approach are two folds. (1) By calculating the time consuming part of wave propagation in advance, the warning system is able to do rapid forecasting for nearfield tsunamis. (2) By determining the coefficients on an event by event basis, the system is flexible to cover all scenario earthquakes with a reasonable size of database.
