本研究分為海嘯速算系統之建置和分析1867年基隆海嘯事件兩部份。在海嘯速算系統建置部份中，其目標為建立一套可靠、快速且低成本之海嘯預警系統，本研究以COMCOT (Cornell Multi-grid Coupled Tsunami Model)海嘯數值模式為核心進行開發，該模式經OpenMP平行處理提升計算速度，本模式整合Yen and Ma, 2011提出之地震尺度關係式並建立海溝走向資料庫協，助模式於地震發生後第一時間震源機制解尚未明確時進行海嘯模擬。本研究以shell script使整套流程能自動化執行，本文章以2011年日本海嘯事件來進行準確性較驗和結果展示，整套流程可於3分鐘內完成10小時之模擬。 在分析1867年基隆海嘯事件部份中，本研究發展一套有系統之方法分析與還原此事件。該場海嘯事件暗示台灣北海岸三座核電廠具有潛在海嘯危機。過去研究普遍認為該事件是由山腳斷層一場規模Mw7.0之地震所引發，然而並無任何證據顯示如此規模之地震可造成7公尺高之海嘯波高。考慮台灣北海岸複雜之海底地形和旺盛之火山活動，本研究認為此海嘯事件之波高極可能是受地震或是海底火山噴發產生之海底山崩所加劇。為進行有系統之分析，並評估可能之潛在海嘯來源，本研究提出海嘯源逆向追蹤法(Tsunami Reverse Tracking Method)尋找對北海岸具衝擊性之海嘯源。潛在海嘯源逆向追蹤法的理論建立在線性假設下，海嘯波的傳遞具有雙向性，並使用COMCOT求解線性淺水波方程式，快速排除不可能之海嘯源。本研究更進一步提出影響強度分析法(Impact Intensity Analysis)，除了量化各潛在海嘯源對研究區域之威脅強度外，更可以針對海嘯源逆向追蹤法於近岸處呈現模糊處進行補足。最後依海嘯源逆向追蹤法和影響強度分析法之結果設計情境案例模擬，本文展示七處海底山崩和五座火山情境案例，模擬結果與歷史文獻資料波高作比對，海底山崩情境中最有可能者來自於棉花峽谷、基隆海谷或基隆陸棚，火山情境最有可能者來自於基隆市西北方35公里處之火山。 ;This study was divided in two part, design a fast tsunami warning system and analysis the 1867 Keelung tsunami event. In first part, this study aims to develop reliable, fast, and low-cost system which is able to predict the tsunami wave height automatically based on the preliminary earthquake parameters. The Cornell Multi-grid Coupled Tsunami Model (COMCOT) was chosen as the kernel as it had been widely validated. The original source code has been parallelized by OpenMP to speed up the computing capability. We added a sophisticated source-scaling relationship proposed by Yen and Ma (2011) and a strike-dip trench database for generating tsunami source in the beginning stage of the tsunami event. A shell script was coded to execute the entire process automatically. We demonstrate the accuracy and performance by the 2011 Japan tsunami event. The warning products can be obtained in 3 minutes for a 10-hour simulation. In second part, this study uses a systematic method to analyse and reconstruct the 1867 Keelung event. The 1867 Keelung tsunami event is important to Taiwan because it indicates that three nuclear power plants nearby are under the threat of tsunami attack. Previous studies consider that this tsunami might be generated by an Mw7.0 earthquake along the Shanchiao Fault. However, there is no evidence showing the relationship between this mild seismic activities and the 7-m large tsunami wave height. We aimed to find out the potential tsunami source through the numerical analysis. Considering the steep bathymetry and intense volcanic activity along the Keelung coast, the tsunami might be triggered by not only an earthquake, but also by a submarine landslide or by a volcanic eruption which were able to increase the tsunami height dramatically. However, numbers of scenarios impeded the careful analysis. For this, we developed the Tsunami Reverse Tracking Method (TRTM) based on the linear hypothesis of tsunami wave propagation, to narrow down the possible source locations of tsunami. The Cornell Multi-grid Coupled Tsunami Model (COMCOT) was adopted for solving the shallow water equations. We also developed an Impact Intensity Analysis (IIA) method to quantify the tsunami impact from each discretized computational domain by calculating the maximum wave height. After that, a series of scenario studies were performed. Each scenario has to satisfy the geological feature and the simulated tsunami has to agree with the wave height recorded in the literatures. 7 landslide scenarios and 5 volcano scenario are showed in this study. The result shows most possible landslide scenario of the 1867 tsunami event was from Mein-Hwa Canyon, Keelung sea valley or Keelung Shelf. The most possible volcano scenario is from 35 km northwest of Keelung city.
